1 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
4 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6 (ix86_option_override_internal): Set MASK_VZEROUPPER if
7 neither -mzeroupper nor -mno-zeroupper is used and
8 TARGET_EMIT_VZEROUPPER is set.
9 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
10 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
12 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
14 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
15 folding of vector compares.
16 (fold_build_vec_cmp): New helper function.
17 (fold_compare_helper): New helper function.
18 (builtin_function_type): Add compare builtins to the list of functions
19 having unsigned arguments. Cosmetic updates to comment indentation.
20 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
21 the not+eq combination.
23 2017-11-15 Bin Cheng <bin.cheng@arm.com>
25 PR tree-optimization/82726
26 PR tree-optimization/70754
27 * tree-predcom.c (order_drefs_by_pos): New function.
28 (combine_chains): Move code setting has_max_use_after to...
29 (try_combine_chains): ...here. New parameter. Sort combined chains
30 according to position information.
31 (tree_predictive_commoning_loop): Update call to above function.
32 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
34 2017-11-15 Bin Cheng <bin.cheng@arm.com>
36 PR tree-optimization/82726
38 2017-01-23 Bin Cheng <bin.cheng@arm.com>
40 PR tree-optimization/70754
41 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
42 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
43 combined stmt before it if not NULL.
44 (combine_chains): Process refs reversely and compute dominance point
48 2017-02-23 Bin Cheng <bin.cheng@arm.com>
50 PR tree-optimization/79663
51 * tree-predcom.c (combine_chains): Process refs in reverse order
52 only for ZERO length chains, and add explaining comment.
54 2017-11-15 Tamar Christina <tamar.christina@arm.com>
56 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
57 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
58 * doc/invoke.texi (ARM Options): Add armv8.3-a.
60 2017-11-15 Tamar Christina <tamar.christina@arm.com>
62 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
64 2017-11-15 Martin Liska <mliska@suse.cz>
66 * tree-cfg.c (pass_warn_function_return::execute):
67 Compare warn_return_type for greater than zero.
69 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
73 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
74 to return true on Xeon and not on Xeon Phi.
75 (ix86_check_avx256_register): Changed to ...
76 (ix86_check_avx_upper_register): ... this. Add extra check for
77 VALID_AVX512F_REG_OR_XI_MODE.
78 (ix86_avx_u128_mode_needed): Changed
79 ix86_check_avx256_register to ix86_check_avx_upper_register.
80 (ix86_check_avx256_stores): Changed to ...
81 (ix86_check_avx_upper_stores): ... this. Changed
82 ix86_check_avx256_register to ix86_check_avx_upper_register.
83 (ix86_avx_u128_mode_after): Changed
84 avx_reg256_found to avx_upper_reg_found. Changed
85 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
86 (ix86_avx_u128_mode_entry): Changed
87 ix86_check_avx256_register to ix86_check_avx_upper_register.
88 (ix86_avx_u128_mode_exit): Ditto.
89 * config/i386/i386.h: (host_detect_local_cpu): New define.
91 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
93 * config/arm/xgene1.md (xgene1): Split into automatons
94 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
95 (xgene1_f_load): Adjust reservations and/or types.
96 (xgene1_f_store): Likewise.
97 (xgene1_load_pair): Likewise.
98 (xgene1_store_pair): Likewise.
99 (xgene1_fp_load1): Likewise.
100 (xgene1_load1): Likewise.
101 (xgene1_store1): Likewise.
102 (xgene1_move): Likewise.
103 (xgene1_alu): Likewise.
104 (xgene1_simd): Likewise.
105 (xgene1_bfm): Likewise.
106 (xgene1_neon_load1): Likewise.
107 (xgene1_neon_store1): Likewise.
108 (xgene1_neon_logic): Likewise.
109 (xgene1_neon_st1): Likewise.
110 (xgene1_neon_ld1r): Likewise.
111 (xgene1_alu_cond): Added.
112 (xgene1_shift_reg): Likwise.
113 (xgene1_bfx): Likewise.
114 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
116 2017-11-15 Jakub Jelinek <jakub@redhat.com>
119 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
121 (can_widen_mult_without_libcall): New function.
122 (expand_mul_overflow): If only checking unsigned mul overflow,
123 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
124 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
125 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
126 (expand_DIVMOD): Formatting fix.
127 * expmed.h (expand_mult): Add NO_LIBCALL argument.
128 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
129 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
131 PR tree-optimization/82977
132 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
133 constructed temporary to strlen_to_stridx.put.
135 2017-11-15 Martin Liska <mliska@suse.cz>
137 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
138 * configure: Regenerate.
140 2017-11-15 Martin Liska <mliska@suse.cz>
143 * config/sh/sh-mem.cc: Use proper probability for
146 2017-11-14 Jeff Law <law@redhat.com>
148 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
149 the red zone for stack_clash_protection_final_dynamic_probe targets
150 when the total dynamic stack size is zero bytes.
152 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
153 blocks is post order.
155 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
157 * dumpfile.h (TDF_COMPARE_DEBUG): New.
158 * final.c (rest_of_clean_state): Set it for the
159 -fcompare-debug dump.
160 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
161 class when TDF_COMPARE_DEBUG is set.
163 * dwarf2out.c (gen_producer_string): Discard
166 2017-11-15 Joseph Myers <joseph@codesourcery.com>
169 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
170 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
171 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
172 (__TGMATH_REAL_2_3): Remove macros.
173 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
174 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
176 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
178 (remquo): Define using __TGMATH_REAL_3.
180 2017-11-14 Jeff Law <law@redhat.com>
182 * vr-values.c: New file with contents extracted from tree-vrp.c.
183 * Makefile.in (OBJS): Add vr-values.o
184 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
185 (set_value_range, set_and_canonicalize_value_range): Likewise.
186 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
187 (value_range_constant_singleton, symbolic_range_p): Likewise.
188 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
189 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
190 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
191 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
192 (find_case_label_range, find_case_label_index): Likewise.
193 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
194 (range_int_cst_singleton_p, value_inside_range): Likewise.
195 (get_single_symbol): Likewise.
196 (switch_update): Move structure definition here.
197 (to_remove_edges, to_update_switch_stmts): Provide externs.
198 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
199 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
200 (vrp_val_is_min, set_value_range): Likewise.
201 (set_and_canonicalize_value_range, copy_value_range): Likewise.
202 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
203 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
204 (range_is_nonnull, range_int_cst_p): Likewwise.
205 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
206 (get_single_symbol, operand_less_p): Likewise
207 (compare_values_warnv, compare_values): Likewise.
208 (value_inside_range, value_range_constant_singleton): Likewise.
209 (zero_nonzero_bitgs_from_vr): Likewise.
210 (extract_range_from_binary_expr_1): Likewise.
211 (overflow_comparison_p): Likewise.
212 (to_remove_edges, to_update_switch_stmts): Likewise.
213 (find_case_label-index, find_case_label_range): Likewise.
214 (switch_update, set_value_range_to_nonnegative): Remove.
215 (set_value_range_to_truthvalue): Likewise.
216 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
217 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
218 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
219 (find_case_label_ranges, test_for_singularity): Likewise.
220 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
221 (x_vr_values): Move to its remaining use site.
223 2017-11-10 Jeff Law <law@redhat.com>
225 * vr-values.h (VR_INITIALIZER): Move #define here.
226 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
227 * Makefile.in (OBJS): Add tree-evrp.o
228 * tree-vrp.h (assert_info): Move structure definition here.
229 (set_value_range_to_varying): Prototype.
230 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
231 (infer_value_range, register_edge_assert_for): Likewise.
232 (stmt_interesting_for_vrp): Likewise.
233 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
234 (set_value_range_to_varying): No longer static.
235 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
236 (infer_value_range, register_edge_assert_for): Likewise.
238 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
240 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
241 generate the XXBRD instruction.
243 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
244 (rs6000_builtin_type_compatible): Treat _Float128 and long double
245 as being compatible if -mabi=ieeelongdouble.
246 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
247 to setup float128 built-ins with hardware support.
248 (BU_FLOAT128_HW_2): Likewise.
249 (BU_FLOAT128_HW_3): Likewise.
250 (BU_FLOAT128_HW_VSX_1): Likewise.
251 (BU_FLOAT128_HW_VSX_2): Likewise.
252 (scalar_extract_expq): Change float128 built-in functions to
253 accommodate having both KFmode and TFmode functions. Use the
254 KFmode variant as the default.
255 (scalar_extract_sigq): Likewise.
256 (scalar_test_neg_qp): Likewise.
257 (scalar_insert_exp_q): Likewise.
258 (scalar_insert_exp_qp): Likewise.
259 (scalar_test_data_class_qp): Likewise.
260 (sqrtf128_round_to_odd): Delete processing the round to odd
261 built-in functions as special built-in functions, and define them
262 as float128 built-ins. Use the KFmode variant as the default.
263 (truncf128_round_to_odd): Likewise.
264 (addf128_round_to_odd): Likewise.
265 (subf128_round_to_odd): Likewise.
266 (mulf128_round_to_odd): Likewise.
267 (divf128_round_to_odd): Likewise.
268 (fmaf128_round_to_odd): Likewise.
269 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
270 support for KFmode and TFmode xststdcqp calls.
271 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
272 point, switch the built-in handlers for the get/set float128
273 exponent, get float128 mantissa, float128 test built-ins, and the
274 float128 round to odd built-in functions. Eliminate creating the
275 float128 round to odd built-in functions as special built-ins.
276 (rs6000_init_builtins): Eliminate special creation of the float128
277 round to odd built-in functions.
278 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
279 function insns to support both TFmode and KFmode variants.
280 (xsxsigqp_<mode>): Likewise.
281 (xsiexpqpf_<mode>): Likewise.
282 (xsiexpqp_<mode>): Likewise.
283 (xststdcqp_<mode>): Likewise.
284 (xststdcnegqp_<mode>): Likewise.
285 (xststdcqp_<mode>): Likewise.
287 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
289 * tree-ssa-threadupdate.c (compute_path_counts): Remove
290 unused path_in_freq_ptr parameter.
291 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
293 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
295 * ipa-inline.c (edge_badness): Dump sreal frequency.
296 (compute_inlined_call_time): Match natural implementaiton ...
297 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
298 forgotten division by CGRAPH_FREQ_BASE.
300 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
302 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
303 Solaris 11. Update comment.
304 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
306 * config/sol2.h (STARTFILE_SPEC): Likewise.
307 * configure: Regenerate.
309 2017-11-14 Carl Love <cel@us.ibm.com>
311 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
312 le_ and be_ prefixes to swap* variables. Remove
313 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
316 2017-11-14 Jason Merrill <jason@redhat.com>
318 Support GTY((cache)) on hash_map.
319 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
320 (ggc_cache_remove): Override it instead of ggc_mx.
321 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
322 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
323 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
324 (simple_cache_map_traits): Override maybe_mx.
325 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
326 (hash_map): Friend gt_cleare_cache.
327 (gt_cleare_cache): New.
328 * tree.h (tree_cache_traits): New hash_map traits class.
329 (tree_cache_map): New typedef.
331 2017-11-14 Richard Biener <rguenther@suse.de>
333 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
334 paramter and handling.
335 (cleanup_control_flow_bb): Likewise.
336 (cleanup_control_flow_pre): New helper performing a DFS walk
337 to call cleanup_control_flow_bb in PRE order.
338 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
339 via cleanup_control_flow_pre.
341 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
343 * config/aarch64/aarch64-simd.md
344 (aarch64_simd_bsl<mode>_internal): Remove DImode.
345 (*aarch64_simd_bsl<mode>_alt): Likewise.
346 (aarch64_simd_bsldi_internal): New.
347 (aarch64_simd_bsldi_alt): Likewise.
349 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
351 * tracer.c (better_p): Do not compare frequencies.
352 * reg-stack.c (better_edge): Likewise.
353 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
356 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
358 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
359 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
360 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
361 * lto-streamer-in.c (input_function): Use update_max_bb_count.
362 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
363 * predict.c (maybe_hot_frequency_p): Inline to ...
364 (maybe_hot_count_p): ... here; rewrite to counts.
365 (counts_to_freqs): Rename to ...
366 (update_max_bb_count): ... this one.
367 (expensive_function_p): Use counts.
368 (estimate_bb_frequencies): Update.
369 (rebuild_frequencies): Update.
370 * predict.h (counts_to_freqs): Rename to ...
371 (update_max_bb_count): ... this one.
372 * profile.c (compute_branch_probabilities): Add debug info
373 * tree-inline.c (expand_call_inline): Update debug info.
374 (optimize_inline_calls): Use update_max_bb_count..
375 (tree_function_versioning): Use update_max_bb_count..
376 * value-prof.c (gimple_value_profile_transformations):
377 Do not use update_max_bb_count.
379 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
381 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
382 always use frequencies.
384 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
386 * bb-reorder.c: Remove frequencies from comments.
387 (better_edge_p): Use profile counts.
388 (find_traces): Dump profile counts.
389 (rotate_loop): Use profile counts.
390 (find_traces_1_round): Likewise.
391 (connect_better_edge_p): Use counts instead of probabilities for
393 (copy_bb_p): Drop early check for non-0 frequency.
394 (sanitize_hot_paths): Update comments.
396 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
398 * ipa-split.c (struct split_point): Add count.
399 (consider_split): Do not compute incoming frequency; compute incoming
400 count and store it to split_point.
401 (split_function): Set count of the call to split part correctly.
403 2017-11-13 Carl Love <cel@us.ibm.com>
405 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
407 2017-11-13 Tom Tromey <tom@tromey.com>
409 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
411 2017-11-13 Carl Love <cel@us.ibm.com>
413 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
414 Add support for builtins:
415 unsigned int vec_first_{,miss}_match_{,or_eos}index,
416 vector {un,}signed {char,int,short},
417 vector {un,}signed {char,int,short}) arguments.
418 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
419 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
420 Add BU_P9V_AV_2 expansions for the builtins.
421 * config/rs6000/altivec.h (vec_first_match_index,
422 vec_first_mismatch_index, vec_first_match_or_eos_index,
423 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
424 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
425 new extern declaration.
426 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
427 * config/rs6000/vsx.md (first_match_index_<mode>,
428 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
429 first_mismatch_or_eos_index_<mode>): Add define expand.
430 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
431 * doc/extend.texi: Update the built-in documenation file for the new
434 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
436 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
437 into the min/max operations for _Float<N> and _Float<N>X types.
439 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
442 * dwarf2out.c (do_eh_frame): New static variable.
443 (dwarf2out_begin_prologue): Set it.
444 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
446 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
448 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
450 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
452 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
455 * debug.h (dwarf2out_do_eh_frame): Declare.
456 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
457 (dwarf2out_do_frame): Use it.
458 (dwarf2out_do_cfi_asm): Likewise.
459 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
460 (dwarf2out_assembly_start): Likewise.
461 (dwarf2out_begin_prologue): Fix comment.
462 * toplev.c (compile_file): Always call dwarf2out_frame_finish
463 if the target needs either debug or unwind DWARF2 info.
464 * lto-opts.c (lto_write_options): Do not save -fexceptions,
465 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
466 -ftrapping-math, -ftrapv and -fwrapv.
468 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
470 * cgraph.c (cgraph_edge::sreal_frequency): New function.
471 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
472 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
473 (estimate_edge_size_and_time): Likewise.
474 (ipa_merge_fn_summary_after_inlining): Likewise.
475 * ipa-inline.c (cgraph_freq_base_rec): Remove.
476 (compute_uninlined_call_time): Use sreal_frequency.
477 (compute_inlined_call_time): Likewise.
478 (ipa_inline): Do not initialize cgraph_freq_base_rec.
479 * profile-count.c: Include sreal.h.
480 (profile_count::to_sreal_scale): New.
481 * profile-count.h: Forward declare sreal.
482 (profile_count::to_sreal_scale): Declare.
484 2017-11-13 Nathan Sidwell <nathan@acm.org>
486 * diagnostic.c (maybe_line_and_column): New.
487 (diagnostic_get_location_text): Use it.
488 (diagnostic_report_current_module): Likewise.
489 (test_diagnostic_get_location_text): Add tests.
491 2017-11-13 Luis Machado <luis.machado@linaro.org>
493 * doc/md.texi (Specifying processor pipeline description): Fix
494 incorrect latency for the div instruction example.
496 2017-11-13 Jakub Jelinek <jakub@redhat.com>
498 PR tree-optimization/78821
499 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
500 that bit_not_p is the same.
501 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
502 (split_group): Count precisely bit_not_p bits in each statement.
503 (invert_op): New function.
504 (imm_store_chain_info::output_merged_store): Use invert_op to
505 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
506 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
508 2017-11-13 Martin Liska <mliska@suse.cz>
510 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
511 (struct source_info): Likewise.
512 (add_branch_counts): Likewise.
513 (add_line_counts): Likewise.
514 (function_summary): Likewise.
515 (output_intermediate_line): Likewise.
516 (generate_results): Likewise.
518 2017-11-13 Martin Liska <mliska@suse.cz>
520 * gcov.c (struct block_info): Remove typedef for block_t.
521 (struct line_info): Likewise.
522 (line_info::has_block): Likewise.
523 (EXIT_BLOCK): Likewise.
526 (get_cycles_count): Likewise.
527 (process_file): Likewise.
528 (read_graph_file): Likewise.
529 (solve_flow_graph): Likewise.
530 (find_exception_blocks): Likewise.
531 (add_line_counts): Likewise.
532 (accumulate_line_info): Likewise.
533 (output_line_details): Likewise.
535 2017-11-13 Martin Liska <mliska@suse.cz>
537 * gcov.c (struct arc_info): Remove typedef for arc_t.
538 (struct line_info): Likewise.
539 (add_branch_counts): Likewise.
540 (output_branch_count): Likewise.
541 (function_info::~function_info): Likewise.
543 (output_intermediate_line): Likewise.
544 (read_graph_file): Likewise.
545 (solve_flow_graph): Likewise.
546 (find_exception_blocks): Likewise.
547 (add_line_counts): Likewise.
548 (accumulate_line_info): Likewise.
549 (output_line_details): Likewise.
550 (output_function_details): Likewise.
552 2017-11-13 Martin Liska <mliska@suse.cz>
554 * gcov.c (struct function_info): Remove typedef for function_t.
555 (struct source_info): Likewise.
556 (source_info::get_functions_at_location): Likewise.
557 (solve_flow_graph): Likewise.
558 (find_exception_blocks): Likewise.
559 (add_line_counts): Likewise.
560 (output_intermediate_file): Likewise.
561 (process_file): Likewise.
562 (generate_results): Likewise.
563 (release_structures): Likewise.
564 (read_graph_file): Likewise.
565 (read_count_file): Likewise.
566 (accumulate_line_counts): Likewise.
567 (output_lines): Likewise.
569 2017-11-13 Martin Liska <mliska@suse.cz>
571 * gcov.c (function_info::function_info): Remove num_counts
572 and add vector<gcov_type>.
573 (function_info::~function_info): Use the vector.
574 (process_file): Likewise.
575 (read_graph_file): Likewise.
576 (read_count_file): Likewise.
577 (solve_flow_graph): Likewise.
579 2017-11-13 Martin Liska <mliska@suse.cz>
581 * gcov.c (function_info::is_artificial): New function.
582 (process_file): Erase all artificial early.
583 (generate_results): Skip as all artificial are already
586 2017-11-13 Martin Liska <mliska@suse.cz>
588 * gcov.c (read_graph_file): Store to global vector of functions.
589 (read_count_file): Iterate the vector.
590 (process_file): Likewise.
591 (generate_results): Likewise.
592 (release_structures): Likewise.
594 2017-11-13 Jakub Jelinek <jakub@redhat.com>
596 PR tree-optimization/82954
597 * gimple-ssa-store-merging.c
598 (imm_store_chain_info::coalesce_immediate_stores): If
599 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
601 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
603 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
604 Upddate call to ENDIAN_LANE_N.
605 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
606 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
607 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
608 and use aarch64_endian_lane_rtx.
609 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
611 2017-11-12 Tom de Vries <tom@codesourcery.com>
613 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
615 2017-11-12 Tom de Vries <tom@codesourcery.com>
617 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
619 2017-11-12 Tom de Vries <tom@codesourcery.com>
621 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
623 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
624 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
625 * defaults.h (ASM_OUTPUT_LABELREF): Same.
627 2017-11-11 Martin Sebor <msebor@redhat.com>
630 * doc/extend.texi (attribute nonstring): Remove spurious argument.
633 * prefic.c (translate_name): Replace strncpy with memcpy to
634 avoid -Wstringop-truncation.
636 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
638 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
640 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
642 * predict.c (maybe_hot_frequency_p): Do not use cfun.
644 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
646 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
649 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
651 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
652 enable generating XXBRH if the value is in a vector register.
653 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
654 value is in a vector register.
655 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
656 register to register bswap64's instead of doing the GPR sequence
657 used on previous machines.
658 (bswapdi2_xxbrd): New insn.
659 (bswapdi2_reg): Disallow on ISA 3.0.
660 (register to register bswap64 splitter): Do not split the insn on
661 ISA 3.0 systems that use XXBRD.
663 2017-11-10 Martin Sebor <msebor@redhat.com>
666 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
668 (find_subframework_file): Same.
670 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
672 * auto-profile.c (afdo_indirect_call): Drop frequency.
673 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
674 (cgraph_node::create_edge): Drop frequency argument.
675 (cgraph_node::create_indirect_edge): Drop frequency argument.
676 (cgraph_edge::make_speculative): Drop frequency arguments.
677 (cgraph_edge::resolve_speculation): Do not update frequencies
678 (cgraph_edge::dump_edge_flags): Do not dump frequency.
679 (cgraph_node::dump): Check consistency in IPA mode.
680 (cgraph_edge::maybe_hot_p): Use IPA counter.
681 (cgraph_edge::verify_count_and_frequency): Rename to ...
682 (cgraph_edge::verify_count): ... this one; drop frequency checking.
683 (cgraph_node::verify_node): Update.
684 * cgraph.h (struct cgraph_edge): Drop frequency.
685 (cgraph_edge::frequency): New function.
686 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
688 (cgraph_edge::rebuild_edges): Likewise.
689 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
690 (duplicate_thunk_for_node): Do not pass frequency.
691 (cgraph_node::create_clone): Scale only counts.
692 (cgraph_node::create_virtual_clone): Do not pass frequency.
693 (cgraph_node::create_edge_including_clones): Do not pass frequency.
694 (cgraph_node::create_version_clone): Do not pass frequency.
695 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
696 (cgraph_node::expand_thunk): Do not pass frequency.
697 (cgraph_node::create_wrapper): Do not pass frequency.
698 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
700 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
701 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
702 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
703 (gather_caller_stats): Use frequency function.
704 (ipcp_cloning_candidate_p): Use frequency function.
705 (ipcp_propagate_stage): Use frequency function.
706 (get_info_about_necessary_edges): Use frequency function.
707 (update_profiling_info): Update only IPA profile.
708 (update_specialized_profile): Use frequency functoin.
709 (perhaps_add_new_callers): Update only IPA profile.
710 * ipa-devirt.c (ipa_devirt): Use IPA profile.
711 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
712 (dump_ipa_call_summary): Use frequency function.
713 (estimate_edge_size_and_time): Use frequency function.
714 (ipa_merge_fn_summary_after_inlining): Use frequency function.
715 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
716 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
717 (update_noncloned_counts): ... ths one; scale counts only.
718 (clone_inlined_nodes): Do not scale frequency.
719 (inline_call): Do not pass frequency.
720 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
721 (compute_inlined_call_time): Use IPA profile.
722 (want_inline_small_function_p): Use IPA profile.
723 (want_inline_self_recursive_call_p): Use IPA profile.
724 (edge_badness): Use IPA profile.
725 (lookup_recursive_calls): Use IPA profile.
726 (recursive_inlining): Do not pass frequency.
727 (resolve_noninline_speculation): Do not update frequency.
728 (inline_small_functions): Collect max of IPA profile.
729 (dump_overall_stats): Dump IPA porfile.
730 (dump_inline_stats): Dump IPA porfile.
731 (ipa_inline): Collect IPA stats.
732 * ipa-inline.h (clone_inlined_nodes): Update prototype.
733 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
734 (ipa_propagate_frequency): Use frequency function.
735 (ipa_profile): Cleanup.
736 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
737 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
738 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
739 (input_node): Do not stream frequency.
740 (input_edge): Do not stream frequency.
741 (merge_profile_summaries): Scale only IPA profiles.
742 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
743 * predict.c (drop_profile): Do not recompute frequency.
744 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
745 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
746 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
747 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
748 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
749 * tree-inline.c (copy_bb): Do not scale frequency.
750 (expand_call_inline): Do not scale frequency.
751 (tree_function_versioning): Do not scale frequency.
752 * ubsan.c (ubsan_create_edge): Do not pass frequency.
754 2017-11-10 Julia Koval <julia.koval@intel.com>
756 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8,
757 _mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8,
758 _mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8,
759 _mm256_maskz_gf2p8affine_epi64_epi8,
760 _mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8
761 _mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
762 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi,
763 __builtin_ia32_vgf2p8affineqb_v32qi,
764 __builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
765 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
767 2017-11-10 Tamar Christina <tamar.christina@arm.com>
771 (arm_option_override): Refactor.
772 (arm_option_reconfigure_globals): New.
773 (arm_options_perform_arch_sanity_checks): New.
774 * config/arm/arm-protos.h (arm_option_reconfigure_globals): New prototype.
775 (arm_options_perform_arch_sanity_checks): Likewise
777 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
779 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
781 * rs6000/rs6000.md ("type" attr): Add qmul.
782 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
783 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
784 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
786 2017-11-10 Martin Sebor <msebor@redhat.com>
789 * builtins.c (compute_objsize): Handle arrays that
790 compute_builtin_object_size likes to fail for. Make extern.
791 * builtins.h (compute_objsize): Declare.
792 (check_strncpy_sizes): New function.
793 (expand_builtin_strncpy): Call check_strncpy_sizes.
794 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
795 -Wstringop-truncation.
796 (gimple_fold_builtin_strncat): Same.
797 * gimple.c (gimple_build_call_from_tree): Set call location.
798 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
799 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
800 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
801 (handle_builtin_strlen): Use strlen_to_stridx.
802 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
804 Use strlen_to_stridx.
805 (pass_strlen::execute): Release strlen_to_stridx.
806 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
807 (-Wstringop-truncation): Document new option.
809 2017-11-10 Martin Liska <mliska@suse.cz>
811 PR gcov-profile/82702
812 * gcov.c (main): Handle intermediate files in a different
814 (get_gcov_intermediate_filename): New function.
815 (output_gcov_file): Remove support of intermediate files.
816 (generate_results): Allocate intermediate file.
817 (release_structures): Clean-up properly fn_end.
818 (output_intermediate_file): Start iterating with line 1.
820 2017-11-10 Jakub Jelinek <jakub@redhat.com>
822 PR tree-optimization/82929
823 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
824 ops_swapped_p non-static data member.
825 (store_immediate_info::store_immediate_info): Clear it.
826 (imm_store_chain_info::coalesce_immediate_stores): If swapping
827 ops set ops_swapped_p.
828 (count_multiple_uses): Handle ops_swapped_p.
830 2017-11-10 Martin Liska <mliska@suse.cz>
832 * coverage.c (coverage_init): Stream information about
833 support of has_unexecuted_blocks.
834 * doc/gcov.texi: Document that.
835 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
836 * gcov.c (read_graph_file): Likewise.
837 (output_line_beginning): Fix a small issue with
840 2017-11-10 Bin Cheng <bin.cheng@arm.com>
842 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
843 reference of trivial component.
845 2017-11-10 Jakub Jelinek <jakub@redhat.com>
848 * gimple-ssa-store-merging.c
849 (pass_store_merging::terminate_all_aliasing_chains): For
850 gimple_store_p stmts also call refs_output_dependent_p.
852 PR rtl-optimization/82913
853 * compare-elim.c (try_merge_compare): Punt if def_insn is not
856 2017-11-09 Jeff Law <law@redhat.com>
858 * vr-values.h: New file with vr_values class.
859 * tree-vrp.c: Include vr-values.h
860 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
861 data objects into the vr_values class.
862 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
863 (get_value_range): Make it a member function within vr_values class.
864 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
865 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
866 (op_with_constant_singleton_value_range): Likewise.
867 (extract_range_for_var_from_comparison_expr): Likewise.
868 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
869 (extract_range_from_binary_expr): Likewise.
870 (extract_range_from_unary_expr): Likewise.
871 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
872 (check_for_binary_op_overflow, extract_range_basic): Likewise.
873 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
874 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
875 (compare_name_with_value, compare_names): Likewise.
876 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
877 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
878 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
879 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
880 (extract_range_from_phi_node): Likewise.
881 (simplify_truth_ops_using_ranges): Likewise.
882 (simplify_div_or_mod_using_ranges): Likewise.
883 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
884 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
885 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
886 (simplify_float_conversion_using_ranges): Likewise.
887 (simplify_internal_call_using_ranges): Likewise.
888 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
889 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
890 poitner via x_vr_values for calls into gimple folder.
891 (vrp_initialize_lattice): Make this the vr_values ctor.
892 (vrp_free_lattice): Make this the vr_values dtor.
893 (set_vr_value): New function.
894 (class vrp_prop): Add vr_values data member. Add various member
895 functions as well as member functions that delegate to vr_values.
896 (check_array_ref): Make a member function within vrp_prop class.
897 (search_for_addr_array, vrp_initialize): Likewise.
898 (vrp_finalize): Likewise. Revamp to avoid direct access to
899 vr_value, values_propagated, etc.
900 (check_array_bounds): Extract vrp_prop class instance pointer from
901 walk info structure. Use it to call member functions.
902 (check_all_array_refs): Make a member function within vrp_prop class.
903 Smuggle class instance pointer via walk info structure.
904 (x_vr_values): New local static.
905 (vrp_valueize): Use x_vr_values to get class instance.
906 (vr_valueize_1): Likewise.
907 (class vrp_folder): Add vr_values data member. Add various member
908 functions as well as member functions that delegate to vr_values.
909 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
910 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
911 class instance from vr_values. Use it to call member functions.
912 (vrp_dom_walker): Add vr_values data member.
913 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
914 instance via x_vr_values.
915 (identify_jump_threads): Accept vr_values as argument. Store
916 it into the walker structure.
917 (evrp_dom_walker): Add vr_values class data member. Add various
919 (evrp_dom_walker::try_find_new_range): Use vr_values data
920 member to access the memory allocator.
921 (evrp_dom_walker::before_dom_children): Store vr_values class
922 instance into the vrp_folder class.
923 (evrp_dom_walker::push_value_range): Rework to avoid direct
924 access to num_vr_values and vr_value.
925 (evrp_dom_walker::pop_value_range): Likewise.
926 (execute_early_vrp): Remove call to vrp_initialize_lattice.
927 Use vr_values to get to dump_all_value_ranges member function.
928 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
929 and simplify_cond_using_ranges_2 via vrp_prop class instance.
930 Pass vr_values class instance down to identify_jump_threads.
931 Remove call to vrp_free_lattice.
932 (debug_all_value_ranges): Remove.
934 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
935 (vrp_folder): Likewise.
937 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
938 Get it from the existing bitmap instead.
939 (vrp_intersect_ranges_1): Likewise.
941 2017-11-09 Jakub Jelinek <jakub@redhat.com>
943 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
945 (store_immediate_info::store_immediate_info): Add bitnotp argument,
947 (imm_store_chain_info::coalesce_immediate_stores): Break group
948 if bit_not_p is different.
949 (count_multiple_uses, split_group,
950 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
951 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
952 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
953 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
954 to store_immediate_info ctor.
956 2017-11-09 Jim Wilson <jimw@sifive.com>
958 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
959 (scan_prog_file): Likewise.
961 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
963 * bb-reorder.c (max_entry_frequency): Remove.
964 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
965 connect_traces, push_to_next_round_p): Remove prototypes.
966 (find_traces_1_round): Use counts only.
967 (push_to_next_round_p): Likewise.
968 (find_traces): Likewise.
969 (rotate_loop): Likewise.
970 (find_traces_1_round): Likewise.
971 (connect_traces): Likewise.
972 (edge_order): Likewise.
974 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
976 * config/arm/arm.c (output_return_instruction): Add comments to
977 indicate requirement for cmse_nonsecure_entry return to account
978 for the size of clearing instruction output here.
979 (thumb_exit): Likewise.
980 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
981 return in hardfloat mode.
983 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
985 * config/rs6000/rs6000.c (machine_function): Add a bool,
986 "toc_is_wrapped_separately".
987 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
988 if it wasn't explicitly set or unset, we are optimizing for speed, and
989 doing separate shrink-wrapping.
990 (rs6000_get_separate_components): Enable the TOC component if
991 saving the TOC register in the prologue.
992 (rs6000_components_for_bb): Handle the TOC component.
993 (rs6000_emit_prologue_components): Store the TOC register where needed.
994 (rs6000_set_handled_components): Mark TOC as handled, if handled.
995 (rs6000_emit_prologue): Don't save the TOC if that is already done.
997 2017-11-09 Martin Jambor <mjambor@suse.cz>
999 * ipa-param-manipulation.c: New file.
1000 * ipa-param-manipulation.h: Likewise.
1001 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
1002 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
1003 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
1004 (ipa_parm_adjustment): Likewise.
1005 (ipa_parm_adjustment_vec): Likewise.
1006 (ipa_get_vector_of_formal_parms): Moved declaration to
1007 ipa-param-manipulation.h.
1008 (ipa_get_vector_of_formal_parm_types): Likewise.
1009 (ipa_modify_formal_parameters): Likewise.
1010 (ipa_modify_call_arguments): Likewise.
1011 (ipa_combine_adjustments): Likewise.
1012 (ipa_dump_param_adjustments): Likewise.
1013 (ipa_modify_expr): Likewise.
1014 (ipa_get_adjustment_candidate): Likewise.
1015 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1016 ipa-param-manipulation.c.
1017 (ipa_get_vector_of_formal_parm_types): Likewise.
1018 (ipa_modify_formal_parameters): Likewise.
1019 (ipa_modify_call_arguments): Likewise.
1020 (ipa_modify_expr): Likewise.
1021 (get_ssa_base_param): Likewise.
1022 (ipa_get_adjustment_candidate): Likewise.
1023 (index_in_adjustments_multiple_times_p): Likewise.
1024 (ipa_combine_adjustments): Likewise.
1025 (ipa_dump_param_adjustments): Likewise.
1026 * tree-sra.c: Also include ipa-param-manipulation.h
1027 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
1030 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1031 Alan Hayward <alan.hayward@arm.com>
1032 David Sherwood <david.sherwood@arm.com>
1034 * doc/sourcebuild.texi (vect_masked_store): Document.
1036 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1037 Alan Hayward <alan.hayward@arm.com>
1038 David Sherwood <david.sherwood@arm.com>
1040 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
1042 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1043 Alan Hayward <alan.hayward@arm.com>
1044 David Sherwood <david.sherwood@arm.com>
1046 * doc/sourcebuild.texi (vect_variable_length): Document.
1048 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1049 Alan Hayward <alan.hayward@arm.com>
1050 David Sherwood <david.sherwood@arm.com>
1052 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
1054 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1055 Alan Hayward <alan.hayward@arm.com>
1056 David Sherwood <david.sherwood@arm.com>
1058 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1060 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1061 Alan Hayward <alan.hayward@arm.com>
1062 David Sherwood <david.sherwood@arm.com>
1064 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1065 previously undocumented selectors.
1066 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1068 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1070 * doc/rtl.texi (const_vector): Say that elements can be
1071 const_wide_ints too.
1072 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1073 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1074 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1075 * optabs.c (expand_vector_broadcast): Likewise.
1077 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1078 Alan Hayward <alan.hayward@arm.com>
1079 David Sherwood <david.sherwood@arm.com>
1081 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1082 scaled index even if the unscaled address was invalid.
1083 Don't increase the complexity of using a scale in that case.
1085 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1086 Alan Hayward <alan.hayward@arm.com>
1087 David Sherwood <david.sherwood@arm.com>
1089 * doc/rtl.texi: Rewrite the subreg rules so that they partition
1090 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1091 UNITS_PER_WORD bytes.
1092 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1093 based on REGMODE_NATURAL_SIZE of the inner mode.
1094 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1095 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1096 * expmed.c (lowpart_bit_field_p): Divide the value up into
1097 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1098 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1099 whether something is likely to occupy more than one register.
1101 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1104 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1105 profile_count::adjust_for_ipa_scaling.
1106 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1107 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1109 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1111 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1113 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1114 (split_group): Add total_orig and total_new arguments, estimate the
1115 number of statements related to the store group without store merging
1116 and with store merging.
1117 (imm_store_chain_info::output_merged_store): Adjust split_group
1118 callers, punt if estimated number of statements with store merging
1119 is not smaller than estimated number of statements without it.
1121 (handled_load): Remove has_single_use checks.
1122 (pass_store_merging::process_store): Likewise.
1124 2017-11-09 Richard Biener <rguenther@suse.de>
1126 PR tree-optimization/82902
1127 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1129 2017-11-09 Martin Liska <mliska@suse.cz>
1132 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1135 2017-11-09 Martin Liska <mliska@suse.cz>
1137 PR tree-optimization/82669
1138 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1140 2017-11-09 Martin Liska <mliska@suse.cz>
1142 PR gcov-profile/48463
1143 * coverage.c (coverage_begin_function): Output also end locus
1144 of a function and information whether the function is
1146 * gcov-dump.c (tag_function): Parse and print the information.
1147 * gcov.c (INCLUDE_MAP): Add include.
1148 (INCLUDE_SET): Likewise.
1149 (struct line_info): Move earlier in the source file because
1150 of vector<line_info> in function_info structure.
1151 (line_info::line_info): Likewise.
1152 (line_info::has_block): Likewise.
1153 (struct source_info): Add new member index.
1154 (source_info::get_functions_at_location): New function.
1155 (function_info::group_line_p): New function.
1156 (output_intermediate_line): New function.
1157 (output_intermediate_file): Use the mentioned function.
1158 (struct function_start): New.
1159 (struct function_start_pair_hash): Likewise.
1160 (process_file): Add code that identifies group functions.
1161 Assign lines either to global or function scope.
1162 (generate_results): Skip artificial functions.
1163 (find_source): Assign index for each source file.
1164 (read_graph_file): Read new flag artificial and end_line.
1165 (add_line_counts): Assign it either to global of function scope.
1166 (accumulate_line_counts): Isolate core of the function to
1167 accumulate_line_info and call it for both function and global
1169 (accumulate_line_info): New function.
1170 (output_line_beginning): Fix GNU coding style.
1171 (print_source_line): New function.
1172 (output_line_details): Likewise.
1173 (output_function_details): Likewise.
1174 (output_lines): Iterate both source (global) scope and function
1176 (struct function_line_start_cmp): New class.
1177 * doc/gcov.texi: Reflect changes in documentation.
1179 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1182 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1183 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1184 and similarly for not instead of neg.
1186 2017-11-08 Andi Kleen <ak@linux.intel.com>
1188 * config/i386/i386.opt: Add -mforce-indirect-call.
1189 * config/i386/predicates.md: Check for flag_force_indirect_call.
1190 * doc/invoke.texi: Document -mforce-indirect-call
1192 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1194 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1196 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1197 config/riscv/riscv.c (predict.h): New include.
1198 (riscv_slow_unaligned_access_p): No longer static.
1199 (riscv_block_move_straight): Add require.
1200 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1202 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1205 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1206 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1207 nonimmediate_operand predicate for operand 1 instead of
1210 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1212 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1214 * config/aarch64/constraints.md (Uml): New constraint.
1215 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1218 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1220 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1221 of two vec_duplicates into a vec_concat.
1223 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1225 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1226 Simplify vec_merge of vec_duplicate and vec_concat.
1227 * config/aarch64/constraints.md (Utq): New constraint.
1228 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1231 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1233 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1234 Simplify vec_merge of vec_duplicate and const_vector.
1235 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1237 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1238 mode iterator. Update predicate on operand 1 to
1239 handle non-const_vec constants. Delete constraints.
1240 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1242 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1244 PR tree-optimization/78821
1245 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1247 (store_operand_info::store_operand_info): Initialize it to false.
1248 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1249 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1250 store in the group, and if chain_info is non-NULL, to ignore altogether
1252 (compatible_load_p): Fail if bit_not_p does not match.
1253 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1254 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1255 (pass_store_merging::process_store): Adjust
1256 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1257 call terminate_all_aliasing_chains newly when adding a store into
1258 a chain with non-NULL chain_info.
1260 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1262 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1264 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1266 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1268 (aarch64_layout_frame): Initialise emit_frame_chain.
1269 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1270 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1272 2017-11-08 Martin Liska <mliska@suse.cz>
1274 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1275 of gimple_call_internal_p.
1277 2017-11-07 Tom de Vries <tom@codesourcery.com>
1279 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1281 2017-11-07 Tom de Vries <tom@codesourcery.com>
1283 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1286 2017-11-08 Martin Liska <mliska@suse.cz>
1289 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1291 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1293 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1295 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1298 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1299 (*cmp<mode>_ccz_1): New insn with $k alternative.
1302 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1303 RTX_COMM_COMPARE as commutative as well.
1304 (ix86_binary_operator_ok): Formatting fix.
1305 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1306 *<code><mode>3<mask_name><round_saeonly_name>,
1307 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1308 *<s>mul<mode>3_highpart<mask_name>,
1309 *vec_widen_umult_even_v16si<mask_name>,
1310 *vec_widen_umult_even_v8si<mask_name>,
1311 *vec_widen_umult_even_v4si<mask_name>,
1312 *vec_widen_smult_even_v16si<mask_name>,
1313 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1314 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1315 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1316 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1317 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1318 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1319 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1320 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1321 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1322 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1323 ix86_binary_operator_ok. Formatting fixes.
1324 (*<plusminus_insn><mode>3<mask_name><round_name>,
1325 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1328 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1330 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1331 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1332 eq and ne if TARGET_ISEL.
1333 (cmp): New code_iterator.
1334 (UNS, UNSU_, UNSIK): New code_attrs.
1335 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1336 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1338 ("eq<mode>3"): ... this.
1339 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1341 ("ne<mode>3"): ... this.
1343 2017-11-07 Julia Koval <julia.koval@intel.com>
1346 * common/config/i386/i386-common.c
1347 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1348 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1349 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1350 * config/i386/i386.opt: Ditto.
1351 * config/i386/i386.c (ix86_target_string): Ditto.
1352 (ix86_option_override_internal): Ditto.
1353 (ix86_init_mpx_builtins): Move MPX to args2.
1354 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1355 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1356 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1357 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1358 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1359 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1360 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
1362 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
1365 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
1366 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
1367 (zero-extendsidi peephole2): Remove peephole.
1369 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1372 * stor-layout.c: Include attribs.h.
1373 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
1374 explicit lookup of "aligned" attribute.
1376 2017-11-07 Andrew Waterman <andrew@sifive.com>
1378 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
1379 (riscv_expand_block_move): Likewise.
1380 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1382 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1383 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1384 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1385 (riscv_adjust_block_mem): Likewise.
1386 (riscv_block_move_loop): Likewise.
1387 (riscv_expand_block_move): Likewise.
1388 gcc/config/riscv/riscv.md (movmemsi): New pattern.
1390 2017-11-07 Michael Clark <michaeljclark@mac.com>
1392 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
1393 (MUSL_DYNAMIC_LINKER): Likewise.
1395 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
1397 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
1398 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
1401 2017-11-07 Richard Biener <rguenther@suse.de>
1403 * match.pd: Fix build.
1405 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
1406 Jackson Woodruff <jackson.woodruff@arm.com>
1408 PR tree-optimization/71026
1409 * match.pd: Canonicalize negate in division.
1411 2017-11-07 Sudakshina Das <sudi.das@arm.com>
1414 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
1416 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1418 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
1419 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
1421 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1423 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
1424 non-scalar integral types.
1425 * match.pd (negate_expr_p): Handle MINUS_EXPR.
1426 (-(A-B), -(~A)): New transformations.
1428 2017-11-07 Tom de Vries <tom@codesourcery.com>
1430 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
1431 semicolon after "do {} while (0)".
1432 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1433 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1434 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1435 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1436 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1438 2017-11-07 Tom de Vries <tom@codesourcery.com>
1440 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
1441 after "do {} while (0)".
1442 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1443 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1444 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1445 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1446 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1448 2017-11-07 Tom de Vries <tom@codesourcery.com>
1451 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
1453 (arm_rtx_costs_internal): Add missing semicolon after
1454 HANDLE_NARROW_SHIFT_ARITH call.
1456 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1458 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1459 disable isel if it was not set explicitly.
1461 2017-11-06 James Bowman <james.bowman@ftdichip.com>
1463 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
1465 (add_type_attribute) likewise.
1467 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
1469 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
1470 of struct ix86_frame.
1471 (ix86_initial_elimination_offset): Likewise.
1472 (ix86_expand_split_stack_prologue): Likewise.
1474 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
1476 * tree-vrp.h (enum value_range_type): Update stale comment.
1478 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1479 Alan Hayward <alan.hayward@arm.com>
1480 David Sherwood <david.sherwood@arm.com>
1482 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
1483 (aarch64_expand_vec_perm_const): Take the number of units too.
1484 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
1485 (aarch64_expand_vec_perm_const): Likewise.
1486 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
1487 (vec_perm<mode>): Update accordingly.
1489 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1490 Alan Hayward <alan.hayward@arm.com>
1491 David Sherwood <david.sherwood@arm.com>
1493 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
1494 Take the number of units too.
1495 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
1496 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
1497 but check for a vector mode before rather than after the call.
1498 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
1499 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
1500 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
1501 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
1502 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
1503 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
1504 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
1505 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
1506 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
1507 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
1508 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
1509 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
1510 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
1511 (aarch64_sqdmull2_n<mode>): Update accordingly.
1513 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1514 Alan Hayward <alan.hayward@arm.com>
1515 David Sherwood <david.sherwood@arm.com>
1517 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
1518 the number of units too.
1519 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
1520 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
1521 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
1522 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
1523 (vec_store_lanesxi<mode>): Update accordingly.
1525 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1526 Alan Hayward <alan.hayward@arm.com>
1527 David Sherwood <david.sherwood@arm.com>
1529 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
1530 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
1531 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
1532 of units rather than the mode.
1533 * config/aarch64/iterators.md (nunits): New mode attribute.
1534 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
1535 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
1536 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
1537 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
1538 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1539 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
1540 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
1541 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
1542 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
1543 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
1544 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1545 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
1546 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
1547 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
1548 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1549 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1550 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
1551 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
1552 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
1553 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
1554 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
1555 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
1556 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
1557 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1558 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
1559 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
1560 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1561 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
1562 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
1563 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
1564 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
1565 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
1566 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
1567 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
1568 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
1569 (aarch64_simd_vec_setv2di): Likewise.
1571 2017-11-06 Carl Love <cel@us.ibm.com>
1573 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
1575 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
1576 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
1578 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
1580 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
1581 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
1582 Add power 8 macro expansions.
1583 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
1584 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
1585 power 8 instructions. (VSX_XXBR): Add iterator.
1587 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
1589 * config/arm/arm.md (predicable_short_it): Change default to "no",
1590 improve documentation, remove uses that are identical to the default.
1591 (enabled_for_depr_it): Rename to enabled_for_short_it.
1592 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
1593 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
1594 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
1595 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
1596 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
1598 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
1601 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
1602 float128 helper macros, which are no longer used after deleting
1603 the old 'q' built-in functions, and moving the round to odd
1604 built-in functions to being special built-in functions.
1605 (BU_FLOAT128_2): Likewise.
1606 (BU_FLOAT128_1_HW): Likewise.
1607 (BU_FLOAT128_2_HW): Likewise.
1608 (BU_FLOAT128_3_HW): Likewise.
1609 (FABSQ): Delete old 'q' built-in functions.
1610 (COPYSIGNQ): Likewise.
1611 (SQRTF128_ODD): Move round to odd built-in functions to be
1612 special built-in functions, so that we can handle
1613 -mabi=ieeelongdouble.
1614 (TRUNCF128_ODD): Likewise.
1615 (ADDF128_ODD): Likewise.
1616 (SUBF128_ODD): Likewise.
1617 (MULF128_ODD): Likewise.
1618 (DIVF128_ODD): Likewise.
1619 (FMAF128_ODD): Likewise.
1620 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
1621 built-in names to 'f128'.
1622 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
1623 old 'q' built-in functions, as the machine independent code for
1624 'f128' built-in functions handles this.
1625 (rs6000_expand_builtin): Add expansion for float128 round to odd
1626 functions, keying off on -mabi=ieeelongdouble of whether to use
1627 the KFmode or TFmode variant.
1628 (rs6000_init_builtins): Initialize the _Float128 round to odd
1630 * doc/extend.texi (PowerPC Built-in Functions): Document the old
1631 _Float128 'q' built-in functions are now mapped into the new
1632 'f128' built-in functions.
1634 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
1636 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
1637 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
1639 2017-11-06 David Malcolm <dmalcolm@redhat.com>
1642 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
1643 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
1644 * toplev.c: Include "ipa-fnsummary.h".
1645 (toplev::finalize): Call ipa_fnsummary_c_finalize.
1647 2017-11-06 Jakub Jelinek <jakub@redhat.com>
1649 PR tree-optimization/82838
1650 * gimple-ssa-store-merging.c
1651 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
1652 on a separate gimple_seq which is then appended to seq.
1654 2017-11-06 Jeff Law <law@redhat.com>
1657 * config/i386/i386.c (PROBE_INTERVAL): Remove.
1658 (get_probe_interval): New functions.
1659 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
1660 (ix86_adjust_stack_and_probe): Likewise.
1661 (output_adjust_stack_and_probe): Likewise.
1662 (ix86_emit_probe_stack_range): Likewise.
1663 (ix86_expand_prologue): Likewise.
1665 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1667 PR tree-optimization/82816
1668 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
1669 if the modes of the two types are the same.
1670 (convert_plusminus_to_widen): Likewise.
1672 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1674 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
1676 (usadv16qi): New define_expand.
1677 (usadv8hi): New define_expand.
1679 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
1682 * ipa-inline-transform.c (update_noncloned_frequencies): Always
1684 (inline_transform): Likewise.
1685 * predict.c (counts_to_freqs): Remove useless conditional.
1686 * profile-count.h (profile_count::apply_scale): Move sanity check.
1687 * tree-inline.c (copy_bb): Always scale.
1688 (copy_cfg_body): Likewise.
1690 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
1693 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
1695 (*cmp_ite0): Add enabled_for_depr_it attribute.
1696 (*cmp_ite1): Likewise.
1698 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1700 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
1703 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1705 * tree-vrp.c (vrp_int_const_binop): Return true on success and
1706 return the value by pointer.
1707 (extract_range_from_multiplicative_op_1): Update accordingly.
1708 Return as soon as an operation fails.
1710 2017-11-05 Tom de Vries <tom@codesourcery.com>
1713 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
1714 (DEF_SANITIZER_BUILTIN): ... here.
1715 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
1716 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
1718 2017-11-05 Tom de Vries <tom@codesourcery.com>
1721 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
1723 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
1724 ASM_OUTPUT_BEFORE_CASE_LABEL call.
1725 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
1727 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1728 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1729 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1731 2017-11-05 Tom de Vries <tom@codesourcery.com>
1734 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
1737 2017-11-04 Michael Clark <michaeljclark@mac.com>
1739 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
1740 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
1742 (*addsi3_extended): Likewise.
1743 (*addsi3_extended2): Likewise.
1744 (<optab>si3): Likewise.
1745 (<optab>di3): Likewise.
1746 (<optab><mode>3): Likewise.
1747 (<*optabe>si3_internal): Likewise.
1748 (zero_extendqi<SUPERQI:mode>2): Likewise.
1749 (*add<mode>hi3): Likewise.
1750 (*xor<mode>hi3): Likewise.
1751 (<optab>di3): Likewise.
1752 (*<optab>si3_extend): Likewise.
1753 (*sge<u>_<X:mode><GPR:mode>): Likewise.
1754 (*slt<u>_<X:mode><GPR:mode>): Likewise.
1755 (*sle<u>_<X:mode><GPR:mode>): Likewise.
1757 2017-11-04 Andrew Waterman <andrew@sifive.com>
1759 * config/riscv/riscv.c (riscv_option_override): Conditionally set
1760 TARGET_STRICT_ALIGN based upon -mtune argument.
1762 2017-11-04 Andrew Waterman <andrew@sifive.com>
1764 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
1766 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
1768 config/i386/i386.c (choose_basereg): Use optional scratch
1769 register and add assertion.
1770 (x86_emit_outlined_ms2sysv_save): Use scratch register when
1771 needed, and don't allocate stack.
1772 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
1773 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
1774 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
1776 2017-11-03 Jeff Law <law@redhat.com>
1778 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
1779 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
1780 to probe at the start of a noreturn function.
1782 2017-11-03 Jakub Jelinek <jakub@redhat.com>
1784 PR tree-optimization/78821
1785 * gimple-ssa-store-merging.c: Update the file comment.
1786 (MAX_STORE_ALIAS_CHECKS): Define.
1787 (struct store_operand_info): New type.
1788 (store_operand_info::store_operand_info): New constructor.
1789 (struct store_immediate_info): Add rhs_code and ops data members.
1790 (store_immediate_info::store_immediate_info): Add rhscode, op0r
1791 and op1r arguments to the ctor, initialize corresponding data members.
1792 (struct merged_store_group): Add load_align_base and load_align
1794 (merged_store_group::merged_store_group): Initialize them.
1795 (merged_store_group::do_merge): Update them.
1796 (merged_store_group::apply_stores): Pick the constant for
1797 encode_tree_to_bitpos from one of the two operands, or skip
1798 encode_tree_to_bitpos if neither operand is a constant.
1799 (class pass_store_merging): Add process_store method decl. Remove
1800 bool argument from terminate_all_aliasing_chains method decl.
1801 (pass_store_merging::terminate_all_aliasing_chains): Remove
1802 var_offset_p argument and corresponding handling.
1803 (stmts_may_clobber_ref_p): New function.
1804 (compatible_load_p): New function.
1805 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
1806 if there is overlap and rhs_code is not INTEGER_CST. For
1807 non-overlapping stores terminate group if rhs is not mergeable.
1808 (get_alias_type_for_stmts): Change first argument from
1809 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
1810 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
1811 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
1813 (get_location_for_stmts): Change first argument from
1814 auto_vec<gimple *> & to vec<gimple *> &.
1815 (struct split_store): Remove orig_stmts data member, add orig_stores.
1816 (split_store::split_store): Create orig_stores rather than orig_stmts.
1817 (find_constituent_stmts): Renamed to ...
1818 (find_constituent_stores): ... this. Change second argument from
1819 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
1820 to info structures rather than the statements.
1821 (split_group): Rename ALLOW_UNALIGNED argument to
1822 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
1823 it. Adjust find_constituent_stores caller.
1824 (imm_store_chain_info::output_merged_store): Handle rhs_code other
1825 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
1826 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
1827 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
1828 (mem_valid_for_store_merging): New function.
1829 (handled_load): New function.
1830 (pass_store_merging::process_store): New method.
1831 (pass_store_merging::execute): Use process_store method. Adjust
1832 terminate_all_aliasing_chains caller.
1834 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1836 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
1837 Return true for more constants, symbols and label references.
1838 (aarch64_valid_floating_const): Remove unused function.
1840 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1843 * config/aarch64/aarch64.c (aarch64_layout_frame):
1844 Undo forcing of LR at bottom of frame.
1846 2017-11-03 Jeff Law <law@redhat.com>
1849 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
1850 for int_registers_saved.
1852 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
1853 extracted from tree-ssa-dom.c.
1854 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
1855 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
1856 (record_equivalences_from_incoming_edge): Add additional argument
1857 to single_pred_edge_ignoring_loop_edges call.
1858 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
1859 (uncprop_dom_walker::before_dom_children): Add additional argument
1860 to single_pred_edge_ignoring_loop_edges call.
1861 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
1862 single_pred_edge_ignoring_loop_edges rather than open coding.
1863 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
1865 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
1867 * match.pd (-(-A)): Rewrite.
1869 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1871 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
1872 (rs6000_emit_int_cmove): New declaration.
1873 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
1874 (rs6000_emit_sISEL): Delete.
1875 (rs6000_emit_int_cmove): Make non-static.
1876 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
1877 instead of rs6000_emit_sISEL.
1879 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
1881 * asan.c (create_cond_insert_point): Maintain profile.
1882 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
1884 * basic-block.h (struct basic_block_def): Remove frequency.
1885 (EDGE_FREQUENCY): Use to_frequency
1886 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
1888 (find_traces): Update to use to_frequency.
1889 (find_traces_1_round): Likewise; use only IPA counts.
1890 (bb_to_key): Likewise.
1891 (connect_traces): Use IPA counts only.
1892 (copy_bb_p): Update to use to_frequency.
1893 (fix_up_crossing_landing_pad): Likewise.
1894 (sanitize_hot_paths): Likewise.
1895 * bt-load.c (basic_block_freq): Likewise.
1896 * cfg.c (init_flow): Set count_max to uninitialized.
1897 (check_bb_profile): Remove frequencies; check counts.
1898 (dump_bb_info): Do not dump frequencies.
1899 (update_bb_profile_for_threading): Update counts only.
1900 (scale_bbs_frequencies_int): Likewise.
1901 (MAX_SAFE_MULTIPLIER): Remove.
1902 (scale_bbs_frequencies_gcov_type): Update counts only.
1903 (scale_bbs_frequencies_profile_count): Update counts only.
1904 (scale_bbs_frequencies): Update counts only.
1905 * cfg.h (struct control_flow_graph): Add count-max.
1906 (update_bb_profile_for_threading): Update prototype.
1907 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
1908 (find_many_sub_basic_blocks): Likewise.
1909 * cfgcleanup.c (try_forward_edges): Likewise.
1910 (try_crossjump_to_edge): Likewise.
1911 * cfgexpand.c (expand_gimple_cond): Likewise.
1912 (expand_gimple_tailcall): Likewise.
1913 (construct_init_block): Likewise.
1914 (construct_exit_block): Likewise.
1915 * cfghooks.c (verify_flow_info): Check consistency of counts.
1916 (dump_bb_for_graph): Do not dump frequencies.
1917 (split_block_1): Do not update frequencies.
1918 (split_edge): Do not update frequencies.
1919 (make_forwarder_block): Do not update frequencies.
1920 (duplicate_block): Do not update frequencies.
1921 (account_profile_record): Do not update frequencies.
1922 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
1923 for global heuristics.
1924 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
1925 (expected_loop_iterations_unbounded): Use counts only.
1926 * cfgloopmanip.c (scale_loop_profile): Simplify.
1927 (create_empty_loop_on_edge): Simplify
1929 (duplicate_loop_to_header_edge): Simplify
1930 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
1931 (update_br_prob_note): Take care of removing note when profile
1933 (relink_block_chain): Do not dump frequency.
1934 (rtl_account_profile_record): Use to_frequency.
1935 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
1936 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
1937 (cgraph_update_edges_for_call_stmt_node): Likewise.
1938 (cgraph_edge::verify_count_and_frequency): Update.
1939 (cgraph_node::verify_node): Temporarily disable frequency verification.
1940 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
1941 to_cgraph_frequency.
1942 (cgraph_edge::rebuild_edges): Convert to ipa counts.
1943 * cgraphunit.c (init_lowered_empty_function): Do not initialize
1945 (cgraph_node::expand_thunk): Update profile.
1946 * except.c (dw2_build_landing_pads): Do not update frequency.
1947 * final.c (compute_alignments): Use to_frequency.
1948 (dump_basic_block_info): Do not dump frequency.
1949 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
1950 (dump_gimple_bb_header): Do not dump frequency.
1951 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
1953 * gimple-streamer-in.c (input_bb): Do not stream frequency.
1954 * gimple-streamer-out.c (output_bb): Do not stream frequency.
1955 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
1956 (init_before_recovery): Do not update frequency.
1957 (sched_create_recovery_edges): Do not update frequency.
1958 * hsa-gen.c (convert_switch_statements): Do not update frequency.
1959 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
1960 (ipa_cp_c_finalize): Set max_count to uninitialized.
1961 * ipa-fnsummary.c (get_minimal_bb): Use counts.
1962 (param_change_prob): Use counts.
1963 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
1965 * ipa-split.c (consider_split): Use to_frequency.
1966 (split_function): Use to_frequency.
1967 * ira-build.c (loop_compare_func): Likewise.
1968 (mark_loops_for_removal): Likewise.
1969 (mark_all_loops_for_removal): Likewise.
1970 * loop-doloop.c (doloop_modify): Do not update frequency.
1971 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
1973 * lto-streamer-in.c (input_function): Update count_max.
1974 * omp-expand.c (expand_omp_taskreg): Update count_max.
1975 * omp-simd-clone.c (simd_clone_adjust): Update profile.
1976 * predict.c (maybe_hot_frequency_p): Use to_frequency.
1977 (maybe_hot_count_p): Use ipa counts only.
1978 (maybe_hot_bb_p): Simplify.
1979 (maybe_hot_edge_p): Simplify.
1980 (probably_never_executed): Do not take frequency argument.
1981 (probably_never_executed_bb_p): Do not pass frequency.
1982 (probably_never_executed_edge_p): Likewise.
1983 (combine_predictions_for_bb): Check that profile is nonzero.
1984 (propagate_freq): Do not set frequency.
1985 (drop_profile): Simplify.
1986 (counts_to_freqs): Simplify.
1987 (expensive_function_p): Use to_frequency.
1988 (propagate_unlikely_bbs_forward): Simplify.
1989 (determine_unlikely_bbs): Simplify.
1990 (estimate_bb_frequencies): Add hack to silence graphite issues.
1991 (compute_function_frequency): Use ipa counts.
1992 (pass_profile::execute): Update.
1993 (rebuild_frequencies): Use counts only.
1994 (force_edge_cold): Use counts only.
1995 * profile-count.c (profile_count::dump): Dump new count types.
1996 (profile_count::differs_from_p): Check compatiblity.
1997 (profile_count::to_frequency): New function.
1998 (profile_count::to_cgraph_frequency): New function.
1999 * profile-count.h (struct function): Declare.
2000 (enum profile_quality): Add profile_guessed_local and
2001 profile_guessed_global0.
2002 (class profile_proability): Decrease number of bits to 29;
2003 update from_reg_br_prob_note and to_reg_br_prob_note.
2004 (class profile_count: Update comment; decrease number of bits
2005 to 61. Check compatibility.
2006 (profile_count::compatible_p): New private member function.
2007 (profile_count::ipa_p): New member function.
2008 (profile_count::operator<): Handle global zero correctly.
2009 (profile_count::operator>): Handle global zero correctly.
2010 (profile_count::operator<=): Handle global zero correctly.
2011 (profile_count::operator>=): Handle global zero correctly.
2012 (profile_count::nonzero_p): New member function.
2013 (profile_count::force_nonzero): New member function.
2014 (profile_count::max): New member function.
2015 (profile_count::apply_scale): Handle IPA scalling.
2016 (profile_count::guessed_local): New member function.
2017 (profile_count::global0): New member function.
2018 (profile_count::ipa): New member function.
2019 (profile_count::to_frequency): Declare.
2020 (profile_count::to_cgraph_frequency): Declare.
2021 * profile.c (OVERLAP_BASE): Delete.
2022 (compute_frequency_overlap): Delete.
2023 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2024 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2025 * sched-ebb.c (rank): Use counts only.
2026 * shrink-wrap.c (handle_simple_exit): Use counts only.
2027 (try_shrink_wrapping): Use counts only.
2028 (place_prologue_for_one_component): Use counts only.
2029 * tracer.c (find_best_predecessor): Use to_frequency.
2030 (find_trace): Use to_frequency.
2031 (tail_duplicate): Use to_frequency.
2032 * trans-mem.c (expand_transaction): Do not update frequency.
2033 * tree-call-cdce.c: Do not update frequency.
2034 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2035 (gimple_merge_blocks): Likewise.
2036 (gimple_split_edge): Likewise.
2037 (gimple_duplicate_sese_region): Likewise.
2038 (gimple_duplicate_sese_tail): Likewise.
2039 (move_sese_region_to_fn): Likewise.
2040 (gimple_account_profile_record): Likewise.
2041 (insert_cond_bb): Likewise.
2042 * tree-complex.c (expand_complex_div_wide): Likewise.
2043 * tree-eh.c (lower_resx): Update profile.
2044 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2046 (initialize_cfun): Do not initialize frequencies
2047 (freqs_to_counts): Delete.
2048 (copy_cfg_body): Ignore count parameter.
2049 (copy_body): Update.
2050 (expand_call_inline): Update count_max.
2051 (optimize_inline_calls): Update count_max.
2052 (tree_function_versioning): Update count_max.
2053 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2054 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2056 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2057 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2058 (try_peel_loop): Likewise.
2059 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2061 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2062 (tree_transform_and_unroll_loop): Do not use frequencies
2063 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2064 Use reliable prediction only.
2065 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2066 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2067 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2068 probability scaling.
2069 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2070 not update frequency
2071 (any_remaining_duplicated_blocks): Likewise.
2072 (update_profile): Likewise.
2073 (estimated_freqs_path): Delete.
2074 (freqs_to_counts_path): Delete.
2075 (clear_counts_path): Delete.
2076 (ssa_fix_duplicate_block_edges): Likewise.
2077 (duplicate_thread_path): Likewise.
2078 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2079 * tree-tailcall.c (decrease_profile): Do not update frequency.
2080 (eliminate_tail_call): Likewise.
2081 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2082 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2083 (optimize_mask_stores): Likewise.
2084 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2085 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2086 (ubsan_expand_ptr_ifn): Update profile.
2087 * value-prof.c (gimple_ic): Simplify.
2088 * value-prof.h (gimple_ic): Update prototype.
2089 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2090 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2092 (want_inline_self_recursive_call_p): Likewise.
2093 (resolve_noninline_speculation): Only cummulate defined counts.
2094 (inline_small_functions): Use nonzero_p.
2095 (ipa_inline): Do not access freed node.
2097 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2099 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2100 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2102 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2104 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2105 non-legitimate address.
2107 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2109 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2110 (*lt0_<mode>di, *lt0_<mode>si): New.
2112 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2114 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2115 TARGET_PAIRED_FLOAT.
2117 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2118 Jim Wilson <jim.wilson@linaro.org>
2120 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2121 * config/aarch64/aarch64-tune.md: Regenerated.
2122 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2123 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2125 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2127 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2128 (arc_expand_prologue): Restore blink for millicode.
2129 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2131 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2134 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2135 gen_vec_duplicate after forcing the scalar into a register.
2137 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2139 * combine (try_combine): Print the insns input to try_combine to the
2142 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2145 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2146 Remove second argument from aarch64_process_target_attr call.
2147 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2149 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2151 (aarch64_handle_attr_arch): Remove second argument.
2152 (aarch64_handle_attr_cpu): Ditto.
2153 (aarch64_handle_attr_tune): Ditto.
2154 (aarch64_handle_attr_isa_flags): Ditto.
2155 (aarch64_process_one_target_attr): Ditto.
2156 (aarch64_process_target_attr): Ditto.
2157 (aarch64_option_valid_attribute_p): Remove second argument.
2158 on aarch64_process_target_attr call.
2160 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2162 * diagnostic.c: Include "selftest-diagnostic.h".
2163 (selftest::assert_location_text): New function.
2164 (selftest::test_diagnostic_get_location_text): New function.
2165 (selftest::diagnostic_c_tests): Call it.
2167 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2169 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2170 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2171 (class selftest::test_diagnostic_context): Move to...
2172 * selftest-diagnostic.c: New file.
2173 * selftest-diagnostic.h: New file.
2175 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2177 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2178 offset range for FT32B.
2179 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2180 * config/ft32/ft32.md: Add TARGET_NOPM.
2181 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2182 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2184 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2186 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2188 2017-11-02 Jeff Law <law@redhat.com>
2190 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2191 virtual keyword on FINAL OVERRIDE members.
2193 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2194 virtuals together. Add virtual destructor.
2195 (substitute_and_fold_engine): Similarly.
2197 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2199 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2201 2017-11-02 Richard Biener <rguenther@suse.de>
2203 PR tree-optimization/82795
2204 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2206 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2208 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2210 (gcc_GAS_CHECK_FEATURE): Remove.
2211 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2212 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2213 * configure: Regenerate.
2215 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2217 * config/arc/arc.c (hwloop_optimize): Account for empty
2220 2017-11-02 Richard Biener <rguenther@suse.de>
2223 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2224 Truncate ARRAY_REF index and element size.
2226 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2228 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2230 2017-11-01 Jeff Law <law@redhat.com>
2232 * tree-ssa-ccp.c (ccp_folder): New class derived from
2233 substitute_and_fold_engine.
2234 (ccp_folder::get_value): New member function.
2235 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2236 (ccp_fold_stmt): Remove prototype.
2237 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2238 * tree-ssa-copy.c (copy_folder): New class derived from
2239 substitute_and_fold_engine.
2240 (copy_folder::get_value): Renamed from get_value.
2241 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2242 * tree-vrp.c (vrp_folder): New class derived from
2243 substitute_and_fold_engine.
2244 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2245 (vrp_folder::get_value): New member function.
2246 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2247 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2248 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2249 provide a class interface to folder/substitute routines.
2250 (ssa_prop_fold_stmt_fn): Remove typedef.
2251 (ssa_prop_get_value_fn): Likewise.
2252 (subsitute_and_fold): Remove prototype.
2253 (replace_uses_in): Likewise.
2254 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2255 Renamed from replace_uses_in. Call the virtual member function
2256 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2257 (substitute_and_fold_dom_walker): Remove initialization of
2258 data member entries for calbacks. Add substitute_and_fold_engine
2259 member and initialize it.
2260 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2261 member functions for get_value, replace_phi_args_in c
2262 replace_uses_in, and fold_stmt calls.
2263 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2264 substitute_and_fold. Remove assert. Update ctor call.
2266 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2267 (ssa_prop_visit_phi_fn): Likewise.
2268 (class ssa_propagation_engine): New class to provide an interface
2270 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2272 (ssa_prop_visit_phi): Likewise.
2273 (ssa_propagation_engine::simulate_stmt): Moved into class.
2274 Call visit_phi/visit_stmt from the class rather than via
2275 file scoped static variables.
2276 (ssa_propagation_engine::simulate_block): Moved into class.
2277 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2278 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2279 set file scoped statics for the visit_stmt/visit_phi callbacks.
2280 * tree-complex.c (complex_propagate): New class derived from
2281 ssa_propagation_engine.
2282 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2283 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2284 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2286 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2287 ssa_propagation_engine.
2288 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2289 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2290 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2291 * tree-ssa-copy.c (copy_prop): New class derived from
2292 ssa_propagation_engine.
2293 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2294 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2295 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2296 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2297 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2298 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2299 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2301 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2303 PR rtl-optimization/82778
2304 PR rtl-optimization/82597
2305 * compare-elim.c (struct comparison): Add in_a_setter field.
2306 (find_comparison_dom_walker::before_dom_children): Remove killed
2307 bitmap and df_simulate_find_defs call, instead walk the defs.
2308 Compute last_setter and initialize in_a_setter. Merge definitions
2309 with first initialization for a few variables.
2310 (try_validate_parallel): Use insn_invalid_p instead of
2311 recog_memoized. Return insn rather than just the pattern.
2312 (try_merge_compare): Fix up comment. Don't uselessly test if
2313 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2315 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2318 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2319 Alan Hayward <alan.hayward@arm.com>
2320 David Sherwood <david.sherwood@arm.com>
2322 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2323 aarch64_hard_regno_nregs to get the number of registers
2326 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2327 Alan Hayward <alan.hayward@arm.com>
2328 David Sherwood <david.sherwood@arm.com>
2330 * config/aarch64/constraints.md (Upl): Rename to...
2332 * config/aarch64/aarch64.md
2333 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
2336 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2337 Alan Hayward <alan.hayward@arm.com>
2338 David Sherwood <david.sherwood@arm.com>
2340 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2341 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2344 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2345 Alan Hayward <alan.hayward@arm.com>
2346 David Sherwood <david.sherwood@arm.com>
2348 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2349 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2350 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2352 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2353 of a single element.
2354 * config/aarch64/iterators.md: Add a comment above the permute
2355 unspecs to say that they are generated directly by
2356 aarch64_expand_vec_perm_const.
2357 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
2359 2017-11-01 Nathan Sidwell <nathan@acm.org>
2361 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
2363 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2365 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
2366 and medany code models, and describe what they do.
2368 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2370 Revert accidental duplicate:
2372 * combine.c (can_change_dest_mode): Reject changes in
2373 REGMODE_NATURAL_SIZE.
2375 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
2377 PR rtl-optimization/64682
2378 PR rtl-optimization/69567
2379 PR rtl-optimization/69737
2380 PR rtl-optimization/82683
2381 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2382 register mentioned in the note, drop the note, unless it came from I3,
2383 in which case it should go to I3 again.
2385 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2387 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
2388 and return false if not.
2389 (clear_bytes_written_by, live_bytes_read): Update accordingly.
2391 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2393 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
2394 range is known to be empty.
2396 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2397 Alan Hayward <alan.hayward@arm.com>
2398 David Sherwood <david.sherwood@arm.com>
2400 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
2401 and CONST_VECTOR_NUNITS instead of computing the number of units from
2402 the byte sizes of the vector and element.
2403 (simplify_binary_operation_1): Likewise.
2404 (simplify_const_binary_operation): Likewise.
2405 (simplify_ternary_operation): Likewise.
2407 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2408 Alan Hayward <alan.hayward@arm.com>
2409 David Sherwood <david.sherwood@arm.com>
2411 * var-tracking.c (INT_MEM_OFFSET): Replace with...
2412 (int_mem_offset): ...this new function.
2413 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
2414 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
2415 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
2418 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2419 Alan Hayward <alan.hayward@arm.com>
2420 David Sherwood <david.sherwood@arm.com>
2422 * lower-subreg.c (interesting_mode_p): New function.
2423 (compute_costs, find_decomposable_subregs, decompose_register)
2424 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
2425 (resolve_clobber, dump_choices): Use it.
2427 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2428 Alan Hayward <alan.hayward@arm.com>
2429 David Sherwood <david.sherwood@arm.com>
2431 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
2433 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2434 Alan Hayward <alan.hayward@arm.com>
2435 David Sherwood <david.sherwood@arm.com>
2437 * alias.c (find_base_value, find_base_term): Only process integer
2438 truncations. Check the precision rather than the size.
2440 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2441 Alan Hayward <alan.hayward@arm.com>
2442 David Sherwood <david.sherwood@arm.com>
2444 * machmode.h (is_narrower_int_mode): New function
2445 * optabs.c (expand_float, expand_fix): Use it.
2446 * dwarf2out.c (rotate_loc_descriptor): Likewise.
2448 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2449 Alan Hayward <alan.hayward@arm.com>
2450 David Sherwood <david.sherwood@arm.com>
2452 * rtl.h (narrower_subreg_mode): New function.
2453 * ira-color.c (update_costs_from_allocno): Use it.
2455 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2456 Alan Hayward <alan.hayward@arm.com>
2457 David Sherwood <david.sherwood@arm.com>
2459 * optabs-query.h (convert_optab_p): New function, split out from...
2460 (convert_optab_handler): ...here.
2461 (widening_optab_handler): Delete.
2462 (find_widening_optab_handler): Remove permit_non_widening parameter.
2463 (find_widening_optab_handler_and_mode): Likewise. Provide an
2464 override that operates on mode class wrappers.
2465 * optabs-query.c (widening_optab_handler): Delete.
2466 (find_widening_optab_handler_and_mode): Remove permit_non_widening
2467 parameter. Assert that the two modes are the same class and that
2468 the "from" mode is narrower than the "to" mode. Use
2469 convert_optab_handler instead of widening_optab_handler.
2470 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
2471 instead of widening_optab_handler.
2472 * expr.c (expand_expr_real_2): Update calls to
2473 find_widening_optab_handler.
2474 * optabs.c (expand_widen_pattern_expr): Likewise.
2475 (expand_binop_directly): Take the insn_code as a parameter.
2476 (expand_binop): Only call find_widening_optab_handler for
2477 conversion optabs; use optab_handler otherwise. Update calls
2478 to find_widening_optab_handler and expand_binop_directly.
2479 Use convert_optab_handler instead of widening_optab_handler.
2480 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
2481 find_widening_optab_handler and use scalar_mode rather than
2483 (convert_plusminus_to_widen): Likewise.
2485 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2486 Alan Hayward <alan.hayward@arm.com>
2487 David Sherwood <david.sherwood@arm.com>
2489 * machmode.h (fixed_size_mode): New class.
2490 * rtl.h (get_pool_mode): Return fixed_size_mode.
2491 * gengtype.c (main): Add fixed_size_mode.
2492 * target.def (get_raw_result_mode): Return a fixed_size_mode.
2493 (get_raw_arg_mode): Likewise.
2494 * doc/tm.texi: Regenerate.
2495 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
2496 * targhooks.c (default_get_reg_raw_mode): Likewise.
2497 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
2498 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
2499 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
2500 (msp430_get_raw_result_mode): Likewise.
2501 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
2502 * dbxout.c (dbxout_parms): Require fixed-size modes.
2503 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
2504 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2505 * omp-low.c (lower_oacc_reductions): Likewise.
2506 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
2507 (simplify_subreg): Update accordingly.
2508 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
2509 (force_const_mem): Update accordingly. Return NULL_RTX for modes
2510 that aren't fixed-size.
2511 (get_pool_mode): Return a fixed_size_mode.
2512 (output_constant_pool_2): Take a fixed_size_mode.
2514 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2515 Alan Hayward <alan.hayward@arm.com>
2516 David Sherwood <david.sherwood@arm.com>
2518 * doc/rtl.texi (vec_series): Document.
2519 (const): Say that the operand can be a vec_series.
2520 * rtl.def (VEC_SERIES): New rtx code.
2521 * rtl.h (const_vec_series_p_1): Declare.
2522 (const_vec_series_p): New function.
2523 * emit-rtl.h (gen_const_vec_series): Declare.
2524 (gen_vec_series): Likewise.
2525 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
2526 (gen_vec_series): Likewise.
2527 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
2528 * simplify-rtx.c (simplify_unary_operation): Handle negations
2530 (simplify_binary_operation_series): New function.
2531 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
2532 (test_vector_ops_series): New function.
2533 (test_vector_ops): Call it.
2534 * config/powerpcspe/altivec.md (altivec_lvsl): Use
2535 gen_const_vec_series.
2536 (altivec_lvsr): Likewise.
2537 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
2539 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2540 Alan Hayward <alan.hayward@arm.com>
2541 David Sherwood <david.sherwood@arm.com>
2543 * doc/rtl.texi (const): Update description of address constants.
2544 Say that vector constants are allowed too.
2545 * common.md (E, F): Use CONSTANT_P instead of checking for
2547 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
2548 checking for CONST_VECTOR.
2549 * expmed.c (make_tree): Use build_vector_from_val for a CONST
2551 * expr.c (expand_expr_real_2): Check for vector modes instead
2552 of checking for CONST_VECTOR.
2553 * rtl.h (const_vec_p): New function.
2554 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
2555 (unwrap_const_vec_duplicate): Handle them here too.
2557 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2558 David Malcolm <dmalcolm@redhat.com>
2559 Alan Hayward <alan.hayward@arm.com>
2560 David Sherwood <david.sherwood@arm.com>
2562 * rtl.h (vec_duplicate_p): New function.
2563 * selftest-rtl.c (assert_rtx_eq_at): New function.
2564 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
2565 (assert_rtx_eq_at): Declare.
2566 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
2567 * selftest-run-tests.c (selftest::run_tests): Call it.
2568 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
2569 (simplify_unary_operation_1): Recursively handle vector duplicates.
2570 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
2572 (simplify_subreg): Handle subregs of vector duplicates.
2573 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
2574 (selftest::simplify_rtx_c_tests): New functions.
2576 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2577 Alan Hayward <alan.hayward@arm.com>
2578 David Sherwood <david.sherwood@arm.com>
2580 * emit-rtl.h (gen_const_vec_duplicate): Declare.
2581 (gen_vec_duplicate): Likewise.
2582 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
2584 (gen_const_vector): ...here.
2585 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
2586 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
2587 whose elements are all equal.
2588 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
2589 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
2590 (simplify_relational_operation): Likewise.
2591 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
2593 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
2594 (aarch64_expand_vector_init): Likewise.
2595 * config/arm/arm.c (neon_vdup_constant): Likewise.
2596 (neon_expand_vector_init): Likewise.
2597 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
2598 (arm_block_set_unaligned_vect): Likewise.
2599 (arm_block_set_aligned_vect): Likewise.
2600 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
2601 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
2602 (expand_vec_perm_even_odd_pack): Likewise.
2603 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
2604 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
2605 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
2606 gen_const_vec_duplicate.
2607 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
2608 * config/mips/mips.c (mips_gen_const_int_vector): Use
2609 gen_const_vec_duplicate.
2610 (mips_expand_vector_init): Use CONST0_RTX.
2611 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
2612 (define_split): Use gen_const_vec_duplicate.
2613 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
2614 (define_split): Use gen_const_vec_duplicate.
2615 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
2616 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
2617 * config/spu/spu.c (spu_const): Likewise.
2619 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2620 Alan Hayward <alan.hayward@arm.com>
2621 David Sherwood <david.sherwood@arm.com>
2623 * combine.c (can_change_dest_mode): Reject changes in
2624 REGMODE_NATURAL_SIZE.
2626 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
2628 * configure.ac (--enable-libssp): New.
2629 (gcc_cv_libc_provides_ssp): Check for explicit setting before
2630 trying to determine target-specific default. Adjust indentation.
2631 * configure: Regenerated.
2632 * doc/install.texi (Configuration): Expand --disable-libssp
2635 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
2637 config/i386/i386.c (ix86_expand_epilogue): Correct stack
2640 2017-10-31 Martin Jambor <mjambor@suse.cz>
2643 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
2645 2017-10-31 David Malcolm <dmalcolm@redhat.com>
2647 * auto-profile.c (autofdo_source_profile::read): Use
2648 UNKNOWN_LOCATION rather than 0.
2649 * diagnostic-core.h (warning_at_rich_loc): Rename to...
2650 (warning_at): ...this overload.
2651 (warning_at_rich_loc_n): Rename to...
2652 (warning_n): ...this overload.
2653 (error_at_rich_loc): Rename to...
2654 (error_at): ...this overload.
2655 (pedwarn_at_rich_loc): Rename to...
2656 (pedwarn): ...this overload.
2657 (permerror_at_rich_loc): Rename to...
2658 (permerror): ...this overload.
2659 (inform_at_rich_loc): Rename to...
2660 (inform): ...this overload.
2661 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
2662 (diagnostic_n_impl_richloc): Rename to...
2663 (diagnostic_n_impl): ...this rich_location *-based decl.
2664 (inform_at_rich_loc): Rename to...
2665 (inform): ...this, and add an assertion.
2666 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
2667 (warning_at_rich_loc): Rename to...
2668 (warning_at): ...this, and add an assertion.
2669 (warning_at_rich_loc_n): Rename to...
2670 (warning_n): ...this, and add an assertion.
2671 (warning_n): Update location_t-based implementation for removal of
2672 location_t-based diagnostic_n_impl.
2673 (pedwarn_at_rich_loc): Rename to...
2674 (pedwarn): ...this, and add an assertion.
2675 (permerror_at_rich_loc): Rename to...
2676 (permerror): ...this, and add an assertion.
2677 (error_n): Update for removal of location_t-based diagnostic_n_impl.
2678 (error_at_rich_loc): Rename to...
2679 (error_at): ...this, and add an assertion.
2680 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
2681 (driver::do_spec_on_infiles): Likewise.
2682 * substring-locations.c (format_warning_va): Update for renaming
2683 of inform_at_rich_loc.
2685 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2687 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
2688 _Float<N>X built-in functions so that the variant without the
2689 "__builtin_" prefix is only enabled for the GNU C and Objective C
2690 languages when they are in non-strict ANSI/ISO mode.
2691 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
2692 * target.def (floatn_builtin_p): Add a target hook to control
2693 whether _Float<N> and _Float<N>X built-in functions without the
2694 "__builtin_" prefix are enabled, and return true for C and
2695 Objective C in the default hook. Include langhooks.h in
2697 * targhooks.h (default_floatn_builtin_p): Likewise.
2698 * targhooks.c (default_floatn_builtin_p): Likewise.
2699 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
2700 floatn_builtin_p target hook.
2701 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
2703 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
2704 Eric Botcazou <ebotcazou@adacore.com>
2706 PR rtl-optimization/81803
2707 * lra-constraints.c (curr_insn_transform): Also reload the whole
2708 register for a strict subreg no wider than a word if this is for
2709 a WORD_REGISTER_OPERATIONS target.
2711 2017-10-31 Jason Merrill <jason@redhat.com>
2713 * gdbinit.in: Skip over inlines from timevar.h.
2715 2017-10-31 Martin Liska <mliska@suse.cz>
2717 * doc/gcov.texi: Document new option.
2718 * gcov.c (print_usage): Likewise print it.
2719 (process_args): Support the argument.
2720 (format_count): New function.
2721 (format_gcov): Use the function.
2723 2017-10-31 Martin Liska <mliska@suse.cz>
2725 * gcov.c (struct name_map): do not use typedef.
2726 Define operator== and operator<.
2727 (name_search): Remove.
2728 (name_sort): Remove.
2729 (main): Do not allocate names.
2730 (process_file): Add vertical space.
2731 (generate_results): Use std::find.
2732 (release_structures): Do not release memory.
2733 (find_source): Use std::find.
2735 2017-10-31 Martin Liska <mliska@suse.cz>
2737 * gcov.c (struct line_info): Remove it's typedef.
2738 (line_info::line_info): Add proper ctor.
2739 (line_info::has_block): Do not use a typedef.
2740 (struct source_info): Do not use typedef.
2741 (circuit): Likewise.
2742 (get_cycles_count): Likewise.
2743 (output_intermediate_file): Iterate via vector iterator.
2744 (add_line_counts): Use std::vector methods.
2745 (accumulate_line_counts): Likewise.
2746 (output_lines): Likewise.
2748 2017-10-31 Martin Liska <mliska@suse.cz>
2750 * gcov.c (struct source_info): Remove typedef.
2751 (source_info::source_info): Add proper ctor.
2752 (accumulate_line_counts): Use struct, not it's typedef.
2753 (output_gcov_file): Likewise.
2754 (output_lines): Likewise.
2755 (main): Do not allocate an array.
2756 (output_intermediate_file): Use size of vector container.
2757 (process_file): Resize the vector.
2758 (generate_results): Do not preallocate, use newly added vector
2760 (release_structures): Do not release sources.
2761 (find_source): Use vector methods.
2762 (add_line_counts): Do not use typedef.
2764 2017-10-31 Martin Liska <mliska@suse.cz>
2766 * doc/gcov.texi: Document that.
2767 * gcov.c (add_line_counts): Mark lines with a non-executed
2769 (output_line_beginning): Handle such lines.
2770 (output_lines): Pass new argument.
2771 (output_intermediate_file): Print it in intermediate format.
2773 2017-10-31 Martin Liska <mliska@suse.cz>
2775 * color-macros.h: New file.
2776 * diagnostic-color.c: Factor out color related to macros to
2778 * doc/gcov.texi: Document -k option.
2779 * gcov.c (INCLUDE_STRING): Include string.h.
2780 (print_usage): Add -k option.
2781 (process_args): Parse it.
2782 (pad_count_string): New function.
2783 (output_line_beginning): Likewise.
2784 (DEFAULT_LINE_START): New macro.
2785 (output_lines): Support color output.
2787 2017-10-31 Martin Liska <mliska@suse.cz>
2789 PR gcov-profile/82633
2790 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
2791 their interaction with GCOV infrastructure.
2792 * configure.ac: Add -fkeep-{inline,static}-functions to
2794 * configure: Regenerate.
2796 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
2799 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
2801 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
2804 * config/rs6000/rs6000.md (allocate_stack): Force update interval
2805 into a register if it does not fit into an immediate offset field.
2807 2017-10-31 Olivier Hainque <hainque@adacore.com>
2809 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
2811 2017-10-31 Julia Koval <julia.koval@intel.com>
2813 * config.gcc: Add gfniintrin.h.
2814 * config/i386/gfniintrin.h: New.
2815 * config/i386/i386-builtin-types.def
2816 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2817 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2818 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2819 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2820 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2821 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
2822 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
2823 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
2824 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
2825 V64QI_FTYPE_V64QI_V64QI_INT): New types.
2826 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2827 * config/i386/immintrin.h: Include gfniintrin.h.
2828 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
2830 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
2832 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
2834 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2836 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
2837 (arm_ashldi3_1bit): Remove pattern.
2838 (ashrdi3): Remove shift by 1 expansion.
2839 (arm_ashrdi3_1bit): Remove pattern.
2840 (lshrdi3): Remove shift by 1 expansion.
2841 (arm_lshrdi3_1bit): Remove pattern.
2842 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
2843 cost of ashldi3 by 1.
2844 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
2845 (<shift>di3_neon): Likewise.
2847 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2849 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
2850 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
2851 and (*aarch64_simd_mov<VQ:mode>).
2852 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
2853 pattern alternative.
2854 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
2855 attributes to match pattern alternative.
2857 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
2859 * config.gcc (powerpc*-*-*): Add emmintrin.h.
2860 * config/rs6000/emmintrin.h: New file.
2861 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
2863 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2865 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
2866 * (movdi_vfp_cortexa8): Remove pattern.
2868 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2870 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
2872 (Specific, alpha*-dec-osf5.1): Remove.
2873 (Specific, mips-sgi-irix5): Remove.
2874 (Specific, mips-sgi-irix6): Remove.
2876 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2879 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
2880 arguments to clear_bit_region_be.
2882 2017-10-30 Jim Wilson <wilson@tuliptree.org>
2884 * gimplify.c: Include memmodel.h.
2886 2017-10-30 Martin Jambor <mjambor@suse.cz>
2888 * omp-grid.c (grid_attempt_target_gridification): Also insert a
2889 condition whether loop should be executed at all.
2891 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
2893 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2894 gimple folding of vec_madd() intrinsics.
2895 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
2896 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
2897 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
2899 2017-10-30 Richard Biener <rguenther@suse.de>
2901 PR tree-optimization/82762
2903 2017-10-23 Richard Biener <rguenther@suse.de>
2905 PR tree-optimization/82129
2907 2017-08-01 Richard Biener <rguenther@suse.de>
2909 PR tree-optimization/81181
2910 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2911 (compute_antic): ... end of iteration here.
2913 2017-10-30 Joseph Myers <joseph@codesourcery.com>
2915 * doc/invoke.texi (C Dialect Options): Document -std=c17,
2916 -std=iso9899:2017 and -std=gnu17.
2917 * doc/standards.texi (C Language): Document C17 support.
2918 * doc/cpp.texi (Overview): Mention -std=c17.
2919 (Standard Predefined Macros): Document C11 and C17 values of
2920 __STDC_VERSION__. Do not refer to C99 support as incomplete.
2921 * doc/extend.texi (Inline): Do not list individual options for
2922 standards newer than C99.
2923 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
2925 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
2928 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
2930 * asan.c (asan_finish_file): Align asan globals array by shadow
2933 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2936 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
2937 (struct store_immediate_info): Add bitregion_start and bitregion_end
2939 (store_immediate_info::store_immediate_info): Add brs and bre
2940 arguments and initialize bitregion_{start,end} from those.
2941 (struct merged_store_group): Add bitregion_start, bitregion_end,
2942 align_base and mask fields. Drop unnecessary struct keyword from
2943 struct store_immediate_info. Add do_merge method.
2944 (clear_bit_region_be): Use memset instead of loop storing zeros.
2945 (merged_store_group::do_merge): New method.
2946 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
2947 stores as long as the surrounding bitregions have no gaps.
2948 (merged_store_group::merge_overlapping): Use do_merge.
2949 (merged_store_group::apply_stores): Test that bitregion_{start,end}
2950 is byte aligned, rather than requiring that start and width are
2951 byte aligned. Drop unnecessary struct keyword from
2952 struct store_immediate_info. Allocate and populate also mask array.
2953 Make start of the arrays relative to bitregion_start rather than
2954 start and size them according to bitregion_{end,start} difference.
2955 (struct imm_store_chain_info): Drop unnecessary struct keyword from
2956 struct store_immediate_info.
2957 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
2958 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
2959 struct keyword from struct store_immediate_info.
2960 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
2961 between stores as long as the surrounding bitregions have no gaps.
2963 (struct split_store): Add orig non-static data member.
2964 (split_store::split_store): Initialize orig to false.
2965 (find_constituent_stmts): Return store_immediate_info *, non-NULL
2966 if there is exactly a single original stmt. Change stmts argument
2967 to pointer from reference, if NULL, don't push anything to it. Add
2968 first argument, use it to optimize skipping over orig stmts that
2969 are known to be before bitpos already. Simplify.
2970 (split_group): Return unsigned int count how many stores are or
2971 would be needed rather than a bool. Add allow_unaligned argument.
2972 Change split_stores argument from reference to pointer, if NULL,
2973 only do a dry run computing how many stores would be produced.
2974 Rewritten algorithm to use both alignment and misalign if
2975 !allow_unaligned and handle bitfield stores with gaps.
2976 (imm_store_chain_info::output_merged_store): Set start_byte_pos
2977 from bitregion_start instead of start. Compute allow_unaligned
2978 here, if true, do 2 split_group dry runs to compute which one
2979 produces fewer stores and prefer aligned if equal. Punt if
2980 new count is bigger or equal than original before emitting any
2981 statements, rather than during that. Remove no longer needed
2982 new_ssa_names tracking. Replace num_stmts with
2983 split_stores.length (). Use 32-bit stack allocated entries
2984 in split_stores auto_vec. Try to reuse original store lhs/rhs1
2985 if possible. Handle bitfields with gaps.
2986 (pass_store_merging::execute): Ignore bitsize == 0 stores.
2987 Compute bitregion_{start,end} for the stores and construct
2988 store_immediate_info with that. Formatting fixes.
2990 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
2993 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
2994 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
2996 2017-10-29 Jim Wilson <wilson@tuliptree.org>
2998 * gimplify.c: Include tm_p.h.
3000 * common.opt (gcoff): Re-add as ignored option.
3001 (gcoff1, gcoff2, gcoff3): Likewise.
3003 * Makefile.in (OBJS): Delete sdbout.o.
3004 (GTFILES): Delete $(srcdir)/sdbout.c.
3005 * debug.h: Delete sdb_debug_hooks.
3006 * final.c: Delete sdbout.h include.
3007 (final_scan_insn): Delete SDB_DEBUG check.
3008 (rest_of_clean_state): Likewise.
3009 * output.h: Delete sdb_begin_function_line.
3012 * toplev.c: Delete sdbout.h include.
3013 (process_options): Delete SDB_DEBUG check.
3014 * tree-core.h (tree_type_common): Delete pointer field of
3016 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3017 TYPE_SYMTAB_POINTER.
3018 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3019 (TYPE_SYMTAB_IS_POINTER): Delete.
3020 (TYPE_SYMTAB_IS_DIE): Renumber.
3021 * xcoffout.c: Refer to former sdbout.c file.
3022 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3024 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3025 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3026 (Debugging Options): Delete -gcoff.
3027 (-gstabs): Delete SDB reference.
3029 (-gcoff@var{level}): Delete.
3030 * doc/passes.texi (Debugging information output): Delete SDB and
3031 sdbout.c references.
3032 * doc/tm.texi: Regenerate.
3033 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3034 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3036 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3037 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3038 (SDB_DEBUGGING_INFO): Delete.
3039 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3040 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3041 * target.def (output_source_filename): Delete COFF reference.
3043 * common.opt (gcoff): Delete.
3044 (gxcoff+): Update Negative chain.
3045 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3047 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3048 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3049 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3051 (expand_function_start): Change sdb reference to past tense.
3052 (expand_function_end): Change sdb reference to past tense.
3053 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3054 * opts.c (debug_type_names): Delete coff entry.
3055 (common_handle_option): Delete OPT_gcoff case.
3056 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3058 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3059 * config/cris/cris.h: Delete SDB reference in comment.
3060 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3061 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3062 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3063 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3065 (ix86_expand_prologue): Likewise.
3066 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3067 * config/ia64/ia64.h: Likewise.
3068 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3069 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3071 * config/mmix/mmix.h: Likewise.
3072 * config/nds32/nds32.c: Likewise.
3073 * config/stormy/storym16.h: Likewise.
3074 * config/visium/visium.h: Likewise.
3075 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3077 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3079 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3080 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3081 FRAME_POINTER_REGNUM point at high end of local var area.
3083 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3085 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3086 Move comment around. Do not reset best_edge for a copiable
3087 destination if the copy would cause a partition change.
3088 (better_edge_p): Remove redundant check.
3090 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3092 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3094 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3096 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3097 for math functions that have _Float<N> and _Float<N>X variants.
3098 (mathfn_built_in_2): Add support for math functions that have
3099 _Float<N> and _Float<N>X variants.
3100 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3101 (expand_builtin_mathfn_ternary): Add support for fma with
3102 _Float<N> and _Float<N>X variants.
3103 (expand_builtin): Likewise.
3104 (fold_builtin_3): Likewise.
3105 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3106 create math function _Float<N> and _Float<N>X variants as external
3108 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3109 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3110 the __builtin_ prefix and if not strict ansi, without the prefix.
3111 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3112 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3113 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3114 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3115 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3116 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3117 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3118 function signatures for fma _Float<N> and _Float<N>X variants.
3119 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3120 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3121 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3122 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3123 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3124 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3125 * gencfn-macros.c (print_case_cfn): Add support for math functions
3126 that have _Float<N> and _Float<N>X variants.
3127 (print_define_operator_list): Likewise.
3128 (fltfn_suffixes): Likewise.
3130 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3131 for math functions that have _Float<N> and _Float<N>X variants.
3132 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3133 and _Float<N>X variants.
3134 (COPYSIGN): Likewise.
3137 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3138 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3140 (integer_valued_read_call_p): Likewise.
3141 * fold-const-call.c (fold_const_call_ss): Likewise.
3142 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3143 _Float<N> and _Float<N>X variants.
3144 (fold_const_call_ssss): Add support for fma _Float<N> and
3145 _Float<N>X variants.
3146 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3147 support for copysign and fma _Float<N> and _Float<N>X variants.
3148 (backprop::process_builtin_call_use): Likewise.
3149 * tree-call-cdce.c (can_test_argument_range); Add support for
3150 sqrt _Float<N> and _Float<N>X variants.
3151 (edom_only_function): Likewise.
3152 (get_no_error_domain): Likewise.
3153 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3154 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3155 copysign _Float<N> and _Float<N>X variants.
3156 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3157 handled by machine independent code.
3158 (FMAF128): Likewise.
3159 * doc/cpp.texi (Common Predefined Macros): Document defining
3160 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3161 fma _Float<N> and _Float<N>X variants.
3163 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3166 * config/i386/i386-modes.def (CCFPU): Remove definition.
3167 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3168 (ix86_cc_modes_compatible): Ditto.
3169 (ix86_expand_carry_flag_compare): Ditto.
3170 (ix86_expand_int_movcc): Ditto.
3171 (ix86_expand_int_addcc): Ditto.
3172 (ix86_reverse_condition): Ditto.
3173 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3174 Return true/false for unordered/ordered fp comparisons.
3175 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3176 (ix86_prepare_fp_compare_args): Update for rename.
3177 (ix86_expand_fp_compare): Update for rename. Generate unordered
3178 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3179 (ix86_expand_sse_compare_and_jump): Ditto.
3180 * config/i386/predicates.md (fcmov_comparison_operator):
3181 Remove CCFPU mode handling.
3182 (ix86_comparison_operator): Ditto.
3183 (ix86_carry_flag_operator): Ditto.
3184 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3185 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3186 (*cmpu<mode>_cc_i387): Ditto.
3187 (FPCMP): Remove mode iterator.
3188 (unord): Remove mode attribute.
3189 (unord_subst): New define_subst transformation
3190 (unord): New define_subst attribute.
3192 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3193 (*cmpi<unord>xf_i387): Ditto.
3194 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3195 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3196 using unord_subst transformation.
3197 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3198 (round_saeonly): Also handle CCFP mode.
3199 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3200 Remove UNSPEC_SAHF unspec handling.
3202 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3204 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3206 2017-10-27 Jeff Law <law@redhat.com>
3208 * gimple-ssa-sprintf.c: Include domwalk.h.
3209 (class sprintf_dom_walker): New class, derived from dom_walker.
3210 (sprintf_dom_walker::before_dom_children): New function.
3211 (struct call_info): Moved into sprintf_dom_walker class
3212 (compute_formath_length, handle_gimple_call): Likewise.
3213 (sprintf_length::execute): Call the dominator walker rather
3214 than walking the statements.
3216 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3217 gimple statement locations.
3218 (check_array_bounds): Corresponding changes. Get the statement's
3219 location directly from wi->stmt.
3221 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3224 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3226 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3228 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3229 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3231 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3234 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3235 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3236 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3237 maybe_get_pool_constant.
3238 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3241 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3243 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3244 2.26 caveat. Update gas and gld versions.
3245 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3248 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3250 * cgraph.h (set_malloc_flag): Declare.
3251 * cgraph.c (set_malloc_flag_1): New function.
3252 (set_malloc_flag): Likewise.
3253 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3254 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3256 (read_ipa_call_summary): Add support for reading is_return_callee.
3257 (write_ipa_call_summary): Stream is_return_callee.
3258 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3259 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3260 ipa-prop.h, ipa-fnsummary.h.
3261 (pure_const_names): Change to static.
3262 (malloc_state_e): Define.
3263 (malloc_state_names): Define.
3264 (funct_state_d): Add field malloc_state.
3265 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3266 (check_retval_uses): New function.
3267 (malloc_candidate_p): Likewise.
3268 (analyze_function): Add support for malloc attribute.
3269 (pure_const_write_summary): Stream malloc_state.
3270 (pure_const_read_summary): Add support for reading malloc_state.
3271 (dump_malloc_lattice): New function.
3272 (propagate_malloc): New function.
3273 (warn_function_malloc): New function.
3274 (ipa_pure_const::execute): Call propagate_malloc and
3275 ipa_free_fn_summary.
3276 (pass_local_pure_const::execute): Add support for malloc attribute.
3277 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3278 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3280 2017-10-27 Martin Liska <mliska@suse.cz>
3282 PR gcov-profile/82457
3283 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3284 for fork-like functions to be properly instrumented.
3286 2017-10-27 Richard Biener <rguenther@suse.de>
3289 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3290 info when we redirected EH.
3292 2017-10-26 Michael Collison <michael.collison@arm.com>
3294 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3296 (<optab>_trunchf<GPI:mode>2: New pattern.
3297 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3298 * config/aarch64/iterators.md (wv): New mode attribute.
3299 (vf, VF): New mode attributes.
3300 (vgp, VGP): New mode attributes.
3301 (s): Update attribute with SImode and DImode prefixes.
3303 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3305 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3306 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3307 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3308 (nios2_option_overide): Initialize it. Don't allow R0-relative
3309 addressing with PIC.
3310 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3311 (nios2_symbolic_constant_p): Likewise.
3312 (nios2_legitimate_address_p): Likewise.
3313 (nios2_r0rel_section_name_p): New.
3314 (nios2_symbol_ref_in_r0rel_data_p): New.
3315 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3316 (r0rel_constant_p): New.
3317 (nios2_print_operand_address): Handle r0rel_constant_p.
3318 (nios2_cdx_narrow_form_p): Likewise.
3319 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3320 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3321 (Nios II Options): Document -mr0rel-sec.
3323 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3325 * config/nios2/nios2.c: Include xregex.h.
3326 (nios2_gprel_sec_regex): New.
3327 (nios2_option_overide): Initialize it. Don't allow GP-relative
3328 addressing with PIC.
3329 (nios2_small_section_name_p): Check for regex match.
3330 * config/nios2/nios2.opt (mgprel-sec=): New option.
3331 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3332 (Nios II Options): Document -mgprel-sec.
3334 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3336 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3338 2017-10-26 Tom de Vries <tom@codesourcery.com>
3340 PR tree-optimization/82707
3341 * gimple.c (gimple_copy): Fix unsharing of
3342 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3344 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3346 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3347 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3348 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3349 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3350 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3351 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3352 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3353 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3354 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3355 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3356 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3357 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3358 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
3359 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
3360 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
3361 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
3362 _mm512_mask_cmpunord_ps_mask): New intrinsics.
3364 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3366 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
3368 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
3369 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
3370 warning to rs6000.c. Remove the Undocumented flag, since it has
3372 (-mabi=ibmlongdouble): Likewise.
3373 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
3374 already set, set the default format for long double.
3375 (rs6000_debug_reg_global): Print whether long double is IBM or
3377 (rs6000_option_override_internal): Rework setting long double
3378 format. Only warn if the user is changing the long double default
3379 and they did not use -Wno-psabi.
3380 * doc/invoke.texi (PowerPC options): Update the documentation for
3381 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3383 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3384 Alan Hayward <alan.hayward@arm.com>
3385 David Sherwood <david.sherwood@arm.com>
3387 * rtl.h (wider_subreg_mode): New function.
3388 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
3389 rather than an unsigned int *.
3390 * ira-color.c (regno_max_ref_width): Replace with...
3391 (regno_max_ref_mode): ...this new variable.
3392 (coalesced_pseudo_reg_slot_compare): Update accordingly.
3393 Use wider_subreg_mode.
3394 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
3395 rather than an unsigned int *.
3396 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
3397 (process_alt_operands): Likewise.
3398 (invariant_p): Likewise.
3399 * lra-spills.c (assign_mem_slot): Likewise.
3400 (add_pseudo_to_slot): Likewise.
3401 * lra.c (collect_non_operand_hard_regs): Likewise.
3402 (add_regs_to_insn_regno_info): Likewise.
3403 * reload1.c (regno_max_ref_width): Replace with...
3404 (regno_max_ref_mode): ...this new variable.
3405 (reload): Update accordingly. Update call to
3406 ira_sort_regnos_for_alter_reg.
3407 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
3408 (init_eliminable_invariants): Update to use regno_max_ref_mode.
3409 (scan_paradoxical_subregs): Likewise.
3411 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3413 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
3414 (aarch64_frame): Add emit_frame_chain boolean.
3415 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3416 Move eh_return case to aarch64_layout_frame.
3417 (aarch64_layout_frame): Initialize emit_frame_chain.
3418 (aarch64_expand_prologue): Use emit_frame_chain.
3420 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3422 * config/aarch64/aarch64.c (aarch64_layout_frame):
3423 Ensure LR is always stored at the bottom of the callee-saves.
3424 Remove rarely used frame layout which saves callee-saves at top of
3425 frame, so the store of LR can be used as a valid probe in all cases.
3427 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3429 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
3430 Improve unaligned TImode/TFmode base/offset split.
3432 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3433 Alan Hayward <alan.hayward@arm.com>
3434 David Sherwood <david.sherwood@arm.com>
3436 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
3437 * combine.c (find_single_use_1): Likewise.
3438 (expand_field_assignment): Likewise.
3439 (move_deaths): Likewise.
3440 * lra-constraints.c (simplify_operand_subreg): Likewise.
3441 (curr_insn_transform): Likewise.
3442 * lra.c (collect_non_operand_hard_regs): Likewise.
3443 (add_regs_to_insn_regno_info): Likewise.
3444 * rtlanal.c (reg_referenced_p): Likewise.
3445 (covers_regno_no_parallel_p): Likewise.
3447 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3449 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
3450 Don't print any bits outside the precision of the value.
3451 * wide-int.cc (test_printing): Add some new tests.
3453 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3455 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
3456 supports -xbrace_comment option.
3457 * configure: Regenerate.
3458 * config.in: Regenerate.
3459 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
3460 (ASM_CPU_SPEC): Use it.
3462 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3464 * target.def (static_rtx_alignment): New hook.
3465 * targhooks.h (default_static_rtx_alignment): Declare.
3466 * targhooks.c (default_static_rtx_alignment): New function.
3467 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
3468 * doc/tm.texi: Regenerate.
3469 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
3470 instead of targetm.constant_alignment. Remove call to
3472 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3473 (cris_preferred_mininum_alignment): New function, split out from...
3474 (cris_constant_alignment): ...here.
3475 (cris_static_rtx_alignment): New function.
3476 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
3478 (ix86_constant_alignment): ...here.
3479 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3480 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3481 (mmix_static_rtx_alignment): New function.
3482 * config/spu/spu.c (spu_static_rtx_alignment): New function.
3483 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3485 2017-10-26 Tamar Christina <tamar.christina@arm.com>
3488 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
3489 Add flag_trapping_math and flag_fp_int_builtin_inexact.
3491 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
3493 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
3494 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
3495 mark as a sign-extending load.
3496 (local_pic_load_u): Define.
3498 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
3501 * fold-const.c (operand_equal_for_comparison_p): Also return true
3502 if ARG0 is a simple variant of ARG1 with narrower precision.
3503 (fold_ternary_loc): Always pass unstripped operands to the predicate.
3505 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
3507 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
3509 * i386.h (processor_costs): Add gather_static, gather_per_elt,
3510 scatter_static, scatter_per_elt.
3511 * x86-tune-costs.h: Add new cost entries.
3513 2017-10-25 Richard Biener <rguenther@suse.de>
3515 * tree-ssa-sccvn.h (vn_eliminate): Declare.
3516 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
3517 class pass_fre): Move to ...
3518 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
3519 class pass_fre): ... here and adjust for statistics.
3521 2017-10-25 Jakub Jelinek <jakub@redhat.com>
3524 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
3525 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
3526 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
3527 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
3530 2017-10-25 Richard Biener <rguenther@suse.de>
3532 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
3533 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
3534 eliminate_push_avail, eliminate_insert): Move inside...
3535 (class eliminate_dom_walker): ... this class in preparation
3537 (fini_eliminate): Remove by merging with ...
3538 (eliminate): ... this function. Adjust for class changes.
3539 (pass_pre::execute): Remove fini_eliminate call.
3540 (pass_fre::execute): Likewise.
3542 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3545 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
3546 (VPERMI2, VPERMI2I): New mode iterators.
3547 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
3548 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
3550 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
3551 mode iterator. Remove 3 old define_insn patterns.
3552 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
3553 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
3554 VPERMI2 mode iterator, remove the other two expanders.
3555 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
3556 to use VPERMI2 mode iterator, add another alternative for vpermi2*
3557 instructions, remove the other two patterns.
3558 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
3559 mode iterator, remove the other two patterns.
3560 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
3561 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
3562 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
3563 and adjust argument order accordingly.
3564 (ix86_expand_vec_perm): Adjust caller.
3565 (expand_vec_perm_1): Likewise.
3566 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
3567 (expand_vec_perm_vpermt2_vpshub2): ... this.
3568 (ix86_expand_vec_perm_const_1): Adjust caller.
3569 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
3572 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
3573 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
3574 (vec_shl_<mode>): Remove unused expander.
3575 (avx512bw_<shift_insn><mode>3): New define_insn.
3576 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
3577 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
3579 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
3582 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
3585 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3587 PR rtl-optimization/82396
3588 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
3589 (autopref_multipass_init): Simplify initialization.
3590 (autopref_rank_data): Simplify sort order.
3591 * gcc/sched-int.h (autopref_multipass_data_): Remove
3592 multi_mem_insn_p, min_offset and max_offset.
3594 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3597 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3598 Check special value of flag_omit_frame_pointer.
3599 (aarch64_can_eliminate): Likewise.
3600 (aarch64_override_options_after_change_1): Simplify handling of
3601 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
3603 2017-10-24 Richard Biener <rguenther@suse.de>
3605 PR tree-optimization/82697
3606 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
3607 zero for conditional load and unconditional store.
3609 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3611 * doc/install.texi: Document bootstrap-cet.
3613 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3616 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
3617 ENDBR instruction at function entrance if function is only
3620 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3623 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
3624 patterns to better describe from which operation the CF is computed.
3625 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
3626 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
3627 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
3628 is 0, use _0 suffixed expanders instead of emitting a comparison
3631 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
3633 * config/i386/i386.md(*movsf_internal, *movdf_internal):
3634 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
3636 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
3639 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
3640 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
3641 * loop-iv.c (iv_get_reaching_def): Likewise.
3642 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
3643 variable is promoted and the partition contains undefined values.
3645 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3647 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
3649 (nios2_address_cost): Define.
3650 (nios2_legitimize_address): Recognize (exp + constant) directly.
3651 (TARGET_ADDRESS_COST): Define.
3653 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3655 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
3656 (nios2_symbolic_memory_operand_p): Declare.
3657 (nios2_split_large_constant): Declare.
3658 (nios2_split_symbolic_memory_operand): Declare.
3659 * config/nios2/nios2.c: Adjust includes.
3660 (nios2_symbolic_constant_allowed): New.
3661 (nios2_symbolic_constant_p): New.
3662 (nios2_plus_symbolic_constant_p): New.
3663 (nios2_valid_addr_expr_p): Recognize addresses involving
3665 (nios2_legitimate_address_p): Likewise, also LO_SUM.
3666 (nios2_symbolic_memory_operand_p): New.
3667 (nios2_large_constant_p): New.
3668 (nios2_split_large_constant): New.
3669 (nios2_split_plus_large_constant): New.
3670 (nios2_split_symbolic_memory_operand): New.
3671 (nios2_legitimize_address): Code refactoring. Handle addresses
3672 involving symbolic constants.
3673 (nios2_emit_move_sequence): Likewise.
3674 (nios2_print_operand): Improve error output.
3675 (nios2_print_operand_address): Handle LO_SUM.
3676 (nios2_cdx_narrow_form_p): Likewise.
3677 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
3678 operands involving symbolic constants.
3679 (movhi_internal, movsi_internal): Likewise.
3680 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
3681 (extendhisi2, extendqi<mode>2): Likewise.
3683 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3685 * tree-pass.h (PROP_rtl_split_insns): Define.
3686 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
3688 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3690 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
3692 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3695 * target.def (const_not_ok_for_debug_p): Default to
3696 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
3697 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
3698 * targhooks.c (default_const_not_ok_for_debug_p): New function.
3699 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
3700 which targetm.const_not_ok_for_debug_p returned true.
3701 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
3703 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
3705 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
3706 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
3707 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
3709 (ix86_const_not_ok_for_debug_p): New function.
3710 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
3711 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
3713 2017-10-23 David Malcolm <dmalcolm@redhat.com>
3716 * system.h: Conditionally include "unique-ptr.h" if
3717 INCLUDE_UNIQUE_PTR is defined.
3718 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
3719 of defining INCLUDE_UNIQUE_PTR before including "system.h".
3721 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
3723 * config/rl78/rl78.md: New define_expand "subdi3".
3725 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
3728 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
3729 DF_REF_INSN if DF_REF_INSN_INFO is false.
3731 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
3733 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
3734 xmm_move instead of sse_move.
3735 (sse_store_index): New function.
3736 (ix86_register_move_cost): Be more sensible about mismatch stall;
3737 model AVX moves correctly; make difference between sse->integer and
3739 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
3740 moves; make difference between SSE and AVX.
3741 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
3742 and zmm_move. Increase size of sse load and store tables;
3743 add unaligned load and store tables; add ssemmx_to_integer.
3744 * x86-tune-costs.h: Update all entries according to real
3745 move latencies from Agner Fog's manual and chip documentation.
3747 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3750 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
3751 * config/i386/constraints.md (Wf): New constraint.
3752 * config/i386/i386.md (UNSPEC_SBB): New unspec.
3753 (cmp<dwi>_doubleword): Removed.
3754 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
3755 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
3756 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
3757 expand with cmp<dwi>_doubleword. For LTU and GEU use
3758 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
3760 * common.opt (gcolumn-info): Enable by default.
3761 * doc/invoke.texi (gcolumn-info): Document new default.
3763 2017-10-23 Richard Biener <rguenther@suse.de>
3765 PR tree-optimization/82672
3766 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
3767 Fold the stmt if we propagated into it.
3769 2017-10-23 Richard Biener <rguenther@suse.de>
3771 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
3772 (bitmap_remove_expr_from_set): ... this. All callers call this
3773 for non-constant values.
3774 (bitmap_set_subtract): Rename to...
3775 (bitmap_set_subtract_expressions): ... this. Adjust and
3777 (bitmap_set_contains_value): Remove superfluous check.
3778 (bitmap_set_replace_value): Inline into single caller ...
3779 (bitmap_value_replace_in_set): ... here and simplify.
3780 (dependent_clean): Merge into ...
3781 (clean): ... this using an overload. Adjust.
3782 (prune_clobbered_mems): Adjust.
3783 (compute_antic_aux): Likewise.
3784 (compute_partial_antic_aux): Likewise.
3786 2017-10-23 Richard Biener <rguenther@suse.de>
3788 PR tree-optimization/82129
3790 2017-08-01 Richard Biener <rguenther@suse.de>
3792 PR tree-optimization/81181
3793 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3794 (compute_antic): ... end of iteration here.
3796 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3798 * target.def (starting_frame_offset): New hook.
3799 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
3800 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
3801 * doc/tm.texi.in: Regenerate.
3802 * hooks.h (hook_hwi_void_0): Declare.
3803 * hooks.c (hook_hwi_void_0): New function.
3804 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
3805 STARTING_FRAME_OFFSET.
3806 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
3807 * reload1.c (reload): Likewise.
3808 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
3809 instead of STARTING_FRAME_OFFSET.
3810 * function.c (try_fit_stack_local): Likewise.
3811 (assign_stack_local_1): Likewise
3812 (instantiate_virtual_regs): Likewise.
3813 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
3814 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
3815 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
3816 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
3817 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
3818 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
3819 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
3820 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
3821 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
3822 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
3823 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
3824 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
3825 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
3826 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
3827 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
3828 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
3829 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
3830 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
3831 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
3832 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
3833 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
3834 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
3835 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
3836 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
3837 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
3838 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
3839 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
3840 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
3841 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
3842 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
3843 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
3844 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
3845 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
3846 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
3847 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
3848 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
3849 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
3850 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
3851 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
3852 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
3853 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
3854 * config/avr/avr.c (avr_starting_frame_offset): Make static and
3855 return a HOST_WIDE_INT.
3856 (avr_builtin_setjmp_frame_value): Use it instead of
3857 STARTING_FRAME_OFFSET.
3858 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3859 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
3860 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
3862 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3863 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
3864 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
3865 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3866 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
3867 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
3868 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3869 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
3870 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
3871 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3872 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
3873 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
3875 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3876 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
3877 * config/mips/mips.c (mips_compute_frame_info): Refer to
3878 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3879 (mips_starting_frame_offset): New function.
3880 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3881 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
3882 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
3883 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
3884 and return a HOST_WIDE_INT.
3885 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3886 (mmix_initial_elimination_offset): Refer to
3887 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3888 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
3889 * config/pa/pa.c (pa_starting_frame_offset): New function.
3890 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
3891 (pa_expand_prologue): Likewise.
3892 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3893 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
3894 !FRAME_GROWS_DOWNWARD handling to...
3895 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3896 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
3897 !FRAME_GROWS_DOWNWARD handling to...
3898 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3899 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
3900 !FRAME_GROWS_DOWNWARD handling to...
3901 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3902 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
3904 (rs6000_starting_frame_offset): New function.
3905 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
3906 !FRAME_GROWS_DOWNWARD handling to...
3907 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3908 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
3909 !FRAME_GROWS_DOWNWARD handling to...
3910 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3911 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
3912 !FRAME_GROWS_DOWNWARD handling to...
3913 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3914 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
3915 (rs6000_starting_frame_offset): New function.
3916 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
3917 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
3918 * config/vax/vax.c (vax_starting_frame_offset): New function.
3919 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
3920 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3921 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
3922 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
3923 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3924 * system.h (STARTING_FRAME_OFFSET): Poison.
3926 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3928 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3929 SCALAR_TYPE_MODE instead of TYPE_MODE.
3931 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3932 Alan Hayward <alan.hayward@arm.com>
3933 David Sherwood <david.sherwood@arm.com>
3935 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
3937 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3938 Alan Hayward <alan.hayward@arm.com>
3939 David Sherwood <david.sherwood@arm.com>
3941 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
3943 2017-10-23 Richard Biener <rguenther@suse.de>
3945 PR tree-optimization/82129
3946 * tree-ssa-pre.c (bitmap_set_and): Remove.
3947 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
3948 canonicalizing expressions in the set to those with lowest
3949 ID rather than taking that from the first edge.
3951 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3953 * combine.c (rtx_equal_for_field_assignment_p): Use
3954 byte_lowpart_offset.
3956 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3957 Alan Hayward <alan.hayward@arm.com>
3958 David Sherwood <david.sherwood@arm.com>
3960 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
3961 to a SUBREG_PROMOTED_VAR.
3963 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3964 Alan Hayward <alan.hayward@arm.com>
3965 David Sherwood <david.sherwood@arm.com>
3967 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
3968 (expand_debug_source_expr): Likewise.
3969 * combine.c (combine_simplify_rtx): Likewise.
3970 * cse.c (fold_rtx): Likewise.
3971 * optabs.c (expand_float): Likewise.
3972 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3973 (simplify_binary_operation_1): Likewise.
3975 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3976 Alan Hayward <alan.hayward@arm.com>
3977 David Sherwood <david.sherwood@arm.com>
3979 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
3980 (record_promoted_value): Likewise.
3981 * expr.c (expand_expr_real_2): Likewise.
3982 * ree.c (update_reg_equal_equiv_notes): Likewise.
3983 (combine_set_extension): Likewise.
3984 * rtlanal.c (low_bitmask_len): Likewise.
3985 * simplify-rtx.c (neg_const_int): Likewise.
3986 (simplify_binary_operation_1): Likewise.
3988 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3989 Alan Hayward <alan.hayward@arm.com>
3990 David Sherwood <david.sherwood@arm.com>
3992 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
3993 * regcprop.c (maybe_mode_change): Likewise.
3994 * reload1.c (alter_reg): Likewise.
3996 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3998 * inchash.h (inchash::hash::add_wide_int): New function.
3999 * lto-streamer-out.c (hash_tree): Use it.
4001 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4003 * inchash.h (inchash::hash::add_wide_int): Rename to...
4004 (inchash::hash::add_hwi): ...this.
4005 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
4006 (polymorphic_call_target_hasher::hash): Likewise.
4007 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
4008 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
4009 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
4010 * lto-streamer-out.c (hash_tree): Likewise.
4011 * optc-save-gen.awk: Likewise.
4012 * tree.c (add_expr): Likewise.
4014 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4017 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4018 for ordered inequality comparisons even with TARGET_IEEE_FP.
4020 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4023 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4024 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4025 Expand with cmp<dwi>_doubleword.
4027 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4029 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4030 List CET intrinsics.
4031 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4032 specific to -fcf-protection option.
4034 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4036 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4037 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4038 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4039 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4040 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4041 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4042 (extra_objs): Add cet.o for Linux/x86 targets.
4043 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4044 * config/i386/cet.c: New file.
4045 * config/i386/cetintrin.h: Likewise.
4046 * config/i386/t-cet: Likewise.
4047 * config/i386/cpuid.h (bit_SHSTK): New.
4048 (bit_IBT): Likewise.
4049 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4050 pass IBT and SHSTK bits.
4051 * config/i386/i386-builtin-types.def
4052 (VOID_FTYPE_UNSIGNED_PVOID): New.
4053 (VOID_FTYPE_UINT64_PVOID): Likewise.
4054 * config/i386/i386-builtin.def: Add CET intrinsics.
4055 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4056 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4057 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4058 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4060 * config/i386/i386.c (rest_of_insert_endbranch): New.
4061 (pass_data_insert_endbranch): Likewise.
4062 (pass_insert_endbranch): Likewise.
4063 (make_pass_insert_endbranch): Likewise.
4064 (ix86_notrack_prefixed_insn_p): Likewise.
4065 (ix86_target_string): Add -mibt, -mshstk flags.
4066 (ix86_option_override_internal): Add flag_cf_protection
4068 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4069 (ix86_print_operand): Add 'notrack' prefix output.
4070 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4071 (ix86_expand_builtin): Expand CET intrinsics.
4072 (x86_output_mi_thunk): Add 'endbranch' instruction.
4073 * config/i386/i386.h (TARGET_IBT): New.
4074 (TARGET_IBT_P): Likewise.
4075 (TARGET_SHSTK): Likewise.
4076 (TARGET_SHSTK_P): Likewise.
4077 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4078 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4079 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4080 (builtin_setjmp_setup): New pattern.
4081 (builtin_longjmp): Likewise.
4082 (rdssp<mode>): Likewise.
4083 (incssp<mode>): Likewise.
4084 (saveprevssp): Likewise.
4085 (rstorssp): Likewise.
4086 (wrss<mode>): Likewise.
4087 (wruss<mode>): Likewise.
4088 (setssbsy): Likewise.
4089 (clrssbsy): Likewise.
4090 (nop_endbr): Likewise.
4091 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4093 * config/i386/immintrin.h: Include <cetintrin.h>.
4094 * config/i386/linux-common.h
4095 (file_end_indicate_exec_stack_and_cet): New prototype.
4096 (TARGET_ASM_FILE_END): New.
4098 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4100 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4101 latencies instead of having separate table; make difference between
4102 integer and float costs.
4103 * i386.h (processor_costs): Remove scalar_stmt_cost,
4104 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4105 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4107 * x86-tune-costs.h: Remove entries which has been removed in
4108 procesor_costs from all tables; make cond_taken_branch_cost
4109 and cond_not_taken_branch_cost COST_N_INSNS based.
4111 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4113 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4115 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4117 * config/i386/i386.md (isa): Remove fma_avx512f.
4118 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4119 <avx512>_fmadd_<mode>_mask3<round_name>,
4120 <avx512>_fmsub_<mode>_mask<round_name>,
4121 <avx512>_fmsub_<mode>_mask3<round_name>,
4122 <avx512>_fnmadd_<mode>_mask<round_name>,
4123 <avx512>_fnmadd_<mode>_mask3<round_name>,
4124 <avx512>_fnmsub_<mode>_mask<round_name>,
4125 <avx512>_fnmsub_<mode>_mask3<round_name>,
4126 <avx512>_fmaddsub_<mode>_mask<round_name>,
4127 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4128 <avx512>_fmsubadd_<mode>_mask<round_name>,
4129 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4130 (*vec_widen_umult_even_v16si<mask_name>,
4131 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4132 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4134 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4136 * extend.texi: Add 'nocf_check' documentation.
4137 * gimple.texi: Add second parameter to
4138 gimple_build_call_from_tree.
4139 * invoke.texi: Add -fcf-protection documentation.
4140 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4142 2017-10-20 Richard Biener <rguenther@suse.de>
4144 PR tree-optimization/82473
4145 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4146 the largest input type.
4148 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4150 * c-attribs.c (handle_nocf_check_attribute): New function.
4151 (c_common_attribute_table): Add 'nocf_check' handling.
4152 * gimple-parser.c: Add second argument NULL to
4153 gimple_build_call_from_tree.
4154 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4155 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4157 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4158 * common.opt: Add fcf-protection flag.
4159 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4160 * flag-types.h: Add enum cf_protection_level.
4161 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4162 Add 'nocf_check' attribute propagation to gimple call.
4163 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4164 (gimple_build_call_from_tree): Update prototype.
4165 (gimple_call_nocf_check_p): New function.
4166 (gimple_call_set_nocf_check): Likewise.
4167 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4168 * ipa-icf.c: Add nocf_check attribute in statement hash.
4169 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4170 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4171 * toplev.c (process_options): Add flag_cf_protection handling.
4173 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4175 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4177 2017-10-20 Richard Biener <rguenther@suse.de>
4179 PR tree-optimization/82603
4180 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4181 remove false predicated stores.
4183 2017-10-20 Richard Biener <rguenther@suse.de>
4185 * graphite-isl-ast-to-gimple.c
4186 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4187 Remove return value and simplify, dump copied stmt after lhs
4189 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4190 Reduce dump verbosity.
4191 (gsi_insert_earliest): Likewise.
4192 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4193 * graphite.c (print_global_statistics): Adjust dumping.
4194 (print_graphite_scop_statistics): Likewise.
4195 (print_graphite_statistics): Do not dump loops here.
4196 (graphite_transform_loops): But here.
4198 2017-10-20 Nicolas Roche <roche@adacore.com>
4200 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4201 * configure: Regenerate.
4203 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4206 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4207 functions when optimizing replace GIMPLE_RETURN stmts with
4208 calls to __builtin_unreachable ().
4211 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4212 for -fsanitize=thread link of executables.
4213 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4214 link of executables.
4217 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4219 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4220 define_insns for logical vector shifts to use VI248_AVX512BW
4221 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4222 condition, useless isa and prefix attributes. Change the first
4223 2 of these define_insns to ...
4224 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4225 define_insn for avx512vl.
4226 (<shift_insn><mode>3): ... and this, new define_insn without
4227 masking for non-avx512vl.
4230 * config/i386/sse.md (*andnot<mode>3,
4231 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4232 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4233 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4234 not applied use empty suffix even for TARGET_AVX512VL.
4235 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4236 is applied, supply evex,evex or evex,evex,evex instead of just
4239 2017-10-20 Julia Koval <julia.koval@intel.com>
4241 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4242 (OPTION_MASK_ISA_GFNI_UNSET): New.
4243 (ix86_handle_option): Handle OPT_mgfni.
4244 * config/i386/cpuid.h (bit_GFNI): New.
4245 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4246 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4247 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4248 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4249 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4250 * config/i386/i386.opt: Add mgfni.
4252 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4254 * config/msp430/msp430.c (msp430_option_override): Disable
4255 -fdelete-null-pointer-checks.
4256 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4258 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4260 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4261 of x87 and SSE instructions.
4263 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4265 * asan.c (create_cond_insert_point): Do not update edge count.
4266 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4267 (afdo_propagate_circuit): Likewise.
4268 (afdo_calculate_branch_prob): Likewise.
4269 (afdo_annotate_cfg): Likewise.
4270 * basic-block.h (struct edge_def): Remove count.
4271 (edge_def::count): New accessor.
4272 * bb-reorder.c (rotate_loop): Update.
4273 (find_traces_1_round): Update.
4274 (connect_traces): Update.
4275 (sanitize_hot_paths): Update.
4276 * cfg.c (unchecked_make_edge): Update.
4277 (make_single_succ_edge): Update.
4278 (check_bb_profile): Update.
4279 (dump_edge_info): Update.
4280 (update_bb_profile_for_threading): Update.
4281 (scale_bbs_frequencies_int): Update.
4282 (scale_bbs_frequencies_gcov_type): Update.
4283 (scale_bbs_frequencies_profile_count): Update.
4284 (scale_bbs_frequencies): Update.
4285 * cfganal.c (connect_infinite_loops_to_exit): Update.
4286 * cfgbuild.c (compute_outgoing_frequencies): Update.
4287 (find_many_sub_basic_blocks): Update.
4288 * cfgcleanup.c (try_forward_edges): Update.
4289 (try_crossjump_to_edge): Update
4290 * cfgexpand.c (expand_gimple_cond): Update
4291 (expand_gimple_tailcall): Update
4292 (construct_exit_block): Update
4293 * cfghooks.c (verify_flow_info): Update
4294 (redirect_edge_succ_nodup): Update
4295 (split_edge): Update
4296 (make_forwarder_block): Update
4297 (duplicate_block): Update
4298 (account_profile_record): Update
4299 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4300 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4301 * cfgloopmanip.c (scale_loop_profile): Update.
4303 (lv_adjust_loop_entry_edge): Update.
4304 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4305 (force_nonfallthru_and_redirect): Update.
4306 (purge_dead_edges): Update.
4307 (rtl_flow_call_edges_add): Update.
4308 * cgraphunit.c (init_lowered_empty_function): Update.
4309 (cgraph_node::expand_thunk): Update.
4310 * gimple-pretty-print.c (dump_probability): Update.
4311 (dump_edge_probability): Update.
4312 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4313 * haifa-sched.c (sched_create_recovery_edges): Update.
4314 * hsa-gen.c (convert_switch_statements): Update.
4315 * ifcvt.c (dead_or_predicable): Update.
4316 * ipa-inline-transform.c (inline_transform): Update.
4317 * ipa-split.c (split_function): Update.
4318 * ipa-utils.c (ipa_merge_profiles): Update.
4319 * loop-doloop.c (add_test): Update.
4320 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4321 * lto-streamer-in.c (input_cfg): Update.
4322 (input_function): Update.
4323 * lto-streamer-out.c (output_cfg): Update.
4324 * modulo-sched.c (sms_schedule): Update.
4325 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4326 * predict.c (maybe_hot_edge_p): Update.
4327 (unlikely_executed_edge_p): Update.
4328 (probably_never_executed_edge_p): Update.
4329 (dump_prediction): Update.
4330 (drop_profile): Update.
4331 (propagate_unlikely_bbs_forward): Update.
4332 (determine_unlikely_bbs): Update.
4333 (force_edge_cold): Update.
4334 * profile.c (compute_branch_probabilities): Update.
4335 * reg-stack.c (better_edge): Update.
4336 * shrink-wrap.c (handle_simple_exit): Update.
4337 * tracer.c (better_p): Update.
4338 * trans-mem.c (expand_transaction): Update.
4339 (split_bb_make_tm_edge): Update.
4340 * tree-call-cdce.c: Update.
4341 * tree-cfg.c (gimple_find_sub_bbs): Update.
4342 (gimple_split_edge): Update.
4343 (gimple_duplicate_sese_region): Update.
4344 (gimple_duplicate_sese_tail): Update.
4345 (gimple_flow_call_edges_add): Update.
4346 (insert_cond_bb): Update.
4347 (execute_fixup_cfg): Update.
4348 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4349 * tree-complex.c (expand_complex_div_wide): Update.
4350 * tree-eh.c (lower_resx): Update.
4351 (unsplit_eh): Update.
4352 (cleanup_empty_eh_move_lp): Update.
4353 * tree-inline.c (copy_edges_for_bb): Update.
4354 (freqs_to_counts): Update.
4355 (copy_cfg_body): Update.
4356 * tree-ssa-dce.c (remove_dead_stmt): Update.
4357 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4358 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4359 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4360 (unloop_loops): Update.
4361 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4362 * tree-ssa-loop-split.c (connect_loops): Update.
4363 (split_loop): Update.
4364 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4365 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4366 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4367 * tree-ssa-reassoc.c (branch_fixup): Update.
4368 * tree-ssa-tail-merge.c (replace_block_by): Update.
4369 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4370 (compute_path_counts): Update.
4371 (update_profile): Update.
4372 (recompute_probabilities): Update.
4373 (update_joiner_offpath_counts): Update.
4374 (estimated_freqs_path): Update.
4375 (freqs_to_counts_path): Update.
4376 (clear_counts_path): Update.
4377 (ssa_fix_duplicate_block_edges): Update.
4378 (duplicate_thread_path): Update.
4379 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4380 (case_bit_test_cmp): Update.
4381 (collect_switch_conv_info): Update.
4382 (gen_inbound_check): Update.
4383 (do_jump_if_equal): Update.
4384 (emit_cmp_and_jump_insns): Update.
4385 * tree-tailcall.c (decrease_profile): Update.
4386 (eliminate_tail_call): Update.
4387 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4388 (vect_do_peeling): Update.
4389 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4390 * ubsan.c (ubsan_expand_null_ifn): Update.
4391 (ubsan_expand_ptr_ifn): Update.
4392 * value-prof.c (gimple_divmod_fixed_value): Update.
4393 (gimple_mod_pow2): Update.
4394 (gimple_mod_subtract): Update.
4395 (gimple_ic): Update.
4396 (gimple_stringop_fixed_value): Update.
4398 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4401 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
4403 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
4405 PR rtl-optimization/82395
4406 * ira-color.c (allocno_priority_compare_func): Fix comparison step
4407 based on non_spilled_static_chain_regno_p.
4409 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4411 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
4412 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
4413 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
4415 2017-10-19 Martin Sebor <msebor@redhat.com>
4417 PR tree-optimization/82596
4418 * tree.c (array_at_struct_end_p): Handle STRING_CST.
4420 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4422 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
4423 (get_mem_refs_of_builtin_call): Likewise.
4424 * builtins.c (expand_builtin_apply): Adjust call to
4425 allocate_dynamic_stack_space.
4426 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
4427 the third argument to allocate_dynamic_stack_space, otherwise -1.
4428 (expand_builtin): Deal with all alloca variants.
4429 (is_inexpensive_builtin): Likewise.
4430 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
4431 * calls.c (special_function_p): Deal with all alloca variants.
4432 (initialize_argument_information): Adjust call to
4433 allocate_dynamic_stack_space.
4434 (expand_call): Likewise.
4435 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
4436 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
4437 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
4438 use it for the stack usage computation.
4439 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
4440 * function.c (gimplify_parameters): Call build_alloca_call_expr.
4441 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
4442 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
4443 (in_loop_p): Remove first argument and useless check.
4444 (pass_walloca::execute): Remove useless test and adjust call to above.
4445 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
4446 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
4447 (gimplify_call_expr): Deal with all alloca variants.
4448 * hsa-gen.c (gen_hsa_alloca): Likewise.
4449 (gen_hsa_insns_for_call): Likewise.
4450 * ipa-pure-const.c (special_builtin_state): Likewise.
4451 * tree-chkp.c (chkp_build_returned_bound): Likewise.
4452 * tree-object-size.c (alloc_object_size): Likewise.
4453 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4454 (call_may_clobber_ref_p_1): Likewise.
4455 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
4456 (ccp_fold_stmt): Likewise.
4457 (optimize_stack_restore): Likewise.
4458 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
4459 (mark_all_reaching_defs_necessary_1): Likewise.
4460 (propagate_necessity): Likewise.
4461 (eliminate_unnecessary_stmts): Likewise.
4462 * tree.c (build_common_builtin_nodes): Build
4463 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
4464 (build_alloca_call_expr): New function.
4465 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
4466 (CASE_BUILT_IN_ALLOCA): Likewise.
4467 (build_alloca_call_expr): Declare.
4468 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
4470 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4473 * dwarf2out.c (new_die_raw): New static inline function.
4474 (new_die): Use it to create the DIE.
4475 (add_AT_external_die_ref): Likewise.
4476 (clone_die): Likewise.
4477 (clone_as_declaration): Likewise.
4478 (dwarf2out_vms_debug_main_pointer): Likewise.
4479 (base_type_die): Likewise. Remove early return for corner cases.
4480 Do not call add_pubtype on the DIE here.
4481 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
4482 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
4483 typedefs for base types with DW_AT_endianity. Make sure a DIE with
4484 native order exists for base types, attach the DIE manually and call
4485 add_pubtype on it. Do not equate a reverse order DIE to the type.
4487 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
4489 * config/arm/arm.c (align_ok_ldrd_strd): New function.
4490 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
4492 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
4494 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4496 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
4497 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
4498 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
4499 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
4500 * opts.c (sanitizer_opts): Add builtin.
4501 * ubsan.c (instrument_builtin): New function.
4502 (pass_ubsan::execute): Call it.
4503 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
4504 * doc/invoke.texi: Document -fsanitize=builtin.
4506 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
4507 builtins, store max (log2 (align), 0) into uchar field instead of
4508 align into uptr field.
4509 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
4510 store uchar 0 field instead of uptr 0 field.
4511 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
4512 instead of passing one address of struct with 2 locations pass
4513 two addresses of structs with 1 location each.
4514 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
4515 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
4516 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
4517 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
4518 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
4519 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
4520 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
4521 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
4523 2017-10-19 Martin Liska <mliska@suse.cz>
4526 * file-find.c (remove_prefix): Remove.
4527 * file-find.h (remove_prefix): Likewise.
4528 * gcc-ar.c: Remove smartness of lookup.
4530 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
4532 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
4533 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
4534 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
4536 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4539 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
4540 (setcc + and to xor + setcc): New peephole2.
4542 2017-10-19 Tom de Vries <tom@codesourcery.com>
4544 * doc/sourcebuild.texi (Test Directives, Variants of
4545 dg-require-support): Add dg-require-stack-size.
4547 2017-10-19 Martin Liska <mliska@suse.cz>
4550 * gimplify.c (gimplify_decl_expr): Do not instrument variables
4551 that have a large alignment.
4552 (gimplify_target_expr): Likewise.
4554 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
4556 PR rtl-optimization/82602
4557 * ira.c (rtx_moveable_p): Return false for volatile asm.
4559 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
4562 * config/i386/i386-modes.def (CCGZ): New CC mode.
4563 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
4564 * config/i386/predicates.md (ix86_comparison_operator):
4566 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4567 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
4568 with double-word subtraction.
4569 (put_condition_code): Handle CCGZmode.
4571 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
4573 * wide-int.cc (debug (const wide_int &)): New.
4574 (debug (const wide_int *)): New.
4575 (debug (const widest_int &)): New.
4576 (debug (const widest_int *)): New.
4578 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
4581 * lra-constraints.c (curr_insn_transform): Use non-input operand
4582 instead of output one for matched reload.
4584 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4586 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
4587 (tree-ssa-loop-ivopts.h): New header file.
4588 (struct builtin_info): New fields.
4589 (classify_builtin_1): Compute and record base and offset parts for
4590 memset builtin partition by calling strip_offset.
4591 (offset_cmp, fuse_memset_builtins): New functions.
4592 (finalize_partitions): Fuse adjacent memset partitions by calling
4594 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
4595 Expose the interface.
4596 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
4598 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4600 PR tree-optimization/82574
4601 * tree-loop-distribution.c (find_single_drs): New parameter. Check
4602 that data reference must be executed exactly once per iteration
4603 against the outermost loop in nest.
4604 (classify_partition): Update call to above function.
4606 2017-10-18 Richard Biener <rguenther@suse.de>
4608 PR tree-optimization/82591
4609 * graphite.c (graphite_transform_loops): Move code gen message
4611 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4612 Here. Handle scop_to_isl_ast failing.
4613 (scop_to_isl_ast): Limit the number of ISL operations.
4615 2017-10-18 Richard Biener <rguenther@suse.de>
4617 * graphite-isl-ast-to-gimple.c
4618 (translate_isl_ast_to_gimple::set_rename): Simplify.
4619 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
4620 (graphite_copy_stmts_from_block): ... here.
4621 (copy_bb_and_scalar_dependences): Simplify.
4622 (add_parameters_to_ivs_params): Canonicalize.
4623 (generate_entry_out_of_ssa_copies): Simplify.
4624 * graphite-sese-to-poly.c (extract_affine_name): Simplify
4625 by passing in ISL dimension.
4626 (parameter_index_in_region_1): Rename to ...
4627 (parameter_index_in_region): ... this.
4628 (extract_affine): Adjust assert, pass down parameter index.
4629 (add_param_constraints): Use range-info when available.
4630 (build_scop_context): Adjust.
4631 * sese.c (new_sese_info): Adjust.
4632 (free_sese_info): Likewise.
4633 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
4634 Remove unused typedefs.
4635 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
4637 2017-10-18 Martin Liska <mliska@suse.cz>
4639 * combine.c (simplify_compare_const): Add gcc_fallthrough.
4641 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4643 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
4644 (s390_sched_init): Do not reset s390_sched_state if we entered the
4645 current basic block via a fallthru edge and all others are unlikely.
4647 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4649 * config/s390/s390.c (NUM_SIDES): New variable.
4650 (LONGRUNNING_THRESHOLD): New variable.
4651 (LATENCY_FACTOR): New variable.
4652 (s390_sched_score): Decrease score for long-running instructions on
4654 (s390_sched_variable_issue): Perform bookkeeping for long-running
4657 2017-10-18 Richard Biener <rguenther@suse.de>
4659 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
4660 Simplify with removal of the parameter rename map.
4661 (set_rename): Likewise.
4662 (should_copy_to_new_region): Likewise.
4663 (graphite_copy_stmts_from_block): Likewise.
4664 (copy_bb_and_scalar_dependences): Remove initialization of
4665 unused copied_bb_map.
4667 (copy_internal_parameters): Likewise.
4668 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
4669 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
4670 Use INTEGRAL_TYPE_P.
4671 (parameter_index_in_region_1): Rename to ...
4672 (assign_parameter_index_in_region): ... this. Assert we have
4673 a parameter we handle.
4674 (scan_tree_for_params): Adjust.
4675 * sese.h (parameter_rename_map_t): Remove.
4676 (struct sese_info_t): Remove unused parameter_rename_map and
4677 copied_bb_map members.
4678 * sese.c (new_sese_info): Adjust.
4679 (free_sese_info): Likewise.
4681 2017-10-18 Martin Liska <mliska@suse.cz>
4684 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
4685 on an abnormal edge.
4687 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
4689 * doc/invoke.texi (ffunction-sections and fdata-sections):
4692 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4694 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
4695 the use statement can throw internally.
4697 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4699 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
4700 any RTX present on the RHS of a SET.
4701 * compare-elim.c (try_eliminate_compare): Restore comment.
4703 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4705 * langhooks.h (struct lang_hooks): Document that tree_size langhook
4706 may be also called on tcc_type nodes.
4707 * langhooks.c (lhd_tree_size): Likewise.
4709 2017-10-17 David Malcolm <dmalcolm@redhat.com>
4711 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
4712 format_warning_at_substring.
4713 (maybe_warn): Convert source_range * param to a location_t. Pass
4714 UNKNOWN_LOCATION rather than NULL to fmtwarn.
4715 (format_directive): Remove code to extract source_ranges and
4716 source_range * in favor of just a location_t.
4717 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
4719 * substring-locations.c (format_warning_va): Convert
4720 source_range * param to a location_t.
4721 (format_warning_at_substring): Likewise.
4722 * substring-locations.h (format_warning_va): Likewise.
4723 (format_warning_at_substring): Likewise.
4725 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
4727 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
4729 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
4730 and scatter/gather ops.
4732 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
4733 vec_gather_load and vec_scatter_store.
4734 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
4735 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
4736 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
4737 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
4738 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
4739 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
4741 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
4743 * reg-stack.c (compare_for_stack_reg): Add bool argument.
4744 Detect FTST instruction and handle its register pops. Only pop
4745 second operand if can_pop_second_op is true.
4746 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
4747 set can_pop_second_op to false in the compare_for_stack_reg call.
4749 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
4750 output_fp_compare for stack register operands.
4751 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
4752 instructions here. Do not emit stack register pops here. Assert
4753 that FCOMPP pops next to top stack register. Rewrite function.
4755 2017-10-17 Nathan Sidwell <nathan@acm.org>
4758 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
4759 use DECL_ASSEMBLER_NAME_RAW.
4762 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
4765 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
4766 Wilco Dijkstra <wilco.dijkstra@arm.com>
4768 * builtins.c (expand_builtin_update_setjmp_buf): Add a
4769 converstion to Pmode from the buf_addr.
4771 2017-10-17 Richard Biener <rguenther@suse.de>
4773 * graphite-dependences.c (scop_get_reads_and_writes): Change
4774 output parameters to references.
4776 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
4778 PR 71026/tree-optimization
4779 * fold-const.c (distribute_real_division): Removed.
4780 (fold_binary_loc): Remove calls to distribute_real_divison.
4782 2017-10-17 Richard Biener <rguenther@suse.de>
4784 * graphite-scop-detection.c
4785 (scop_detection::stmt_has_simple_data_refs_p): Always use
4786 the full nest as region.
4787 (try_generate_gimple_bb): Likewise.
4788 * sese.c (scalar_evolution_in_region): Simplify now that
4789 SCEV can handle instantiation in regions.
4790 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
4791 in the non-loop part of a function if requested.
4793 2017-10-17 Richard Biener <rguenther@suse.de>
4795 PR tree-optimization/82563
4796 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
4798 (graphite_regenerate_ast_isl): Call it.
4799 * graphite-scop-detection.c (build_scops): Remove entry edge split.
4801 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4803 PR tree-optimization/82549
4804 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
4805 Formatting fixes. Instead of calling make_bit_field_ref with negative
4808 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
4810 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
4811 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
4812 _mm_maskz_reduce_ss): New.
4813 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
4814 __builtin_ia32_reducess_mask): Ditto..
4815 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
4816 * config/i386/sse.md (reduces<mode>): Renamed to ...
4817 (reduces<mode><mask_scalar_name>): ... this.
4818 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
4820 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
4821 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
4823 2017-10-16 David Malcolm <dmalcolm@redhat.com>
4825 * Makefile.in (OBJS): Add unique-ptr-tests.o.
4826 * selftest-run-tests.c (selftest::run_tests): Call
4827 selftest::unique_ptr_tests_cc_tests.
4828 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
4829 * unique-ptr-tests.cc: New file.
4831 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
4834 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4836 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4837 (make_hard_regno_born, make_hard_regno_dead): Update
4838 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
4840 2017-10-16 Jeff Law <law@redhat.com>
4842 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4844 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
4846 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
4848 2017-10-16 Olivier Hainque <hainque@adacore.com>
4850 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
4851 with_cpu if we were configured for an e500v2 target cpu name.
4853 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
4855 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
4856 * doc/invoke.texi: Document +nodsp as a valid extension for
4859 2017-10-16 Martin Liska <mliska@suse.cz>
4861 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
4862 (test_set_range): Likewise.
4863 (test_range_functions): Rename to ...
4864 (test_bit_in_range): ... this.
4865 (sbitmap_c_tests): Add new test.
4867 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4869 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
4871 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
4872 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
4874 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4876 * config/aarch64/aarch64-builtins.c
4877 (aarch64_types_quadopu_lane_qualifiers): New.
4878 (TYPES_QUADOPU_LANE): New.
4879 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
4880 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
4881 (aarch64_<sur>dot_laneq<vsi2qi>): New.
4882 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
4883 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
4884 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
4885 (Vdottype, DOTPROD): New.
4886 (sur): Add SDOT and UDOT.
4888 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4890 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
4891 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
4892 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
4894 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
4895 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
4896 Enable TARGET_DOTPROD.
4897 (cortex-a75.cortex-a55): Likewise.
4898 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
4900 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4902 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
4903 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
4905 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
4907 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
4908 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
4909 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
4910 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
4911 * config/arm/types.md (neon_dot, neon_dot_q): New.
4912 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
4914 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4916 * config/arm/arm.h (TARGET_DOTPROD): New.
4917 * config/arm/arm.c (arm_arch_dotprod): New.
4918 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
4919 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
4920 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
4921 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
4922 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
4923 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
4924 * doc/invoke.texi (armv8.2-a): Document dotprod
4926 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4928 * i386.c (ix86_vec_cost): New function.
4929 (ix86_rtx_costs): Handle vector operations better.
4930 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
4931 * x86-tune-costs.h: Add new costs to all tables.
4933 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4935 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
4937 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
4938 divsd, sqrtss and sqrtsd
4939 * x86-tune-costs.h: Add new entries to all costs.
4940 (znver1_cost): Fix to match real instruction latencies.
4942 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4943 Michael Collison <michael.collison@arm.com>
4945 * compare-elim.c: Include emit-rtl.h.
4946 (can_merge_compare_into_arith): New function.
4947 (try_validate_parallel): Likewise.
4948 (try_merge_compare): Likewise.
4949 (try_eliminate_compare): Call the above when no previous clobber
4951 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
4954 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4958 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
4959 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
4963 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
4964 to be any operand_equal_p operands. For & (B - 1) require
4965 B to be power of 2. Recognize
4966 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
4968 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4971 * optabs.c (expand_memory_blockage): Fix call of
4972 targetm.have_memory_blockage.
4974 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4977 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
4978 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
4979 objects to extra_objs instead of overwriting it.
4981 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4983 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
4984 Use any_fp_register_operand as operand[3] predicate. Simplify
4985 equality test for operands[2] and operands[4] memory location.
4986 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
4987 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
4988 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
4989 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
4990 any_fp_register_operand as operand[1] predicate. Simplify
4991 equality test for operands[0] and operands[3] memory location.
4992 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
4993 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
4994 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
4996 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4998 * target-insns.def: Add memory_blockage.
4999 * optabs.c (expand_memory_blockage): New function.
5000 (expand_asm_memory_barrier): Rename ...
5001 (expand_asm_memory_blockage): ... to this.
5002 (expand_mem_thread_fence): Call expand_memory_blockage
5003 instead of expand_asm_memory_barrier.
5004 (expand_mem_singnal_fence): Ditto.
5005 (expand_atomic_load): Ditto.
5006 (expand_atomic_store): Ditto.
5007 * doc/md.texi (Standard Pattern Names For Generation):
5008 Document memory_blockage instruction pattern.
5010 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
5012 * config/rl78/rl78.c (rl78_emit_libcall): New function.
5013 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
5014 * config/rl78/rl78.md: New define_expand "adddi3".
5016 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5018 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5021 2017-10-13 Jeff Law <law@redhat.com>
5023 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5025 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5028 * internal-fn.c (expand_mul_overflow): If both operands have
5029 the same highpart of -1 or 0 and the topmost bit of lowpart
5030 is different, overflow is if res <= 0 rather than res < 0.
5032 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5034 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5035 TARGET_P9_VECTOR code for unaligned_load case.
5037 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5039 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5041 2017-10-13 Nathan Sidwell <nathan@acm.org>
5043 * tree-core.h (tree_contains_struct): Make bool.
5044 * tree.c (tree_contains_struct): Likewise.
5045 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5046 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5047 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5048 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5049 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5051 2017-10-13 Richard Biener <rguenther@suse.de>
5053 * graphite-isl-ast-to-gimple.c
5054 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5055 parameters and dominance check.
5056 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5057 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5058 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5059 Do not update SSA form here or do intermediate IL verification.
5060 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5061 (graphite_initialize): Remove check on the number of loops in
5062 the function and inline into graphite_transform_loops.
5063 (graphite_finalize): Inline into graphite_transform_loops.
5064 (graphite_transform_loops): Perform SSA update and IL verification
5066 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5068 2017-10-13 Richard Biener <rguenther@suse.de>
5070 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5071 graphite_expression_type_precision): Avoid global constructor
5073 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5074 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5075 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5076 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5077 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5078 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5080 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5083 * config/i386/i386.h (ix86_red_zone_size): New.
5084 * config/i386/i386.md (push peephole2s): Replace
5085 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5087 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5088 Alan Hayward <alan.hayward@arm.com>
5089 David Sherwood <david.sherwood@arm.com>
5091 * combine.c (can_change_dest_mode): Reject changes in
5092 REGMODE_NATURAL_SIZE.
5094 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5095 Alan Hayward <alan.hayward@arm.com>
5096 David Sherwood <david.sherwood@arm.com>
5098 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5099 (expand_debug_source_expr): Likewise.
5100 * combine.c (combine_simplify_rtx): Likewise.
5101 * cse.c (fold_rtx): Likewise.
5102 * fwprop.c (canonicalize_address): Likewise.
5103 * targhooks.c (default_shift_truncation_mask): Likewise.
5105 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5106 Alan Hayward <alan.hayward@arm.com>
5107 David Sherwood <david.sherwood@arm.com>
5109 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5110 (widened_mode): Likewise.
5111 (expand_unop): Likewise.
5112 * ree.c (transform_ifelse): Likewise.
5113 (merge_def_and_ext): Likewise.
5114 (combine_reaching_defs): Likewise.
5115 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5117 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5118 Alan Hayward <alan.hayward@arm.com>
5119 David Sherwood <david.sherwood@arm.com>
5121 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5122 * combine.c (gen_lowpart_for_combine): Likewise.
5123 * dwarf2out.c (rtl_for_decl_location): Likewise.
5124 * final.c (alter_subreg): Likewise.
5125 * rtlhooks.c (gen_lowpart_general): Likewise.
5126 (gen_lowpart_if_possible): Likewise.
5128 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5129 Alan Hayward <alan.hayward@arm.com>
5130 David Sherwood <david.sherwood@arm.com>
5132 * calls.c (expand_call): Use subreg_lowpart_offset.
5133 * cse.c (cse_insn): Likewise.
5134 * regcprop.c (copy_value): Likewise.
5135 (copyprop_hardreg_forward_1): Likewise.
5137 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5140 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5141 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5142 =Q constraints to +Q and into insn condition add check
5143 that operands[0] and operands[1] are equal.
5144 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5145 =Q constraints to +Q and into insn condition add check
5146 that operands[0] is equal to either operands[1] or operands[2].
5149 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5150 instead of handling MINUS_EXPR twice (once for each argument),
5151 canonicalize operand order and handle just once, use rtype where
5152 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5155 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5156 any values of __C while still being pattern recognizable as a simple
5159 2017-10-13 Richard Biener <rguenther@suse.de>
5161 PR tree-optimization/82451
5163 2017-10-02 Richard Biener <rguenther@suse.de>
5165 PR tree-optimization/82355
5166 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5167 a mapping for the enclosing loop but avoid generating one for
5169 (copy_bb_and_scalar_dependences): Remove premature codegen
5170 error on PHIs in blocks duplicated into multiple places.
5171 * graphite-scop-detection.c
5172 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5173 in the region use it as loop and nest to analyze the DR in.
5174 (try_generate_gimple_bb): Likewise.
5175 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5176 (add_loop_constraints): For blocks in a loop not in the region
5177 create a dimension with a single iteration.
5178 * sese.h (gbb_loop_at_index): Remove assert.
5180 * cfgloop.c (loop_preheader_edge): For the loop tree root
5181 return the single successor of the entry block.
5182 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5183 Reset the SCEV hashtable and niters.
5184 * graphite-scop-detection.c
5185 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5186 assert that we only have POLYNOMIAL_CHREC that vary in loops
5187 contained in the region.
5188 (scop_detection::graphite_can_represent_expr): Adjust.
5189 (scop_detection::stmt_has_simple_data_refs_p): For loops
5190 not in the region set loop to NULL. The nest is now the
5191 entry edge to the region.
5192 (try_generate_gimple_bb): Likewise.
5193 * sese.c (scalar_evolution_in_region): Adjust for
5194 instantiate_scev change.
5195 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5196 Make nest parameter the edge into the region.
5197 (create_data_ref): Likewise.
5198 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5199 entry edge into a region and adjust instantiate_scev calls.
5200 (create_data_ref): Likewise.
5201 (graphite_find_data_references_in_stmt): Likewise.
5202 (find_data_references_in_stmt): Pass the loop preheader edge
5203 from the nest argument.
5204 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5205 parameter the edge into the region.
5206 (instantiate_parameters): Use the loop preheader edge as entry.
5207 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5209 (get_instantiated_value_entry): Make instantiate_below parameter
5210 the edge into the region.
5211 (instantiate_scev_name): Likewise. Adjust dominance checks,
5212 when we cannot use loop-based instantiation instantiate by
5213 walking use-def chains.
5214 (instantiate_scev_poly): Adjust.
5215 (instantiate_scev_binary): Likewise.
5216 (instantiate_scev_convert): Likewise.
5217 (instantiate_scev_not): Likewise.
5218 (instantiate_array_ref): Remove.
5219 (instantiate_scev_3): Likewise.
5220 (instantiate_scev_2): Likewise.
5221 (instantiate_scev_1): Likewise.
5222 (instantiate_scev_r): Do not blindly handle N-operand trees.
5223 Do not instantiate array-refs. Handle all constants and invariants.
5224 (instantiate_scev): Make instantiate_below parameter
5225 the edge into the region.
5226 (resolve_mixers): Use the loop preheader edge for the region
5227 parameter to instantiate_scev_r.
5228 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5230 2017-10-13 Richard Biener <rguenther@suse.de>
5232 PR tree-optimization/82525
5233 * graphite-isl-ast-to-gimple.c
5234 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5236 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5237 Fail code generation when we cannot represent the isl integer.
5238 (binary_op_to_tree): Elide modulo operations that are no-ops
5239 in the type we code generate. Remove now superfluous code
5242 2017-10-13 Richard Biener <rguenther@suse.de>
5244 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5245 (scop_detection::harmful_loop_in_region): Remove premature
5246 IV type restriction.
5247 (scop_detection::graphite_can_represent_scev): We can handle
5248 pointer IVs just fine.
5250 2017-10-13 Alan Modra <amodra@gmail.com>
5252 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5253 "Clobbers and Scratch Registers". Add paragraph on
5254 alternative to clobbers for scratch registers and OpenBLAS
5257 2017-10-13 Alan Modra <amodra@gmail.com>
5259 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5260 example of a memory input for a string of known length. Move
5261 commentary out of table. Add a number of new examples
5262 covering array memory inputs.
5264 2017-10-12 Martin Liska <mliska@suse.cz>
5266 PR tree-optimization/82493
5267 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5268 (test_range_functions): New function.
5269 (sbitmap_c_tests): Likewise.
5270 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5271 * selftest.h (sbitmap_c_tests): New function.
5273 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5275 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5277 * config/rs6000/amo.h: Fix spacing issue.
5279 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5282 * config/i386/i386.md (*ashl<mode>3_mask_1,
5283 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5284 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5287 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5289 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5290 (profile_probability): Set max_probability
5291 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5293 * profile-count.c (profile_probability::differs_from_p): Do not
5294 rely on max_probaiblity == 10000
5296 2017-10-12 Jeff Law <law@redhat.com>
5298 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5301 2017-10-12 Martin Sebor <msebor@redhat.com>
5305 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5306 (handle_alias_pairs): Call it.
5307 * common.opt (-Wattribute-alias): New option.
5308 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5309 * doc/invoke.texi (-Wattribute-alias): Document.
5311 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5314 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5316 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5318 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5319 (make_hard_regno_born, make_hard_regno_dead): Update
5320 bb_killed_pseudos and bb_gen_pseudos.
5322 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5324 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5326 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5328 * config/alpha/alpha.c (alpha_split_conditional_move):
5329 Use std::swap instead of manually swapping.
5330 (alpha_stdarg_optimize_hook): Ditto.
5331 (alpha_canonicalize_comparison): Ditto.
5333 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5335 * tree-loop-distribution.c (struct builtin_info): New struct.
5336 (struct partition): Refactor fields into struct builtin_info.
5337 (partition_free): Free struct builtin_info.
5338 (build_size_arg_loc, build_addr_arg_loc): Delete.
5339 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5340 information from struct builtin_info.
5341 (find_single_drs): New function refactored from classify_partition.
5342 Also moved builtin validity checks to this function.
5343 (compute_access_range, alloc_builtin): New functions.
5344 (classify_builtin_st, classify_builtin_ldst): New functions.
5345 (classify_partition): Refactor code into functions find_single_drs,
5346 classify_builtin_st and classify_builtin_ldst.
5347 (distribute_loop): Don't do runtime alias check when distributing
5349 (find_seed_stmts_for_distribution): New function.
5350 (pass_loop_distribution::execute): Refactor code finding seed
5351 stmts into above function. Support distribution for the innermost
5352 two-level loop nest. Adjust dump information.
5354 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5356 * tree-loop-distribution.c: Adjust the general comment.
5357 (NUM_PARTITION_THRESHOLD): New macro.
5358 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
5359 (classify_partition): Skip builtin pattern of loop nest's inner loop.
5360 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
5361 in call to build_partition_graph.
5362 (finalize_partitions): New parameter. Make loop distribution more
5363 conservative by fusing more partitions.
5364 (distribute_loop): Don't do runtime alias check in case of loop nest
5366 (find_seed_stmts_for_distribution): New function.
5367 (prepare_perfect_loop_nest): New function.
5368 (pass_loop_distribution::execute): Refactor code finding seed stmts
5369 and loop nest into above functions. Support loop nest distribution.
5370 Adjust dump information accordingly.
5372 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5374 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
5375 and set PTYPE_SEQUENTIAL for merged partition.
5377 2017-10-12 Richard Biener <rguenther@suse.de>
5379 PR tree-optimization/69728
5381 2017-09-19 Richard Biener <rguenther@suse.de>
5383 PR tree-optimization/69728
5384 * graphite-sese-to-poly.c (schedule_error): New global.
5385 (add_loop_schedule): Handle empty domain by failing the
5387 (build_original_schedule): Handle schedule_error.
5389 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
5390 domain by returning an unchanged schedule.
5392 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5394 * genrecog.c (validate_pattern): For VEC_SELECT verify that
5395 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
5397 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
5399 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
5402 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5405 * expr.c (store_field): Don't optimize away bitsize == 0 store
5406 from CALL_EXPR with addressable return type.
5408 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5410 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
5411 * config/rs6000/rs6000.md (sel): Delete mode attribute.
5412 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
5413 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
5414 TARGET_ISEL instead of TARGET_ISEL<sel>.
5416 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
5418 * config/rs6000/rs6000.c
5419 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
5421 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5423 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
5425 (reg_or_cint_operand): Fix comment.
5426 (reg_or_zero_operand): New predicate.
5427 * config/rs6000/rs6000-protos.h (output_isel): Delete.
5428 * config/rs6000/rs6000.c (output_isel): Delete.
5429 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
5430 instead of reg_or_cint_operand. Output instruction directly (not via
5432 (isel_unsigned_<mode>): Ditto.
5433 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
5434 gpc_reg_operand. Add an instruction alternative for this. Output
5435 instruction directly.
5436 (*isel_reversed_unsigned_<mode>): Ditto.
5438 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5440 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
5441 (TARGET_CANONICALIZE_COMPARISON): Define.
5443 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
5446 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
5447 Check whether the dest is REG before adding REG_EQUIV note.
5449 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5452 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5454 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5455 (make_hard_regno_born, make_hard_regno_dead): Update
5456 bb_killed_pseudos and bb_gen_pseudos.
5458 2017-10-11 Nathan Sidwell <nathan@acm.org>
5460 * incpath.h (enum incpath_kind): Name enum, prefix values.
5461 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
5462 * incpath.c (heads, tails): Use INC_MAX.
5463 (add_env_var_paths, add_standard_paths): Use incpath_kind.
5464 (merge_include_chains, split_quote_chain,
5465 register_include_chains): Update incpath_kind names.
5466 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
5467 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
5469 (add_framework_path, darwin_register_objc_includes): Likewise.
5470 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
5472 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5474 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5475 Do not use float_operator operator predicate.
5476 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5477 * config/i386/predicates.md (float_operator): Remove predicate.
5479 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5481 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
5482 (*jccxf_i387): Ditto.
5483 (*jcc<mode>_i387): Ditto.
5484 (*jccu<mode>_i387): Ditto.
5485 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
5486 (*jcc_*_i387 splitters): Remove.
5487 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
5488 * config/i386/i386.c (ix86_split_fp_branch): Remove.
5489 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
5492 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5494 * profile-count.h (slow_safe_scale_64bit): New function.
5495 (safe_scale_64bit): New inline.
5496 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
5497 * profile-count.c: Include wide-int.h
5498 (slow_safe_scale_64bit): New.
5500 2017-10-11 Nathan Sidwell <nathan@acm.org>
5502 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
5503 HAS_DECL_ASSEMBLER_NAME_P.
5504 * gimple-expr.c (gimple_decl_printable_name: Check
5505 HAS_DECL_ASSEMBLER_NAME_P too.
5506 * ipa-utils.h (type_in_anonymous_namespace_p): Check
5507 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
5508 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
5509 * passes.c (rest_of_decl_compilation): Check
5510 HAS_DECL_ASSEMBLER_NAME_P too.
5511 * recog.c (verify_changes): Likewise.
5512 * tree-pretty-print.c (dump_decl_name): Likewise.
5513 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
5515 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
5516 (SET_DECL_ASSEMBLER_NAME): Use it.
5517 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
5518 (COPY_DECL_ASSEMBLER_NAME): Likewise.
5519 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
5521 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5523 * config.gcc (i386, x86_64): Add extra objects.
5524 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
5525 (ix86_min_insn_size): Declare.
5526 (ix86_issue_rate): Declare.
5527 (ix86_adjust_cost): Declare.
5528 (ia32_multipass_dfa_lookahead): Declare.
5529 (ix86_macro_fusion_p): Declare.
5530 (ix86_macro_fusion_pair_p): Declare.
5531 (ix86_bd_has_dispatch): Declare.
5532 (ix86_bd_do_dispatch): Declare.
5533 (ix86_core2i7_init_hooks): Declare.
5534 (ix86_atom_sched_reorder): Declare.
5535 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
5536 (COSTS_N_BYTES): Move to x86-tune-costs.h.
5537 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
5538 (rip_relative_addr_p): Rename to ...
5539 (ix86_rip_relative_addr_p): ... this one; export.
5540 (memory_address_length): Update.
5541 (ix86_issue_rate): Move to x86-tune-sched.c.
5542 (ix86_flags_dependent): Move to x86-tune-sched.c.
5543 (ix86_agi_dependent): Move to x86-tune-sched.c.
5544 (exact_dependency_1): Move to x86-tune-sched.c.
5545 (exact_store_load_dependency): Move to x86-tune-sched.c.
5546 (ix86_adjust_cost): Move to x86-tune-sched.c.
5547 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
5548 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
5549 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
5550 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
5551 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
5552 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
5553 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
5554 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
5555 (min_insn_size): Rename to ...
5556 (ix86_min_insn_size): ... this one; export.
5557 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
5558 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
5559 (core2i7_first_cycle_multipass_backtrack): Move to
5560 x86-tune-sched-core.c.
5561 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
5562 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
5563 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
5564 (ix86_avoid_jump_mispredicts): Update.
5565 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
5566 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
5567 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
5568 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
5569 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
5570 (MAX_INSN): Move to ix86-tune-sched-bd.c.
5571 (MAX_IMM): Move to ix86-tune-sched-bd.c.
5572 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
5573 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
5574 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
5575 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
5576 (MAX_STORE): Move to ix86-tune-sched-bd.c.
5577 (BIG): Move to ix86-tune-sched-bd.c.
5578 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
5579 (enum insn_path): Move to ix86-tune-sched-bd.c.
5580 (get_mem_group): Move to ix86-tune-sched-bd.c.
5581 (is_cmp): Move to ix86-tune-sched-bd.c.
5582 (dispatch_violation): Move to ix86-tune-sched-bd.c.
5583 (is_branch): Move to ix86-tune-sched-bd.c.
5584 (is_prefetch): Move to ix86-tune-sched-bd.c.
5585 (init_window): Move to ix86-tune-sched-bd.c.
5586 (allocate_window): Move to ix86-tune-sched-bd.c.
5587 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
5588 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
5589 (process_end_window): Move to ix86-tune-sched-bd.c.
5590 (allocate_next_window): Move to ix86-tune-sched-bd.c.
5591 (find_constant): Move to ix86-tune-sched-bd.c.
5592 (get_num_immediates): Move to ix86-tune-sched-bd.c.
5593 (has_immediate): Move to ix86-tune-sched-bd.c.
5594 (get_insn_path): Move to ix86-tune-sched-bd.c.
5595 (get_insn_group): Move to ix86-tune-sched-bd.c.
5596 (count_num_restricted): Move to ix86-tune-sched-bd.c.
5597 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
5598 (add_insn_window): Move to ix86-tune-sched-bd.c.
5599 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
5600 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
5601 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
5602 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
5603 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
5604 (do_dispatch): Move to ix86-tune-sched-bd.c.
5605 (has_dispatch): Move to ix86-tune-sched-bd.c.
5606 * config/i386/t-i386: Add new object files.
5607 * config/i386/x86-tune-costs.h: New file.
5608 * config/i386/x86-tune-sched-atom.c: New file.
5609 * config/i386/x86-tune-sched-bd.c: New file.
5610 * config/i386/x86-tune-sched-core.c: New file.
5611 * config/i386/x86-tune-sched.c: New file.
5613 2017-10-11 Liu Hao <lh_mouse@126.com>
5615 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
5616 the generic version below instead.
5617 (should_colorize): Recognize Windows consoles as terminals
5619 * pretty-print.c [__MINGW32__] (write_all): New function.
5620 [__MINGW32__] (find_esc_head): Likewise.
5621 [__MINGW32__] (find_esc_terminator): Likewise.
5622 [__MINGW32__] (eat_esc_sequence): Likewise.
5623 [__MINGW32__] (mingw_ansi_fputs): New function that handles
5625 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
5628 2017-10-11 Richard Biener <rguenther@suse.de>
5630 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
5631 Properly call analyze_scalar_evolution with the loop of the stmt.
5633 2017-10-11 Richard Biener <rguenther@suse.de>
5635 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
5636 * tree-core.h (tree_base): Add chrec_var union member.
5637 * tree.h (CHREC_VAR): Remove.
5638 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
5639 * tree-chrec.h (build_polynomial_chrec): Adjust.
5640 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
5641 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
5643 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
5645 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
5646 * match.pd: ... here.
5647 ((T) X == (T) Y): Relax condition.
5649 2017-10-11 Bin Cheng <bin.cheng@arm.com>
5651 PR tree-optimization/82472
5652 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
5654 (break_alias_scc_partitions): Update postorder number.
5656 2017-10-11 Martin Liska <mliska@suse.cz>
5659 * opts.c (parse_no_sanitize_attribute): Do not use error_value
5661 * opts.h (parse_no_sanitize_attribute): Remove last argument.
5663 2017-10-11 Martin Liska <mliska@suse.cz>
5665 * print-rtl.c (print_insn): Move declaration of idbuf
5666 to same scope as name.
5668 2017-10-11 Martin Liska <mliska@suse.cz>
5673 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5676 2017-10-11 Martin Liska <mliska@suse.cz>
5679 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5682 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
5684 * config.gcc (default_gnu_indirect_function): Default to yes for
5685 arm*-*-linux* with glibc.
5687 2017-10-11 Richard Biener <rguenther@suse.de>
5689 * tree-scalar-evolution.c (get_scalar_evolution): Handle
5690 default-defs and types we do not want to analyze.
5691 (interpret_loop_phi): Replace unreachable code with an assert.
5692 (compute_scalar_evolution_in_loop): Remove and inline ...
5693 (analyze_scalar_evolution_1): ... here, replacing condition with
5694 what makes the intent clearer. Remove handling of cases
5695 get_scalar_evolution now handles.
5697 2017-10-10 Jim Wilson <wilson@tuliptree.org>
5699 PR rtl-optimization/81434
5700 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
5701 comment for main loop. In sched_group_found if, also add checks for
5702 pass and min_cost_group.
5704 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
5706 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
5707 (rs6000_insn_cost): New function.
5708 * config/rs6000/rs6000.md (cost): New attribute.
5710 2017-10-10 Jakub Jelinek <jakub@redhat.com>
5711 H.J. Lu <hongjiu.lu@intel.com>
5715 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
5716 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
5717 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
5718 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
5719 to OPTION_MASK_ISA_AVX512VL - builtins that have both
5720 OPTION_MASK_ISA_MMX and some other bit set require both
5721 mmx and the ISAs without the mmx bit.
5722 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
5723 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
5724 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
5725 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
5726 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
5727 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
5728 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
5729 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
5730 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
5731 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
5732 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
5733 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
5734 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
5735 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
5736 Add OPTION_MASK_ISA_MMX.
5738 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
5740 * config.gcc (armv7*-*-freebsd*): New target.
5741 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
5743 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
5745 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
5746 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
5749 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
5751 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
5752 a template parameter.
5753 (WIDE_INT_REF_FOR): Update accordingly.
5754 * tree.h (wi::int_traits <const_tree>): Delete.
5755 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
5756 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
5757 (wi::tree_to_wide_ref): New typedef.
5758 (wi::to_wide): New function.
5759 * calls.c (get_size_range): Use wi::to_wide when operating on
5761 * cgraph.c (cgraph_node::create_thunk): Likewise.
5762 * config/i386/i386.c (ix86_data_alignment): Likewise.
5763 (ix86_local_alignment): Likewise.
5764 * dbxout.c (stabstr_O): Likewise.
5765 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
5766 * expr.c (const_vector_from_tree): Likewise.
5767 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
5768 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
5769 (fold_negate_expr_1, int_const_binop_1, const_binop)
5770 (fold_convert_const_int_from_real, optimize_bit_field_compare)
5771 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
5772 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
5773 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
5774 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
5775 (fold_not_const, round_up_loc): Likewise.
5776 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
5777 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
5778 (alloca_call_type): Likewise.
5779 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
5780 * godump.c (go_output_typedef): Likewise.
5781 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
5782 * internal-fn.c (get_min_precision): Likewise.
5783 * ipa-cp.c (ipcp_store_vr_results): Likewise.
5784 * ipa-polymorphic-call.c
5785 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
5786 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
5787 (ipa_modify_call_arguments): Likewise.
5788 * match.pd: Likewise.
5789 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
5790 * print-tree.c (print_node_brief, print_node): Likewise.
5791 * stmt.c (expand_case): Likewise.
5792 * stor-layout.c (layout_type): Likewise.
5793 * tree-affine.c (tree_to_aff_combination): Likewise.
5794 * tree-cfg.c (group_case_labels_stmt): Likewise.
5795 * tree-data-ref.c (dr_analyze_indices): Likewise.
5796 (prune_runtime_alias_test_list): Likewise.
5797 * tree-dump.c (dequeue_and_dump): Likewise.
5798 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
5799 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
5800 * tree-pretty-print.c (dump_generic_node): Likewise.
5801 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
5802 (simple_iv_with_niters): Likewise.
5803 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
5804 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
5805 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
5806 * tree-ssa-loop-niter.c (split_to_var_and_offset)
5807 (refine_value_range_using_guard, number_of_iterations_ne_max)
5808 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
5809 (get_cst_init_from_scev, record_nonwrapping_iv)
5810 (scev_var_range_cant_overflow): Likewise.
5811 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
5812 * tree-ssa-pre.c (compute_avail): Likewise.
5813 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
5814 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
5815 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
5816 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
5817 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
5818 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
5819 (dump_case_nodes, try_switch_expansion): Likewise.
5820 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
5821 (vect_do_peeling): Likewise.
5822 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
5823 * tree-vect-stmts.c (vectorizable_load): Likewise.
5824 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
5825 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
5826 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
5827 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
5828 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
5829 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
5830 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
5831 (evrp_dom_walker::before_dom_children): Likewise.
5832 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
5833 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
5834 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
5835 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
5836 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
5837 (get_range_pos_neg): Likewise.
5838 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
5839 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
5840 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
5841 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
5842 * config/avr/avr.c (avr_fold_builtin): Likewise.
5843 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
5844 * config/msp430/msp430.c (msp430_attr): Likewise.
5845 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
5846 * config/powerpcspe/powerpcspe-c.c
5847 (altivec_resolve_overloaded_builtin): Likewise.
5848 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
5849 (rs6000_expand_ternop_builtin): Likewise.
5850 * config/rs6000/rs6000-c.c
5851 (altivec_resolve_overloaded_builtin): Likewise.
5852 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
5853 (rs6000_expand_ternop_builtin): Likewise.
5854 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
5856 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5858 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
5859 when copying loop nest with only one inner loop.
5861 2017-10-10 Richard Biener <rguenther@suse.de>
5863 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
5864 blocks if SCEV is active.
5865 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
5867 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
5868 (scev_initialize): Assert we are not yet initialized.
5870 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5872 * tree-loop-distribution.c (generate_loops_for_partition): Remove
5873 inner loop's exit stmt by making it always exit the loop, otherwise
5874 we would generate an infinite empty loop.
5876 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5878 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
5879 renaming variables in new preheader if it's deleted.
5881 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5883 * tree-loop-distribution.c (struct partition): Remove unused field
5884 loops of the structure.
5885 (partition_alloc, partition_free): Ditto.
5886 (build_rdg_partition_for_vertex): Ditto.
5888 2017-10-09 Jeff Law <law@redhat.com>
5890 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
5891 return type to match prototype and documentation.
5893 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5895 * config/rs6000/rs6000.c (processor_costs): Move to ...
5896 * config/rs6000/rs6000.h: ... here.
5897 (rs6000_cost): Declare.
5899 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
5901 * except.c (setjmp_fn): New global variable.
5902 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
5903 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
5904 if DONT_USE_BUILTIN_SETJMP is defined.
5906 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5908 * target.def (insn_cost): New hook.
5909 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
5910 * doc/tm.texi: Regenerate.
5911 * rtlanal.c (insn_cost): Use the new hook.
5913 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5915 * combine.c (combine_validate_cost): Compute the new insn_cost,
5916 not just pattern_cost.
5917 (try_combine): Adjust comment.
5919 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5921 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
5923 * combine.c (uid_insn_cost): Adjust comment.
5924 (combine_validate_cost): Adjust comment. Use pattern_cost instead
5926 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
5927 * dse.c (find_shift_sequence): Ditto.
5928 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
5929 (bb_valid_for_noce_process_p): Use pattern_cost.
5930 * rtl.h (insn_rtx_cost): Delete.
5931 (pattern_cost): New prototype.
5932 (insn_cost): New prototype.
5933 * rtlanal.c (insn_rtx_cost): Rename to...
5934 (pattern_cost): ... this.
5937 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
5939 * config/i386/i386.md (*jcc_2): Remove insn pattern.
5940 (*jcc<mode>_0_r_i387): Ditto.
5941 (*jccxf_r_i387): Ditto.
5942 (*jcc<mode>_r_i387): Ditto.
5943 (*jccu<mode>_r_i387): Ditto.
5944 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
5945 (*jcc): Rename from *jcc_1.
5947 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5949 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
5950 deferred rescans after the lvx/stvx recombination pre-pass.
5952 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
5954 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
5955 memory operation instruction support.
5956 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
5957 (rs6000-ibm-aix[789]*): Likewise.
5958 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
5959 Document new functions.
5961 2017-10-09 Richard Biener <rguenther@suse.de>
5963 PR tree-optimization/82397
5964 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
5965 equality only for semantically equal trees.
5967 2017-10-09 Richard Biener <rguenther@suse.de>
5969 PR tree-optimization/82449
5970 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
5971 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
5972 allow constant addresses.
5973 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
5976 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5978 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
5981 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5984 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
5987 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5990 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
5992 2017-10-09 Jakub Jelinek <jakub@redhat.com>
5995 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
5996 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
5998 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
6000 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
6001 (WI_BINARY_PREDICATE_RESULT): Likewise.
6002 (wi::binary_traits::operator_result): New type.
6003 (wi::binary_traits::predicate_result): Likewise.
6004 (generic_wide_int::operator~, unary generic_wide_int::operator-)
6005 (generic_wide_int::operator==, generic_wide_int::operator!=)
6006 (generic_wide_int::operator&, generic_wide_int::and_not)
6007 (generic_wide_int::operator|, generic_wide_int::or_not)
6008 (generic_wide_int::operator^, generic_wide_int::operator+
6009 (binary generic_wide_int::operator-, generic_wide_int::operator*):
6011 (operator~, unary operator-, operator==, operator!=, operator&)
6012 (operator|, operator^, operator+, binary operator-, operator*): New
6014 * expr.c (get_inner_reference): Use wi::bit_and_not.
6015 * fold-const.c (fold_binary_loc): Likewise.
6016 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6017 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6018 (bit_value_binop): Likewise.
6019 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6020 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6021 (extract_range_from_binary_expr_1): Likewise.
6022 (masked_increment): Likewise.
6023 (simplify_bit_ops_using_ranges): Likewise.
6025 2017-10-09 Martin Jambor <mjambor@suse.cz>
6028 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6029 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6030 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6032 (hsa_fixup_mov_insn_type): New function.
6033 (hsa_op_with_type::get_in_type): Use it.
6034 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6035 immediates in an assert.
6036 (hsa_op_with_type::extend_int_to_32bit): New method.
6037 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6038 types. Convert to dest type if necessary.
6039 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6040 (reg_for_gimple_ssa): Pass false as min32int to
6041 hsa_type_for_scalar_tree_type.
6042 (gen_hsa_addr): Fixup type when creating addresable temporary.
6043 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6044 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6045 necessary. Call hsa_fixup_mov_insn_type.
6046 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6047 extend operands and convert to dest type if necessary.
6048 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6049 to dest type if necessary.
6050 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6051 if conversion nt necessary and size matches.
6052 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6053 to dest type if necessary.
6054 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6055 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6057 (gen_hsa_clrsb): Likewise.
6058 (gen_hsa_ffs): Likewise.
6059 (gen_hsa_divmod): Extend operands and convert to dest type if
6061 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6063 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6065 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6066 Remove empty default arguments. Use a brace block as output
6068 (conditional return): Ditto.
6070 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6071 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6073 (group_ending_nop): Ditto.
6074 (doloop_end): Ditto.
6075 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6076 (splitters for those): Ditto.
6078 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6080 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6081 a conditional jump (and the compare for it) so that pc_rtx is the
6083 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6084 for the deleted and renamed ctr<mode>_internal[234] patterns.
6085 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6086 Delete second conditional return pattern.
6087 (ctr<mode>_internal2): Delete this second bdnz pattern.
6088 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6089 (ctr<mode>_internal4): Delete this second bdz pattern.
6091 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6093 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6094 (always_initialized_rtx_for_ssa_name_p): New predicate.
6095 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6096 (finish_out_of_ssa): Free new field of SA.
6097 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6098 * tree-ssa-coalesce.c: Include tree-ssa.h.
6099 (get_parm_default_def_partitions): Remove extern keyword.
6100 (get_undefined_value_partitions): New function.
6101 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6102 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6104 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6106 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6108 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6110 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6112 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6113 for targets that preffer 128bit.
6115 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6117 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6119 2017-10-08 Olivier Hainque <hainque@adacore.com>
6121 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6122 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6124 (thumb_set_return_address): Likewise.
6126 2017-10-08 Olivier Hainque <hainque@adacore.com>
6128 * common/config/arm/arm-common.c (arm_except_unwind_info):
6129 Handle DWARF2_UNWIND_INFO.
6131 2017-10-07 Michael Collison <michael.collison@arm.com>
6133 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6136 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6138 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6139 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6140 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6141 defined, force the creation of a new block for a dispatch label.
6143 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6145 * invoke.texi (Wsuggest-attribute=cold): Document.
6146 * common.opt (Wsuggest-attribute=cold): New
6147 * ipa-pure-const.c (warn_function_cold): New function.
6148 * predict.c (compute_function_frequency): Use it.
6149 * predict.h (warn_function_cold): Declare.
6151 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6153 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6156 2017-10-06 Martin Liska <mliska@suse.cz>
6158 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6159 keyword for member functions.
6160 (struct sanopt_tree_couple): New struct.
6161 (struct sanopt_tree_couple_hash): New function.
6162 (struct sanopt_ctx): Add new hash_map.
6163 (has_dominating_ubsan_ptr_check): New function.
6164 (record_ubsan_ptr_check_stmt): Likewise.
6165 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6166 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6167 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6169 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6172 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6173 aarch64_simd_valid_immediate on CONST_VECTORs.
6174 (aarch64_reg_or_bic_imm): Likewise.
6176 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6178 PR rtl-optimization/82396
6179 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6181 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6183 * graphite-dependences.c (scop_get_reads): Move code to...
6184 (scop_get_must_writes): Move code to...
6185 (scop_get_may_writes): Move code to...
6186 (scop_get_reads_and_writes): ... here.
6187 (scop_get_dependences): Call scop_get_reads_and_writes.
6189 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6191 PR tree-optimization/82434
6192 * fold-const.h (can_native_encode_type_p,
6193 can_native_encode_string_p): Remove.
6194 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6195 don't encode anything, just return what would be otherwise returned.
6196 (native_encode_fixed, native_encode_complex, native_encode_vector):
6198 (native_encode_string): Likewise. Inline by hand
6199 can_native_encode_string_p.
6200 (can_native_encode_type_p): Remove.
6201 (can_native_encode_string_p): Remove.
6202 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6203 STRING_CSTs using can_native_encode_string_p, test all
6204 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6205 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6206 argument from native_encode_expr.
6207 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6208 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6211 2017-10-06 Richard Biener <rguenther@suse.de>
6213 PR tree-optimization/82397
6214 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6215 operand_equal_p but rely on data_ref_compare_tree for detecting
6217 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6218 to match up with dr_group_sort_cmp.
6220 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6223 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6225 * config/s390/s390-builtin-types.def: Regenerate.
6227 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6230 * config/s390/s390-builtin-types.def: Regenerate.
6231 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6232 Change flag from B_VXE to B_VX.
6233 (s390_vec_min_dbl): Remove B_VXE flag.
6235 2017-10-06 Richard Biener <rguenther@suse.de>
6237 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6238 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6239 translate_isl_ast_to_gimple::is_valid_rename,
6240 translate_isl_ast_to_gimple::get_rename,
6241 translate_isl_ast_to_gimple::get_def_bb_for_const,
6242 translate_isl_ast_to_gimple::get_new_name,
6243 translate_isl_ast_to_gimple::collect_all_ssa_names,
6244 translate_isl_ast_to_gimple::copy_loop_phi_args,
6245 translate_isl_ast_to_gimple::collect_all_ssa_names,
6246 translate_isl_ast_to_gimple::copy_loop_phi_args,
6247 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6248 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6249 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6250 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6251 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6252 translate_isl_ast_to_gimple::copy_cond_phi_args,
6253 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6254 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6255 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6256 translate_isl_ast_to_gimple::rename_uses,
6257 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6258 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6259 (set_rename_for_each_def): Likewise.
6260 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6261 here. Handle rewriting SCEV analyzable uses here.
6262 (copy_bb_and_scalar_dependences): Generate code for PHI
6264 (graphite_regenerate_ast_isl): Adjust.
6265 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6266 (add_write, add_read): New functions.
6267 (build_cross_bb_scalars_def): Use it and simplify.
6268 (build_cross_bb_scalars_use): Likewise.
6269 (graphite_find_cross_bb_scalar_vars): Inline into...
6270 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6271 simulating out-of-SSA. Compute liveout and add dependencies.
6272 (build_scops): Force an empty entry block.
6273 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6275 (sese_build_liveouts): Declare.
6276 (sese_trivially_empty_bb_p): Likewise.
6277 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6278 compute liveout and debug_liveout.
6279 (sese_bad_liveouts_use): Remove.
6280 (sese_reset_debug_liveouts_bb): Likewise.
6281 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6282 (sese_build_liveouts): Build liveout and debug_liveout and store
6284 (new_sese_info): Adjust.
6285 (free_sese_info): Likewise.
6286 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6287 do not build liveout here.
6288 (move_sese_in_condition): Adjust region entry.
6289 (scev_analyzable_p): Match up with chrec_apply requirements.
6290 (sese_trivially_empty_bb_p): New.
6291 * tree-into-ssa.c (get_reaching_def): Properly support generating
6292 default-defs for incremental rewrite of anonymous names.
6294 2017-10-06 Richard Biener <rguenther@suse.de>
6296 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6297 precision do not perform modulo reduction.
6299 2017-10-06 Richard Biener <rguenther@suse.de>
6301 PR tree-optimization/82436
6302 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6303 conservatively choose the vectorization factor when checking
6304 whether we can perform the required load permutation.
6305 (vect_transform_slp_perm_load): Assert when we may not fail.
6307 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6309 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6310 message for incompatible -msdata=* and -mcall-* options.
6312 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6314 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6315 rate for post-reload scheduling.
6317 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6319 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6321 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6323 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6324 to improve monte carlo in scimark.
6326 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6328 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6329 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6330 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6331 pentium4_cost, nocona_cost): Set reassociation width to 1.
6332 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6333 width to 2 for fp operations and 1 otherwise.
6334 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6336 (atom_cost): Set reassociation width to 2.
6337 (slm_cost, generic_cost): Set fp reassociation width
6338 to 2 and 1 otherwise.
6339 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6340 (core_cost): Set fp reassociation width to 4 and vector to 2.
6341 (ix86_reassociation_width): Rewrite using cost table; special case
6342 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6343 and TARGET_AVX128_OPTIMAL.
6344 * config/i386/i386.h (processor_costs): Add
6345 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6346 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6347 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6348 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6349 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6350 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6352 2017-10-05 Nathan Sidwell <nathan@acm.org>
6354 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6356 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6358 * config/arm/arm.c (arm_test_fpu_data): New.
6359 (arm_run_selftests): Call arm_test_fpu_data.
6361 2017-10-04 Nathan Sidwell <nathan@acm.org>
6363 * toplev.c (toplev::main): Remove excess parens on pretty_printer
6365 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
6367 2017-10-04 Sudakshina Das <sudi.das@arm.com>
6369 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
6370 check type for aarch64_simd_valid_immediate.
6371 (aarch64_output_simd_mov_immediate): Update prototype.
6372 (aarch64_simd_valid_immediate): Update prototype.
6373 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
6374 support for ORR-immediate.
6375 (and<mode>3): modified pattern to add support for BIC-immediate.
6376 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
6377 now checks for valid immediate for BIC and ORR based on new enum
6379 (aarch64_output_simd_mov_immediate): Function now used to output
6380 BIC/ORR imm as well based on new enum argument.
6381 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6383 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6384 (aarch64_reg_or_bic_imm): Likewise.
6386 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6388 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
6389 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
6391 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6395 PR rtl-optimization/82396
6396 * haifa-sched.c (autopref_multipass_init): Simplify
6398 (autopref_rank_data): Simplify sort order.
6399 * sched-int.h (autopref_multipass_data_): Remove
6400 multi_mem_insn_p, min_offset and max_offset.
6402 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6404 * doc/sourcebuild.texi: Document vect_peeling_profitable.
6406 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6408 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
6411 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6413 * doc/sourcebuild.texi: Document vect_long_mult.
6415 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
6417 PR tree-optimization/82413
6418 * fold-const.c (build_range_check): Use widest_int when comparing
6419 the maximum ETYPE value with HIGH.
6421 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6423 PR rtl-optimization/82396
6424 * haifa-sched.c (autopref_multipass_init): Simplify
6426 (autopref_rank_data): Simplify sort order.
6427 * sched-int.h (autopref_multipass_data_): Remove
6428 multi_mem_insn_p, min_offset and max_offset.
6430 2017-10-04 Jakub Jelinek <jakub@redhat.com>
6432 PR tree-optimization/82381
6433 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
6434 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
6437 PR tree-optimization/82374
6438 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
6439 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
6440 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
6441 current_function_decl to the new decl.
6443 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6445 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
6446 helper macro for IEEE float128 hardware built-in functions.
6447 (SQRTF128_ODD): Add built-in functions with the round-to-odd
6449 (TRUNCF128_ODD): Likewise.
6450 (ADDF128_ODD): Likewise.
6451 (SUBF128_ODD): Likewise.
6452 (MULF128_ODD): Likewise.
6453 (DIVF128_ODD): Likewise.
6454 (FMAF128_ODD): Likewise.
6455 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
6456 UNSPEC_TRUNC_ROUND_TO_ODD.
6457 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
6458 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
6459 floating point round to odd instructions.
6460 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
6461 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
6462 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
6463 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
6464 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
6465 (trunc<mode>sf2_hw): Change the truncate with round to odd
6466 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
6467 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
6468 to odd hardware instructions.
6469 (sub<mode>3_odd): Likewise.
6470 (mul<mode>3_odd): Likewise.
6471 (div<mode>3_odd): Likewise.
6472 (sqrt<mode>2_odd): Likewise.
6473 (fma<mode>4_odd): Likewise.
6474 (fms<mode>4_odd): Likewise.
6475 (nfma<mode>4_odd): Likewise.
6476 (nfms<mode>4_odd): Likewise.
6477 (trunc<mode>df2_odd): Change the truncate with round to odd
6478 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
6480 * doc/extend.texi (PowerPC built-in functions): Update documentation
6481 for existing IEEE float128-bit built-in functions. Add built-in
6482 functions that generate the IEEE 128-bit floating point round to
6485 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
6487 PR rtl-optimization/77729
6488 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
6489 to (X&(C1&~C2))|C2 transformations.
6491 2017-10-03 Martin Jambor <mjambor@suse.cz>
6493 PR tree-optimization/82363
6494 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
6495 mismatch, mark lacc written regardless of racc.
6497 2017-10-03 Jakub Jelinek <jakub@redhat.com>
6499 PR tree-optimization/82381
6500 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
6501 stmt_to_insert nor wheather SSA_NAMEs are default defs.
6502 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
6503 fallthrough into reassoc_stmt_dominates_stmt_p.
6506 * combine.c (combine_instructions): Don't combine in unreachable
6509 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6512 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
6513 function to not use the have_cpu variable. Do not set cpu_index,
6514 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
6516 (rs6000_valid_attribute_p): Remove duplicate initializations of
6517 old_optimize and func_optimize.
6518 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
6519 (rs6000_activate_target_options): Make global.
6520 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
6523 2017-10-02 Jakub Jelinek <jakub@redhat.com>
6525 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
6526 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
6527 Set *poffset to 0 and *psize and *pmax_size to -1 if
6528 *poffset + *psize overflows in HOST_WIDE_INT.
6530 PR tree-optimization/82387
6531 PR tree-optimization/82388
6532 PR tree-optimization/82389
6533 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
6534 instead of live_bytes non-NULL.
6536 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
6539 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
6542 2017-10-02 Richard Biener <rguenther@suse.de>
6544 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
6545 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
6546 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
6548 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6550 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
6551 requested precision matches the type's.
6552 * calls.c (alloc_max_size): Calculate the new candidate size as
6553 a widest_int and use wi::to_widest when comparing it with the
6554 current candidate size.
6555 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
6556 zero rather than integer_zero_node.
6557 * match.pd: Check for a no-op conversion before using wi::add
6558 rather than after. Use tree_to_uhwi when summing small shift
6559 counts into an unsigned int.
6561 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6562 Alan Hayward <alan.hayward@arm.com>
6563 David Sherwood <david.sherwood@arm.com>
6566 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
6567 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
6568 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
6569 POINTER_AND_FP_REGS.
6571 2017-10-02 Richard Biener <rguenther@suse.de>
6573 PR tree-optimization/82355
6574 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6575 a mapping for the enclosing loop but avoid generating one for
6577 (copy_bb_and_scalar_dependences): Remove premature codegen
6578 error on PHIs in blocks duplicated into multiple places.
6579 * graphite-scop-detection.c
6580 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6581 in the region use it as loop and nest to analyze the DR in.
6582 (try_generate_gimple_bb): Likewise.
6583 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6584 (add_loop_constraints): For blocks in a loop not in the region
6585 create a dimension with a single iteration.
6586 * sese.h (gbb_loop_at_index): Remove assert.
6588 2017-10-01 Kevin Buettner <kevinb@redhat.com>
6590 * omp-expand.c (adjust_context_scope): New function.
6591 (expand_parallel_call): Call adjust_context_scope.
6593 2017-10-01 Jeff Law <law@redhat.com>
6595 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
6596 dom_opt_dom_walker class with direct access to private members.
6597 Add comments. Call test_for_singularity.
6598 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
6599 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
6600 m_dummy_cond anymore.
6601 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
6603 (pass_dominator:execute): Build the dummy_cond here and pass it
6604 to the dom_opt_dom_walker ctor.
6605 (test_for_singularity): New function.
6607 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
6608 Maya Rashish <coypu@sdf.org>
6610 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
6611 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
6612 (alpha*-*-netbsd*) Use nbsd_tm_file.
6613 (arm*-*-netbsdelf*) Likewise.
6614 (i[34567]86-*-netbsdelf*) Likewise.
6615 (x86_64-*-netbsd*) Likewise.
6616 (mips*-*-netbsd*) Likewise.
6617 (powerpc-*-netbsd*) Likewise.
6618 (sh*-*-netbsd*) Likewise.
6619 (sparc-*-netbsdelf*) Likewise.
6620 (sparc64-*-netbsd*) Likewise.
6621 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
6623 (vax-*-netbsdelf*) Likewise.
6624 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
6625 (UINT_FAST8_TYPE) Likewise.
6626 (INT_FAST16_TYPE) Check CHAR_FAST16.
6627 (UINT_FAST16_TYPE) Likewise.
6629 2017-09-30 Jakub Jelinek <jakub@redhat.com>
6632 * config/i386/i386.md
6633 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
6634 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
6635 *divmodsi4_zext_2): New define_insn_and_split.
6636 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
6637 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
6638 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
6639 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
6640 New define_insn_and_split.
6641 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
6642 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
6643 operands[1] having DImode when mode is SImode.
6645 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
6646 always SImode for DIV and MOD in REG_EQUAL notes.
6648 2017-09-29 Yury Gribov <tetra2005@gmail.com>
6651 * match.pd: Fix handling of NaNs in pattern.
6653 2017-09-29 Jeff Law <law@redhat.com>
6655 * sbitmap.c (bitmap_bit_in_range_p): New function.
6656 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
6657 * tree-ssa-dse.c (live_bytes_read): New function.
6658 (dse_classify_store): Ignore reads of dead bytes.
6660 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
6661 typos and whitespace errors.
6662 * config/i386/predicates.md (address_no_seg_operand): Likewise.
6663 * config/s390/s390.c (s390_emit_prologue): Likewise.
6665 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6668 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
6669 with a symbol for LRA.
6671 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6673 PR rtl-optimization/82338
6674 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
6676 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
6678 * genmodes.c (calc_wider_mode): Suppress qsort macro.
6679 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
6680 (qsort_chk): Declare.
6681 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
6682 (qsort_chk): New function.
6684 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6686 PR tree-optimization/82337
6687 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
6688 phi definition if the PHI result appears in an abnormal PHI.
6689 (find_basis_for_base_expr): Don't record a basis if the LHS of the
6690 basis appears in an abnormal PHI.
6692 2017-09-29 Richard Biener <rguenther@suse.de>
6694 * graphite-isl-ast-to-gimple.c
6695 (translate_isl_ast_to_gimple::set_codegen_error): New function.
6696 (binary_op_to_tree): Use it.
6697 (get_rename_from_scev): Likewise.
6698 (copy_loop_phi_nodes): Likewise.
6699 (copy_bb_and_scalar_dependences): Likewise.
6700 (translate_pending_phi_nodes): Likewise.
6702 2017-09-29 Jakub Jelinek <jakub@redhat.com>
6705 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
6706 for movabsq $(i32 << shift), r64.
6708 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
6710 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
6711 index when encoding %esp as %rsp to avoid 0x67 prefix.
6713 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
6715 * config/i386/i386.md (*movsf_internal, *movdf_internal):
6716 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
6718 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
6720 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
6721 Extensions with more than 16 double VFP registers.
6722 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
6723 to_clear_mask and all code related to it. Replace the remaining
6724 entry by a sbitmap and adapt code accordingly.
6726 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
6728 * brig-builtins.def: Change pure attributes to const.
6730 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6732 * config.gcc (default_gnu_indirect_function): Default to yes for
6733 sparc*-*-linux* with glibc.
6735 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6737 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
6738 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
6739 when creating .init_array and .fini_array sections with priority
6742 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
6745 * config/aarch64/aarch64.c
6746 (aarch64_builtin_support_vector_misalignment): Always return false
6747 when misalignment is unknown.
6749 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
6751 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
6752 this function to return false if the definition used by the swap
6753 instruction is artificial, or if the memory address from which the
6754 constant value is loaded is not represented by a base address held
6755 in a register or if the base address register is a frame or stack
6756 pointer. Additionally, return false if the base address of the
6757 loaded constant is a SYMBOL_REF but is not considered to be a
6759 (replace_swapped_load_constant): New function.
6760 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
6761 loaded constant vector with a load of a swapped constant vector.
6763 2017-09-27 Carl Love <cel@us.ibm.com>
6765 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
6766 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
6767 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
6770 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
6772 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
6773 first, always call autopref_rank_data otherwise.
6775 2017-09-27 Richard Biener <rguenther@suse.de>
6777 * graphite-scop-detection.c (find_scop_parameters): Move
6778 loop bound handling ...
6779 (gather_bbs::before_dom_children): ... here, avoiding the need
6780 to build scop_info->loop_nest.
6781 (record_loop_in_sese): Remove.
6782 * sese.h (sese_info_t::loop_nest): Remove.
6783 * sese.c (new_sese_info): Do not allocate loop_nest.
6784 (free_sese_info): Do not free loop_nest.
6786 2017-09-27 Jakub Jelinek <jakub@redhat.com>
6789 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
6790 lhs from calls if the lhs has addressable type.
6792 2017-09-27 Richard Biener <rguenther@suse.de>
6794 * graphite.h (scop::max_alias_set): New member.
6795 * graphite-scop-detection.c: Remove references to non-existing
6796 --param in comments.
6797 (build_alias_sets): Record the maximum alias set used for drs.
6798 (build_scops): Support zero as unlimited for
6799 --param graphite-max-arrays-per-scop.
6800 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
6802 (build_poly_sr_1): ... here. Compute alias set based on the
6803 maximum alias set used for drs rather than
6804 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
6806 2017-09-27 Richard Biener <rguenther@suse.de>
6808 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
6809 --param loop-block-tile-size=0 to disable tiling.
6811 2017-09-27 Richard Biener <rguenther@suse.de>
6813 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
6814 (graphite-max-nb-scop-params): Document special value zero.
6815 * domwalk.h (dom_walker::STOP): New symbolical constant.
6816 (dom_walker::dom_walker): Add optional parameter for bb to
6818 (dom_walker::~dom_walker): Declare.
6819 (dom_walker::before_dom_children): Document STOP return value.
6820 (dom_walker::m_user_bb_to_rpo): New member.
6821 (dom_walker::m_bb_to_rpo): Likewise.
6822 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
6823 mapping here if not provided by the user.
6824 (dom_walker::~dom_walker): Free bb to RPO mapping if not
6825 provided by the user.
6826 (dom_walker::STOP): Define.
6827 (dom_walker::walk): Do not compute bb to RPO mapping here.
6828 Support STOP return value from before_dom_children to stop
6830 * graphite-optimize-isl.c (optimize_isl): If the schedule
6831 is the same still generate code if -fgraphite-identity
6832 or -floop-parallelize-all are given.
6833 * graphite-scop-detection.c: Include cfganal.h.
6834 (gather_bbs::gather_bbs): Get and pass through bb to RPO
6836 (gather_bbs::before_dom_children): Return STOP for BBs
6838 (build_scops): Compute bb to RPO mapping and pass it to
6839 the domwalk. Treat --param graphite-max-nb-scop-params=0
6840 as not limiting the number of params.
6841 * graphite.c (graphite_initialize): Remove limit on the
6842 number of basic-blocks in a function.
6843 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
6844 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
6845 default value of 10.
6847 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6849 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
6850 Adjust code to eliminate needing to do the shift right 32-bits
6851 operation after XSCVDPSPN.
6853 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6855 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
6856 ((X / Y) != 0 -> X >= Y): Likewise.
6858 2017-09-26 Carl Love <cel@us.ibm.com>
6860 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
6861 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
6862 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
6863 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
6864 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
6865 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
6866 definitions and overloading.
6867 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
6868 statement for P9V_BUILTIN_XST_LEN_R.
6869 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
6870 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
6871 define_expand and define_insn for the instructions and builtins.
6872 * doc/extend.texi: Update the built-in documentation file for the new
6874 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
6875 define_insn for the instructions
6877 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
6880 * gcc/config/netbsd-protos.h: New file.
6881 * gcc/config/netbsd.c: New file.
6882 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
6883 * gcc/config/t-netbsd: New file.
6884 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
6885 (tmake_file) Add t-netbsd.
6886 (extra_objs) Add netbsd.o.
6888 2017-09-26 Janus Weil <janus@gcc.gnu.org>
6892 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
6894 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6896 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
6897 sign extension from a vector register to a GPR by doing a 32-bit
6898 direct move and then an EXTSW.
6899 (extendsi<mode>2 splitter): Likewise.
6900 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
6901 right or vector extract after doing XSCVDPSPN. Use
6902 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
6904 (movdi_from_sf_zero_ext): Likewise.
6905 (reload_gpr_from_vsxsf): Likewise.
6906 (p8_mfvsrd_4_disf): Delete, no longer used.
6907 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
6908 then needing to move the SFmode to a GPR to use the XSCVDPSP
6909 instruction instead of FRSP and XSCVDPSPN.
6910 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
6911 it is adjacent to the other XSCVSPDP insns.
6912 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
6913 SFmode to be in traditional Altivec registers.
6914 (vsx_xscvdpspn): Eliminate useless alternative constraint.
6915 (vsx_xscvspdpn): Likewise.
6916 (vsx_xscvspdpn_scalar): Likewise.
6918 2017-09-26 Martin Jambor <mjambor@suse.cz>
6920 * tree-sra.c (compare_access_positions): Put integral types first,
6921 stabilize sorting of integral types, remove conditions putting
6922 non-full-precision integers last.
6923 (sort_and_splice_var_accesses): Disable scalarization if a
6924 non-integert would be represented by a non-full-precision integer.
6926 2017-09-26 Joseph Myers <joseph@codesourcery.com>
6928 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
6929 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
6930 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
6931 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
6932 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
6933 conditionals inside the function instead of around it. Call
6934 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
6935 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
6937 2017-09-26 Richard Biener <rguenther@suse.de>
6939 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
6941 (scop_detection::build_scop_breadth): ... this. Removed.
6942 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
6943 (scop_detection::harmful_stmt_in_bb): Likewise.
6944 (scop_detection::graphite_can_represent_stmt): Likewise.
6945 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
6946 (scop_detection::can_represent_loop): Remove recursion, fold in ...
6947 (scop_detection::can_represent_loop_1): ... this. Removed.
6948 (scop_detection::harmful_loop_in_region): Simplify after inlining
6949 the above and remove more quadraticness.
6950 (build_scops): Adjust.
6951 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
6954 2017-09-26 Jakub Jelinek <jakub@redhat.com>
6957 * config/i386/i386.c (ix86_print_operand_address_as): Only test
6958 REGNO (base) == SP_REG if base is a REG.
6961 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
6962 if it is different SSA_NAME.
6963 (optimize_range_tests_cmp_bitwise): New function.
6964 (optimize_range_tests): Call it.
6966 2017-09-26 Richard Biener <rguenther@suse.de>
6968 PR tree-optimization/82321
6969 * graphite.c (canonicalize_loop_closed_ssa): Properly check
6970 for the def being inside the loop.
6972 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6974 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
6976 * config/s390/s390-builtins.def: Fix constraint on op4.
6978 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6980 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
6981 independent expanders.
6982 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
6983 ("vec_ordered", "vec_unordered"): New expanders.
6985 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6987 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
6990 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6992 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
6993 vec_unpacks_lo_v16qi.
6994 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
6996 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6998 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
6999 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
7000 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
7002 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7004 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
7006 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
7008 ("*vec_slb<mode>"): New insn pattern.
7009 ("vec_shr_<mode>"): New expander.
7010 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
7011 and force the shift count operand to V16QImode.
7012 ("vec_srb<mode>"): Set shift count mode to V16QI.
7014 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7016 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7017 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7018 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7020 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7023 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7025 2017-09-26 Richard Biener <rguenther@suse.de>
7027 PR tree-optimization/82320
7028 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7031 2017-09-25 Jeff Law <law@redhat.com>
7033 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7034 prototype for new argument.
7035 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7036 mostly extracted from rs6000_emit_allocate_stack.
7037 (rs6000_emit_probe_stack_range_stack_clash): New function.
7038 (rs6000_emit_allocate_stack): Call
7039 rs6000_emit_probe_stack_range_stack_clash as needed.
7040 (rs6000_emit_probe_stack_range): Add additional argument
7041 to call to gen_probe_stack_range{si,di}.
7042 (output_probe_stack_range): New.
7043 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7044 (output_probe_stack_range_stack_clash): New.
7045 (rs6000_emit_prologue): Emit notes into dump file as requested.
7046 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7047 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7048 Add additional operand and pass it to output_probe_stack_range.
7050 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7052 PR tree-optimization/82163
7053 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7054 (checking_verify_loop_closed_ssa): New parameter.
7055 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7056 (check_loop_closed_ssa_stmt): Delete.
7057 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7058 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7059 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7062 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7064 * brig-builtins.def: Treat HSAIL barrier builtins as
7065 setjmp/longjump style functions.
7067 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7069 * target.def (constant_alignment): New hook.
7070 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7071 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7072 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7073 * doc/tm.texi: Regenerate.
7074 * targhooks.h (default_constant_alignment): Declare.
7075 (constant_alignment_word_strings): Likewise.
7076 * targhooks.c (default_constant_alignment): New function.
7077 (constant_alignment_word_strings): Likewise.
7078 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7079 instead of CONSTANT_ALIGNMENT.
7080 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7081 (force_const_mem): Likewise.
7082 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7083 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7084 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7085 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7086 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7088 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7089 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7090 constant_alignment_word_strings.
7091 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7092 (CONSTANT_ALIGNMENT): Likewise.
7093 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7094 (arm_constant_alignment): New function.
7095 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7096 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7097 constant_alignment_word_strings.
7098 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7099 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7100 constant_alignment_word_strings.
7101 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7102 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7103 (cris_constant_alignment): New function.
7104 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7105 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7106 (epiphany_constant_alignment): New function.
7107 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7108 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7109 constant_alignment_word_strings.
7110 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7111 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7112 constant_alignment_word_strings.
7113 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7114 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7115 constant_alignment_word_strings.
7116 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7117 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7118 * config/i386/i386.c (ix86_constant_alignment): Make static.
7119 Use the same interface as the target hook.
7120 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7121 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7122 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7123 constant_alignment_word_strings.
7124 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7125 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7126 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7127 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7128 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7129 constant_alignment_word_strings.
7130 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7131 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7132 constant_alignment_word_strings.
7133 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7134 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7135 constant_alignment_word_strings.
7136 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7137 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7139 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7140 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7141 * config/mips/mips.c (mips_constant_alignment): New function.
7142 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7143 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7144 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7145 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7146 (mmix_constant_alignment): Make static. Use the same interface
7148 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7149 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7150 constant_alignment_word_strings.
7151 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7152 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7153 constant_alignment_word_strings.
7154 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7155 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7156 constant_alignment_word_strings.
7157 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7158 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7159 (rs6000_constant_alignment): New function.
7160 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7161 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7162 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7163 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7164 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7165 (rs6000_constant_alignment): New function.
7166 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7167 * config/s390/s390.c (s390_constant_alignment): New function.
7168 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7169 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7170 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7171 constant_alignment_word_strings.
7172 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7173 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7174 (sparc_constant_alignment): New function.
7175 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7176 * config/spu/spu.c (spu_constant_alignment): New function.
7177 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7178 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7179 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7180 constant_alignment_word_strings.
7181 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7182 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7183 constant_alignment_word_strings.
7184 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7185 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7186 constant_alignment_word_strings.
7187 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7188 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7189 (visium_constant_alignment): New function.
7190 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7191 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7192 (xtensa_constant_alignment): New function.
7193 * system.h (CONSTANT_ALIGNMENT): Poison.
7195 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7197 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7198 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7199 (rs6000_builtin_valid_without_lhs): New helper function.
7200 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7201 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7203 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7205 * target.h (vec_perm_indices): Use unsigned short rather than
7207 (auto_vec_perm_indices): Likewise.
7208 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7209 Use unsigned int rather than unsigned char.
7210 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7212 2017-09-25 Richard Biener <rguenther@suse.de>
7214 * cfgloop.h (sort_sibling_loops): Declare.
7215 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7216 (sort_sibling_loops): New function sorting the sibling loop list
7218 * graphite.c (graphite_transform_loops): Sort sibling loops.
7220 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7222 * target.def (vec_perm_const_ok): Change sel parameter to
7224 * optabs-query.c (can_vec_perm_p): Update accordingly.
7225 * doc/tm.texi: Regenerate.
7226 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7227 auto_vec_perm_indices and remove separate nelt field.
7228 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7229 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7230 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7231 (aarch64_expand_vec_perm_const): Update accordingly.
7232 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7233 to vec_perm_indices.
7234 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7235 auto_vec_perm_indices and remove separate nelt field.
7236 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7237 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7238 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7240 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7241 to vec_perm_indices.
7242 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7243 sel to vec_perm_indices.
7244 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7245 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7246 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7248 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7251 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7254 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7255 on the FUNCTION_DECL function context if it has a DIE that is a
7258 2017-09-25 Richard Biener <rguenther@suse.de>
7260 PR tree-optimization/82285
7261 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7264 2017-09-25 Tom de Vries <tom@codesourcery.com>
7268 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7271 2017-09-25 Richard Biener <rguenther@suse.de>
7273 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7274 ISL errors other than isl_error_quota happen. Dump if the
7275 schedule is the same.
7276 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7277 errors instead of aborting inside ISL.
7279 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7282 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7283 of libgcc_eh for m64.
7284 * config/i386/darwin64.h: Likewise.
7286 2017-09-25 Richard Biener <rguenther@suse.de>
7289 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7290 attribute for incomplete types nor twice for complete ones.
7292 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7295 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7296 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7299 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7302 * config/i386/i386.opt (mprefer-avx256): Use
7303 ix86_target_flags variable.
7304 * config/i386/i386.c (ix86_target_string): Move
7305 -mprefer-avx256 to flag2_opts.
7307 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7310 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7311 and x != -1 | y != -1 into (x & y) != -1.
7313 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7315 * config.gcc: Add new case statement to set
7316 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7317 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7318 s390x-*-linux* case statements. Added aarch64 to the list of
7319 supported architectures.
7321 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7323 PR tree-optimization/82289
7324 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7325 STMT_VINFO_RELEVANT_P.
7327 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7328 Alan Hayward <alan.hayward@arm.com>
7329 David Sherwood <david.sherwood@arm.com>
7331 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7332 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7333 (extract_range_from_binary_expr_1): Don't call
7334 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7336 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7337 Alan Hayward <alan.hayward@arm.com>
7338 David Sherwood <david.sherwood@arm.com>
7340 * target.def (preferred_vector_alignment): New hook.
7341 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7343 * doc/tm.texi: Regenerate.
7344 * targhooks.h (default_preferred_vector_alignment): Declare.
7345 * targhooks.c (default_preferred_vector_alignment): New function.
7346 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7348 (DR_TARGET_ALIGNMENT): New macro.
7349 (aligned_access_p): Update commentary.
7350 (vect_known_alignment_in_bytes): New function.
7351 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7353 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7354 Calculate the misalignment based on the target alignment rather than
7356 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7357 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7358 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
7359 the target alignment, rather than masking the element misalignment
7360 with the number of elements in a vector. Also use the target
7361 alignment when calculating the maximum number of peels.
7362 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
7363 instead of TYPE_ALIGN_UNIT.
7364 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
7365 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
7366 (vect_create_addr_base_for_vector_ref): Update call accordingly.
7367 (vect_create_data_ref_ptr): Likewise.
7368 (vect_setup_realignment): Realign by ANDing with
7369 -DR_TARGET_MISALIGNMENT.
7370 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
7371 the number of peels based on DR_TARGET_ALIGNMENT.
7372 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
7373 with the guaranteed alignment boundary when deciding whether
7375 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
7376 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
7377 (ensure_base_align): Remove stmt_info parameter. Get the
7378 target base alignment from DR_TARGET_ALIGNMENT.
7379 (vectorizable_store): Update call accordingly. Interpret
7380 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7382 (vectorizable_load): Likewise.
7384 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7385 Alan Hayward <alan.hayward@arm.com>
7386 David Sherwood <david.sherwood@arm.com>
7388 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
7389 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
7390 (vect_enhance_data_refs_alignment): Likewise.
7392 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7394 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7395 error. Only quit immediately if parsing is complete.
7396 (BEGIN): Initialize fatal_err and parse_done.
7397 (begin fpu, end fpu): Check number of arguments.
7398 (begin arch, end arch): Likewise.
7399 (begin cpu, end cpu): Likewise.
7400 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7401 (optalias): Likewise.
7403 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7405 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7406 * config/arm/arm-isa.h: Delete. Move definitions to ...
7407 * arm-cpus.in: ... here. Use new feature and fgroup values.
7408 * config/arm/arm.c (arm_option_override): Use lower case for feature
7410 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7411 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7412 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7414 (print_isa_bits_for): New function.
7415 (gen_isa): New function.
7416 (gen_comm_data): Use print_isa_bits_for.
7417 (define feature): New keyword.
7418 (define fgroup): New keyword.
7419 * config/arm/t-arm (TM_H): Remove.
7420 (GTM_H): Add arm-isa.h.
7421 (arm-isa.h): Add rule to generate file.
7422 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7423 case for feature bit names.
7425 2017-09-22 Richard Biener <rguenther@suse.de>
7427 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
7429 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
7430 print of no dependency loops ...
7431 * graphite.c (graphite_transform_loops): ... here.
7432 (canonicalize_loop_closed_ssa_form): Work from inner to outer
7434 (same_close_phi_node, remove_duplicate_close_phi,
7435 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
7436 (canonicalize_loop_closed_ssa): ... here and simplify.
7437 * graphite-optimize-isl.c: Include tree-vectorizer.h.
7438 (optimize_isl): Use dump_printf_loc to tell when we stopped
7439 optimizing because of an ISL timeout.
7441 2017-09-22 Richard Biener <rguenther@suse.de>
7443 PR tree-optimization/82291
7444 * tree-if-conv.c (predicate_mem_writes): Make sure to
7445 remove writes in blocks predicated with false.
7447 2017-09-22 Richard Biener <rguenther@suse.de>
7449 * sese.c: Include cfganal.h.
7450 (if_region_set_false_region): Remove.
7451 (create_if_region_on_edge): Likewise.
7452 (move_sese_in_condition): Re-implement without destroying
7455 2017-09-22 Richard Biener <rguenther@suse.de>
7457 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
7458 Verify both BBs contain loop PHI nodes before dispatching to
7460 (graphite_regenerate_ast_isl): Do not recompute dominators,
7461 do not verify three times. Restructure for clarity.
7462 * graphite-scop-detection.c (same_close_phi_node,
7463 remove_duplicate_close_phi, make_close_phi_nodes_unique,
7464 defined_in_loop_p, canonicalize_loop_closed_ssa,
7465 canonicalize_loop_closed_ssa_form): Simplify, remove excess
7466 checking and SSA rewrite, move to ...
7467 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
7468 (graphite_initialize): Do not pass in ctx, do not reset the
7469 SCEV cache, compute only dominators.
7470 (graphite_transform_loops): Allocate ISL ctx after
7471 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
7472 Maintain post-dominators only around build_scops.
7473 * sese.c (if_region_set_false_region): Make static. Free
7474 and recompute dominators.
7475 (move_sese_in_condition): Assert we don't get called with
7476 post-dominators computed.
7477 * sese.h (if_region_set_false_region): Remove.
7479 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
7481 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
7482 mode attribute for TARGET_AVX512VL.
7484 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
7486 * config/i386/i386.opt (mprefer-avx256): New option.
7487 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
7489 (ix86_preferred_simd_mode): Return 256-bit AVX modes
7490 for TARGET_PREFER_AVX256.
7491 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
7493 2017-09-21 Jeff Law <law@redhat.com>
7495 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
7496 Fix dump output if the only stack space is for pushed registers.
7498 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7500 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
7503 2017-09-21 Martin Sebor <msebor@redhat.com>
7506 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
7507 code (in C++) or code that triggers warnings.
7509 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
7511 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
7513 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
7515 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
7516 * sched-rgn.c: Ditto.
7517 * sel-sched-ir.c: Ditto.
7519 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
7521 * toplev.h (set_random_seed): Adjust return type.
7522 * toplev.c (init_local_tick): Move eager initialization of random_seed
7523 to get_random_seed. Adjust comment.
7524 (init_random_seed): Inline to get_random_seed, delete.
7525 (get_random_seed): Initialize random_seed lazily.
7526 (set_random_seed): Do not return previous value.
7527 (print_switch_value): Do not call get_random_seed.
7529 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
7531 * cgraph.c (delete_function_version): New, broken out from...
7532 (cgraph_node::delete_function_version): ...here. Rename to
7533 cgraph_node::delete_function_version_by_decl. Update all uses.
7534 (cgraph_node::remove): Call delete_function_version.
7536 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7539 * tree-inline.c (expand_call_inline): Emit clobber stmts for
7540 VAR_DECLs to which addressable non-volatile parameters are mapped
7541 and for id->retvar after the return value assignment. Clear
7542 id->retval and id->retbnd after inlining.
7544 2017-09-21 Richard Biener <rguenther@suse.de>
7546 PR tree-optimization/82276
7547 PR tree-optimization/82244
7548 * tree-vrp.c (build_assert_expr_for): Set
7549 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
7551 (remove_range_assertions): Revert earlier change.
7553 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
7556 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
7558 2017-09-21 Richard Biener <rguenther@suse.de>
7560 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7561 Restore valid IL after code generation errors.
7562 * graphite.c (graphite_transform_loops): Diagnose code
7563 generation issues as MSG_MISSED_OPTIMIZATION and continue
7564 with processing SCOPs.
7566 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7567 Alan Hayward <alan.hayward@arm.com>
7568 David Sherwood <david.sherwood@arm.com>
7570 * calls.c (compute_argument_addresses): Use simplify_gen_binary
7571 rather than choosing between plus_constant and gen_rtx_<CODE>.
7572 * expr.c (emit_push_insn): Likewise.
7573 (expand_expr_real_2): Likewise.
7575 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7576 Alan Hayward <alan.hayward@arm.com>
7577 David Sherwood <david.sherwood@arm.com>
7579 * loop-unroll.c (split_iv): Call copy_rtx on the step.
7581 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7582 Alan Hayward <alan.hayward@arm.com>
7583 David Sherwood <david.sherwood@arm.com>
7585 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
7586 calling tree_to_uhwi.
7588 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7589 Alan Hayward <alan.hayward@arm.com>
7590 David Sherwood <david.sherwood@arm.com>
7592 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
7593 INTEGER_CST rather than a negative test for ADDR_EXPR.
7595 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7596 Alan Hayward <alan.hayward@arm.com>
7597 David Sherwood <david.sherwood@arm.com>
7599 * tree-vrp.c (extract_range_from_binary_expr_1): Check
7600 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
7602 2017-09-21 Richard Biener <rguenther@suse.de>
7604 PR tree-optimization/71351
7605 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
7606 graphite_create_new_loop_guard): Remove, fold remaining parts
7608 (translate_isl_ast_node_for): ... here and simplify.
7610 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7613 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
7614 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
7615 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
7616 alternative always use QI mode, for -Os imov (=R,R) alternative
7617 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
7620 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7621 Jeff Law <law@redhat.com>
7623 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
7624 (allocate_stack_space): New function, partially extracted from
7626 (s390_emit_prologue): Track offset to most recent stack probe.
7627 Code to allocate space moved into allocate_stack_space.
7628 Dump actions when no stack is allocated.
7629 (s390_prologue_plus_offset): New function.
7630 (s390_emit_stack_probe): Likewise.
7632 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
7634 * common.opt (Wa, Wl, Wp, g, gz=): Add
7636 (gno-column-info): Remove.
7637 (gcolumn-info): Drop RejectNegative.
7639 (gno-record-gcc-switches): Remove.
7640 (grecord-gcc-switches): Drop RejectNegative.
7641 (gno-split-dwarf): Remove.
7642 (gsplit-dwarf): Drop RejectNegative.
7643 (gno-strict-dwarf): Remove.
7644 (gstrict-dwarf): Drop RejectNegative.
7645 * config/darwin.opt (gfull, gused): Add RejectNegative.
7646 * dwarf2out.c (gen_producer_string): Drop
7647 gno-record-gcc-switches handler.
7648 * optc-gen.awk: Add g to prefixes with negative forms.
7649 * opts-common.c (remapping_prefix_p): New.
7650 (find_opt): Check it.
7651 (generate_canonical_option): Test g prefix.
7652 (option_map): Add -gno- mapping.
7653 (add_misspelling_candidates): Check remapping_prefix_p.
7655 2017-09-20 Jeff Law <law@redhat.com>
7657 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
7658 thinko in stack clash protection support.
7660 * explow.c (compute_stack_clash_protection_loop_data): Use
7661 CONST_INT_P instead of explicit test. Verify object is a
7662 CONST_INT_P before looking at INTVAL.
7663 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
7664 instead of explicit test.
7666 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7669 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
7670 address instead of to r1 and r11.
7672 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
7674 * config.gcc: Support "knm".
7675 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
7676 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7678 * config/i386/i386.c (m_KNM): Define.
7679 (processor_target_table): Add "knm".
7681 (ix86_option_override_internal): Add "knm".
7682 (ix86_issue_rate): Add PROCESSOR_KNM.
7683 (ix86_adjust_cost): Ditto.
7684 (ia32_multipass_dfa_lookahead): Ditto.
7685 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
7686 (fold_builtin_cpu): Add M_INTEL_KNM.
7687 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
7688 (processor_type): Add PROCESSOR_KNM.
7689 * config/i386/x86-tune.def: Add m_KNM.
7690 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
7692 2017-09-20 Richard Biener <rguenther@suse.de>
7694 PR tree-optimization/80213
7695 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
7696 are allowed in empty BBs as well.
7697 (canonicalize_loop_closed_ssa): Also look for other complex
7699 (scop_detection::get_sese): Include the loop-closed PHI block
7701 (scop_detection::merge_sese): Remove code adding extra blocks.
7702 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
7703 (build_scops): Assert the final returned scop is invalid.
7705 2017-09-20 Richard Biener <rguenther@suse.de>
7707 PR tree-optimization/82264
7708 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
7710 (vn_phi_lookup): Likewise.
7711 (vn_phi_insert): Likewise.
7713 2017-09-20 Jakub Jelinek <jakub@redhat.com>
7715 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
7716 that fits into uhwi or shwi, add DW_AT_const_value regardless
7717 of early_dwarf without going through RTL, using add_AT_unsigned
7720 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
7721 (DEBUG_LTO_ABBREV_SECTION): Likewise.
7722 (DEBUG_LTO_MACINFO_SECTION): Likewise.
7723 (DEBUG_MACRO_SECTION): Likewise.
7724 (DEBUG_LTO_MACRO_SECTION): Likewise.
7725 (DEBUG_STR_DWO_SECTION): Likewise.
7726 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
7727 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
7728 (DEBUG_LTO_DWO_LINE_SECTION): Define.
7729 (DEBUG_LTO_LINE_STR_SECTION): Define.
7730 (init_sections_and_labels): Initialize debug_line_str_section
7731 variable. Initialize debug_loc_section for -gdwarf-5 to
7732 DEBUG_LOCLISTS_SECTION. Formatting fixes.
7734 2017-09-20 Richard Biener <rguenther@suse.de>
7736 * graphite-sese-to-poly.c (extract_affine): Properly handle
7737 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
7739 2017-09-20 Richard Biener <rguenther@suse.de>
7741 PR tree-optimization/81373
7742 * graphite-scop-detection.c (build_cross_bb_scalars_def):
7743 Force SESE live-out defs to be handled even if they are
7746 2017-09-19 Jeff Law <law@redhat.com>
7748 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
7749 nothing for stack adjustments with REG_STACK_CHECK.
7750 * sched-deps.c (parse_add_or_inc): Reject insns with
7751 REG_STACK_CHECK from dependency breaking.
7752 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
7753 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
7754 * reg-notes.def (STACK_CHECK): New note.
7756 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
7757 (ix86_expand_prologue): Dump stack clash info as needed.
7758 Call ix86_adjust_stack_and_probe_stack_clash as needed.
7760 * function.c (dump_stack_clash_frame_info): New function.
7761 * function.h (dump_stack_clash_frame_info): Prototype.
7762 (enum stack_clash_probes): New enum.
7764 * config/alpha/alpha.c (alpha_expand_prologue): Also check
7765 flag_stack_clash_protection.
7766 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
7767 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
7768 (arm_frame_pointer_required): Likewise.
7769 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
7770 (ia64_expand_prologue): Likewise.
7771 * config/mips/mips.c (mips_expand_prologue): Likewise.
7772 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
7773 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7774 (sparc_flat_expand_prologue): Likewise.
7775 * config/spu/spu.c (spu_expand_prologue): Likewise.
7777 * explow.c: Include "params.h".
7778 (anti_adjust_stack_and_probe_stack_clash): New function.
7779 (get_stack_check_protect): Likewise.
7780 (compute_stack_clash_protection_loop_data): Likewise.
7781 (emit_stack_clash_protection_loop_start): Likewise.
7782 (emit_stack_clash_protection_loop_end): Likewise.
7783 (allocate_dynamic_stack_space): Use get_stack_check_protect.
7784 Use anti_adjust_stack_and_probe_stack_clash.
7785 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
7786 (emit_stack_clash_protection_loop_start): Likewise.
7787 (emit_stack_clash_protection_loop_end): Likewise.
7788 * rtl.h (get_stack_check_protect): Prototype.
7789 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
7790 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
7791 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
7793 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
7795 * doc/tm.texi: Rebuilt.
7796 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
7797 get_stack_check_protect.
7798 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
7799 * config/arm/arm.c (arm_expand_prologue): Likewise.
7800 (arm_frame_pointer_required): Likewise.
7801 * config/i386/i386.c (ix86_expand_prologue): Likewise.
7802 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
7803 * config/mips/mips.c (mips_expand_prologue): Likewise.
7804 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
7805 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
7806 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7807 (sparc_flat_expand_prologue): Likewise.
7809 * common.opt (-fstack-clash-protection): New option.
7810 * flag-types.h (enum stack_check_type): Note difference between
7811 -fstack-check= and -fstack-clash-protection.
7812 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
7813 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
7814 * toplev.c (process_options): Issue warnings/errors for cases
7815 not handled with -fstack-clash-protection.
7816 * doc/invoke.texi (-fstack-clash-protection): Document new option.
7817 (-fstack-check): Note additional problem with -fstack-check=generic.
7818 Note that -fstack-check is primarily for Ada and refer users
7819 to -fstack-clash-protection for stack-clash-protection.
7820 Document new params for stack clash protection.
7822 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
7824 * config/i386/i386.c (ix86_split_long_move): Do not handle
7825 address used for LEA in a special way.
7827 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
7829 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
7831 2017-09-19 Martin Sebor <msebor@redhat.com>
7834 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
7835 of incompatible types.
7837 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
7839 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7840 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
7841 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7842 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
7844 2017-09-19 Richard Biener <rguenther@suse.de>
7846 PR tree-optimization/82244
7847 * tree-vrp.c (remove_range_assertions): Do not propagate
7848 a constant to abnormals but replace the assert with a copy.
7850 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
7852 PR rtl-optimization/57878
7853 PR rtl-optimization/68988
7854 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
7855 avoidance test involving non_reload_pseudos. Move frequency test
7856 below the general fragmentation avoidance test.
7858 2017-09-19 Richard Biener <rguenther@suse.de>
7860 PR tree-optimization/69728
7861 * graphite-sese-to-poly.c (schedule_error): New global.
7862 (add_loop_schedule): Handle empty domain by failing the
7864 (build_original_schedule): Handle schedule_error.
7866 2017-09-19 Richard Biener <rguenther@suse.de>
7868 * graphite-scop-detection.c (scop_detection::can_represent_loop):
7869 Do not iterate to sibling loops but only to siblings of inner
7872 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
7875 * config/m68k/m68k.md (moveq feeding equality comparison): Check
7876 that the registers are different.
7878 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
7880 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
7881 to processor_model and "amdfam17h" to arch_names_table.
7882 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
7884 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7887 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
7889 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7890 Alan Hayward <alan.hayward@arm.com>
7891 David Sherwood <david.sherwood@arm.com>
7893 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
7895 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
7897 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
7898 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
7899 vect_schedule_slp_instance.
7900 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
7901 Update call to vect_slp_analyze_node_operations. Simplify return
7903 (vect_slp_analyze_bb_1): Update call accordingly.
7904 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
7905 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
7906 (vect_schedule_slp): Update call accordingly.
7908 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7909 Alan Hayward <alan.hayward@arm.com>
7910 David Sherwood <david.sherwood@arm.com>
7912 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
7913 with types that aren't in fact scalar.
7915 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7917 * tree-vect-slp.c (vect_record_max_nunits): New function,
7919 (vect_build_slp_tree_1): ...here.
7920 (vect_build_slp_tree_2): Call it for phis too.
7922 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7924 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
7925 to vect_get_vec_def_for_operand when getting the mask operand.
7927 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7928 Alan Hayward <alan.hayward@arm.com>
7929 David Sherwood <david.sherwood@arm.com>
7931 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
7934 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7935 Alan Hayward <alan.hayward@arm.com>
7936 David Sherwood <david.sherwood@arm.com>
7938 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
7939 calculation for vector booleans.
7941 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7942 Alan Hayward <alan.hayward@arm.com>
7943 David Sherwood <david.sherwood@arm.com>
7945 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
7947 (vect_transform_stmt): ...here.
7948 (vect_analyze_stmt): Use it instead of calling
7949 vectorizable_live_operation directly.
7951 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
7953 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
7954 non-SIMT targets in acc vector loops.
7956 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
7958 * configure.ac: Add arc and check if assembler supports gdwarf2.
7959 * configure: Regenerate.
7961 2017-09-18 Richard Biener <rguenther@suse.de>
7963 PR tree-optimization/82220
7964 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
7965 epilogue niters from the min_profitable_iters compute.
7967 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7970 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
7971 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
7972 (ix86_init_pic_reg): Revert 2017-09-01 changes.
7974 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
7977 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
7978 switching to a new text section.
7980 2017-09-18 Richard Biener <rguenther@suse.de>
7982 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
7984 (build_alias_set): Reject aliases with no access function.
7986 2017-09-18 Richard Biener <rguenther@suse.de>
7988 PR tree-optimization/79622
7989 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
7991 (build_cross_bb_scalars_use): Likewise.
7993 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
7995 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
7997 2017-09-18 Alan Modra <amodra@gmail.com>
8000 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
8001 stack_pointer_rtx for count 0. Update comments. Break up
8002 large rtl expression.
8004 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
8006 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
8007 Increase to 20 bytes.
8008 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
8009 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
8010 or avx version of the stub.
8012 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
8015 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8016 compute the minimum stack alignment. Also update preferred stack
8017 boundary for leaf functions.
8019 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8021 PR tree-optimization/82228
8022 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8025 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8027 * common/config/nds32/nds32-common.c
8028 (nds32_option_optimization_table): Refine formatting.
8029 (nds32_option_optimization_table): Use -fsched-pressure and
8030 -fomit-frame-pointer for specific optimization level.
8032 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8034 * config/nds32/nds32.c: Refine formatting and comments.
8035 * config/nds32/nds32.h: Likewise.
8036 * config/nds32/nds32.md: Likewise.
8037 * config/nds32/nds32-cost.c: Likewise.
8038 * config/nds32/nds32-isr.c: Likewise.
8039 * config/nds32/nds32-md-auxiliary.c: Likewise.
8040 * config/nds32/nds32-multiple.md: Likewise.
8041 * config/nds32/nds32-predicates.c: Likewise.
8043 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8044 Jakub Jelinek <jakub@redhat.com>
8046 Add support for -std=c++2a.
8047 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8049 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8051 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8054 * doc/extend.texi (Common Function Attributes): Add
8055 references to ARM, AArch64, and S/390 specific attributes.
8056 (Function Specific Option Pragmas): Add AArch64 and S/390
8057 to list of back ends that support the target pragma.
8059 2017-09-15 Nathan Sidwell <nathan@acm.org>
8061 * doc/standards.texi: Fix C++17 description. Update URLs for
8064 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8066 * common.opt (Wcast-align=strict): New warning option.
8067 * doc/invoke.texi: Document -Wcast-align=strict.
8069 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8071 * cgraph.h (cgraph_thunk_info): Add comments.
8072 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8073 assert for VIRTUAL_* arguments stricter.
8075 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8077 PR tree-optimization/71026
8078 * match.pd: Move RDIV patterns from fold-const.c
8079 * fold-const.c (distribute_real_division): Removed.
8080 (fold_binary_loc): Remove calls to distribute_real_divison.
8082 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8084 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8085 c++1z and gnu++1z as deprecated. Change other references to
8086 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8087 Change -Wc++1z-compat to -Wc++17-compat.
8088 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8089 * dwarf2out.c (highest_c_language): Handle C++17.
8090 (gen_compile_unit_die): Likewise.
8092 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8094 PR rtl-optimization/82192
8095 * combine.c (make_extraction): Don't look through non-paradoxical
8096 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8099 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8100 Alan Hayward <alan.hayward@arm.com>
8101 David Sherwood <david.sherwood@arm.com>
8103 * target.def (function_arg_offset): New hook.
8104 * targhooks.h (default_function_arg_offset): Declare.
8105 * targhooks.c (default_function_arg_offset): New function.
8106 * function.c (locate_and_pad_parm): Use
8107 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8108 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8109 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8110 * doc/tm.texi: Regenerate.
8111 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8112 * config/spu/spu.c (spu_function_arg_offset): New function.
8113 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8114 * system.h (FUNCTION_ARG_OFFSET): Poison.
8116 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8117 Alan Hayard <alan.hayward@arm.com>
8118 David Sherwood <david.sherwood@arm.com>
8120 * target.def (truly_noop_truncation): New hook.
8121 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8122 than TRULY_NOOP_TRUNCATION.
8123 * hooks.h (hook_bool_uint_uint_true): Declare.
8124 * hooks.c (hook_bool_uint_uint_true): New function.
8125 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8126 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8127 * doc/tm.texi: Regenerate.
8128 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8129 rather than TRULY_NOOP_TRUNCATION in comments.
8130 (simplify_comparison): Likewise.
8131 (record_truncated_value): Likewise.
8132 * expmed.c (extract_bit_field_1): Likewise.
8133 (extract_split_bit_field): Likewise.
8134 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8135 instead of TRULY_NOOP_TRUNCATION.
8136 * function.c (assign_parm_setup_block): Likewise.
8137 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8138 * rtlhooks.c: Include target.h.
8139 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8140 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8141 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8142 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8143 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8144 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8145 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8146 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8147 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8148 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8149 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8150 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8151 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8152 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8153 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8154 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8155 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8156 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8157 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8158 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8159 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8160 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8161 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8162 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8163 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8164 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8165 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8166 rather than TRULY_NOOP_TRUNCATION in comments.
8167 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8168 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8169 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8170 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8171 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8172 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8173 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8174 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8175 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8176 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8177 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8178 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8179 rather than TRULY_NOOP_TRUNCATION in comments.
8180 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8181 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8182 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8183 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8184 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8185 TRULY_NOOP_TRUNCATION condition.
8186 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8187 (TRULY_NOOP_TRUNCATION): Delete.
8188 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8189 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8190 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8191 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8192 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8193 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8194 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8195 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8196 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8197 rather than TRULY_NOOP_TRUNCATION in comments.
8198 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8199 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8200 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8201 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8202 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8203 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8205 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8208 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8209 (*cmp_ior): Likewise.
8210 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8211 (*ior_scc_scc_cmp): Likewise.
8212 (*and_scc_scc): Likewise.
8213 (*and_scc_scc_cmp): Likewise.
8215 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8216 Alan Hayard <alan.hayward@arm.com>
8217 David Sherwood <david.sherwood@arm.com>
8219 * target.def (can_change_mode_class): New hook.
8220 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8221 (hard_regno_nregs): Likewise.
8222 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8223 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8224 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8225 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8226 (LOAD_EXTEND_OP): Update accordingly.
8227 * doc/tm.texi: Regenerate.
8228 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8229 CANNOT_CHANGE_MODE_CLASS.
8230 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8231 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8232 * combine.c (simplify_set): Update accordingly.
8233 * emit-rtl.c (validate_subreg): Likewise.
8234 * recog.c (general_operand): Likewise.
8235 * regcprop.c (mode_change_ok): Likewise.
8236 * reload1.c (choose_reload_regs): Likewise.
8237 (inherit_piecemeal_p): Likewise.
8238 * rtlanal.c (simplify_subreg_regno): Likewise.
8239 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8240 instead of CANNOT_CHANGE_MODE_CLASS.
8241 (reload_cse_simplify_operands): Likewise.
8242 * reload.c (push_reload): Use targetm.can_change_mode_class
8243 instead of CANNOT_CHANGE_MODE_CLASS.
8244 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8245 REG_CANNOT_CHANGE_MODE_P.
8246 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8247 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8248 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8249 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8250 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8251 (arm_can_change_mode_class): New function.
8252 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8253 than CANNOT_CHANGE_MODE_CLASS in comments.
8254 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8255 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8256 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8257 (ix86_can_change_mode_class): ...this new function, inverting the
8258 sense of the return value.
8259 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8260 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8261 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8262 (ia64_can_change_mode_class): New function.
8263 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8264 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8265 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8266 (m32c_can_change_mode_class): ...this new function, inverting the
8267 sense of the return value.
8268 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8269 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8270 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8271 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8272 (mips_can_change_mode_class): ...this new function, inverting the
8273 sense of the return value.
8274 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8275 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8276 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8277 (msp430_can_change_mode_class): New function.
8278 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8279 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8280 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8281 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8282 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8283 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8284 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8285 (pa_cannot_change_mode_class): Replace with...
8286 (pa_can_change_mode_class): ...this new function, inverting the
8287 sense of the return value.
8288 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8289 than CANNOT_CHANGE_MODE_CLASS in comments.
8290 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8291 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8292 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8293 (pdp11_cannot_change_mode_class): Replace with...
8294 (pdp11_can_change_mode_class): ...this new function, inverting the
8295 sense of the return value.
8296 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8297 * config/powerpcspe/powerpcspe-protos.h
8298 (rs6000_cannot_change_mode_class_ptr): Delete.
8299 * config/powerpcspe/powerpcspe.c
8300 (rs6000_cannot_change_mode_class_ptr): Delete.
8301 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8302 (rs6000_option_override_internal): Assign to
8303 targetm.can_change_mode_class instead of
8304 rs6000_cannot_change_mode_class_ptr.
8305 (rs6000_cannot_change_mode_class): Replace with...
8306 (rs6000_can_change_mode_class): ...this new function, inverting the
8307 sense of the return value.
8308 (rs6000_debug_cannot_change_mode_class): Replace with...
8309 (rs6000_debug_can_change_mode_class): ...this new function.
8310 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8311 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8312 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8313 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8314 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8316 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8317 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8318 (rs6000_option_override_internal): Assign to
8319 targetm.can_change_mode_class instead of
8320 rs6000_cannot_change_mode_class_ptr.
8321 (rs6000_cannot_change_mode_class): Replace with...
8322 (rs6000_can_change_mode_class): ...this new function, inverting the
8323 sense of the return value.
8324 (rs6000_debug_cannot_change_mode_class): Replace with...
8325 (rs6000_debug_can_change_mode_class): ...this new function.
8326 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8327 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8328 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8329 (s390_can_change_mode_class): ...this new function, inverting the
8330 sense of the return value.
8331 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8332 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8333 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8334 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8335 (sh_cannot_change_mode_class): Replace with...
8336 (sh_can_change_mode_class): ...this new function, inverting the
8337 sense of the return value.
8338 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8339 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8340 (sparc_can_change_mode_class): New function.
8341 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8342 * config/spu/spu.c (spu_can_change_mode_class): New function.
8343 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8344 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8345 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8346 (visium_can_change_mode_class): New function.
8347 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8349 2017-09-15 Richard Biener <rguenther@suse.de>
8351 PR tree-optimization/82217
8352 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8353 but not undefined case.
8355 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8358 * postreload.c (reload_cse_simplify_operands): Skip
8359 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
8361 2017-09-15 Richard Biener <rguenther@suse.de>
8363 PR tree-optimization/68823
8364 * graphite-scop-detection.c (build_alias_set): If we have a
8365 possible dependence check whether we can handle them by just
8366 looking at the DRs DR_ACCESS_FNs.
8367 (build_scops): If build_alias_set fails, fail the SCOP.
8369 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
8371 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
8372 to support float128 built-in functions that require the ISA 3.0
8374 (BU_FLOAT128_3_HW): Likewise.
8375 (SQRTF128): Add support for the IEEE 128-bit square root and fma
8377 (FMAF128): Likewise.
8379 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8380 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8381 floating point instructions.
8382 (rs6000_invalid_builtin): Likewise.
8383 (rs6000_builtin_mask_names): Likewise.
8384 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8385 (RS6000_BTM_FLOAT128_HW): Likewise.
8386 (RS6000_BTM_COMMON): Likewise.
8387 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
8389 * doc/extend.texi (RS/6000 built-in functions): Document the
8390 IEEE 128-bit floating point square root and fused multiply-add
8393 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
8395 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
8396 reg (r2) isn't in the set of registers defined in the prologue.
8398 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8399 Alan Hayward <alan.hayward@arm.com>
8400 David Sherwood <david.sherwood@arm.com>
8402 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
8403 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
8404 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
8405 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
8406 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
8408 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
8409 max_vectorization_factor.
8410 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
8412 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8413 Alan Hayward <alan.hayward@arm.com>
8414 David Sherwood <david.sherwood@arm.com>
8416 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
8417 (vect_worthwhile_without_simd_p): Declare.
8418 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
8419 (vectorizable_reduction): Use it.
8420 * tree-vect-stmts.c (vectorizable_shift): Likewise.
8421 (vectorizable_operation): Likewise.
8423 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8424 Alan Hayward <alan.hayward@arm.com>
8425 David Sherwood <david.sherwood@arm.com>
8427 * tree-vectorizer.h (vect_get_num_copies): New function.
8428 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
8429 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8430 (vectorizable_induction): Likewise.
8431 (vectorizable_live_operation): Likewise.
8432 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8433 (vectorizable_bswap): Likewise.
8434 (vectorizable_call): Likewise.
8435 (vectorizable_conversion): Likewise.
8436 (vectorizable_assignment): Likewise.
8437 (vectorizable_shift): Likewise.
8438 (vectorizable_operation): Likewise.
8439 (vectorizable_store): Likewise.
8440 (vectorizable_load): Likewise.
8441 (vectorizable_condition): Likewise.
8442 (vectorizable_comparison): Likewise.
8443 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
8445 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8446 Alan Hayward <alan.hayward@arm.com>
8447 David Sherwood <david.sherwood@arm.com>
8449 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
8450 of vect_init_vector.
8452 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8453 Alan Hayward <alan.hayward@arm.com>
8454 David Sherwood <david.sherwood@arm.com>
8456 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
8457 an inline wrapper that provides a location.
8458 (gimple_build_vector): Likewise.
8459 * gimple-fold.c (gimple_build_vector_from_val): New function.
8460 (gimple_build_vector): Likewise.
8461 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
8462 functions to build the initial value. Always return a gimple value.
8463 (get_initial_defs_for_reduction): Likewise. Only compute
8465 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
8466 vect_init_vector on the results from get_initial_def(s)_for_reduction.
8467 (vectorizable_induction): Use gimple_build_vector rather than
8470 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8471 Alan Hayward <alan.hayward@arm.com>
8472 David Sherwood <david.sherwood@arm.com>
8474 * target.h (vec_perm_indices): New typedef.
8475 (auto_vec_perm_indices): Likewise.
8476 * optabs-query.h: Include target.h
8477 (can_vec_perm_p): Take a vec_perm_indices *.
8478 * optabs-query.c (can_vec_perm_p): Likewise.
8479 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
8480 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8481 * tree-vect-generic.c (lower_vec_perm): Likewise.
8482 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
8483 (vect_grouped_load_supported): Likewise.
8484 (vect_shift_permute_load_chain): Likewise.
8485 (vect_permute_store_chain): Use auto_vec_perm_indices.
8486 (vect_permute_load_chain): Likewise.
8487 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
8488 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
8489 Update uses of can_vec_perm_p.
8490 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
8491 mode with a number of elements. Take a vec_perm_indices *.
8492 (vect_create_epilog_for_reduction): Update accordingly.
8493 Use auto_vec_perm_indices.
8494 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
8495 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
8496 (vect_transform_slp_perm_load): Likewise.
8497 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
8498 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
8499 (vect_gen_perm_mask_checked): Likewise.
8500 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
8501 (vect_gen_perm_mask_checked): Likewise.
8502 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
8503 (vectorizable_store): Likewise.
8504 (vectorizable_load): Likewise.
8505 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
8506 (vectorizable_bswap): Likewise.
8508 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8509 Alan Hayward <alan.hayward@arm.com>
8510 David Sherwood <david.sherwood@arm.com>
8512 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
8513 * tree.c (build_vector): Likewise.
8514 (build_vector_from_ctor): Update accordingly.
8515 (build_vector_from_val): Likewise.
8516 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8517 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8518 * tree-vect-generic.c (add_rshift): Likewise.
8519 (expand_vector_divmod): Likewise.
8520 (optimize_vector_constructor): Likewise.
8521 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
8522 (vect_transform_slp_perm_load): Likewise.
8523 (vect_schedule_slp_instance): Likewise.
8524 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
8525 (vectorizable_call): Likewise.
8526 (vect_gen_perm_mask_any): Likewise. Add elements in order.
8527 * expmed.c (make_tree): Likewise.
8528 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
8529 a vector passed to build_vector.
8530 (fold_convert_const): Likewise.
8531 (exact_inverse): Likewise.
8532 (fold_ternary_loc): Likewise.
8533 (fold_relational_const): Likewise.
8534 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
8535 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
8536 (const_unop): Likewise. Store the reduction accumulator in a
8537 variable rather than an array.
8538 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
8539 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
8540 the new vector, rather than constructing it after the input arrays.
8541 (native_interpret_vector): Use auto_vec<tree> when building
8542 a vector passed to build_vector. Add elements in order.
8543 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
8544 auto_vec<tree> when building a vector passed to build_vector.
8545 (vect_create_epilog_for_reduction): Likewise.
8546 (vectorizable_induction): Likewise.
8547 (get_initial_def_for_reduction): Likewise. Fix indentation of
8549 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
8551 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
8552 passed to build_vector.
8554 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8555 Alan Hayward <alan.hayward@arm.com>
8556 David Sherwood <david.sherwood@arm.com>
8558 * tree-core.h (tree_base::u): Add an "nelts" field.
8559 (tree_vector): Use VECTOR_CST_NELTS as the length.
8560 * tree.c (tree_size): Likewise.
8561 (make_vector): Initialize VECTOR_CST_NELTS.
8562 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
8563 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
8564 TYPE_VECTOR_SUBPARTS.
8565 * expr.c (const_vector_mask_from_tree): Consistently use "units"
8566 as the number of units, setting it from VECTOR_CST_NELTS.
8567 (const_vector_from_tree): Likewise.
8568 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
8569 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
8570 (fold_negate_expr_1): Likewise.
8571 (fold_convert_const): Likewise.
8572 (const_binop): Likewise. Differentiate the number of output and
8574 (const_unop): Likewise.
8575 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
8576 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
8577 in cases that did the opposite.
8579 2017-09-14 Richard Biener <rguenther@suse.de>
8581 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
8584 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
8586 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
8588 2017-09-14 Jakub Jelinek <jakub@redhat.com>
8591 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
8592 if and where to split a bb, except for splitting before debug insn
8593 sequences followed by non-label real insn. Delete debug insns
8594 in between basic blocks.
8596 * combine.c (make_compound_operation_int): Formatting fixes.
8598 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
8599 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8600 * config/netbsd.h (LINK_EH_SPEC): Likewise.
8601 * config/sol2.h (LINK_EH_SPEC): Likewise.
8602 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8603 * config/s390/linux.h (LINK_SPEC): Likewise.
8604 * config/freebsd.h (LINK_EH_SPEC): Likewise.
8605 * config/openbsd.h (LINK_EH_SPEC): Likewise.
8606 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8607 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
8608 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
8609 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8610 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
8611 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
8613 2017-09-13 Jakub Jelinek <jakub@redhat.com>
8615 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
8617 (ENDFILE_LINUX_SPEC): Likewise.
8618 (LINK_EH_SPEC): Likewise.
8619 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
8620 (LINK_OS_LINUX_SPEC32): Likewise.
8621 (LINK_OS_LINUX_SPEC64): Likewise.
8622 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
8623 (LINK_OS_LINUX_SPEC): Likewise.
8625 2017-09-13 Martin Liska <mliska@suse.cz>
8628 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
8629 CASE_HIGH is NULL_TREE.
8631 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8632 Alan Hayward <alan.hayward@arm.com>
8633 David Sherwood <david.sherwood@arm.com>
8635 * target.def (secondary_memory_needed): New hook.
8636 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
8637 instead of SECONDARY_MEMORY_NEEDED.
8638 (secondary_memory_needed_mode): Likewise.
8639 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
8640 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
8641 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
8642 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
8643 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
8644 * doc/tm.texi: Regenerate.
8645 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
8646 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
8647 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8648 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
8649 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
8650 * config/i386/i386.c (inline_secondary_memory_needed): Put the
8651 mode argument first and change the reg_class arguments to reg_class_t.
8652 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
8653 Make static. Update the call to inline_secondary_memory_needed.
8654 (ix86_register_move_cost): Update the call to
8655 inline_secondary_memory_needed.
8656 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8657 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
8659 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
8660 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8662 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
8663 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
8664 * config/mips/mips.c (mips_secondary_memory_needed): Make static
8665 and match hook interface. Add comment from mips.h.
8666 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8667 * config/mmix/mmix.md (truncdfsf2): Refer to
8668 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8670 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
8671 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
8672 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8673 (pa_secondary_memory_needed): New function.
8674 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
8675 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
8676 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8677 (pdp11_secondary_memory_needed): Make static and match hook interface.
8678 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
8679 * config/powerpcspe/powerpcspe-protos.h
8680 (rs6000_secondary_memory_needed_ptr): Delete.
8681 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
8683 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8684 (rs6000_option_override_internal): Assign to
8685 targetm.secondary_memory_needed rather than
8686 rs6000_secondary_memory_needed_ptr.
8687 (rs6000_secondary_memory_needed): Match hook interface.
8688 (rs6000_debug_secondary_memory_needed): Likewise.
8689 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
8690 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
8691 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
8692 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8693 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
8694 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
8696 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
8697 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8698 (rs6000_option_override_internal): Assign to
8699 targetm.secondary_memory_needed rather than
8700 rs6000_secondary_memory_needed_ptr.
8701 (rs6000_secondary_memory_needed): Match hook interface.
8702 (rs6000_debug_secondary_memory_needed): Likewise.
8703 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
8704 * config/s390/s390.c (s390_secondary_memory_needed): New function.
8705 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8706 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
8707 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8708 (sparc_secondary_memory_needed): New function.
8709 * lra-constraints.c (check_and_process_move): Refer to
8710 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8712 (curr_insn_transform): Likewise.
8713 (process_alt_operands): Use targetm.secondary_memory_needed
8714 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8715 (check_secondary_memory_needed_p): Likewise.
8716 (choose_split_class): Likewise.
8717 * reload.c: Unconditionally include code that was previously
8718 conditional on SECONDARY_MEMORY_NEEDED.
8719 (push_secondary_reload): Use targetm.secondary_memory_needed
8720 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8721 (push_reload): Likewise.
8722 * reload1.c: Unconditionally include code that was previously
8723 conditional on SECONDARY_MEMORY_NEEDED.
8724 (choose_reload_regs): Use targetm.secondary_memory_needed
8725 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8726 (gen_reload): Likewise.
8727 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
8729 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8730 Alan Hayward <alan.hayward@arm.com>
8731 David Sherwood <david.sherwood@arm.com>
8733 * target.def (secondary_memory_needed_mode): New hook:
8734 * targhooks.c (default_secondary_memory_needed_mode): Declare.
8735 * targhooks.h (default_secondary_memory_needed_mode): New function.
8736 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
8737 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
8738 * doc/tm.texi: Regenerate.
8739 * lra-constraints.c (check_and_process_move): Use
8740 targetm.secondary_memory_needed_mode instead of
8741 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
8742 (curr_insn_transform): Likewise.
8743 * reload.c (get_secondary_mem): Likewise.
8744 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8745 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
8747 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8748 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8749 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
8750 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8751 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
8753 * config/powerpcspe/powerpcspe-protos.h
8754 (rs6000_secondary_memory_needed_mode): Delete.
8755 * config/powerpcspe/powerpcspe.c
8756 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8757 (rs6000_secondary_memory_needed_mode): Make static.
8758 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8759 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
8761 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8763 (rs6000_secondary_memory_needed_mode): Make static.
8764 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8765 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
8766 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8767 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8768 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8770 (sparc_secondary_memory_needed_mode): New function.
8771 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
8773 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
8775 * config/aarch64/constraints.md (Umq): New constraint.
8776 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
8778 (mov<mode>): Update condition.
8780 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8782 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
8783 when bitposition is the same.
8785 2017-09-13 Richard Biener <rguenther@suse.de>
8787 * dwarf2out.c (output_die_symbol): Remove.
8788 (output_die): Do not output a DIEs symbol.
8790 2017-09-13 Richard Biener <rguenther@suse.de>
8793 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
8794 default-def to avoid breaking iterator update with the weird
8795 interaction with cgraph_update_edges_for_call_stmt_node.
8797 2017-09-13 Richard Biener <rguenther@suse.de>
8799 * tree-cfg.c (verify_gimple_assign_binary): Add verification
8800 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
8801 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
8802 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
8804 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
8806 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
8807 Disable pc relative literal load irrespective of
8808 TARGET_FIX_ERR_A53_84341 for default.
8810 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
8812 * config/sparc/sparc.c (output_return): Output the source location of
8813 the insn in the delay slot, if any.
8814 (output_sibcall): Likewise.
8816 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
8819 * common.opt (-static-pie): New alias.
8820 (shared): Negate static-pie.
8821 (-no-pie): Update help text.
8823 (static-pie): New option.
8824 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
8825 -static-pie support.
8826 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
8827 (LINK_EH_SPEC): Likewise.
8828 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8829 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8830 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8831 * gcc.c (LINK_COMMAND_SPEC): Likewise.
8832 (init_gcc_specs): Likewise.
8833 (init_spec): Likewise.
8834 (display_help): Update help message for -pie.
8835 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
8838 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
8840 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
8841 (movdi_aarch64): Likewise.
8842 (movti_aarch64): Likewise.
8844 2017-09-12 Simon Wright <simon@pushface.org>
8847 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
8848 calculation of the minor version, always output as 0.
8850 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8853 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8854 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
8855 on it early, rather than manual conversion late. For
8856 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
8857 instead of performing manual conversion.
8859 2017-09-12 Carl Love <cel@us.ibm.com>
8861 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
8862 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
8864 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8865 VMULOSW): Add definitions.
8866 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8867 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8868 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8869 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8870 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8872 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8874 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
8876 (movti_aarch64): Likewise.
8877 (movdf_aarch64): Likewise.
8878 (movtf_aarch64): Likewise.
8879 (load_pairdi): Likewise.
8880 (store_pairdi): Likewise.
8881 (load_pairdf): Likewise.
8882 (store_pairdf): Likewise.
8883 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
8884 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
8885 (ldr_got_small_<mode>): Likewise.
8886 (ldr_got_small_28k_<mode>): Likewise.
8887 (ldr_got_tiny): Likewise.
8888 * config/aarch64/iterators.md (ldst_sz): New.
8889 (ldpstp_sz): Likewise.
8890 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
8892 (thunderx_load): Split load_8 to load_16.
8893 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
8895 (thunderx2t99_storepair_basic): Split store_8 to store_16.
8896 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
8897 (xgene1_store_pair): Split store_8 to store_16.
8898 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
8899 (falkor_st_0_st_sd): Split store_8 to store_16.
8901 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8903 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
8904 and store1/2/3/4 to store_4/8/12/16.
8905 * config/aarch64/aarch64.md: Update for rename.
8906 * config/arm/arm.md: Likewise.: Likewise.
8907 * config/arm/arm.c: Likewise.
8908 * config/arm/thumb1.md: Likewise.
8909 * config/arm/thumb2.md: Likewise.
8910 * config/arm/vfp.md: Likewise.
8911 * config/arm/arm-generic.md: Likewise.
8912 * config/arm/arm1020e.md: Likewise.
8913 * config/arm/arm1026ejs.md: Likewise.
8914 * config/arm/arm1136jfs.md: Likewise.
8915 * config/arm/arm926ejs.md: Likewise.
8916 * config/arm/cortex-a15.md: Likewise.
8917 * config/arm/cortex-a17.md: Likewise.
8918 * config/arm/cortex-a5.md: Likewise.
8919 * config/arm/cortex-a53.md: Likewise.
8920 * config/arm/cortex-a57.md: Likewise.
8921 * config/arm/cortex-a7.md: Likewise.
8922 * config/arm/cortex-a8.md: Likewise.
8923 * config/arm/cortex-a9.md: Likewise.
8924 * config/arm/cortex-m4.md: Likewise.
8925 * config/arm/cortex-m7.md: Likewise.
8926 * config/arm/cortex-r4.md: Likewise.
8927 * config/arm/exynos-m1.md: Likewise.
8928 * config/arm/fa526.md: Likewise.
8929 * config/arm/fa606te.md: Likewise.
8930 * config/arm/fa626te.md: Likewise.
8931 * config/arm/fa726te.md: Likewise.
8932 * config/arm/fmp626.md: Likewise.
8933 * config/arm/iwmmxt.md: Likewise.
8934 * config/arm/ldmstm.md: Likewise.
8935 * config/arm/marvell-pj4.md: Likewise.
8936 * config/arm/xgene1.md: Likewise.
8937 * config/aarch64/thunderx.md: Likewise.
8938 * config/aarch64/thunderx2t99.md: Likewise.
8939 * config/aarch64/falkor.md: Likewise.
8941 2017-09-12 Martin Liska <mliska@suse.cz>
8943 * attribs.c (private_lookup_attribute): New function.
8944 * attribs.h (private_lookup_attribute): Declared here.
8945 (lookup_attribute): Called from this place.
8947 2017-09-12 Richard Biener <rguenther@suse.de>
8949 PR tree-optimization/82157
8950 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
8951 stmts with side-effects.
8953 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8954 Alan Hayward <alan.hayward@arm.com>
8955 David Sherwood <david.sherwood@arm.com>
8957 * target.def (hard_regno_nregs): New hook.
8958 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
8959 * targhooks.h (default_hard_regno_nregs): Declare.
8960 * targhooks.c (default_hard_regno_nregs): New function.
8961 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
8962 (TARGET_HARD_REGNO_NREGS): ...this hook.
8963 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
8964 (CLASS_MAX_NREGS): Likewise.
8965 * doc/tm.texi: Regenerate.
8966 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
8967 instead of HARD_REGNO_NREGS.
8968 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
8969 HARD_REGNO_NREGS in the comment.
8970 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
8971 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
8972 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
8973 Return an unsigned int.
8974 (TARGET_HARD_REGNO_NREGS): Redefine.
8975 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
8976 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
8977 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8978 (arc_hard_regno_nregs): New function.
8979 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
8980 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
8981 (arm_hard_regno_nregs): New function.
8982 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
8983 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
8984 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
8985 (TARGET_HARD_REGNO_NREGS): Redefine.
8986 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
8987 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
8988 (HARD_REGNO_NREGS): Delete.
8989 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
8990 (cr16_hard_regno_nregs): New function.
8991 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
8992 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
8993 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
8994 (cris_hard_regno_nregs): New function.
8995 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
8996 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
8997 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8998 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
8999 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
9000 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
9001 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
9002 (frv_hard_regno_nregs): Make static. Take and return an
9004 (frv_class_max_nregs): Remove outdated copy of documentation.
9005 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
9006 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
9007 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
9008 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
9009 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
9010 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
9011 (TARGET_HARD_REGNO_NREGS): Redefine.
9012 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
9013 (CLASS_MAX_NREGS): Update comment.
9014 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
9015 (ia64_hard_regno_nregs): New function.
9016 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9017 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9018 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9019 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9020 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9022 (m32c_hard_regno_nregs): Likewise. Make static.
9023 (TARGET_HARD_REGNO_NREGS): Redefine.
9024 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9025 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9026 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9027 (m68k_hard_regno_nregs): New function.
9028 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9029 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9030 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9031 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9032 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9033 Take and return an unsigned int.
9034 (TARGET_HARD_REGNO_NREGS): Redefine.
9035 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9036 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9037 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9038 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9039 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9040 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9041 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9042 (msp430_hard_regno_nregs): Make static. Take and return an
9044 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9045 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9046 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9047 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9048 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9049 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9050 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9051 (TARGET_HARD_REGNO_NREGS): Redefine.
9052 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9053 (PA_HARD_REGNO_NREGS): ...this.
9054 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9055 (PA_HARD_REGNO_NREGS): ...this.
9056 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9057 (pa_hard_regno_nregs): New function.
9058 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9059 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9060 (pdp11_hard_regno_nregs): New function.
9061 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9062 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9063 (rs6000_hard_regno_nregs_hook): New function.
9064 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9065 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9066 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9067 Take and return an unsigned int. Move earlier in file.
9068 (TARGET_HARD_REGNO_NREGS): Redefine.
9069 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9070 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9071 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9072 (rl78_hard_regno_nregs): Make static. Take and return an
9074 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9075 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9076 (rs6000_hard_regno_nregs_hook): New function.
9077 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9078 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9079 (TARGET_HARD_REGNO_NREGS): Redefine.
9080 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9081 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9082 instead of HARD_REGNO_NREGS.
9083 (s390_hard_regno_nregs): New function.
9084 (s390_hard_regno_mode_ok): Add comment from s390.h.
9085 (TARGET_HARD_REGNO_NREGS): Redefine.
9086 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9087 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9088 (sh_hard_regno_nregs): New function.
9089 (sh_pass_in_reg_p): Use it.
9090 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9091 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9092 (sparc_hard_regno_nregs): New function.
9093 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9094 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9095 (spu_function_arg_advance): Use it, supplying a valid register number.
9096 (TARGET_HARD_REGNO_NREGS): Redefine.
9097 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9098 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9099 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9100 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9101 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9102 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9103 (CLASS_MAX_NREGS): Remove copy of old documentation.
9104 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9105 (visium_hard_regno_nregs): New function.
9106 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9107 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9108 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9109 xtensa_hard_regno_nregs): New function.
9110 * system.h (HARD_REGNO_NREGS): Poison.
9112 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9114 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9115 hard_regno_nregs instead of HARD_REGNO_NREGS.
9116 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9117 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9118 (c6x_expand_epilogue): Likewise.
9119 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9120 (frv_read_iacc_argument): Likewise.
9121 * config/sh/sh.c: Include regs.h.
9122 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9123 (regs_used): Likewise.
9124 (output_stack_adjust): Likewise.
9125 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9126 * expmed.c: Include regs.h.
9127 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9128 * ree.c: Include regs.h.
9129 (combine_reaching_defs): Use hard_regno_nregs instead of
9131 (add_removable_extension): Likewise.
9133 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9135 * regs.h (hard_regno_nregs): Turn into a function.
9136 (end_hard_regno): Update accordingly.
9137 * caller-save.c (setup_save_areas): Likewise.
9138 (save_call_clobbered_regs): Likewise.
9139 (replace_reg_with_saved_mem): Likewise.
9140 (insert_restore): Likewise.
9141 (insert_save): Likewise.
9142 * combine.c (can_change_dest_mode): Likewise.
9143 (move_deaths): Likewise.
9144 (distribute_notes): Likewise.
9145 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9146 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9147 (rs6000_split_multireg_move): Likewise.
9148 (rs6000_register_move_cost): Likewise.
9149 (rs6000_memory_move_cost): Likewise.
9150 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9151 (rs6000_split_multireg_move): Likewise.
9152 (rs6000_register_move_cost): Likewise.
9153 (rs6000_memory_move_cost): Likewise.
9154 * cselib.c (cselib_reset_table): Likewise.
9155 (cselib_lookup_1): Likewise.
9156 * emit-rtl.c (set_mode_and_regno): Likewise.
9157 * function.c (aggregate_value_p): Likewise.
9158 * ira-color.c (setup_profitable_hard_regs): Likewise.
9159 (check_hard_reg_p): Likewise.
9160 (calculate_saved_nregs): Likewise.
9161 (assign_hard_reg): Likewise.
9162 (improve_allocation): Likewise.
9163 (calculate_spill_cost): Likewise.
9164 * ira-emit.c (modify_move_list): Likewise.
9165 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9166 (ira_hard_reg_in_set_p): Likewise.
9167 * ira.c (setup_reg_mode_hard_regset): Likewise.
9168 (clarify_prohibited_class_mode_regs): Likewise.
9169 (check_allocation): Likewise.
9170 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9171 (lra_setup_reg_renumber): Likewise.
9172 (setup_try_hard_regno_pseudos): Likewise.
9173 (spill_for): Likewise.
9174 (assign_hard_regno): Likewise.
9175 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9176 * lra-constraints.c (in_class_p): Likewise.
9177 (lra_constraint_offset): Likewise.
9178 (simplify_operand_subreg): Likewise.
9179 (lra_constraints): Likewise.
9180 (split_reg): Likewise.
9181 (split_if_necessary): Likewise.
9182 (invariant_p): Likewise.
9183 (inherit_in_ebb): Likewise.
9184 * lra-lives.c (process_bb_lives): Likewise.
9185 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9186 (get_hard_regs): Likewise.
9187 (do_remat): Likewise.
9188 * lra-spills.c (assign_spill_hard_regs): Likewise.
9189 * mode-switching.c (create_pre_exit): Likewise.
9190 * postreload.c (reload_combine_recognize_pattern): Likewise.
9191 * recog.c (peep2_find_free_register): Likewise.
9192 * regcprop.c (kill_value_regno): Likewise.
9193 (set_value_regno): Likewise.
9194 (copy_value): Likewise.
9195 (maybe_mode_change): Likewise.
9196 (find_oldest_value_reg): Likewise.
9197 (copyprop_hardreg_forward_1): Likewise.
9198 * regrename.c (check_new_reg_p): Likewise.
9199 (regrename_do_replace): Likewise.
9200 * reload.c (push_reload): Likewise.
9201 (combine_reloads): Likewise.
9202 (find_dummy_reload): Likewise.
9203 (operands_match_p): Likewise.
9204 (find_reloads): Likewise.
9205 (find_equiv_reg): Likewise.
9206 (reload_adjust_reg_for_mode): Likewise.
9207 * reload1.c (count_pseudo): Likewise.
9208 (count_spilled_pseudo): Likewise.
9209 (find_reg): Likewise.
9210 (clear_reload_reg_in_use): Likewise.
9211 (free_for_value_p): Likewise.
9212 (allocate_reload_reg): Likewise.
9213 (choose_reload_regs): Likewise.
9214 (reload_adjust_reg_for_temp): Likewise.
9215 (emit_reload_insns): Likewise.
9216 (delete_output_reload): Likewise.
9217 * rtlanal.c (subreg_get_info): Likewise.
9218 * sched-deps.c (sched_analyze_reg): Likewise.
9219 * sel-sched.c (init_regs_for_mode): Likewise.
9220 (mark_unavailable_hard_regs): Likewise.
9221 (choose_best_reg_1): Likewise.
9222 (verify_target_availability): Likewise.
9223 * valtrack.c (dead_debug_insert_temp): Likewise.
9224 * var-tracking.c (track_loc_p): Likewise.
9225 (emit_note_insn_var_location): Likewise.
9226 * varasm.c (make_decl_rtl): Likewise.
9227 * reginfo.c (choose_hard_reg_mode): Likewise.
9228 (init_reg_modes_target): Refer directly to
9229 this_target_regs->x_hard_regno_nregs.
9231 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9233 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9234 instead of hard_regno_nregs.
9236 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9238 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9239 end_hard_regno instead of hard_regno_nregs.
9240 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9241 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9242 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9243 * ira-color.c (improve_allocation): Likewise.
9244 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9245 * lra-lives.c (mark_regno_live): Likewise.
9246 (mark_regno_dead): Likewise.
9247 * lra-remat.c (operand_to_remat): Likewise.
9248 * lra.c (collect_non_operand_hard_regs): Likewise.
9249 * postreload.c (reload_combine_note_store): Likewise.
9250 (move2add_valid_value_p): Likewise.
9251 * reload.c (regno_clobbered_p): Likewise.
9253 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9255 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9257 * config/v850/v850.c (v850_reorg): Likewise.
9258 * reload.c (refers_to_regno_for_reload_p): Likewise.
9259 (find_equiv_reg): Likewise.
9260 * reload1.c (reload_reg_reaches_end_p): Likewise.
9262 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9264 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9266 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9267 * config/arm/arm.c (output_move_neon): Likewise.
9268 (arm_attr_length_move_neon): Likewise.
9269 (neon_split_vcombine): Likewise.
9270 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9271 (c6x_mark_reg_written): Likewise.
9272 (c6x_dwarf_register_span): Likewise.
9273 * config/i386/i386.c (ix86_save_reg): Likewise.
9274 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9275 (rws_access_reg): Likewise.
9276 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9277 * mode-switching.c (create_pre_exit): Likewise.
9278 * ree.c (combine_reaching_defs): Likewise.
9279 (add_removable_extension): Likewise.
9280 * regcprop.c (find_oldest_value_reg): Likewise.
9281 (copyprop_hardreg_forward_1): Likewise.
9282 * reload.c (reload_inner_reg_of_subreg): Likewise.
9283 (push_reload): Likewise.
9284 (combine_reloads): Likewise.
9285 (find_dummy_reload): Likewise.
9286 (reload_adjust_reg_for_mode): Likewise.
9287 * reload1.c (find_reload_regs): Likewise.
9288 (forget_old_reloads_1): Likewise.
9289 (reload_reg_free_for_value_p): Likewise.
9290 (reload_adjust_reg_for_temp): Likewise.
9291 (emit_reload_insns): Likewise.
9292 (delete_output_reload): Likewise.
9293 * sel-sched.c (choose_best_reg_1): Likewise.
9294 (choose_best_pseudo_reg): Likewise.
9296 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9297 Alan Hayward <alan.hayward@arm.com>
9298 David Sherwood <david.sherwood@arm.com>
9300 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9301 * target.def (slow_unaligned_access): New hook.
9302 * targhooks.h (default_slow_unaligned_access): Declare.
9303 * targhooks.c (default_slow_unaligned_access): New function.
9304 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9305 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9306 * doc/tm.texi: Regenerate.
9307 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9308 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9309 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9311 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9312 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9314 (rs6000_slow_unaligned_access): New function.
9315 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9316 (expand_block_compare): Likewise.
9317 (expand_strn_compare): Likewise.
9318 (rs6000_rtx_costs): Likewise.
9319 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9320 (riscv_slow_unaligned_access): Likewise.
9321 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9322 (riscv_slow_unaligned_access_p): ...this and make static.
9323 (riscv_option_override): Update accordingly.
9324 (riscv_slow_unaligned_access): New function.
9325 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9326 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9327 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9328 (rs6000_slow_unaligned_access): New function.
9329 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9330 (rs6000_rtx_costs): Likewise.
9331 * config/rs6000/rs6000-string.c (expand_block_compare)
9332 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9333 of SLOW_UNALIGNED_ACCESS.
9334 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9335 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9336 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9337 of SLOW_UNALIGNED_ACCESS.
9338 * expmed.c (simple_mem_bitfield_p): Likewise.
9339 * expr.c (alignment_for_piecewise_move): Likewise.
9340 (emit_group_load_1): Likewise.
9341 (emit_group_store): Likewise.
9342 (copy_blkmode_from_reg): Likewise.
9343 (emit_push_insn): Likewise.
9344 (expand_assignment): Likewise.
9345 (store_field): Likewise.
9346 (expand_expr_real_1): Likewise.
9347 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9348 * lra-constraints.c (simplify_operand_subreg): Likewise.
9349 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9350 * gimple-ssa-store-merging.c: Likewise in block comment at start
9352 * tree-ssa-strlen.c: Include target.h.
9353 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9354 of SLOW_UNALIGNED_ACCESS.
9355 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9357 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9359 PR rtl-optimization/82185
9360 * expmed.c (emit_store_flag_int): Only test tem if it has been
9363 2017-09-12 Richard Biener <rguenther@suse.de>
9366 * match.pd ((FTYPE) N CMP CST): Fix typo.
9368 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
9370 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
9372 (mips_near_type_p): Add 'short_call' attribute as a synonym
9374 * doc/extend.texi (short_call): Document new function attribute.
9376 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9379 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9380 assertion check that in the condition.
9381 (get_atomic_generic_size): Likewise. Before testing if parameter
9382 has pointer type, if it has array type, call for C++
9383 default_conversion to perform array-to-pointer conversion.
9385 2017-09-12 Richard Biener <rguenther@suse.de>
9387 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
9388 for operations we cannot scalarize.
9390 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
9392 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
9393 vectors heap vectors. Clean up comments.
9394 Make visited_bbs a reference.
9395 (profitable_jump_thread_path): Make GC
9396 vectors heap vectors. Clean up comments.
9398 (convert_and_register_jump_thread_path): Make GC vectors heap
9400 (check_subpath_and_update_thread_path): Same. Clean up comments.
9401 Make visited_bbs a reference.
9402 (handle_phi): Abstract common code to to
9403 register_jump_thread_path_if_profitable.
9404 Rename VAR_BB to DEF_BB.
9406 Make GC vectors heap vectors.
9407 Make visited_bbs a reference.
9408 (handle_assignment): Same.
9409 (register_jump_thread_path_if_profitable): New.
9410 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
9412 Make GC vectors heap vectors. Clean up comments.
9413 Make visited_bbs a reference.
9414 (find_jump_threads_backwards): Make visited_bbs live in the stack.
9415 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
9418 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
9421 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
9422 words of E_DImode object are reachable by xtensa_uimm8x4 access.
9424 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
9426 Revert r251800 and r251799.
9428 2017-09-11 Martin Jambor <mjambor@suse.cz>
9431 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
9432 arguments in advance.
9433 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
9434 use it to find predecessor edges.
9435 (naive_outof_ssa): Collect vector of predecessors.
9437 2017-09-08 Jason Merrill <jason@redhat.com>
9439 PR c++/70029 - ICE with ref-qualifier and -flto
9440 * langhooks.h (struct lang_hooks_for_types): Add
9441 copy_lang_qualifiers.
9442 * attribs.c (build_type_attribute_qual_variant): Use it.
9443 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
9445 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
9446 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
9448 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
9451 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
9452 (*mulsi3_sp64): New instruction.
9453 (mulsi3): New expander.
9455 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
9457 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
9459 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9461 * sancov.c: Include memmodel.h.
9463 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
9466 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
9469 2017-09-07 Carl Love <cel@us.ibm.com>
9471 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
9472 the sldi instruction.
9474 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
9476 * sancov.c: Include tm_p.h.
9478 2017-09-07 Jakub Jelinek <jakub@redhat.com>
9481 * output.h (switch_to_other_text_partition): New declaration.
9482 * varasm.c (switch_to_other_text_partition): New function.
9483 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
9484 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
9485 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
9486 to the other text partition before emitting LCL label and switch back
9487 after emitting the word after it.
9489 2017-09-07 Richard Biener <rguenther@suse.de>
9491 * passes.def (pass_split_crit_edges): Remove instance before PRE.
9492 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
9493 critical edges here, after loop init.
9494 (pass_data_pre): Remove PROP_no_crit_edges flags.
9495 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
9496 for valueization of call args to avoid leaking VN_TOP.
9497 (visit_use): Assert we do not visit default defs.
9498 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
9499 Use error_mark_node to more easily detect leaking VN_TOP.
9500 All default-defs are varying, not VN_TOP. Mark them visited.
9501 (run_scc_vn): Make code match comment.
9503 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9505 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
9506 OPTION_MASK_FLOAT128_KEYWORD.
9507 (POWERPC_MASKS): Likewise.
9508 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
9509 support for the -mfloat128-type option, and make -mfloat128
9510 default on PowerPC Linux systems. Define or undefine
9511 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
9512 Define __float128 to be __ieee128 if IEEE 128-bit support is
9513 enabled, or undefine it.
9514 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
9515 Delete defining __FLOAT128_TYPE__.
9516 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
9517 -mfloat128-type option and make -mfloat128 default on PowerPC
9519 (TARGET_FLOAT128_TYPE): Likewise.
9520 (-mfloat128-type): Likewise.
9521 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9522 Delete the -mfloat128-type option and make -mfloat128 default on
9523 PowerPC Linux systems. Always use __ieee128 to be the keyword for
9524 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
9525 128-bit floating point is enabled. Change tests from using
9526 -mfloat128-type to -mfloat128.
9527 (rs6000_mangle_type): Use the correct mangling for the __float128
9528 type even if normal long double is restricted to 64-bits.
9529 (floatn_mode): Enable the _Float128 type by default on VSX Linux
9531 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
9532 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
9533 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
9535 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
9536 documentation for -mfloat128.
9538 2017-09-06 Olivier Hainque <hainque@adacore.com>
9540 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
9542 2017-09-06 Wish Wu <wishwu007@gmail.com>
9543 Jakub Jelinek <jakub@redhat.com>
9545 * asan.c (initialize_sanitizer_builtins): Add
9546 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
9547 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
9548 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
9549 BT_FN_VOID_UINT64_PTR variables.
9550 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
9551 (BT_FN_VOID_UINT16_UINT16): Likewise.
9552 (BT_FN_VOID_UINT32_UINT32): Likewise.
9553 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
9554 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
9555 (BT_FN_VOID_UINT64_PTR): Likewise.
9556 * common.opt (flag_sanitize_coverage): New variable.
9557 (fsanitize-coverage=trace-pc): Remove.
9558 (fsanitize-coverage=): Add.
9559 * flag-types.h (enum sanitize_coverage_code): New enum.
9560 * fold-const.c (fold_range_test): Disable non-short-circuit
9561 optimization if flag_sanitize_coverage.
9562 (fold_truth_andor): Likewise.
9563 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
9564 * opts.c (COVERAGE_SANITIZER_OPT): Define.
9565 (coverage_sanitizer_opts): New array.
9566 (get_closest_sanitizer_option): Add OPTS argument, handle also
9567 OPT_fsanitize_coverage_.
9568 (parse_sanitizer_options): Adjusted to also handle
9569 OPT_fsanitize_coverage_.
9570 (common_handle_option): Add OPT_fsanitize_coverage_.
9571 * sancov.c (instrument_comparison, instrument_switch): New function.
9572 (sancov_pass): Add trace-cmp support.
9573 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
9574 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
9575 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
9576 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
9577 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
9578 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
9579 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
9580 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
9581 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
9582 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
9584 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9586 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9587 error. Only quit immediately if parsing is complete.
9588 (BEGIN): Initialize fatal_err and parse_done.
9589 (begin fpu, end fpu): Check number of arguments.
9590 (begin arch, end arch): Likewise.
9591 (begin cpu, end cpu): Likewise.
9592 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9593 (optalias): Likewise.
9595 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9597 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9598 * config/arm/arm-isa.h: Delete. Move definitions to ...
9599 * arm-cpus.in: ... here. Use new feature and fgroup values.
9600 * config/arm/arm.c (arm_option_override): Use lower case for feature
9602 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9603 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9604 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9606 (print_isa_bits_for): New function.
9607 (gen_isa): New function.
9608 (gen_comm_data): Use print_isa_bits_for.
9609 (define feature): New keyword.
9610 (define fgroup): New keyword.
9611 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
9612 (arm-isa.h): Add rule to generate file.
9613 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9614 case for feature bit names.
9616 2017-09-06 Richard Biener <rguenther@suse.de>
9618 * tree-ssa-pre.c (NECESSARY): Remove.
9619 (create_expression_by_pieces): Do not touch pass-local flags.
9620 (insert_into_preds_of_block): Likewise.
9621 (do_pre_regular_insertion): Likewise.
9622 (eliminate_insert): Likewise.
9623 (eliminate_dom_walker::before_dom_children): Likewise.
9624 (fini_eliminate): Do not look at inserted_exprs.
9625 (mark_operand_necessary): Remove.
9626 (remove_dead_inserted_code): Replace with simple work-list
9627 algorithm based on inserted_exprs and SSA uses.
9628 (pass_pre::execute): Re-order fini_eliminate and
9629 remove_dead_inserted_code.
9631 2017-09-06 Olivier Hainque <hainque@adacore.com>
9633 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
9634 for VxWorks 7. Adjust surrounding comments.
9636 2017-09-06 Richard Biener <rguenther@suse.de>
9638 * gimple-ssa-strength-reduction.c
9639 (find_candidates_dom_walker::before_dom_children): Also allow
9642 2017-09-06 Richard Biener <rguenther@suse.de>
9644 PR tree-optimization/82108
9645 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
9646 for gap in the non-permutation SLP case.
9648 2017-09-06 Martin Jambor <mjambor@suse.cz>
9650 PR tree-optimization/82078
9651 * tree-sra.c (sort_and_splice_var_accesses): Move call to
9652 add_access_to_work_queue...
9653 (build_accesses_from_assign): ...here.
9654 (propagate_all_subaccesses): Make sure racc is the group
9655 representative, if there is one.
9657 2017-09-06 Jakub Jelinek <jakub@redhat.com>
9660 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
9663 2017-09-06 Richard Biener <rguenther@suse.de>
9665 * gimple-ssa-strength-reduction.c
9666 (find_candidates_dom_walker::before_doom_children): Use a
9667 type and not a mode check.
9669 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
9672 * config/arm/predicates.md (arm_general_adddi_operand): Create new
9674 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
9676 2017-09-05 Jeff Law <law@redhat.com>
9678 PR tree-optimization/64910
9679 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
9680 cases where we have 3 or more operands.
9682 2017-09-05 Jakub Jelinek <jakub@redhat.com>
9685 * omp-low.c (lower_omp_for): Recompute tree invariant if
9686 gimple_omp_for_initial/final is ADDR_EXPR.
9689 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
9690 into gimple val before gimplification fo the COND_EXPR.
9692 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
9694 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
9695 REGION_COPY argument.
9696 (thread_through_all_blocks): Remove unused argument to
9697 duplicate_thread_path.
9699 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9700 Alan Hayward <alan.hayward@arm.com>
9701 David Sherwood <david.sherwood@arm.com>
9703 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
9704 Take a scalar_mode rather than a machine_mode.
9705 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9706 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
9707 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9708 (aarch64_gen_adjusted_ldpstp): Likewise.
9709 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
9711 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9712 Alan Hayward <alan.hayward@arm.com>
9713 David Sherwood <david.sherwood@arm.com>
9715 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
9716 Take a scalar_int_mode instead of a machine_mode.
9717 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9718 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9719 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9720 (aarch64_simd_attr_length_rglist): Delete.
9721 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
9722 a scalar_int_mode instead of a machine_mode.
9723 (aarch64_add_offset): Likewise.
9724 (aarch64_internal_mov_immediate): Likewise
9725 (aarch64_add_constant_internal): Likewise.
9726 (aarch64_add_constant): Likewise.
9727 (aarch64_movw_imm): Likewise.
9728 (aarch64_rtx_arith_op_extract_p): Likewise.
9729 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9730 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9731 Remove assert that the mode isn't a vector.
9732 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9733 (aarch64_expand_mov_immediate): Update calls after above changes.
9734 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
9735 (aarch64_and_bitmask_imm): Check for scalar integer modes.
9736 (aarch64_move_imm): Likewise.
9737 (aarch64_can_const_movi_rtx_p): Likewise.
9738 (aarch64_strip_extend): Likewise.
9739 (aarch64_extr_rtx_p): Likewise.
9740 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
9741 a CONST_INT when the mode parameter is VOIDmode.
9742 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
9744 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9746 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
9747 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9748 (bitwise_type_for_mode): Update accordingly.
9750 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9752 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
9753 * stor-layout.c (mode_for_size_tree): Likewise.
9754 (mode_for_array): Update accordingly.
9755 (layout_decl): Likewise.
9756 (compute_record_mode): Likewise. Only set the mode once.
9758 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9760 * target.def (get_mask_mode): Change return type to opt_mode.
9762 * doc/tm.texi: Regenerate.
9763 * targhooks.h (default_get_mask_mode): Return an opt_mode.
9764 * targhooks.c (default_get_mask_mode): Likewise.
9765 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
9766 * optabs-query.c (can_vec_mask_load_store_p): Update use of
9767 targetm.get_mask_mode.
9768 * tree.c (build_truth_vector_type): Likewise.
9770 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9772 * machmode.h (mode_for_vector): Return an opt_mode.
9773 * stor-layout.c (mode_for_vector): Likewise.
9774 (mode_for_int_vector): Update accordingly.
9775 (layout_type): Likewise.
9776 * config/i386/i386.c (emit_memmov): Likewise.
9777 (ix86_expand_set_or_movmem): Likewise.
9778 (ix86_expand_vector_init): Likewise.
9779 (ix86_get_mask_mode): Likewise.
9780 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
9782 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
9783 * expmed.c (extract_bit_field_1): Likewise.
9784 * expr.c (expand_expr_real_2): Likewise.
9785 * optabs-query.c (can_vec_perm_p): Likewise.
9786 (can_vec_mask_load_store_p): Likewise.
9787 * optabs.c (expand_vec_perm): Likewise.
9788 * targhooks.c (default_get_mask_mode): Likewise.
9789 * tree-vect-stmts.c (vectorizable_store): Likewise.
9790 (vectorizable_load): Likewise.
9791 (get_vectype_for_scalar_type_and_size): Likewise.
9793 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9795 * machmode.h (mode_for_int_vector): New function.
9796 * stor-layout.c (mode_for_int_vector): Likewise.
9797 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
9798 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
9799 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
9800 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
9801 (s390_expand_vcond): Likewise.
9803 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9805 * machmode.h (opt_machine_mode): New type.
9806 (opt_mode<T>): Allow construction from anything that can be
9808 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
9809 (mode_for_size): Return an opt_machine_mode.
9810 * stor-layout.c (mode_for_size): Likewise.
9811 (mode_for_size_tree): Update call accordingly.
9812 (bitwise_mode_for_mode): Likewise.
9813 (make_fract_type): Likewise.
9814 (make_accum_type): Likewise.
9815 * caller-save.c (replace_reg_with_saved_mem): Update call
9817 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9818 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9819 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9820 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9821 * expmed.c (extract_bit_field_1): Likewise.
9822 * reload.c (get_secondary_mem): Likewise.
9823 * varasm.c (assemble_integer): Likewise.
9824 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
9827 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9829 * machmode.h (decimal_float_mode_for_size): New function.
9830 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
9831 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
9832 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
9833 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
9834 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
9835 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
9837 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9839 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
9840 (get_builtin_sync_mode): Likewise.
9841 (expand_ifn_atomic_compare_exchange): Likewise.
9842 (expand_builtin_atomic_clear): Likewise.
9843 (expand_builtin_atomic_test_and_set): Likewise.
9844 (fold_builtin_atomic_always_lock_free): Likewise.
9845 * calls.c (compute_argument_addresses): Likewise.
9846 (emit_library_call_value_1): Likewise.
9847 (store_one_arg): Likewise.
9848 * combine.c (combine_instructions): Likewise.
9849 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
9850 * config/arm/arm.c (arm_function_value): Likewise.
9851 (aapcs_allocate_return_reg): Likewise.
9852 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
9853 * config/i386/i386.c (construct_container): Likewise.
9854 (ix86_gimplify_va_arg): Likewise.
9855 (ix86_expand_sse_cmp): Likewise.
9856 (emit_memmov): Likewise.
9857 (emit_memset): Likewise.
9858 (expand_small_movmem_or_setmem): Likewise.
9859 (ix86_expand_pextr): Likewise.
9860 (ix86_expand_pinsr): Likewise.
9861 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
9862 * config/microblaze/microblaze.c (microblaze_block_move_straight):
9864 * config/mips/mips.c (mips_function_value_1) Likewise.
9865 (mips_block_move_straight): Likewise.
9866 (mips_expand_ins_as_unaligned_store): Likewise.
9867 * config/powerpcspe/powerpcspe.c
9868 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9869 (rs6000_darwin64_record_arg_flush): Likewise.
9870 * config/rs6000/rs6000.c
9871 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9872 (rs6000_darwin64_record_arg_flush): Likewise.
9873 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
9874 (sparc_function_value_1): Likewise.
9875 * config/spu/spu.c (adjust_operand): Likewise.
9876 (spu_emit_branch_or_set): Likewise.
9877 (arith_immediate_p): Likewise.
9878 * emit-rtl.c (gen_lowpart_common): Likewise.
9879 * expr.c (expand_expr_real_1): Likewise.
9880 * function.c (assign_parm_setup_block): Likewise.
9881 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9882 * reload1.c (alter_reg): Likewise.
9883 * stor-layout.c (mode_for_vector): Likewise.
9884 (layout_type): Likewise.
9886 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9888 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
9889 (spu_convert_move): Likewise.
9890 * lower-subreg.c (resolve_simple_move): Likewise.
9892 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9895 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
9896 define_insn to a define_expand.
9897 (altivec_vsum2sws_direct): New define_insn.
9898 (altivec_vsumsws): Convert from a define_insn to a define_expand.
9900 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9902 * config/arm/arm.c (arm_option_params_internal): Improve setting of
9905 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
9910 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
9911 realign stack if stack alignment needed is less than incoming
9914 2017-09-05 Marek Polacek <polacek@redhat.com>
9917 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
9920 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9922 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
9924 2017-09-05 Richard Biener <rguenther@suse.de>
9926 PR tree-optimization/82084
9927 * fold-const.c (can_native_encode_string_p): Handle wide characters.
9929 2017-09-05 Richard Biener <rguenther@suse.de>
9931 PR tree-optimization/82102
9932 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
9934 2017-09-05 Martin Liska <mliska@suse.cz>
9936 PR tree-optimization/82032
9937 * tree-cfg.c (generate_range_test): New function.
9938 * tree-cfg.h (generate_range_test): Declared here.
9939 * tree-cfgcleanup.c (convert_single_case_switch): New function.
9940 (cleanup_control_expr_graph): Use it.
9941 * tree-switch-conversion.c (try_switch_expansion): Remove
9943 (emit_case_nodes): Use generate_range_test.
9945 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9948 * config/i386/i386.md (*<btsc><mode>_mask): Add
9949 TARGET_USE_BT to insn constraint.
9950 (*btr<mode>_mask): Ditto.
9952 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
9954 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
9955 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
9957 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
9960 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
9961 TARGET_NEON and TARGET_IWMMXT.
9962 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
9963 TARGET_NEON and TARGET_IWMMXT.
9964 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
9966 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9968 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
9969 (ix86_rewrite_tls_address): Ditto.
9970 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
9971 (ix86_rewrite_tls_address_1): Ditto.
9972 (ix86_rewrite_tls_address): Ditto.
9973 * config/i386/predicates.md (tls_address_pattern): New predicate.
9974 * config/i386/i386.md (TLS address splitter): New splitter.
9976 2017-09-04 Richard Biener <rguenther@suse.de>
9978 PR tree-optimization/82084
9979 * fold-const.h (can_native_encode_string_p): Declare.
9980 * fold-const.c (can_native_encode_string_p): Factor out from ...
9981 (native_encode_string): ... here.
9982 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
9983 vectorizing stores from constants we later cannot handle.
9985 2017-09-04 Marek Polacek <polacek@redhat.com>
9988 * doc/invoke.texi: Update -Wtautological-compare documentation.
9990 2017-09-04 Jeff Law <law@redhat.com>
9992 PR tree-optimization/64910
9993 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
9994 swap the first and last operand if the last is a constant.
9996 2017-09-04 Marek Polacek <polacek@redhat.com>
9999 * convert.c (do_narrow): When sanitizing signed integer overflows,
10000 bail out for signed types.
10001 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
10003 2017-09-04 Richard Biener <rguenther@suse.de>
10005 PR tree-optimization/82060
10006 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
10007 Move devirtualization after stmt folding and before EH/AB/noreturn
10008 cleanup to get the stmt refs canonicalized. Use a bool instead
10009 of gimple_modified_p since that doesn't work for NOPs. Schedule
10010 NOPs generated by folding for removal.
10012 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10013 Alan Hayward <alan.hayward@arm.com>
10014 David Sherwood <david.sherwood@arm.com>
10016 * coretypes.h (pad_direction): New enum.
10017 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10018 (FUNCTION_ARG_PADDING): Likewise.
10019 * target.def (function_arg_padding): New hook.
10020 * targhooks.h (default_function_arg_padding): Declare.
10021 * targhooks.c (default_function_arg_padding): New function.
10022 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10023 (TARGET_FUNCTION_ARG_PADDING): ...this.
10024 * doc/tm.texi: Regenerate.
10025 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10026 instead of direction.
10027 (compute_argument_addresses): Likewise.
10028 (load_register_parameters): Likewise.
10029 (emit_library_call_value_1): Likewise.
10030 (store_one_arg): Use targetm.calls.function_arg_padding instead
10031 of FUNCTION_ARG_PADDING.
10032 (must_pass_in_stack_var_size_or_pad): Likewise.
10033 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10034 (emit_group_store): Likewise.
10035 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10036 instead of FUNCTION_ARG_PADDING.
10037 (emit_push_insn): Likewise, and propagate enum change throughout
10039 * function.h (direction): Delete.
10040 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10042 * function.c (assign_parm_find_stack_rtl): Likewise.
10043 (assign_parm_setup_block_p): Likewise.
10044 (assign_parm_setup_block): Likewise.
10045 (gimplify_parameters): Likewise.
10046 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10047 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10049 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10050 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10051 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10052 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10053 (aarch64_function_arg_padding): ...this new function.
10054 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10055 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10056 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10057 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10058 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10059 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10060 (arm_pad_arg_upward): Replace with...
10061 (arm_function_arg_padding): ...this new function.
10062 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10064 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10065 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10066 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10067 (ia64_hpux_function_arg_padding): Replace with...
10068 (ia64_function_arg_padding): ...this new function. Use pad_direction
10069 instead of direction. Check for TARGET_HPUX.
10070 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10071 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10072 (iq2000_function_arg_padding): New function.
10073 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10074 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10075 (mips_function_arg_padding): ...this new function.
10076 (mips_pad_reg_upward): Update accordingly.
10077 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10078 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10079 targetm.calls.function_arg_padding.
10080 (FUNCTION_ARG_PADDING): Delete.
10081 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10082 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10083 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10084 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10085 (nios2_block_reg_padding): Return pad_direction instead of direction.
10086 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10087 instead of direction.
10088 (nios2_function_arg_padding): Likewise. Make static.
10089 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10090 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10091 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10092 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10093 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10094 (pa_function_arg_padding): Make static. Return pad_direction instead
10096 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10097 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10098 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10099 instead of direction. Use targetm.calls.function_arg_padding.
10100 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10101 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10102 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10103 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10104 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10106 (function_arg_padding): Rename to...
10107 (rs6000_function_arg_padding): ...this. Make static. Return
10108 pad_direction instead of direction.
10109 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10110 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10111 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10112 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10113 instead of direction. Use targetm.calls.function_arg_padding.
10114 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10115 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10116 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10117 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10118 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10119 (function_arg_padding): Rename to...
10120 (rs6000_function_arg_padding): ...this. Make static. Return
10121 pad_direction instead of direction.
10122 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10123 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10124 * config/s390/s390.c (s390_function_arg_padding): New function.
10125 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10126 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10127 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10128 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10129 (function_arg_padding): Rename to...
10130 (sparc_function_arg_padding): ...this. Make static. Return
10131 pad_direction instead of direction.
10132 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10133 * config/spu/spu.c (spu_function_arg_padding): New function.
10134 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10135 * system.h (FUNCTION_ARG_PADDING): Poison.
10137 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10138 Alan Hayward <alan.hayward@arm.com>
10139 David Sherwood <david.sherwood@arm.com>
10141 * target.def (modes_tieable_p): New hook.
10142 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10143 (TARGET_MODES_TIEABLE_P): ...this.
10144 * doc/tm.texi.in: Regenerate.
10145 * hooks.h (hook_bool_mode_mode_true): Declare.
10146 * hooks.c (hook_bool_mode_mode_true): New function.
10147 * combine.c (subst): Use targetm.modes_tieable_p instead of
10149 * dse.c (find_shift_sequence): Likewise.
10150 * expmed.c (extract_low_bits): Likewise.
10151 * lower-subreg.c: Include target.h.
10152 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10154 * rtlanal.c (rtx_cost): Likewise.
10155 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10156 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10157 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10158 (TARGET_MODES_TIEABLE_P): Redefine.
10159 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10160 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10161 (TARGET_MODES_TIEABLE_P): Redefine.
10162 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10163 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10164 (arc_modes_tieable_p): New function.
10165 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10166 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10167 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10168 (arm_modes_tieable_p): Make static.
10169 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10170 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10171 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10172 (TARGET_MODES_TIEABLE_P): Redefine.
10173 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10174 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10175 (TARGET_MODES_TIEABLE_P): Redefine.
10176 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10177 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10178 (cr16_modes_tieable_p): New function.
10179 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10180 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10181 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10182 (TRULY_NOOP_TRUNCATION): Update comment.
10183 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10184 (TRULY_NOOP_TRUNCATION): Update comment.
10185 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10186 (frv_modes_tieable_p): New function.
10187 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10188 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10189 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10190 (TARGET_MODES_TIEABLE_P): Redefine.
10191 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10192 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10193 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10194 (TARGET_MODES_TIEABLE_P): Redefine.
10195 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10196 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10197 (ia64_modes_tieable_p): New function.
10198 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10199 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10200 (iq2000_modes_tieable_p): New function.
10201 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10202 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10203 (lm32_modes_tieable_p): New function.
10204 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10205 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10206 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10207 (TARGET_MODES_TIEABLE_P): Redefine.
10208 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10209 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10210 (m32r_modes_tieable_p): New function.
10211 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10212 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10213 (m68k_modes_tieable_p): New function.
10214 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10215 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10216 (mcore_modes_tieable_p): New function.
10217 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10218 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10220 (TARGET_MODES_TIEABLE_P): Redefine.
10221 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10222 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10223 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10224 (TARGET_MODES_TIEABLE_P): Redefine.
10225 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10226 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10227 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10228 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10229 (mn10300_modes_tieable_p): ...this and make static.
10230 (TARGET_MODES_TIEABLE_P): Redefine.
10231 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10232 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10233 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10234 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10235 (msp430_modes_tieable_p): Make static.
10236 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10237 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10238 (TARGET_MODES_TIEABLE_P): Redefine.
10239 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10240 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10241 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10242 (TARGET_MODES_TIEABLE_P): Redefine.
10243 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10244 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10245 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10246 (TARGET_MODES_TIEABLE_P): Redefine.
10247 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10248 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10249 (pdp11_modes_tieable_p): New function.
10250 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10251 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10252 (rs6000_modes_tieable_p): New function.
10253 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10254 * config/powerpcspe/powerpcspe.md: Update comment.
10255 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10256 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10257 (TARGET_MODES_TIEABLE_P): Redefine.
10258 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10259 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10260 (rl78_modes_tieable_p): New function.
10261 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10262 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10263 (rs6000_modes_tieable_p): New function.
10264 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10265 * config/rs6000/rs6000.md: Update comment.
10266 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10267 * config/rx/rx.c (rx_modes_tieable_p): New function.
10268 (TARGET_MODES_TIEABLE_P): Redefine.
10269 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10270 * config/s390/s390.c (s390_modes_tieable_p): New function.
10271 (TARGET_MODES_TIEABLE_P): Redefine.
10272 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10273 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10274 (sh_modes_tieable_p): New function.
10275 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10276 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10277 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10278 (sparc_modes_tieable_p): Make static.
10279 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10280 * config/spu/spu.c (spu_modes_tieable_p): New function.
10281 (TARGET_MODES_TIEABLE_P): Redefine.
10282 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10283 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10284 (TARGET_MODES_TIEABLE_P): Redefine.
10285 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10286 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10287 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10288 * config/v850/v850.c (v850_modes_tieable_p): New function.
10289 (TARGET_MODES_TIEABLE_P): Redefine.
10290 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10291 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10292 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10293 (visium_modes_tieable_p): New function.
10294 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10295 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10296 (xtensa_modes_tieable_p): New function.
10297 * system.h (MODES_TIEABLE_P): Poison.
10299 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10300 Alan Hayward <alan.hayward@arm.com>
10301 David Sherwood <david.sherwood@arm.com>
10303 * target.def (hard_regno_mode_ok): New hook.
10304 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10305 (TARGET_HARD_REGNO_MODE_OK): ...this.
10306 * doc/tm.texi.in: Regenerate.
10307 * hooks.h (hook_bool_uint_mode_true): Declare.
10308 * hooks.c (hook_bool_uint_mode_true): New function.
10309 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10310 HARD_REGNO_MODE_OK.
10311 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10312 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10313 instead of HARD_REGNO_MODE_OK.
10314 * caller-save.c: Include target.h.
10315 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10316 HARD_REGNO_MODE_OK.
10317 * combine.c (can_combine_p): Likewise.
10318 (combinable_i3pat): Likewise.
10319 (can_change_dest_mode): Likewise.
10320 * expr.c (init_expr_target): Likewise.
10321 (convert_move): Likewise.
10322 (convert_modes): Likewise.
10323 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10324 (setup_prohibited_mode_move_regs): Likewise.
10325 * ira.h (target_ira): Likewise.
10326 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10327 * lra-constraints.c (process_alt_operands): Likewise.
10328 (split_reg): Likewise.
10329 * recog.c (peep2_find_free_register): Likewise.
10330 * ree.c (combine_reaching_defs): Likewise.
10331 * regcprop.c (maybe_mode_change): Likewise.
10332 * reginfo.c (init_reg_sets_1): Likewise.
10333 (choose_hard_reg_mode): Likewise.
10334 (simplifiable_subregs): Likewise.
10335 * regrename.c (check_new_reg_p): Likewise.
10336 * reload.c (find_valid_class): Likewise.
10337 (find_valid_class_1): Likewise.
10338 (reload_inner_reg_of_subreg): Likewise.
10339 (push_reload): Likewise.
10340 (combine_reloads): Likewise.
10341 (find_dummy_reload): Likewise.
10342 (find_reloads): Likewise.
10343 * reload1.c (find_reg): Likewise.
10344 (set_reload_reg): Likewise.
10345 (allocate_reload_reg): Likewise.
10346 (choose_reload_regs): Likewise.
10347 (reload_adjust_reg_for_temp): Likewise.
10348 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10349 (simplify_subreg_regno): Likewise.
10350 * sel-sched.c (init_regs_for_mode): Likewise.
10351 * varasm.c (make_decl_rtl): Likewise.
10352 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10353 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10354 HARD_REGNO_MODE_OK.
10355 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10356 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10357 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10358 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
10359 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
10360 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10361 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
10362 (arc_mode_class): Delete.
10363 (HARD_REGNO_MODE_OK): Delete.
10364 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10365 (arc_hard_regno_mode_ok): Rename old array to...
10366 (arc_hard_regno_mode_ok_modes): ...this.
10367 (arc_conditional_register_usage): Update accordingly.
10368 (arc_mode_class): Make static.
10369 (arc_hard_regno_mode_ok): New function.
10370 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
10371 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
10372 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10373 (arm_hard_regno_mode_ok): Make static.
10374 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
10375 HARD_REGNO_MODE_OK.
10376 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
10377 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
10378 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10380 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10381 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10382 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10383 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10384 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10385 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10386 * config/bfin/predicates.md (valid_reg_operand): Use
10387 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
10388 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
10389 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
10390 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10391 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
10392 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
10393 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10394 (cr16_hard_regno_mode_ok): Make static and return a bool.
10395 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
10396 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10397 (cris_hard_regno_mode_ok): New function.
10398 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
10399 (epiphany_mode_class): Delete.
10400 (HARD_REGNO_MODE_OK): Delete.
10401 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
10402 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10403 (hard_regno_mode_ok): Rename to...
10404 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
10405 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
10406 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
10407 HARD_REGNO_MODE_OK.
10408 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
10409 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
10410 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10411 (frv_hard_regno_mode_ok): Make static and return a bool.
10412 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
10413 HARD_REGNO_MODE_OK.
10414 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
10415 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
10416 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
10417 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
10419 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10420 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
10421 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
10422 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
10424 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10425 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
10426 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10427 (ia64_hard_regno_mode_ok): New function.
10428 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
10429 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10430 (iq2000_hard_regno_mode_ok): New function.
10431 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
10432 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10433 (lm32_hard_regno_mode_ok): New function.
10434 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
10435 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
10436 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
10437 instead of HARD_REGNO_MODE_OK.
10438 (m32c_hard_regno_ok): Rename to...
10439 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
10440 (m32c_cannot_change_mode_class): Update accordingly.
10441 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10442 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
10443 (m32r_mode_class): Delete.
10444 (HARD_REGNO_MODE_OK): Delete.
10445 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10446 (m32r_hard_regno_mode_ok): Rename to...
10447 (m32r_hard_regno_modes): ...this.
10448 (m32r_mode_class): Make static.
10449 (m32r_hard_regno_mode_ok): New function.
10450 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
10451 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
10452 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10453 (m68k_hard_regno_mode_ok): Make static.
10454 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
10455 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10456 (mcore_hard_regno_mode_ok): New function.
10457 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
10458 (HARD_REGNO_MODE_OK): Delete.
10459 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
10461 (microblaze_hard_regno_mode_ok_p): ...this and make static.
10462 (microblaze_hard_regno_mode_ok): New function.
10463 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10464 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
10465 (mips_hard_regno_mode_ok): Delete.
10466 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
10467 (mips_hard_regno_mode_ok_p): ...this and make static.
10468 (mips_hard_regno_mode_ok_p): Rename to...
10469 (mips_hard_regno_mode_ok_uncached): ...this.
10470 (mips_hard_regno_mode_ok): New function.
10471 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
10472 of HARD_REGNO_MODE_OK.
10473 (mips_option_override): Update after above name changes.
10474 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10475 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
10476 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
10477 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
10478 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
10479 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10480 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
10481 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
10482 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
10483 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10484 (msp430_hard_regno_mode_ok): Make static and return a bool.
10485 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
10486 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
10487 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
10489 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10490 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
10491 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
10492 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
10493 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
10494 (PA_HARD_REGNO_MODE_OK): ...this
10495 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
10496 (PA_HARD_REGNO_MODE_OK): ...this.
10497 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10498 (pa_hard_regno_mode_ok): New function.
10499 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
10500 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10501 (pdp11_hard_regno_mode_ok): New function.
10502 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
10503 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
10505 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
10507 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10508 (rs6000_hard_regno_mode_ok): Rename to...
10509 (rs6000_hard_regno_mode_ok_uncached): ...this.
10510 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10511 (rs6000_hard_regno_mode_ok): New function.
10512 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
10513 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
10514 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
10515 (riscv_hard_regno_mode_ok): ...this and make static.
10516 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10517 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
10518 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
10519 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10520 (rl78_hard_regno_mode_ok): Make static and return bool.
10521 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
10522 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
10524 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
10525 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10526 (rs6000_hard_regno_mode_ok): Rename to...
10527 (rs6000_hard_regno_mode_ok_uncached): ...this.
10528 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10529 (rs6000_hard_regno_mode_ok): New function.
10530 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
10531 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
10532 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10533 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
10534 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
10535 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
10536 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10537 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
10538 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
10539 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10540 (sh_hard_regno_mode_ok): Make static.
10541 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
10542 instead of HARD_REGNO_MODE_OK.
10543 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
10544 (sparc_mode_class): Delete.
10545 (HARD_REGNO_MODE_OK): Delete.
10546 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10547 (hard_regno_mode_classes): Make static.
10548 (sparc_mode_class): Likewise.
10549 (sparc_hard_regno_mode_ok): New function.
10550 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
10551 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
10552 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
10554 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10555 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
10556 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
10557 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
10558 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
10559 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10560 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
10561 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
10562 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10563 (visium_hard_regno_mode_ok): New function.
10564 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
10565 instead of HARD_REGNO_MODE_OK.
10566 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
10567 (HARD_REGNO_MODE_OK): Delete.
10568 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
10569 (xtensa_hard_regno_mode_ok_p): ...this and make static.
10570 (xtensa_option_override): Update accordingly.
10571 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10572 (xtensa_hard_regno_mode_ok): New function.
10573 * system.h (HARD_REGNO_MODE_OK): Poison.
10575 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10576 Alan Hayward <alan.hayward@arm.com>
10577 David Sherwood <david.sherwood@arm.com>
10579 * target.def (hard_regno_call_part_clobbered): New hook.
10580 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
10581 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
10582 * doc/tm.texi: Regenerate.
10583 * hooks.h (hook_bool_uint_mode_false): Declare.
10584 * hooks.c (hook_bool_uint_mode_false): New function.
10585 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10586 * cselib.c (cselib_process_insn): Use
10587 targetm.hard_regno_call_part_clobbered instead of
10588 HARD_REGNO_CALL_PART_CLOBBERED.
10589 * ira-conflicts.c (ira_build_conflicts): Likewise.
10590 * ira-costs.c (ira_tune_allocno_costs): Likewise.
10591 * lra-constraints.c (need_for_call_save_p): Likewise.
10592 * lra-lives.c: Include target.h.
10593 (check_pseudos_live_through_calls): Use
10594 targetm.hard_regno_call_part_clobbered instead of
10595 HARD_REGNO_CALL_PART_CLOBBERED.
10596 * regcprop.c: Include target.h.
10597 (copyprop_hardreg_forward_1): Use
10598 targetm.hard_regno_call_part_clobbered instead of
10599 HARD_REGNO_CALL_PART_CLOBBERED.
10600 * reginfo.c (choose_hard_reg_mode): Likewise.
10601 * regrename.c (check_new_reg_p): Likewise.
10602 * reload.c (find_equiv_reg): Likewise.
10603 * reload1.c (emit_reload_insns): Likewise.
10604 * sched-deps.c (deps_analyze_insn): Likewise.
10605 * sel-sched.c (init_regs_for_mode): Likewise.
10606 (mark_unavailable_hard_regs): Likewise.
10607 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
10608 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10609 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
10611 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10612 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10613 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
10615 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
10617 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10618 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10619 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
10621 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10622 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10623 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
10625 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10626 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
10628 * config/powerpcspe/powerpcspe.c
10629 (rs6000_hard_regno_call_part_clobbered): New function.
10630 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10631 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10632 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
10634 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10635 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10636 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
10638 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10639 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10640 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
10642 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10643 Alan Hayward <alan.hayward@arm.com>
10644 David Sherwood <david.sherwood@arm.com>
10646 * rtl.h (subreg_memory_offset): Declare.
10647 * emit-rtl.c (subreg_memory_offset): New function.
10648 * expmed.c (store_bit_field_1): Use it.
10649 * expr.c (undefined_operand_subword_p): Likewise.
10650 * simplify-rtx.c (simplify_subreg): Likewise.
10652 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
10654 PR rtl-optimization/57448
10657 * optabs.c (expand_atomic_load): Place compiler memory barriers if
10658 using atomic_load pattern.
10659 (expand_atomic_store): Likewise.
10661 2017-09-04 Jakub Jelinek <jakub@redhat.com>
10664 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
10665 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
10666 handling. Use replace_call_with_value with NULL instead of
10667 gsi_replace, unlink_stmt_vdef and release_defs.
10669 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
10672 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
10674 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10677 * rtl.h (emit_library_call_value_1): Declare.
10678 (emit_library_call): Replace declaration with a series of overloads.
10679 Remove the parameter count argument.
10680 (emit_library_call_value): Likewise.
10681 * calls.c (emit_library_call_value_1): Make global. Replace varargs
10682 with an "rtx_mode_t *".
10683 (emit_library_call_value): Delete.
10684 (emit_library_call): Likewise.
10685 * asan.c (asan_emit_stack_protection): Update calls accordingly.
10686 (asan_emit_allocas_unpoison): Likewise.
10687 * builtins.c (expand_builtin_powi): Likewise.
10688 (expand_asan_emit_allocas_unpoison): Likewise.
10689 * cfgexpand.c (expand_main_function): Likewise.
10690 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
10691 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
10692 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
10693 * config/arm/arm.c (arm_trampoline_init): Likewise.
10694 (arm_call_tls_get_addr): Likewise.
10695 (arm_expand_divmod_libfunc): Likewise.
10696 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
10697 (smulsi3_highpart): Likewise.
10698 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
10699 (c6x_expand_compare): Likewise.
10700 (c6x_expand_movmem): Likewise.
10701 * config/frv/frv.c (frv_trampoline_init): Likewise.
10702 * config/i386/i386.c (ix86_trampoline_init): Likewise.
10703 (ix86_expand_divmod_libfunc): Likewise.
10704 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
10705 (ia64_expand_compare): Likewise.
10706 (ia64_profile_hook): Likewise.
10707 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
10708 (nonlocal_goto): Likewise.
10709 (restore_stack_nonlocal): Likewise.
10710 * config/m32r/m32r.c (block_move_call): Likewise.
10711 (m32r_trampoline_init): Likewise.
10712 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
10713 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
10714 (m68k_call_m68k_read_tp): Likewise.
10715 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
10716 (microblaze_expand_divide): Likewise.
10717 * config/mips/mips.h (mips_args): Likewise.
10718 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
10719 (MIPS_ICACHE_SYNC): Likewise.
10720 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
10721 (nios2_trampoline_init): Likewise.
10722 * config/pa/pa.c (hppa_tls_call): Likewise.
10723 (pa_trampoline_init): Likewise.
10724 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
10725 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
10726 (expand_strn_compare): Likewise.
10727 (rs6000_generate_compare): Likewise.
10728 (rs6000_expand_float128_convert): Likewise.
10729 (output_profile_hook): Likewise.
10730 (rs6000_trampoline_init): Likewise.
10731 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
10732 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
10733 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
10734 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
10735 (rs6000_generate_compare): Likewise.
10736 (rs6000_expand_float128_convert): Likewise.
10737 (output_profile_hook): Likewise.
10738 (rs6000_trampoline_init): Likewise.
10739 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
10740 * config/sh/sh.c (sh_trampoline_init): Likewise.
10741 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
10742 (sparc_emit_float_lib_cmp): Likewise.
10743 (sparc32_initialize_trampoline): Likewise.
10744 (sparc64_initialize_trampoline): Likewise.
10745 (sparc_profile_hook): Likewise.
10746 * config/spu/spu.c (ea_load_store): Likewise.
10747 * config/spu/spu.md (floatunssidf2): Likewise.
10748 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
10749 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
10750 * config/visium/visium.c (expand_block_move_4): Likewise.
10751 (expand_block_move_2): Likewise.
10752 (expand_block_move_1): Likewise.
10753 (expand_block_set_4): Likewise.
10754 (expand_block_set_2): Likewise.
10755 (expand_block_set_1): Likewise.
10756 (visium_trampoline_init): Likewise.
10757 (visium_profile_hook): Likewise.
10758 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
10759 (xtensa_setup_frame_addresses): Likewise.
10760 (xtensa_trampoline_init): Likewise.
10761 * except.c (sjlj_emit_function_enter): Likewise.
10762 (sjlj_emit_function_exit): Likewise.
10763 * explow.c (allocate_dynamic_stack_space): Likewise.
10764 (probe_stack_range): Likewise.
10765 * expr.c (convert_mode_scalar): Likewise.
10766 * optabs.c (expand_binop): Likewise.
10767 (expand_twoval_binop_libfunc): Likewise.
10768 (expand_unop): Likewise.
10769 (prepare_cmp_insn): Likewise.
10770 (prepare_float_lib_cmp): Likewise.
10771 (expand_float): Likewise.
10772 (expand_fix): Likewise.
10773 (expand_fixed_convert): Likewise.
10774 (maybe_emit_sync_lock_test_and_set): Likewise.
10775 (expand_atomic_compare_and_swap): Likewise.
10776 (expand_mem_thread_fence): Likewise.
10777 (expand_atomic_fetch_op): Likewise.
10779 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
10781 * doc/generic.texi (OpenACC): Adjust URL.
10782 * doc/invoke.texi (C Dialect Options): Ditto.
10784 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
10786 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
10787 predicate for operand 1. Add (m,<S>) constraint.
10788 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
10789 Prevent memory operand 1 with register operand 2.
10791 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
10793 PR rtl-optimization/82024
10794 * combine.c (try_combine): If the combination result is a PARALLEL,
10795 and we only need to retain the SET in there that would be placed
10796 at I2, check that we can place that at I3 instead, before doing so.
10798 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10801 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
10803 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
10804 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
10807 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
10808 Jeff Law <law@redhat.com>
10810 * varasm.c (bss_initializer_p): Do not put constants into .bss
10811 (categorize_decl_for_section): Handle bss_initializer_p returning
10812 false when DECL_INITIAL is NULL.
10814 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10817 * config/s390/s390.c (s390_can_inline_p): New function.
10819 2017-09-01 Jeff Law <law@redhat.com>
10821 PR tree-optimization/82052
10822 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
10823 Always initialize the returned slot after a hash table miss
10824 when INSERT is true.
10826 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
10828 * config/s390/s390.md (mem_signal_fence): Remove.
10829 * doc/md.texi (mem_signal_fence): Remove.
10830 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
10832 * target-insns.def (mem_signal_fence): Remove.
10834 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10837 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
10840 * asan.c (create_odr_indicator): Strip name encoding from assembler
10841 name before appending it after __odr_asan_.
10843 2017-09-01 Martin Liska <mliska@suse.cz>
10845 PR tree-optimization/82059
10846 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
10847 frequency only when an edge is redirected.
10849 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10851 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
10852 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
10853 (arc_conditional_register_usage): Remove ARC600 lp_count
10855 (arc_file_start): Emit Tag_ARC_CPU_variation.
10856 (arc_can_use_doloop_p): New conditions to use ZOLs.
10857 (hwloop_fail): New function.
10858 (hwloop_optimize): Likewise.
10859 (hwloop_pattern_reg): Likewise.
10860 (arc_doloop_hooks): New struct, to be used with reorg_loops.
10861 (arc_reorg_loops): New function, calls reorg_loops.
10862 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
10863 (arc600_corereg_hazard): Remove ZOL checking, case handled by
10865 (arc_loop_hazard): Remove function, functionality moved into
10867 (arc_hazard): Remove arc_loop_hazard call.
10868 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
10869 into hwloop_optimize.
10870 (arc_label_align): Remove ZOL handling.
10871 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
10872 * config/arc/arc.md (doloop_begin): Remove pattern.
10873 (doloop_begin_i): Likewise.
10874 (doloop_end_i): Likewise.
10875 (doloop_fallback): Likewise.
10876 (doloop_fallback_m): Likewise.
10877 (doloop_end): Reimplement expand.
10878 (arc_lp): New pattern for LP instruction.
10879 (loop_end): New pattern.
10880 (loop_fail): Likewise.
10881 (decrement_and_branch_until_zero): Likewise.
10882 * config/arc/arc.opt (mlpc-width): New option.
10883 * doc/invoke.texi (mlpc-width): Document option.
10885 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10887 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
10888 (arc_ccfsm_advance): Fix checking for delay slots.
10889 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
10891 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10893 * config/arc/arc.md (movqi_insn): Add stores to save constant long
10895 (movhi_insn): Update store instruction constraint which are saving
10896 6-bit short immediates.
10897 (movsi_insn): Consider also short scaled load operations.
10898 (zero_extendhisi2_i): Use Usd constraint instead of T.
10899 (extendhisi2_i): Add q constraint.
10900 (arc_clzsi2): Add type and length attributes.
10901 (arc_ctzsi2): Likewise.
10902 * config/arc/constraints.md (Usc): Update constraint, the
10903 assembler can parse two relocations for a single instruction.
10905 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10907 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
10908 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
10910 2017-08-31 Olivier Hainque <hainque@adacore.com>
10912 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
10913 match as powerpc-wrs-vxworks*.
10915 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
10917 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
10918 register constraint for by-element operand.
10919 (aarch64_mls_elt_merge<mode>): Likewise.
10921 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10923 * config/arc/arc.c (arc_can_follow_jump): Check for short
10926 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10928 * config.gcc: Use g.opt for arc.
10929 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
10930 functionality moved to ...
10931 (legitimate_scaled_address_p): New function, ...here.
10932 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
10933 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
10934 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
10936 (arc_override_options): Handle G option.
10937 (arc_output_pic_addr_const): Correct function definition.
10938 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
10939 (arc_decl_anon_ns_mem_p): Delete.
10940 (arc_in_small_data_p): Overhaul this function to take into
10941 consideration the value given via G option.
10942 (arc_rewrite_small_data_1): Renamed and corrected old
10943 arc_rewrite_small_data function.
10944 (arc_rewrite_small_data): New function.
10945 (small_data_pattern): Don't use pic_offset_table_rtx.
10946 * config/arc/arc.h (CC1_SPEC): Recognize G option.
10947 * config/arc/simdext.md (movmisalignv2hi): Use
10948 prepare_move_operands function.
10950 (movmisalign*): Likewise.
10951 * doc/invoke.texi (ARC options): Document -G option.
10953 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10955 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
10957 * config/arc/arc.c (arc_print_operand): Output scalled address for
10958 sdata whenever is possible.
10959 (arc_in_small_data_p): Allow sdata for 64bit datum when double
10960 load/stores are available.
10961 (compact_sda_memory_operand): Check for the alignment required by
10962 code density instructions.
10963 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
10965 * config/arc/constraints.md (Usd): Update constraint.
10966 (Us0): New constraint.
10967 (Usc): Update constraint.
10969 2017-08-31 Richard Biener <rguenther@suse.de>
10971 PR middle-end/82054
10972 * dwarf2out.c (dwarf2out_early_global_decl): Process each
10973 function only once.
10975 2017-08-31 Tamar Christina <tamar.christina@arm.com>
10977 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
10978 Resize type_signature.
10980 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10981 Alan Hayward <alan.hayward@arm.com>
10982 David Sherwood <david.sherwood@arm.com>
10984 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
10985 subregs whose inner modes can be stored in GPRs.
10986 (aarch64_classify_index): Likewise.
10988 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10989 Alan Hayward <alan.hayward@arm.com>
10990 David Sherwood <david.sherwood@arm.com>
10992 * config/aarch64/iterators.md (V_cmp_result): Rename to...
10993 (V_INT_EQUIV): ...this.
10994 (v_cmp_result): Rename to...
10995 (v_int_equiv): ...this.
10996 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
10997 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
10998 (copysign<mode>3): Likewise.
10999 (aarch64_simd_bsl<mode>_internal): Likewise.
11000 (aarch64_simd_bsl<mode>): Likewise.
11001 (vec_cmp<mode><mode>): Likewise.
11002 (vcond<mode><mode>): Likewise.
11003 (vcond<v_cmp_mixed><mode>): Likewise.
11004 (vcondu<mode><v_cmp_mixed>): Likewise.
11005 (aarch64_cm<optab><mode>): Likewise.
11006 (aarch64_cmtst<mode>): Likewise.
11007 (aarch64_fac<optab><mode>): Likewise.
11008 (vec_perm_const<mode>): Likewise.
11009 (vcond_mask_<mode><v_cmp_result>): Rename to...
11010 (vcond_mask_<mode><v_int_equiv>): ...this.
11011 (vec_cmp<mode><v_cmp_result>): Rename to...
11012 (vec_cmp<mode><v_int_equiv>): ...this.
11014 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11015 Alan Hayward <alan.hayward@arm.com>
11016 David Sherwood <david.sherwood@arm.com>
11018 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11020 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11021 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11022 (UNSPEC_LD4_DREG): New unspecs.
11023 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11024 (aarch64_ld2<mode>_dreg_be): Replace with...
11025 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11027 (aarch64_ld3<mode>_dreg_le)
11028 (aarch64_ld3<mode>_dreg_be): Replace with...
11029 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11031 (aarch64_ld4<mode>_dreg_le)
11032 (aarch64_ld4<mode>_dreg_be): Replace with...
11033 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11036 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11038 PR tree-optimization/81987
11039 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11040 insert an initializer in a location not dominated by the stride
11043 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11045 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11046 on the entire header of the finally block in the fallthru case.
11048 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11050 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11052 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11054 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11055 rs6000_emit_move_from_cr and call renamed function.
11056 (rs6000_emit_prologue): Call renamed functions.
11057 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11058 movesi_from_cr, remove volatile CRs.
11060 2017-08-30 Jon Beniston <jon@beniston.com>
11061 Richard Biener <rguenther@suse.de>
11063 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11064 of VECTOR_MODE_P check.
11065 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11066 element vector types.
11068 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11070 * df.h (df_read_modify_subreg_p): Remove in favor of...
11071 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11072 const_rtx instead of an rtx.
11073 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11074 * df-problems.c (df_word_lr_mark_ref): Likewise.
11075 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11076 (mark_pseudo_reg_dead): Likewise.
11077 (mark_ref_dead): Likewise.
11078 * reginfo.c (init_subregs_of_mode): Likewise.
11079 * sched-deps.c (sched_analyze_1): Likewise.
11080 * df-scan.c (df_def_record_1): Likewise.
11081 (df_uses_record): Likewise.
11082 (df_read_modify_subreg_p): Remove in favor of...
11083 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11084 const_rtx instead of an rtx.
11086 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11087 Alan Hayward <alan.hayward@arm.com>
11088 David Sherwood <david.sherwood@arm.com>
11090 * rtl.h (partial_subreg_p): New function.
11091 * caller-save.c (save_call_clobbered_regs): Use it.
11092 * calls.c (expand_call): Likewise.
11093 * combine.c (combinable_i3pat): Likewise.
11094 (simplify_set): Likewise.
11095 (make_extraction): Likewise.
11096 (make_compound_operation_int): Likewise.
11097 (gen_lowpart_or_truncate): Likewise.
11098 (force_to_mode): Likewise.
11099 (make_field_assignment): Likewise.
11100 (reg_truncated_to_mode): Likewise.
11101 (record_truncated_value): Likewise.
11102 (move_deaths): Likewise.
11103 * cse.c (record_jump_cond): Likewise.
11104 (cse_insn): Likewise.
11105 * cselib.c (cselib_lookup_1): Likewise.
11106 * expmed.c (extract_bit_field_using_extv): Likewise.
11107 * function.c (assign_parm_setup_reg): Likewise.
11108 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11109 * ira-build.c (create_insn_allocnos): Likewise.
11110 * lra-coalesce.c (merge_pseudos): Likewise.
11111 * lra-constraints.c (match_reload): Likewise.
11112 (simplify_operand_subreg): Likewise.
11113 (curr_insn_transform): Likewise.
11114 * lra-lives.c (process_bb_lives): Likewise.
11115 * lra.c (new_insn_reg): Likewise.
11116 (lra_substitute_pseudo): Likewise.
11117 * regcprop.c (mode_change_ok): Likewise.
11118 (maybe_mode_change): Likewise.
11119 (copyprop_hardreg_forward_1): Likewise.
11120 * reload.c (push_reload): Likewise.
11121 (find_reloads): Likewise.
11122 (find_reloads_subreg_address): Likewise.
11123 * reload1.c (alter_reg): Likewise.
11124 (eliminate_regs_1): Likewise.
11125 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11127 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11129 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11130 back to if statements, including unpack.
11132 2017-08-30 Martin Liska <mliska@suse.cz>
11134 PR inline-asm/82001
11135 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11137 (func_checker::compare_asm_inputs_outputs): ... this function.
11138 (func_checker::compare_gimple_asm): Use the function to compare
11139 also ASM constrains.
11140 * ipa-icf-gimple.h: Rename the function.
11142 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11143 Alan Hayward <alan.hayward@arm.com>
11144 David Sherwood <david.sherwood@arm.com>
11146 * coretypes.h (complex_mode): New type.
11147 * gdbhooks.py (build_pretty_printer): Handle it.
11148 * machmode.h (complex_mode): New class.
11149 (complex_mode::includes_p): New function.
11150 (is_complex_int_mode): Likewise.
11151 (is_complex_float_mode): Likewise.
11152 * genmodes.c (get_mode_class): Handle complex mode classes.
11153 * function.c (expand_function_end): Use is_complex_int_mode.
11155 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11156 Alan Hayward <alan.hayward@arm.com>
11157 David Sherwood <david.sherwood@arm.com>
11159 * coretypes.h (scalar_mode_pod): New typedef.
11160 * gdbhooks.py (build_pretty_printer): Handle it.
11161 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11162 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11163 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11164 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11165 as_a <scalar_mode>.
11167 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11168 Alan Hayward <alan.hayward@arm.com>
11169 David Sherwood <david.sherwood@arm.com>
11171 * machmode.h (mode_for_vector): Take a scalar_mode instead
11173 * stor-layout.c (mode_for_vector): Likewise.
11174 * explow.c (promote_mode): Use as_a <scalar_mode>.
11175 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11177 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11178 Alan Hayward <alan.hayward@arm.com>
11179 David Sherwood <david.sherwood@arm.com>
11181 * target.def (preferred_simd_mode): Take a scalar_mode
11182 instead of a machine_mode.
11183 * targhooks.h (default_preferred_simd_mode): Likewise.
11184 * targhooks.c (default_preferred_simd_mode): Likewise.
11185 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11186 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11187 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11188 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11189 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11190 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11191 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11192 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11194 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11195 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11196 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11197 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11198 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11199 * doc/tm.texi: Regenerate.
11200 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11203 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11204 Alan Hayward <alan.hayward@arm.com>
11205 David Sherwood <david.sherwood@arm.com>
11207 * target.def (scalar_mode_supported_p): Take a scalar_mode
11208 instead of a machine_mode.
11209 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11210 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11211 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11212 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11213 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11214 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11215 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11216 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11217 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11218 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11219 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11220 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11221 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11222 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11224 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11225 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11226 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11227 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11228 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11230 * doc/tm.texi: Regenerate.
11232 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11233 Alan Hayward <alan.hayward@arm.com>
11234 David Sherwood <david.sherwood@arm.com>
11236 * coretypes.h (opt_scalar_mode): New typedef.
11237 * gdbhooks.py (build_pretty_printers): Handle it.
11238 * machmode.h (mode_iterator::get_2xwider): Add overload for
11240 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11242 * expr.c (convert_mode_scalar): Likewise.
11243 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11244 * optabs.c (expand_float): Likewise.
11245 (expand_fix): Likewise.
11246 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11248 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11249 Alan Hayward <alan.hayward@arm.com>
11250 David Sherwood <david.sherwood@arm.com>
11252 * optabs.c (expand_float): Explicitly check for scalars before
11253 using a branching expansion.
11254 (expand_fix): Likewise.
11256 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11257 Alan Hayward <alan.hayward@arm.com>
11258 David Sherwood <david.sherwood@arm.com>
11260 * expr.c (convert_mode): Split scalar handling out into...
11261 (convert_mode_scalar): ...this new function. Treat the modes
11264 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11265 Alan Hayward <alan.hayward@arm.com>
11266 David Sherwood <david.sherwood@arm.com>
11268 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11270 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11272 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11273 Alan Hayward <alan.hayward@arm.com>
11274 David Sherwood <david.sherwood@arm.com>
11276 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11277 rather than a machine_mode.
11278 (fixed_from_string): Likewise.
11279 (fixed_convert): Likewise.
11280 (fixed_convert_from_int): Likewise.
11281 (fixed_convert_from_real): Likewise.
11282 (real_convert_from_fixed): Likewise.
11283 * fixed-value.c (fixed_from_double_int): Likewise.
11284 (fixed_from_string): Likewise.
11285 (fixed_convert): Likewise.
11286 (fixed_convert_from_int): Likewise.
11287 (fixed_convert_from_real): Likewise.
11288 (real_convert_from_fixed): Likewise.
11289 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
11291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11292 Alan Hayward <alan.hayward@arm.com>
11293 David Sherwood <david.sherwood@arm.com>
11295 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11296 of separate mode class checks. Do not allow vector modes here.
11297 (immed_wide_int_const): Use as_a <scalar_mode>.
11298 * explow.c (trunc_int_for_mode): Likewise.
11299 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11300 (wi::shwi): Likewise.
11301 (wi::min_value): Likewise.
11302 (wi::max_value): Likewise.
11303 * dwarf2out.c (loc_descriptor): Likewise.
11304 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11305 for CONST_WIDE_INT.
11307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11308 Alan Hayward <alan.hayward@arm.com>
11309 David Sherwood <david.sherwood@arm.com>
11311 * tree.h (SCALAR_TYPE_MODE): New macro.
11312 * expr.c (expand_expr_addr_expr_1): Use it.
11313 (expand_expr_real_2): Likewise.
11314 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11315 (fold_convert_const_fixed_from_int): Likewise.
11316 (fold_convert_const_fixed_from_real): Likewise.
11317 (native_encode_fixed): Likewise
11318 (native_encode_complex): Likewise
11319 (native_encode_vector): Likewise.
11320 (native_interpret_fixed): Likewise.
11321 (native_interpret_real): Likewise.
11322 (native_interpret_complex): Likewise.
11323 (native_interpret_vector): Likewise.
11324 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11325 (simd_clone_adjust_argument_types): Likewise.
11326 (simd_clone_init_simd_arrays): Likewise.
11327 (simd_clone_adjust): Likewise.
11328 * stor-layout.c (layout_type): Likewise.
11329 * tree.c (build_minus_one_cst): Likewise.
11330 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11331 * tree-inline.c (estimate_move_cost): Likewise.
11332 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11333 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11334 (vectorizable_reduction): Likewise.
11335 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11336 (vect_recog_mixed_size_cond_pattern): Likewise.
11337 (check_bool_pattern): Likewise.
11338 (adjust_bool_pattern): Likewise.
11339 (search_type_for_mask_1): Likewise.
11340 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11341 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11342 (vectorizable_load): Likewise.
11343 (vectorizable_store): Likewise.
11344 * ubsan.c (ubsan_encode_value): Likewise.
11345 * varasm.c (output_constant): Likewise.
11347 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11348 Alan Hayward <alan.hayward@arm.com>
11349 David Sherwood <david.sherwood@arm.com>
11351 * coretypes.h (scalar_mode): New class.
11352 * machmode.h (scalar_mode): Likewise.
11353 (scalar_mode::includes_p): New function.
11354 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11355 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11356 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11357 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11358 * expmed.c (store_bit_field_1): Likewise.
11359 (extract_bit_field_1): Likewise.
11360 * expr.c (write_complex_part): Likewise.
11361 (read_complex_part): Likewise.
11362 (emit_move_complex_push): Likewise.
11363 (expand_expr_real_2): Likewise.
11364 * function.c (assign_parm_setup_reg): Likewise.
11365 (assign_parms_unsplit_complex): Likewise.
11366 * optabs.c (expand_binop): Likewise.
11367 * rtlanal.c (subreg_get_info): Likewise.
11368 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11369 * varasm.c (output_constant_pool_2): Likewise.
11371 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11372 Alan Hayward <alan.hayward@arm.com>
11373 David Sherwood <david.sherwood@arm.com>
11375 * expmed.c (extract_high_half): Use scalar_int_mode and remove
11377 (expmed_mult_highpart_optab): Likewise.
11378 (expmed_mult_highpart): Likewise.
11380 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11381 Alan Hayward <alan.hayward@arm.com>
11382 David Sherwood <david.sherwood@arm.com>
11384 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11385 instead of a machine_mode.
11386 (builtin_memset_read_str): Likewise.
11387 * builtins.c (c_readstr): Likewise.
11388 (builtin_memcpy_read_str): Likewise.
11389 (builtin_strncpy_read_str): Likewise.
11390 (builtin_memset_read_str): Likewise.
11391 (builtin_memset_gen_str): Likewise.
11392 (expand_builtin_signbit): Use scalar_int_mode for local variables.
11393 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
11394 instead of a machine_mode.
11395 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
11397 (make_extraction): Likewise.
11398 (try_widen_shift_mode): Take and return scalar_int_modes instead
11400 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
11401 a scalar_int_mode instead of a machine_mode.
11402 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
11403 (avr_addr_space_pointer_mode): Likewise.
11404 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
11405 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
11406 (msp430_unwind_word_mode): Likewise.
11407 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
11408 (spu_addr_space_pointer_mode): Likewise.
11409 (spu_addr_space_address_mode): Likewise.
11410 (spu_libgcc_cmp_return_mode): Likewise.
11411 (spu_libgcc_shift_count_mode): Likewise.
11412 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
11413 (rl78_addr_space_pointer_mode): Likewise.
11414 (fl78_unwind_word_mode): Likewise.
11415 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
11417 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
11418 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
11419 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
11420 (mips_valid_pointer_mode): Likewise.
11421 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
11422 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
11423 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
11425 (ft32_addr_space_address_mode): Likewise.
11426 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
11427 scalar_int_mode instead of a machine_mode.
11428 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
11430 (m32c_addr_space_address_mode): Likewise.
11431 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
11432 (rs6000_eh_return_filter_mode): Likewise.
11433 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
11434 (rs6000_eh_return_filter_mode): Likewise.
11435 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
11436 (s390_libgcc_shift_count_mode): Likewise.
11437 (s390_unwind_word_mode): Likewise.
11438 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
11440 * target.def (mode_rep_extended): Likewise.
11441 (valid_pointer_mode): Likewise.
11442 (addr_space.valid_pointer_mode): Likewise.
11443 (eh_return_filter_mode): Return a scalar_int_mode rather than
11445 (libgcc_cmp_return_mode): Likewise.
11446 (libgcc_shift_count_mode): Likewise.
11447 (unwind_word_mode): Likewise.
11448 (addr_space.pointer_mode): Likewise.
11449 (addr_space.address_mode): Likewise.
11450 * doc/tm.texi: Regenerate.
11451 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
11453 (do_jump): Use scalar_int_mode for local variables.
11454 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
11455 rather than a machine_mode.
11456 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
11457 (scompare_loc_descriptor_wide): Likewise.
11458 (scompare_loc_descriptor_narrow): Likewise.
11459 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
11461 * except.c (sjlj_emit_dispatch_table): Likewise.
11462 (expand_builtin_eh_copy_values): Likewise.
11463 * explow.c (convert_memory_address_addr_space_1): Likewise.
11464 Take a scalar_int_mode rather than a machine_mode.
11465 (convert_memory_address_addr_space): Take a scalar_int_mode rather
11466 than a machine_mode.
11467 (memory_address_addr_space): Use scalar_int_mode for local variables.
11468 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
11469 rather than a machine_mode.
11470 * expmed.c (mask_rtx): Likewise.
11471 (init_expmed_one_conv): Likewise.
11472 (expand_mult_highpart_adjust): Likewise.
11473 (extract_high_half): Likewise.
11474 (expmed_mult_highpart_optab): Likewise.
11475 (expmed_mult_highpart): Likewise.
11476 (expand_smod_pow2): Likewise.
11477 (expand_sdiv_pow2): Likewise.
11478 (emit_store_flag_int): Likewise.
11479 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
11481 (extract_low_bits): Likewise.
11482 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
11484 * expr.c (pieces_addr::adjust): Likewise.
11485 (can_store_by_pieces): Likewise.
11486 (store_by_pieces): Likewise.
11487 (clear_by_pieces_1): Likewise.
11488 (expand_expr_addr_expr_1): Likewise.
11489 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
11490 (expand_expr_real_1): Likewise.
11491 (try_casesi): Likewise.
11492 * final.c (shorten_branches): Likewise.
11493 * fold-const.c (fold_convert_const_int_from_fixed): Change the
11494 type of "mode" to machine_mode.
11495 * internal-fn.c (expand_arith_overflow_result_store): Take a
11496 scalar_int_mode rather than a machine_mode.
11497 (expand_mul_overflow): Use scalar_int_mode for local variables.
11498 * loop-doloop.c (doloop_modify): Likewise.
11499 (doloop_optimize): Likewise.
11500 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
11501 than a machine_mode.
11502 (expand_doubleword_shift_condmove): Likewise.
11503 (expand_doubleword_shift): Likewise.
11504 (expand_doubleword_clz): Likewise.
11505 (expand_doubleword_popcount): Likewise.
11506 (expand_doubleword_parity): Likewise.
11507 (expand_absneg_bit): Use scalar_int_mode for local variables.
11508 (prepare_float_lib_cmp): Likewise.
11509 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
11510 rather than a machine_mode.
11511 (convert_memory_address_addr_space): Likewise.
11512 (get_mode_bounds): Likewise.
11513 (get_address_mode): Return a scalar_int_mode rather than a
11515 * rtlanal.c (get_address_mode): Likewise.
11516 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
11517 than a machine_mode.
11518 * targhooks.c (default_mode_rep_extended): Likewise.
11519 (default_valid_pointer_mode): Likewise.
11520 (default_addr_space_valid_pointer_mode): Likewise.
11521 (default_eh_return_filter_mode): Return a scalar_int_mode rather
11522 than a machine_mode.
11523 (default_libgcc_cmp_return_mode): Likewise.
11524 (default_libgcc_shift_count_mode): Likewise.
11525 (default_unwind_word_mode): Likewise.
11526 (default_addr_space_pointer_mode): Likewise.
11527 (default_addr_space_address_mode): Likewise.
11528 * targhooks.h (default_eh_return_filter_mode): Likewise.
11529 (default_libgcc_cmp_return_mode): Likewise.
11530 (default_libgcc_shift_count_mode): Likewise.
11531 (default_unwind_word_mode): Likewise.
11532 (default_addr_space_pointer_mode): Likewise.
11533 (default_addr_space_address_mode): Likewise.
11534 (default_mode_rep_extended): Take a scalar_int_mode rather than
11536 (default_valid_pointer_mode): Likewise.
11537 (default_addr_space_valid_pointer_mode): Likewise.
11538 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
11540 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
11541 rather than a machine_mode.
11542 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
11543 for local variables.
11544 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11545 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
11546 than a machine_mode.
11548 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11549 Alan Hayward <alan.hayward@arm.com>
11550 David Sherwood <david.sherwood@arm.com>
11552 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
11553 the mode argument to scalar_int_mode.
11554 (do_jump_by_parts_zero_rtx): Likewise.
11555 (do_jump_by_parts_equality_rtx): Likewise.
11556 (do_jump_by_parts_greater): Take a mode argument.
11557 (do_jump_by_parts_equality): Likewise.
11558 (do_jump_1): Update calls accordingly.
11560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11561 Alan Hayward <alan.hayward@arm.com>
11562 David Sherwood <david.sherwood@arm.com>
11564 * is-a.h (safe_dyn_cast): New function.
11565 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
11566 (jump_table_for_label): Likewise.
11567 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
11568 instead of an rtx_insn *.
11569 (shorten_branches): Use dyn_cast instead of LABEL_P and
11570 JUMP_TABLE_DATA_P. Use jump_table_for_label and
11571 rtx_jump_table_data::get_data_mode.
11572 (final_scan_insn): Likewise.
11574 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11575 Alan Hayward <alan.hayward@arm.com>
11576 David Sherwood <david.sherwood@arm.com>
11578 * combine.c (try_combine): Use is_a <scalar_int_mode> when
11579 trying to combine a full-register integer set with a subreg
11582 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11583 Alan Hayward <alan.hayward@arm.com>
11584 David Sherwood <david.sherwood@arm.com>
11586 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
11587 that is always either address_mode or pointer_mode.
11589 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11590 Alan Hayward <alan.hayward@arm.com>
11591 David Sherwood <david.sherwood@arm.com>
11593 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
11594 when the two are known to be equal.
11596 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11597 Alan Hayward <alan.hayward@arm.com>
11598 David Sherwood <david.sherwood@arm.com>
11600 * simplify-rtx.c (simplify_const_unary_operation): Use
11601 is_a <scalar_int_mode> instead of checking for a nonzero
11602 precision. Forcibly convert op_mode to a scalar_int_mode
11603 in that case. More clearly differentiate the operand and
11604 result modes and use the former when deciding what the value
11605 of a count-bits operation should be. Use is_int_mode instead
11606 of checking for a MODE_INT. Remove redundant check for whether
11607 this mode has a zero precision.
11609 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11610 Alan Hayward <alan.hayward@arm.com>
11611 David Sherwood <david.sherwood@arm.com>
11613 * optabs.c (widen_leading): Change the type of the mode argument
11614 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
11615 (widen_bswap): Likewise.
11616 (expand_parity): Likewise.
11617 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
11618 (expand_ffs): Likewise.
11619 (epand_unop): Check for scalar integer modes before calling the
11622 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11623 Alan Hayward <alan.hayward@arm.com>
11624 David Sherwood <david.sherwood@arm.com>
11626 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
11628 (expand_expr_real_1): Update call accordingly.
11630 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11631 Alan Hayward <alan.hayward@arm.com>
11632 David Sherwood <david.sherwood@arm.com>
11634 * expmed.c (store_bit_field_using_insv): Add op0_mode and
11635 value_mode arguments. Use scalar_int_mode internally.
11636 (store_bit_field_1): Rename the new integer mode from imode
11637 to op0_mode and use it instead of GET_MODE (op0). Update calls
11638 to store_split_bit_field, store_bit_field_using_insv and
11639 store_fixed_bit_field.
11640 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
11641 Use scalar_int_mode internally. Use a bit count rather than a mode
11642 when calculating the largest bit size for get_best_mode.
11643 Update calls to store_split_bit_field and store_fixed_bit_field_1.
11644 (store_fixed_bit_field_1): Add mode and value_mode arguments.
11645 Remove assertion that OP0 has a scalar integer mode.
11646 (store_split_bit_field): Add op0_mode and value_mode arguments.
11647 Update calls to extract_fixed_bit_field.
11648 (extract_bit_field_using_extv): Add an op0_mode argument.
11649 Use scalar_int_mode internally.
11650 (extract_bit_field_1): Rename the new integer mode from imode to
11651 op0_mode and use it instead of GET_MODE (op0). Update calls to
11652 extract_split_bit_field, extract_bit_field_using_extv and
11653 extract_fixed_bit_field.
11654 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
11655 to extract_split_bit_field and extract_fixed_bit_field_1.
11656 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
11657 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
11658 on the target mode.
11659 (extract_split_bit_field): Add an op0_mode argument. Update call
11660 to extract_fixed_bit_field.
11662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11663 Alan Hayward <alan.hayward@arm.com>
11664 David Sherwood <david.sherwood@arm.com>
11666 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
11667 * explow.c (hard_function_value): Likewise.
11668 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
11669 convert_to_mode call outside the loop.
11670 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
11671 for the mode iterator. Require the mode specified by max_pieces
11673 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
11675 (copy_blkmode_to_reg): Likewise.
11676 (set_storage_via_setmem): Likewise.
11677 * optabs.c (prepare_cmp_insn): Likewise.
11678 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11679 * stor-layout.c (finish_bitfield_representative): Likewise.
11681 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11682 Alan Hayward <alan.hayward@arm.com>
11683 David Sherwood <david.sherwood@arm.com>
11685 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
11686 * expr.c (convert_move): Use them.
11687 (convert_modes): Likewise.
11688 (store_expr_with_bounds): Likewise.
11690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11691 Alan Hayward <alan.hayward@arm.com>
11692 David Sherwood <david.sherwood@arm.com>
11694 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
11695 parameter for the mode of "x". Remove the "known_x", "known_mode"
11696 and "known_ret" arguments. Change the type of the mode argument
11697 to scalar_int_mode.
11698 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
11699 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
11700 (reg_num_sign_bit_copies_for_combine): Likewise.
11701 * rtlanal.c (nonzero_bits1): Likewise.
11702 (num_sign_bit_copies1): Likewise.
11703 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
11704 (reg_num_sign_bit_copies_general): Likewise.
11705 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
11706 (reg_nonzero_bits_general): Likewise.
11708 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11709 Alan Hayward <alan.hayward@arm.com>
11710 David Sherwood <david.sherwood@arm.com>
11712 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
11713 than in subroutines. Return 1 for non-integer modes.
11714 (cached_num_sign_bit_copies): Change the type of the mode parameter
11715 to scalar_int_mode.
11716 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
11717 classes. Handle CONST_INT_P first and then check whether X also
11718 has a scalar integer mode. Check the same thing for inner registers
11719 of a SUBREG and for values that are being extended or truncated.
11721 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11722 Alan Hayward <alan.hayward@arm.com>
11723 David Sherwood <david.sherwood@arm.com>
11725 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
11726 in subroutines. Return the mode mask for non-integer modes.
11727 (cached_nonzero_bits): Change the type of the mode parameter
11728 to scalar_int_mode.
11729 (nonzero_bits1): Likewise. Remove early exit for other mode
11730 classes. Handle CONST_INT_P first and then check whether X
11731 also has a scalar integer mode.
11733 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11734 Alan Hayward <alan.hayward@arm.com>
11735 David Sherwood <david.sherwood@arm.com>
11737 * expr.c (widest_int_mode_for_size): Make the comment match the code.
11738 Return a scalar_int_mode and assert that the size is greater than
11740 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
11741 (op_by_pieces_d::op_by_pieces_d): Likewise.
11742 (op_by_pieces_d::run): Likewise.
11743 (can_store_by_pieces): Likewise.
11745 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11746 Alan Hayward <alan.hayward@arm.com>
11747 David Sherwood <david.sherwood@arm.com>
11749 * combine.c (extract_left_shift): Add a mode argument and update
11751 (make_compound_operation_int): Change the type of the mode parameter
11752 to scalar_int_mode and update the call to extract_left_shift.
11754 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11755 Alan Hayward <alan.hayward@arm.com>
11756 David Sherwood <david.sherwood@arm.com>
11758 * combine.c (simplify_and_const_int): Change the type of the mode
11759 parameter to scalar_int_mode.
11760 (simplify_and_const_int_1): Likewise. Update recursive call.
11762 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11763 Alan Hayward <alan.hayward@arm.com>
11764 David Sherwood <david.sherwood@arm.com>
11766 * combine.c (simplify_compare_const): Check that the mode is a
11767 scalar_int_mode (rather than VOIDmode) before testing its
11769 (simplify_comparison): Move COMPARISON_P handling out of the
11770 loop and restrict the latter part of the loop to scalar_int_modes.
11771 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
11772 and when considering SUBREG_REGs. Use is_int_mode instead of
11773 checking GET_MODE_CLASS against MODE_INT.
11775 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11776 Alan Hayward <alan.hayward@arm.com>
11777 David Sherwood <david.sherwood@arm.com>
11779 * combine.c (try_widen_shift_mode): Move check for equal modes to...
11780 (simplify_shift_const_1): ...here. Use scalar_int_mode for
11781 shift_unit_mode and for modes involved in scalar shifts.
11783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11784 Alan Hayward <alan.hayward@arm.com>
11785 David Sherwood <david.sherwood@arm.com>
11787 * combine.c (force_int_to_mode): New function, split out from...
11788 (force_to_mode): ...here. Keep xmode up-to-date and use it
11789 instead of GET_MODE (x).
11791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11792 Alan Hayward <alan.hayward@arm.com>
11793 David Sherwood <david.sherwood@arm.com>
11795 * optabs-query.h (extraction_insn::struct_mode): Change type to
11796 opt_scalar_int_mode and update comment.
11797 (extraction_insn::field_mode): Change type to scalar_int_mode.
11798 (extraction_insn::pos_mode): Likewise.
11799 * combine.c (make_extraction): Update accordingly.
11800 * optabs-query.c (get_traditional_extraction_insn): Likewise.
11801 (get_optab_extraction_insn): Likewise.
11802 * recog.c (simplify_while_replacing): Likewise.
11803 * expmed.c (narrow_bit_field_mem): Change the type of the mode
11804 parameter to opt_scalar_int_mode.
11806 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11807 Alan Hayward <alan.hayward@arm.com>
11808 David Sherwood <david.sherwood@arm.com>
11810 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
11811 to a scalar_int_mode instead of a machine_mode.
11812 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
11813 (get_best_mode): Return a boolean and use a pointer argument to store
11814 the selected mode. Replace the limit mode parameter with a bit limit.
11815 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
11816 for the values returned by bit_field_mode_iterator::next_mode.
11817 (store_bit_field): Update call to get_best_mode.
11818 (store_fixed_bit_field): Likewise.
11819 (extract_fixed_bit_field): Likewise.
11820 * expr.c (optimize_bitfield_assignment_op): Likewise.
11821 * fold-const.c (optimize_bit_field_compare): Likewise.
11822 (fold_truth_andor_1): Likewise.
11823 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
11824 Update for new type of m_mode.
11825 (get_best_mode): As above.
11827 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11828 Alan Hayward <alan.hayward@arm.com>
11829 David Sherwood <david.sherwood@arm.com>
11831 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
11832 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
11833 (store_bit_field): Check is_a <scalar_int_mode> before calling
11834 strict_volatile_bitfield_p.
11835 (extract_bit_field): Likewise.
11837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11838 Alan Hayward <alan.hayward@arm.com>
11839 David Sherwood <david.sherwood@arm.com>
11841 * target.def (cstore_mode): Return a scalar_int_mode.
11842 * doc/tm.texi: Regenerate.
11843 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
11844 * targhooks.h (default_cstore_mode): Likewise.
11845 * targhooks.c (default_cstore_mode): Likewise, using a forced
11847 * expmed.c (emit_cstore): Expect the target of the cstore to be
11850 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11851 Alan Hayward <alan.hayward@arm.com>
11852 David Sherwood <david.sherwood@arm.com>
11854 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
11856 (niter_desc): Likewise mode.
11857 (iv_analyze): Add a mode parameter.
11859 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
11860 and change its type to scalar_int_mode.
11861 * loop-iv.c: Update commentary at head of file.
11862 (iv_constant): Pass the mode paraeter before the rtx it describes
11863 and change its type to scalar_int_mode. Remove VOIDmode handling.
11864 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
11865 (iv_extend): Likewise.
11866 (shorten_into_mode): Likewise.
11867 (iv_add): Use scalar_int_mode.
11868 (iv_mult): Likewise.
11869 (iv_shift): Likewise.
11870 (canonicalize_iv_subregs): Likewise.
11871 (get_biv_step_1): Pass the outer_mode parameter before the rtx
11872 it describes and change its mode to scalar_int_mode. Also change
11873 the type of the returned inner_mode to scalar_int_mode.
11874 (get_biv_step): Likewise, turning outer_mode from a pointer
11875 into a direct parameter. Update call to get_biv_step_1.
11876 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
11877 iv_constant and get_biv_step.
11878 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
11879 and change its type to scalar_int_mode. Don't initialise iv->mode
11880 to VOIDmode and remove later checks for its still being VOIDmode.
11881 Update calls to iv_analyze_op and iv_analyze_expr. Check
11882 is_a <scalar_int_mode> when changing the mode under consideration.
11883 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
11884 Update call to iv_analyze_expr.
11885 (iv_analyze_op): Add a mode parameter. Reject subregs whose
11886 inner register is not also a scalar_int_mode. Update call to
11888 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
11889 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
11890 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
11891 separate mode class checks. Update calls to iv_analyze. Remove
11892 fix-up of VOIDmodes after iv_analyze_biv.
11893 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
11894 don't have a scalar_int_mode. Update call to biv_p.
11896 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11897 Alan Hayward <alan.hayward@arm.com>
11898 David Sherwood <david.sherwood@arm.com>
11900 * cfgexpand.c (convert_debug_memory_address): Use
11901 as_a <scalar_int_mode>.
11902 * combine.c (expand_compound_operation): Likewise.
11903 (make_extraction): Likewise.
11904 (change_zero_ext): Likewise.
11905 (simplify_comparison): Likewise.
11906 * cse.c (cse_insn): Likewise.
11907 * dwarf2out.c (minmax_loc_descriptor): Likewise.
11908 (mem_loc_descriptor): Likewise.
11909 (loc_descriptor): Likewise.
11910 * expmed.c (init_expmed_one_mode): Likewise.
11911 (synth_mult): Likewise.
11912 (emit_store_flag_1): Likewise.
11913 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
11914 of a comparison with size.
11915 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
11916 (reduce_to_bit_field_precision): Likewise.
11917 * function.c (expand_function_end): Likewise.
11918 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
11919 * loop-doloop.c (doloop_modify): Likewise.
11920 * optabs.c (expand_binop): Likewise.
11921 (expand_unop): Likewise.
11922 (expand_copysign_absneg): Likewise.
11923 (prepare_cmp_insn): Likewise.
11924 (maybe_legitimize_operand): Likewise.
11925 * recog.c (const_scalar_int_operand): Likewise.
11926 * rtlanal.c (get_address_mode): Likewise.
11927 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11928 (simplify_cond_clz_ctz): Likewise.
11929 * tree-nested.c (get_nl_goto_field): Likewise.
11930 * tree.c (build_vector_type_for_mode): Likewise.
11931 * var-tracking.c (use_narrower_mode): Likewise.
11933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11934 Alan Hayward <alan.hayward@arm.com>
11935 David Sherwood <david.sherwood@arm.com>
11937 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
11938 * builtins.c (expand_builtin_signbit): Use it.
11939 * cfgexpand.c (expand_debug_expr): Likewise.
11940 * dojump.c (do_jump): Likewise.
11941 (do_compare_and_jump): Likewise.
11942 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
11943 * expmed.c (make_tree): Likewise.
11944 * expr.c (expand_expr_real_2): Likewise.
11945 (expand_expr_real_1): Likewise.
11946 (try_casesi): Likewise.
11947 * fold-const-call.c (fold_const_call_ss): Likewise.
11948 * fold-const.c (unextend): Likewise.
11949 (extract_muldiv_1): Likewise.
11950 (fold_single_bit_test): Likewise.
11951 (native_encode_int): Likewise.
11952 (native_encode_string): Likewise.
11953 (native_interpret_int): Likewise.
11954 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
11955 * internal-fn.c (expand_addsub_overflow): Likewise.
11956 (expand_neg_overflow): Likewise.
11957 (expand_mul_overflow): Likewise.
11958 (expand_arith_overflow): Likewise.
11959 * match.pd: Likewise.
11960 * stor-layout.c (layout_type): Likewise.
11961 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11962 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
11963 * tree-ssanames.c (get_range_info): Likewise.
11964 * tree-switch-conversion.c (array_value_type) Likewise.
11965 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
11966 (vect_recog_divmod_pattern): Likewise.
11967 (vect_recog_mixed_size_cond_pattern): Likewise.
11968 * tree-vrp.c (extract_range_basic): Likewise.
11969 (simplify_float_conversion_using_ranges): Likewise.
11970 * tree.c (int_fits_type_p): Likewise.
11971 * ubsan.c (instrument_bool_enum_load): Likewise.
11972 * varasm.c (mergeable_string_section): Likewise.
11973 (narrowing_initializer_constant_valid_p): Likewise.
11974 (output_constant): Likewise.
11976 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11977 Alan Hayward <alan.hayward@arm.com>
11978 David Sherwood <david.sherwood@arm.com>
11980 * machmode.h (NARROWEST_INT_MODE): New macro.
11981 * expr.c (alignment_for_piecewise_move): Use it instead of
11982 GET_CLASS_NARROWEST_MODE (MODE_INT).
11983 (push_block): Likewise.
11984 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
11986 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11988 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11989 Alan Hayward <alan.hayward@arm.com>
11990 David Sherwood <david.sherwood@arm.com>
11992 * postreload.c (move2add_valid_value_p): Change the type of the
11993 mode parameter to scalar_int_mode.
11994 (move2add_use_add2_insn): Add a mode parameter and use it instead
11996 (move2add_use_add3_insn): Likewise.
11997 (reload_cse_move2add): Update accordingly.
11999 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12000 Alan Hayward <alan.hayward@arm.com>
12001 David Sherwood <david.sherwood@arm.com>
12003 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
12005 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
12006 * optabs.c (expand_unop): Likewise.
12008 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12009 Alan Hayward <alan.hayward@arm.com>
12010 David Sherwood <david.sherwood@arm.com>
12012 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
12013 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
12014 (popcount_loc_descriptor): Likewise.
12015 (bswap_loc_descriptor): Likewise.
12016 (rotate_loc_descriptor): Likewise.
12017 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12018 calling the functions above.
12020 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12021 Alan Hayward <alan.hayward@arm.com>
12022 David Sherwood <david.sherwood@arm.com>
12024 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12026 (try_combine): Likewise.
12027 (simplify_if_then_else): Likewise.
12028 * cse.c (cse_insn): Likewise.
12029 * dwarf2out.c (mem_loc_descriptor): Likewise.
12030 * emit-rtl.c (gen_lowpart_common): Likewise.
12031 * simplify-rtx.c (simplify_truncation): Likewise.
12032 (simplify_binary_operation_1): Likewise.
12033 (simplify_const_relational_operation): Likewise.
12034 (simplify_ternary_operation): Likewise.
12035 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
12037 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12038 Alan Hayward <alan.hayward@arm.com>
12039 David Sherwood <david.sherwood@arm.com>
12041 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12042 * reload.c (push_reload): Likewise.
12043 (find_reloads): Likewise.
12045 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12046 Alan Hayward <alan.hayward@arm.com>
12047 David Sherwood <david.sherwood@arm.com>
12049 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12050 (make_compound_operation_int): Likewise.
12051 (change_zero_ext): Likewise.
12052 * expr.c (convert_move): Likewise.
12053 (convert_modes): Likewise.
12054 * fwprop.c (forward_propagate_subreg): Likewise.
12055 * loop-iv.c (get_biv_step_1): Likewise.
12056 * optabs.c (widen_operand): Likewise.
12057 * postreload.c (move2add_valid_value_p): Likewise.
12058 * recog.c (simplify_while_replacing): Likewise.
12059 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12060 (simplify_binary_operation_1): Likewise. Remove redundant
12061 mode equality check.
12063 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12064 Alan Hayward <alan.hayward@arm.com>
12065 David Sherwood <david.sherwood@arm.com>
12067 * combine.c (combine_simplify_rtx): Add checks for
12068 is_a <scalar_int_mode>.
12069 (simplify_if_then_else): Likewise.
12070 (make_field_assignment): Likewise.
12071 (simplify_comparison): Likewise.
12072 * ifcvt.c (noce_try_bitop): Likewise.
12073 * loop-invariant.c (canonicalize_address_mult): Likewise.
12074 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12076 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12077 Alan Hayward <alan.hayward@arm.com>
12078 David Sherwood <david.sherwood@arm.com>
12080 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12081 is_a <scalar_int_mode> instead of != BLKmode.
12083 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12084 Alan Hayward <alan.hayward@arm.com>
12085 David Sherwood <david.sherwood@arm.com>
12087 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12088 instead of != VOIDmode.
12089 * combine.c (if_then_else_cond): Likewise.
12090 (change_zero_ext): Likewise.
12091 * dwarf2out.c (mem_loc_descriptor): Likewise.
12092 (loc_descriptor): Likewise.
12093 * rtlanal.c (canonicalize_condition): Likewise.
12094 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12096 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12097 Alan Hayward <alan.hayward@arm.com>
12098 David Sherwood <david.sherwood@arm.com>
12100 * simplify-rtx.c (simplify_binary_operation_1): Use
12101 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12103 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12104 Alan Hayward <alan.hayward@arm.com>
12105 David Sherwood <david.sherwood@arm.com>
12107 * wide-int.h (int_traits<unsigned char>) New class.
12108 (int_traits<unsigned short>) Likewise.
12109 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12110 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12112 * combine.c (set_nonzero_bits_and_sign_copies): Use
12113 is_a <scalar_int_mode>.
12114 (find_split_point): Likewise.
12115 (combine_simplify_rtx): Likewise.
12116 (simplify_logical): Likewise.
12117 (expand_compound_operation): Likewise.
12118 (expand_field_assignment): Likewise.
12119 (make_compound_operation): Likewise.
12120 (extended_count): Likewise.
12121 (change_zero_ext): Likewise.
12122 (simplify_comparison): Likewise.
12123 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12124 (ucompare_loc_descriptor): Likewise.
12125 (minmax_loc_descriptor): Likewise.
12126 (mem_loc_descriptor): Likewise.
12127 (loc_descriptor): Likewise.
12128 * expmed.c (init_expmed_one_mode): Likewise.
12129 * lra-constraints.c (lra_constraint_offset): Likewise.
12130 * optabs.c (prepare_libcall_arg): Likewise.
12131 * postreload.c (move2add_note_store): Likewise.
12132 * reload.c (operands_match_p): Likewise.
12133 * rtl.h (load_extend_op): Likewise.
12134 * rtlhooks.c (gen_lowpart_general): Likewise.
12135 * simplify-rtx.c (simplify_truncation): Likewise.
12136 (simplify_unary_operation_1): Likewise.
12137 (simplify_binary_operation_1): Likewise.
12138 (simplify_const_binary_operation): Likewise.
12139 (simplify_const_relational_operation): Likewise.
12140 (simplify_subreg): Likewise.
12141 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12142 * var-tracking.c (adjust_mems): Likewise.
12143 (prepare_call_arguments): Likewise.
12145 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12146 Alan Hayward <alan.hayward@arm.com>
12147 David Sherwood <david.sherwood@arm.com>
12149 * machmode.h (is_int_mode): New fuction.
12150 * combine.c (find_split_point): Use it.
12151 (combine_simplify_rtx): Likewise.
12152 (simplify_if_then_else): Likewise.
12153 (simplify_set): Likewise.
12154 (simplify_shift_const_1): Likewise.
12155 (simplify_comparison): Likewise.
12156 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12157 * cse.c (notreg_cost): Likewise.
12158 (cse_insn): Likewise.
12159 * cselib.c (cselib_lookup_1): Likewise.
12160 * dojump.c (do_jump_1): Likewise.
12161 (do_compare_rtx_and_jump): Likewise.
12162 * dse.c (get_call_args): Likewise.
12163 * dwarf2out.c (rtl_for_decl_init): Likewise.
12164 (native_encode_initializer): Likewise.
12165 * expmed.c (emit_store_flag_1): Likewise.
12166 (emit_store_flag): Likewise.
12167 * expr.c (convert_modes): Likewise.
12168 (store_field): Likewise.
12169 (expand_expr_real_1): Likewise.
12170 * fold-const.c (fold_read_from_constant_string): Likewise.
12171 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12172 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12173 * optabs.c (expand_binop): Likewise.
12174 (expand_unop): Likewise.
12175 (expand_abs_nojump): Likewise.
12176 (expand_one_cmpl_abs_nojump): Likewise.
12177 * simplify-rtx.c (mode_signbit_p): Likewise.
12178 (val_signbit_p): Likewise.
12179 (val_signbit_known_set_p): Likewise.
12180 (val_signbit_known_clear_p): Likewise.
12181 (simplify_relational_operation_1): Likewise.
12182 * tree.c (vector_type_mode): Likewise.
12184 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12185 Alan Hayward <alan.hayward@arm.com>
12186 David Sherwood <david.sherwood@arm.com>
12188 * machmode.h (smallest_mode_for_size): Fix formatting.
12189 (smallest_int_mode_for_size): New function.
12190 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12191 instead of smallest_mode_for_size.
12192 * combine.c (make_extraction): Likewise.
12193 * config/arc/arc.c (arc_expand_movmem): Likewise.
12194 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12195 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12196 * config/s390/s390.c (s390_expand_insv): Likewise.
12197 * config/sparc/sparc.c (assign_int_registers): Likewise.
12198 * config/spu/spu.c (spu_function_value): Likewise.
12199 (spu_function_arg): Likewise.
12200 * coverage.c (get_gcov_type): Likewise.
12201 (get_gcov_unsigned_t): Likewise.
12202 * dse.c (find_shift_sequence): Likewise.
12203 * expmed.c (store_bit_field_1): Likewise.
12204 * expr.c (convert_move): Likewise.
12205 (store_field): Likewise.
12206 * internal-fn.c (expand_arith_overflow): Likewise.
12207 * optabs-query.c (get_best_extraction_insn): Likewise.
12208 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12209 * stor-layout.c (layout_type): Likewise.
12210 (initialize_sizetypes): Likewise.
12211 * targhooks.c (default_get_mask_mode): Likewise.
12212 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12214 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12215 Alan Hayward <alan.hayward@arm.com>
12216 David Sherwood <david.sherwood@arm.com>
12218 * machmode.h (opt_mode::else_blk): New function.
12219 (int_mode_for_mode): Declare.
12220 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12221 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12223 * cfgexpand.c (expand_debug_expr): Likewise.
12224 * combine.c (gen_lowpart_or_truncate): Likewise.
12225 (gen_lowpart_for_combine): Likewise.
12226 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12227 * config/avr/avr.c (avr_to_int_mode): Likewise.
12228 (avr_out_plus_1): Likewise.
12229 (avr_out_plus): Likewise.
12230 (avr_out_round): Likewise.
12231 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12232 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12233 (s390_expand_vcond): Likewise.
12234 * config/spu/spu.c (spu_split_immediate): Likewise.
12235 (spu_expand_mov): Likewise.
12236 * dse.c (get_stored_val): Likewise.
12237 * expmed.c (store_bit_field_1): Likewise.
12238 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12240 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12241 (extract_low_bits): Likewise.
12242 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12243 handling rather than repeating the check.
12244 (emit_group_store): Likewise.
12245 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12246 * optabs.c (expand_absneg_bit): Likewise.
12247 (expand_copysign_absneg): Likewise.
12248 (expand_copysign_bit): Likewise.
12249 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12250 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12251 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12252 * var-tracking.c (prepare_call_arguments): Likewise.
12253 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12254 int_mode_for_mode instead of mode_for_size.
12255 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12257 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12258 Alan Hayward <alan.hayward@arm.com>
12259 David Sherwood <david.sherwood@arm.com>
12261 * machmode.h (int_mode_for_size): New function.
12262 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12263 instead of mode_for_size.
12264 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12266 * combine.c (expand_field_assignment): Use int_mode_for_size
12267 instead of mode_for_size.
12268 (make_extraction): Likewise.
12269 (simplify_shift_const_1): Likewise.
12270 (simplify_comparison): Likewise.
12271 * dojump.c (do_jump): Likewise.
12272 * dwarf2out.c (mem_loc_descriptor): Likewise.
12273 * emit-rtl.c (init_derived_machine_modes): Likewise.
12274 * expmed.c (flip_storage_order): Likewise.
12275 (convert_extracted_bit_field): Likewise.
12276 * expr.c (copy_blkmode_from_reg): Likewise.
12277 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12278 * internal-fn.c (expand_mul_overflow): Likewise.
12279 * lower-subreg.c (simple_move): Likewise.
12280 * optabs-libfuncs.c (init_optabs): Likewise.
12281 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12282 * tree.c (vector_type_mode): Likewise.
12283 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12284 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12285 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12286 * tree-vect-stmts.c (vectorizable_load): Likewise.
12287 (vectorizable_store): Likewise.
12289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12290 Alan Hayward <alan.hayward@arm.com>
12291 David Sherwood <david.sherwood@arm.com>
12293 * coretypes.h (pod_mode): New type.
12294 (scalar_int_mode_pod): New typedef.
12295 * machmode.h (pod_mode): New class.
12296 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12297 * genmodes.c (emit_mode_int_n): Update accordingly.
12298 * lower-subreg.h (target_lower_subreg): Change type to
12299 scalar_int_mode_pod.
12300 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12301 scalar_int_mode_pod.
12303 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12304 Alan Hayward <alan.hayward@arm.com>
12305 David Sherwood <david.sherwood@arm.com>
12307 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12308 machine_mode to scalar_int_mode.
12309 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12310 (rs6000_option_override_internal): Remove cast to int.
12311 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12312 machine_mode to scalar_int_mode.
12313 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12314 (rs6000_option_override_internal): Remove cast to int.
12315 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12316 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12318 * config/s390/s390.c (s390_expand_builtin): Likewise.
12319 * coretypes.h (scalar_int_mode): New type.
12320 (opt_scalar_int_mode): New typedef.
12321 * machmode.h (scalar_int_mode): New class.
12322 (scalar_int_mode::includes_p): New function.
12323 (byte_mode): Change type to scalar_int_mode.
12324 (word_mode): Likewise.
12325 (ptr_mode): Likewise.
12326 * emit-rtl.c (byte_mode): Likewise.
12327 (word_mode): Likewise.
12328 (ptr_mode): Likewise.
12329 (init_derived_machine_modes): Update accordingly.
12330 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12331 and MODE_PARTIAL_INT.
12332 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12333 opt_scalar_int_mode.
12335 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12336 Alan Hayward <alan.hayward@arm.com>
12337 David Sherwood <david.sherwood@arm.com>
12339 * target.def (libgcc_floating_mode_supported_p): Take a
12341 * doc/tm.texi: Regenerate.
12342 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12344 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12345 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12348 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12349 Alan Hayward <alan.hayward@arm.com>
12350 David Sherwood <david.sherwood@arm.com>
12352 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12353 * doc/tm.texi: Regenerate.
12354 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12355 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12356 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12357 * targhooks.h (default_floatn_mode): Likewise.
12358 * targhooks.c (default_floatn_mode): Likewise.
12359 * tree.c (build_common_tree_nodes): Update accordingly.
12361 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12362 Alan Hayward <alan.hayward@arm.com>
12363 David Sherwood <david.sherwood@arm.com>
12365 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
12366 (mode_iterator::iterate_p): Likewise.
12367 (mode_iterator::get_wider): Likewise.
12368 * expr.c (init_expr_target): Use opt_scalar_float_mode.
12370 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12371 Alan Hayward <alan.hayward@arm.com>
12372 David Sherwood <david.sherwood@arm.com>
12374 * coretypes.h (opt_scalar_float_mode): New typedef.
12375 * machmode.h (float_mode_for_size): New function.
12376 * emit-rtl.c (double_mode): Delete.
12377 (init_emit_once): Use float_mode_for_size.
12378 * stor-layout.c (layout_type): Likewise.
12379 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
12381 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12382 Alan Hayward <alan.hayward@arm.com>
12383 David Sherwood <david.sherwood@arm.com>
12385 * output.h (assemble_real): Take a scalar_float_mode.
12386 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
12387 * config/arm/arm.md (consttable_4): Likewise.
12388 (consttable_8): Likewise.
12389 (consttable_16): Likewise.
12390 * config/mips/mips.md (consttable_float): Likewise.
12391 * config/s390/s390.c (s390_output_pool_entry): Likewise.
12392 * varasm.c (assemble_real): Take a scalar_float_mode.
12393 (output_constant_pool_2): Update accordingly.
12394 (output_constant): Likewise.
12396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12397 Alan Hayward <alan.hayward@arm.com>
12398 David Sherwood <david.sherwood@arm.com>
12400 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
12401 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
12402 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
12403 (native_encode_real): Likewise.
12404 (native_interpret_real): Likewise.
12405 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
12406 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12409 Alan Hayward <alan.hayward@arm.com>
12410 David Sherwood <david.sherwood@arm.com>
12412 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
12413 <scalar_float_mode>. Simplify.
12414 (gen_extend_conv_libfunc): Likewise.
12416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12417 Alan Hayward <alan.hayward@arm.com>
12418 David Sherwood <david.sherwood@arm.com>
12420 * coretypes.h (scalar_float_mode): New type.
12421 * machmode.h (mode_traits::from_int): Use machine_mode if
12422 USE_ENUM_MODES is defined.
12423 (is_a): New function.
12425 (dyn_cast): Likewise.
12426 (scalar_float_mode): New class.
12427 (scalar_float_mode::includes_p): New function.
12428 (is_float_mode): Likewise.
12429 * gdbhooks.py (MachineModePrinter): New class.
12430 (build_pretty_printer): Use it for scalar_float_mode.
12431 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
12432 (format_helper::format_helper): Turn into a template.
12433 * genmodes.c (get_mode_class): New function.
12434 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
12435 or machine_mode if none.
12436 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
12437 as_a <scalar_float_mode>.
12438 * dwarf2out.c (mem_loc_descriptor): Likewise.
12439 (insert_float): Likewise.
12440 (add_const_value_attribute): Likewise.
12441 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12442 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
12443 (expand_unop): Update accordingly.
12444 (expand_abs_nojump): Likewise.
12445 (expand_copysign_absneg): Take a scalar_float_mode.
12446 (expand_copysign_bit): Likewise.
12447 (expand_copysign): Update accordingly.
12449 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12450 Alan Hayward <alan.hayward@arm.com>
12451 David Sherwood <david.sherwood@arm.com>
12453 * coretypes.h (opt_mode): New class.
12454 * machmode.h (opt_mode): Likewise.
12455 (opt_mode::else_void): New function.
12456 (opt_mode::require): Likewise.
12457 (opt_mode::exists): Likewise.
12458 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
12459 (GET_MODE_2XWIDER_MODE): Likewise.
12460 (mode_iterator::get_wider): Update accordingly.
12461 (mode_iterator::get_2xwider): Likewise.
12462 (mode_iterator::get_known_wider): Likewise, turning into a template.
12463 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
12464 forcing a wider mode to exist.
12465 * config/cr16/cr16.h (LONG_REG_P): Likewise.
12466 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12467 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
12468 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
12469 * lower-subreg.c (init_lower_subreg): Likewise.
12470 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
12471 on the final iteration.
12472 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
12473 a wider mode exists before asking for a move pattern.
12474 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
12475 forcing a wider mode to exist.
12476 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
12477 returning false if no such mode exists.
12478 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
12479 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
12480 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
12481 Avoid checking for a MODE_INT if we already know the mode is not a
12483 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
12484 forcing a wider mode to exist.
12485 (expmed_mult_highpart_optab): Likewise.
12486 (expmed_mult_highpart): Likewise.
12487 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
12489 * lto-streamer-in.c (lto_input_mode_table): Likewise.
12490 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
12491 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12492 * internal-fn.c (expand_mul_overflow): Update use of
12493 GET_MODE_2XWIDER_MODE.
12494 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12495 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
12496 GET_MODE_WIDER_MODE.
12497 (convert_plusminus_to_widen): Likewise.
12498 * tree-switch-conversion.c (array_value_type): Likewise.
12499 * var-tracking.c (emit_note_insn_var_location): Likewise.
12500 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12501 Return false inside rather than outside the loop if no wider mode
12503 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
12504 and GET_MODE_2XWIDER_MODE
12505 (can_compare_p): Use else_void.
12506 * gdbhooks.py (OptMachineModePrinter): New class.
12507 (build_pretty_printer): Use it for opt_mode.
12509 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12510 Alan Hayward <alan.hayward@arm.com>
12511 David Sherwood <david.sherwood@arm.com>
12513 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
12514 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
12516 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12517 Alan Hayward <alan.hayward@arm.com>
12518 David Sherwood <david.sherwood@arm.com>
12520 * machmode.h (mode_traits): New structure.
12521 (get_narrowest_mode): New function.
12522 (mode_iterator::start): Likewise.
12523 (mode_iterator::iterate_p): Likewise.
12524 (mode_iterator::get_wider): Likewise.
12525 (mode_iterator::get_known_wider): Likewise.
12526 (mode_iterator::get_2xwider): Likewise.
12527 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
12528 (FOR_EACH_MODE): Likewise.
12529 (FOR_EACH_MODE_FROM): Likewise.
12530 (FOR_EACH_MODE_UNTIL): Likewise.
12531 (FOR_EACH_WIDER_MODE): Likewise.
12532 (FOR_EACH_2XWIDER_MODE): Likewise.
12533 * builtins.c (expand_builtin_strlen): Use new mode iterators.
12534 * combine.c (simplify_comparison): Likewise
12535 * config/i386/i386.c (type_natural_mode): Likewise.
12536 * cse.c (cse_insn): Likewise.
12537 * dse.c (find_shift_sequence): Likewise.
12538 * emit-rtl.c (init_derived_machine_modes): Likewise.
12539 (init_emit_once): Likewise.
12540 * explow.c (hard_function_value): Likewise.
12541 * expmed.c (extract_fixed_bit_field_1): Likewise.
12542 (extract_bit_field_1): Likewise.
12543 (expand_divmod): Likewise.
12544 (emit_store_flag_1): Likewise.
12545 * expr.c (init_expr_target): Likewise.
12546 (convert_move): Likewise.
12547 (alignment_for_piecewise_move): Likewise.
12548 (widest_int_mode_for_size): Likewise.
12549 (emit_block_move_via_movmem): Likewise.
12550 (copy_blkmode_to_reg): Likewise.
12551 (set_storage_via_setmem): Likewise.
12552 (compress_float_constant): Likewise.
12553 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12554 * optabs-query.c (get_best_extraction_insn): Likewise.
12555 * optabs.c (expand_binop): Likewise.
12556 (expand_twoval_unop): Likewise.
12557 (expand_twoval_binop): Likewise.
12558 (widen_leading): Likewise.
12559 (widen_bswap): Likewise.
12560 (expand_parity): Likewise.
12561 (expand_unop): Likewise.
12562 (prepare_cmp_insn): Likewise.
12563 (prepare_float_lib_cmp): Likewise.
12564 (expand_float): Likewise.
12565 (expand_fix): Likewise.
12566 (expand_sfix_optab): Likewise.
12567 * postreload.c (move2add_use_add2_insn): Likewise.
12568 * reg-stack.c (reg_to_stack): Likewise.
12569 * reginfo.c (choose_hard_reg_mode): Likewise.
12570 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12571 * stor-layout.c (mode_for_size): Likewise.
12572 (smallest_mode_for_size): Likewise.
12573 (mode_for_vector): Likewise.
12574 (finish_bitfield_representative): Likewise.
12575 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
12576 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
12577 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12578 * var-tracking.c (prepare_call_arguments): Likewise.
12580 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12581 Alan Hayward <alan.hayward@arm.com>
12582 David Sherwood <david.sherwood@arm.com>
12584 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
12585 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
12586 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
12587 * machmode.h (mode_size): Move earlier in file.
12588 (mode_precision): Likewise.
12589 (mode_inner): Likewise.
12590 (mode_nunits): Likewise.
12591 (mode_unit_size): Likewise.
12592 (unit_unit_precision): Likewise.
12593 (mode_wider): Likewise.
12594 (mode_2xwider): Likewise.
12595 (machine_mode): New class.
12596 (mode_to_bytes): New function.
12597 (mode_to_bits): Likewise.
12598 (mode_to_precision): Likewise.
12599 (mode_to_inner): Likewise.
12600 (mode_to_unit_size): Likewise.
12601 (mode_to_unit_precision): Likewise.
12602 (mode_to_nunits): Likewise.
12603 (GET_MODE_SIZE): Use mode_to_bytes.
12604 (GET_MODE_BITSIZE): Use mode_to_bits.
12605 (GET_MODE_PRECISION): Use mode_to_precision.
12606 (GET_MODE_INNER): Use mode_to_inner.
12607 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
12608 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
12609 (GET_MODE_NUNITS): Use mode_to_nunits.
12610 * system.h (ALWAYS_INLINE): New macro.
12611 * config/powerpcspe/powerpcspe-c.c
12612 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
12613 int for arg1_mode and arg2_mode.
12615 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12616 Alan Hayward <alan.hayward@arm.com>
12617 David Sherwood <david.sherwood@arm.com>
12619 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12620 Prefix mode names with E_ in case statements.
12621 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12622 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
12623 (aarch64_split_simd_move): Likewise.
12624 (aarch64_gen_storewb_pair): Likewise.
12625 (aarch64_gen_loadwb_pair): Likewise.
12626 (aarch64_gen_store_pair): Likewise.
12627 (aarch64_gen_load_pair): Likewise.
12628 (aarch64_get_condition_code_1): Likewise.
12629 (aarch64_constant_pool_reload_icode): Likewise.
12630 (get_rsqrte_type): Likewise.
12631 (get_rsqrts_type): Likewise.
12632 (get_recpe_type): Likewise.
12633 (get_recps_type): Likewise.
12634 (aarch64_gimplify_va_arg_expr): Likewise.
12635 (aarch64_simd_container_mode): Likewise.
12636 (aarch64_emit_load_exclusive): Likewise.
12637 (aarch64_emit_store_exclusive): Likewise.
12638 (aarch64_expand_compare_and_swap): Likewise.
12639 (aarch64_gen_atomic_cas): Likewise.
12640 (aarch64_emit_bic): Likewise.
12641 (aarch64_emit_atomic_swap): Likewise.
12642 (aarch64_emit_atomic_load_op): Likewise.
12643 (aarch64_evpc_trn): Likewise.
12644 (aarch64_evpc_uzp): Likewise.
12645 (aarch64_evpc_zip): Likewise.
12646 (aarch64_evpc_ext): Likewise.
12647 (aarch64_evpc_rev): Likewise.
12648 (aarch64_evpc_dup): Likewise.
12649 (aarch64_gen_ccmp_first): Likewise.
12650 (aarch64_gen_ccmp_next): Likewise.
12651 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12652 (alpha_emit_xfloating_libcall): Likewise.
12653 (emit_insxl): Likewise.
12654 (alpha_arg_type): Likewise.
12655 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
12656 (arc_preferred_simd_mode): Likewise.
12657 (arc_secondary_reload): Likewise.
12658 (get_arc_condition_code): Likewise.
12659 (arc_print_operand): Likewise.
12660 (arc_legitimate_constant_p): Likewise.
12661 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12662 * config/arc/arc.md (casesi_load): Likewise.
12663 (casesi_compact_jump): Likewise.
12664 * config/arc/predicates.md (proper_comparison_operator): Likewise.
12665 (cc_use_register): Likewise.
12666 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12667 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12668 (arm_init_iwmmxt_builtins): Likewise.
12669 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
12670 (neon_expand_vector_init): Likewise.
12671 (arm_attr_length_move_neon): Likewise.
12672 (maybe_get_arm_condition_code): Likewise.
12673 (arm_emit_vector_const): Likewise.
12674 (arm_preferred_simd_mode): Likewise.
12675 (arm_output_iwmmxt_tinsr): Likewise.
12676 (thumb1_output_casesi): Likewise.
12677 (thumb2_output_casesi): Likewise.
12678 (arm_emit_load_exclusive): Likewise.
12679 (arm_emit_store_exclusive): Likewise.
12680 (arm_expand_compare_and_swap): Likewise.
12681 (arm_evpc_neon_vuzp): Likewise.
12682 (arm_evpc_neon_vzip): Likewise.
12683 (arm_evpc_neon_vrev): Likewise.
12684 (arm_evpc_neon_vtrn): Likewise.
12685 (arm_evpc_neon_vext): Likewise.
12686 (arm_validize_comparison): Likewise.
12687 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
12688 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
12689 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
12690 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
12691 (c6x_preferred_simd_mode): Likewise.
12692 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
12693 (epiphany_rtx_costs): Likewise.
12694 * config/epiphany/predicates.md (proper_comparison_operator):
12696 * config/frv/frv.c (condexec_memory_operand): Likewise.
12697 (frv_emit_move): Likewise.
12698 (output_move_single): Likewise.
12699 (output_condmove_single): Likewise.
12700 (frv_hard_regno_mode_ok): Likewise.
12701 (frv_matching_accg_mode): Likewise.
12702 * config/h8300/h8300.c (split_adds_subs): Likewise.
12703 (h8300_rtx_costs): Likewise.
12704 (h8300_print_operand): Likewise.
12705 (compute_mov_length): Likewise.
12706 (output_logical_op): Likewise.
12707 (compute_logical_op_length): Likewise.
12708 (compute_logical_op_cc): Likewise.
12709 (h8300_shift_needs_scratch_p): Likewise.
12710 (output_a_shift): Likewise.
12711 (compute_a_shift_length): Likewise.
12712 (compute_a_shift_cc): Likewise.
12713 (expand_a_rotate): Likewise.
12714 (output_a_rotate): Likewise.
12715 * config/i386/i386.c (classify_argument): Likewise.
12716 (function_arg_advance_32): Likewise.
12717 (function_arg_32): Likewise.
12718 (function_arg_64): Likewise.
12719 (function_value_64): Likewise.
12720 (ix86_gimplify_va_arg): Likewise.
12721 (ix86_legitimate_constant_p): Likewise.
12722 (put_condition_code): Likewise.
12723 (split_double_mode): Likewise.
12724 (ix86_avx256_split_vector_move_misalign): Likewise.
12725 (ix86_expand_vector_logical_operator): Likewise.
12726 (ix86_split_idivmod): Likewise.
12727 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
12728 (ix86_build_const_vector): Likewise.
12729 (ix86_build_signbit_mask): Likewise.
12730 (ix86_match_ccmode): Likewise.
12731 (ix86_cc_modes_compatible): Likewise.
12732 (ix86_expand_branch): Likewise.
12733 (ix86_expand_sse_cmp): Likewise.
12734 (ix86_expand_sse_movcc): Likewise.
12735 (ix86_expand_int_sse_cmp): Likewise.
12736 (ix86_expand_vec_perm_vpermi2): Likewise.
12737 (ix86_expand_vec_perm): Likewise.
12738 (ix86_expand_sse_unpack): Likewise.
12739 (ix86_expand_int_addcc): Likewise.
12740 (ix86_split_to_parts): Likewise.
12741 (ix86_vectorize_builtin_gather): Likewise.
12742 (ix86_vectorize_builtin_scatter): Likewise.
12743 (avx_vpermilp_parallel): Likewise.
12744 (inline_memory_move_cost): Likewise.
12745 (ix86_tieable_integer_mode_p): Likewise.
12746 (x86_maybe_negate_const_int): Likewise.
12747 (ix86_expand_vector_init_duplicate): Likewise.
12748 (ix86_expand_vector_init_one_nonzero): Likewise.
12749 (ix86_expand_vector_init_one_var): Likewise.
12750 (ix86_expand_vector_init_concat): Likewise.
12751 (ix86_expand_vector_init_interleave): Likewise.
12752 (ix86_expand_vector_init_general): Likewise.
12753 (ix86_expand_vector_set): Likewise.
12754 (ix86_expand_vector_extract): Likewise.
12755 (emit_reduc_half): Likewise.
12756 (ix86_emit_i387_round): Likewise.
12757 (ix86_mangle_type): Likewise.
12758 (ix86_expand_round_sse4): Likewise.
12759 (expand_vec_perm_blend): Likewise.
12760 (canonicalize_vector_int_perm): Likewise.
12761 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
12762 (expand_vec_perm_1): Likewise.
12763 (expand_vec_perm_interleave3): Likewise.
12764 (expand_vec_perm_even_odd_pack): Likewise.
12765 (expand_vec_perm_even_odd_1): Likewise.
12766 (expand_vec_perm_broadcast_1): Likewise.
12767 (ix86_vectorize_vec_perm_const_ok): Likewise.
12768 (ix86_expand_vecop_qihi): Likewise.
12769 (ix86_expand_mul_widen_hilo): Likewise.
12770 (ix86_expand_sse2_abs): Likewise.
12771 (ix86_expand_pextr): Likewise.
12772 (ix86_expand_pinsr): Likewise.
12773 (ix86_preferred_simd_mode): Likewise.
12774 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
12775 * config/i386/sse.md (*andnot<mode>3): Likewise.
12776 (<mask_codefor><code><mode>3<mask_name>): Likewise.
12777 (*<code><mode>3): Likewise.
12778 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
12779 (ia64_expand_atomic_op): Likewise.
12780 (ia64_arg_type): Likewise.
12781 (ia64_mode_to_int): Likewise.
12782 (ia64_scalar_mode_supported_p): Likewise.
12783 (ia64_vector_mode_supported_p): Likewise.
12784 (expand_vec_perm_broadcast): Likewise.
12785 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
12786 (iq2000_function_arg_advance): Likewise.
12787 (iq2000_function_arg): Likewise.
12788 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
12789 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
12790 (m68k_libcall_value): Likewise.
12791 (m68k_function_value): Likewise.
12792 (sched_attr_op_type): Likewise.
12793 * config/mcore/mcore.c (mcore_output_move): Likewise.
12794 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
12796 (microblaze_function_arg): Likewise.
12797 * config/mips/mips.c (mips16_build_call_stub): Likewise.
12798 (mips_print_operand): Likewise.
12799 (mips_mode_ok_for_mov_fmt_p): Likewise.
12800 (mips_vector_mode_supported_p): Likewise.
12801 (mips_preferred_simd_mode): Likewise.
12802 (mips_expand_vpc_loongson_even_odd): Likewise.
12803 (mips_expand_vec_unpack): Likewise.
12804 (mips_expand_vi_broadcast): Likewise.
12805 (mips_expand_vector_init): Likewise.
12806 (mips_expand_vec_reduc): Likewise.
12807 (mips_expand_msa_cmp): Likewise.
12808 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
12809 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
12810 (cc_flags_for_mode): Likewise.
12811 * config/msp430/msp430.c (msp430_print_operand): Likewise.
12812 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
12813 (nds32_output_casesi_pc_relative): Likewise.
12814 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12815 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
12816 (nvptx_gen_unpack): Likewise.
12817 (nvptx_gen_pack): Likewise.
12818 (nvptx_gen_shuffle): Likewise.
12819 (nvptx_gen_wcast): Likewise.
12820 (nvptx_preferred_simd_mode): Likewise.
12821 * config/pa/pa.c (pa_secondary_reload): Likewise.
12822 * config/pa/predicates.md (base14_operand): Likewise.
12823 * config/powerpcspe/powerpcspe-c.c
12824 (altivec_resolve_overloaded_builtin): Likewise.
12825 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
12827 (rs6000_preferred_simd_mode): Likewise.
12828 (output_vec_const_move): Likewise.
12829 (rs6000_expand_vector_extract): Likewise.
12830 (rs6000_split_vec_extract_var): Likewise.
12831 (reg_offset_addressing_ok_p): Likewise.
12832 (rs6000_legitimate_offset_address_p): Likewise.
12833 (rs6000_legitimize_address): Likewise.
12834 (rs6000_emit_set_const): Likewise.
12835 (rs6000_const_vec): Likewise.
12836 (rs6000_emit_move): Likewise.
12837 (spe_build_register_parallel): Likewise.
12838 (rs6000_darwin64_record_arg_recurse): Likewise.
12839 (swap_selector_for_mode): Likewise.
12840 (spe_init_builtins): Likewise.
12841 (paired_init_builtins): Likewise.
12842 (altivec_init_builtins): Likewise.
12843 (do_load_for_compare): Likewise.
12844 (rs6000_generate_compare): Likewise.
12845 (rs6000_expand_float128_convert): Likewise.
12846 (emit_load_locked): Likewise.
12847 (emit_store_conditional): Likewise.
12848 (rs6000_output_function_epilogue): Likewise.
12849 (rs6000_handle_altivec_attribute): Likewise.
12850 (rs6000_function_value): Likewise.
12851 (emit_fusion_gpr_load): Likewise.
12852 (emit_fusion_p9_load): Likewise.
12853 (emit_fusion_p9_store): Likewise.
12854 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
12855 (fusion_gpr_mem_load): Likewise.
12856 (fusion_addis_mem_combo_load): Likewise.
12857 (fusion_addis_mem_combo_store): Likewise.
12858 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
12859 (fusion_gpr_mem_load): Likewise.
12860 (fusion_addis_mem_combo_load): Likewise.
12861 (fusion_addis_mem_combo_store): Likewise.
12862 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12864 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
12865 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12866 (rs6000_preferred_simd_mode): Likewise.
12867 (output_vec_const_move): Likewise.
12868 (rs6000_expand_vector_extract): Likewise.
12869 (rs6000_split_vec_extract_var): Likewise.
12870 (reg_offset_addressing_ok_p): Likewise.
12871 (rs6000_legitimate_offset_address_p): Likewise.
12872 (rs6000_legitimize_address): Likewise.
12873 (rs6000_emit_set_const): Likewise.
12874 (rs6000_const_vec): Likewise.
12875 (rs6000_emit_move): Likewise.
12876 (rs6000_darwin64_record_arg_recurse): Likewise.
12877 (swap_selector_for_mode): Likewise.
12878 (paired_init_builtins): Likewise.
12879 (altivec_init_builtins): Likewise.
12880 (rs6000_expand_float128_convert): Likewise.
12881 (emit_load_locked): Likewise.
12882 (emit_store_conditional): Likewise.
12883 (rs6000_output_function_epilogue): Likewise.
12884 (rs6000_handle_altivec_attribute): Likewise.
12885 (rs6000_function_value): Likewise.
12886 (emit_fusion_gpr_load): Likewise.
12887 (emit_fusion_p9_load): Likewise.
12888 (emit_fusion_p9_store): Likewise.
12889 * config/rx/rx.c (rx_gen_move_template): Likewise.
12890 (flags_from_mode): Likewise.
12891 * config/s390/predicates.md (s390_alc_comparison): Likewise.
12892 (s390_slb_comparison): Likewise.
12893 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
12894 (s390_vector_mode_supported_p): Likewise.
12895 (s390_cc_modes_compatible): Likewise.
12896 (s390_match_ccmode_set): Likewise.
12897 (s390_canonicalize_comparison): Likewise.
12898 (s390_emit_compare_and_swap): Likewise.
12899 (s390_branch_condition_mask): Likewise.
12900 (s390_rtx_costs): Likewise.
12901 (s390_secondary_reload): Likewise.
12902 (__SECONDARY_RELOAD_CASE): Likewise.
12903 (s390_expand_cs): Likewise.
12904 (s390_preferred_simd_mode): Likewise.
12905 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
12906 * config/sh/sh.c (sh_print_operand): Likewise.
12907 (dump_table): Likewise.
12908 (sh_secondary_reload): Likewise.
12909 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12910 * config/sh/sh.md (casesi_worker_1): Likewise.
12911 (casesi_worker_2): Likewise.
12912 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
12913 (fcc_comparison_operator): Likewise.
12914 * config/sparc/sparc.c (sparc_expand_move): Likewise.
12915 (emit_soft_tfmode_cvt): Likewise.
12916 (sparc_preferred_simd_mode): Likewise.
12917 (output_cbranch): Likewise.
12918 (sparc_print_operand): Likewise.
12919 (sparc_expand_vec_perm_bmask): Likewise.
12920 (vector_init_bshuffle): Likewise.
12921 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
12922 (spu_vector_mode_supported_p): Likewise.
12923 (spu_expand_insv): Likewise.
12924 (spu_emit_branch_or_set): Likewise.
12925 (spu_handle_vector_attribute): Likewise.
12926 (spu_builtin_splats): Likewise.
12927 (spu_builtin_extract): Likewise.
12928 (spu_builtin_promote): Likewise.
12929 (spu_expand_sign_extend): Likewise.
12930 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
12931 (tilegx_simd_int): Likewise.
12932 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
12933 (tilepro_simd_int): Likewise.
12934 * config/v850/v850.c (const_double_split): Likewise.
12935 (v850_print_operand): Likewise.
12936 (ep_memory_offset): Likewise.
12937 * config/vax/vax.c (vax_rtx_costs): Likewise.
12938 (vax_output_int_move): Likewise.
12939 (vax_output_int_add): Likewise.
12940 (vax_output_int_subtract): Likewise.
12941 * config/visium/predicates.md (visium_branch_operator): Likewise.
12942 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
12943 (visium_print_operand_address): Likewise.
12944 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12945 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
12946 (xtensa_expand_conditional_branch): Likewise.
12947 (xtensa_copy_incoming_a7): Likewise.
12948 (xtensa_output_literal): Likewise.
12949 * dfp.c (decimal_real_maxval): Likewise.
12950 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12952 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12953 Alan Hayward <alan.hayward@arm.com>
12954 David Sherwood <david.sherwood@arm.com>
12956 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
12957 (mode_nunits_inline): Likewise.
12958 (mode_inner_inline): Likewise.
12959 (mode_unit_size_inline): Likewise.
12960 (mode_unit_precision_inline): Likewise.
12961 (emit_insn_modes_h): Likewise. Also emit a #define of the
12963 (emit_mode_wider): Add an E_ prefix to mode names.
12964 (emit_mode_complex): Likewise.
12965 (emit_mode_inner): Likewise.
12966 (emit_mode_adjustments): Likewise.
12967 (emit_mode_int_n): Likewise.
12968 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
12969 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
12970 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
12971 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
12972 (CRC32_BUILTIN, ENTRY): Likewise.
12973 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
12974 (aarch64_pop_regs): Likewise.
12975 (aarch64_process_components): Likewise.
12976 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
12977 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
12978 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
12979 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
12980 * config/arm/arm.c (arm_init_libfuncs): Likewise.
12981 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
12983 * config/i386/i386-builtin.def (pcmpestr): Likewise.
12984 (pcmpistr): Likewise.
12985 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
12986 * config/mmix/mmix.c (mmix_output_condition): Likewise.
12987 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12989 * config/rl78/rl78.c (mduc_regs): Likewise.
12990 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12991 (htm_expand_builtin): Likewise.
12992 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
12993 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
12994 * config/xtensa/xtensa.c (print_operand): Likewise.
12995 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
12996 (NUM_MODE_VECTOR_INT): Likewise.
12997 * genoutput.c (null_operand): Likewise.
12998 (output_operand_data): Likewise.
12999 * genrecog.c (print_parameter_value): Likewise.
13000 * lra.c (debug_operand_data): Likewise.
13002 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13004 * dwarf2out.c (scompare_loc_descriptor_wide)
13005 (scompare_loc_descriptor_narrow): New functions, split out from...
13006 (scompare_loc_descriptor): ...here.
13007 * expmed.c (emit_store_flag_int): New function, split out from...
13008 (emit_store_flag): ...here.
13010 2017-08-30 Richard Biener <rguenther@suse.de>
13012 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
13013 (dwarf2out_early_finish): Move setting of AT_pubnames from
13014 early debug output to early finish.
13016 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13018 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13019 and -mdata-region to the assembler.
13021 2017-08-30 Richard Biener <rguenther@suse.de>
13023 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13025 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13026 (dwarf2out_early_global_decl): Only generate a DIE for the
13027 abstract origin if it doesn't already exist or is a declaration DIE.
13028 (resolve_addr): Do not add the linkage name twice when
13029 generating a stub DIE for the DW_TAG_GNU_call_site target.
13031 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13033 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13034 Use machine_mode rather than int for arg1_mode.
13036 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13039 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13040 that the second argument of the built-in functions to unpack
13041 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13042 switch statement instead a lot of if statements.
13043 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13044 Allow 64-bit values to be in Altivec registers as well as
13045 traditional floating point registers.
13046 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13048 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13050 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13051 MAX_REGS_PER_ADDRESS == 1.
13053 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13055 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13056 * config/i386/i386.c (ix86_option_override_internal): Simplify
13057 setting of opts->x_flag_entry.
13059 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13060 Jakub Jelinek <jakub@redhat.com>
13061 Richard Biener <rguenther@suse.de>
13063 PR tree-optimization/81503
13064 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13065 folded constant fits in the target type; reorder tests for clarity.
13067 2017-08-29 Martin Liska <mliska@suse.cz>
13069 * passes.def: Include pass_lower_switch.
13070 * stmt.c (dump_case_nodes): Remove and move to
13071 tree-switch-conversion.
13072 (case_values_threshold): Likewise.
13073 (expand_switch_as_decision_tree_p): Likewise.
13074 (emit_case_decision_tree): Likewise.
13075 (expand_case): Likewise.
13076 (balance_case_nodes): Likewise.
13077 (node_has_low_bound): Likewise.
13078 (node_has_high_bound): Likewise.
13079 (node_is_bounded): Likewise.
13080 (emit_case_nodes): Likewise.
13081 (struct simple_case_node): New struct.
13082 (add_case_node): Remove.
13083 (emit_case_dispatch_table): Use vector instead of case_list.
13084 (reset_out_edges_aux): Remove.
13085 (compute_cases_per_edge): Likewise.
13086 (expand_case): Build list of simple_case_node.
13087 (expand_sjlj_dispatch_table): Use it.
13088 * tree-switch-conversion.c (struct case_node): Moved from
13089 stmt.c and adjusted.
13090 (emit_case_nodes): Likewise.
13091 (node_has_low_bound): Likewise.
13092 (node_has_high_bound): Likewise.
13093 (node_is_bounded): Likewise.
13094 (case_values_threshold): Likewise.
13095 (reset_out_edges_aux): Likewise.
13096 (compute_cases_per_edge): Likewise.
13097 (add_case_node): Likewise.
13098 (dump_case_nodes): Likewise.
13099 (balance_case_nodes): Likewise.
13100 (expand_switch_as_decision_tree_p): Likewise.
13101 (emit_jump): Likewise.
13102 (emit_case_decision_tree): Likewise.
13103 (try_switch_expansion): Likewise.
13104 (do_jump_if_equal): Likewise.
13105 (emit_cmp_and_jump_insns): Likewise.
13106 (fix_phi_operands_for_edge): New function.
13107 (record_phi_operand_mapping): Likewise.
13108 (class pass_lower_switch): New pass.
13109 (pass_lower_switch::execute): New function.
13110 (make_pass_lower_switch): Likewise.
13111 (conditional_probability):
13112 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13113 * tree-pass.h: Add make_pass_lower_switch.
13115 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13118 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13121 2017-08-29 Richard Biener <rguenther@suse.de>
13123 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13124 we do not add a DW_AT_inline attribute twice.
13125 (gen_subprogram_die): Remove code setting DW_AT_inline on
13126 DECL_ABSTRACT_P nodes.
13128 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13130 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13131 calls to internal functions.
13132 (gimplify_modify_expr): Likewise.
13133 * tree-call-cdce.c (use_internal_fn): Likewise.
13134 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13135 (convert_to_divmod): Set the nothrow flag.
13136 * tree-if-conv.c (predicate_mem_writes): Likewise.
13137 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13138 (vectorizable_call): Likewise.
13139 (vectorizable_store): Likewise.
13140 (vectorizable_load): Likewise.
13141 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13142 (vect_recog_mask_conversion_pattern): Likewise.
13144 2017-08-29 Martin Liska <mliska@suse.cz>
13147 * gcc.c (driver_handle_option): Add new argument.
13148 * opts-common.c (handle_option): Pass
13149 target_option_override_hook.
13150 * opts-global.c (lang_handle_option): Add new option.
13151 (set_default_handlers): Add new argument.
13152 (decode_options): Likewise.
13153 * opts.c (target_handle_option): Likewise.
13154 (common_handle_option): Call target_option_override_hook.
13155 * opts.h (struct cl_option_handler_func): Add hook for
13156 target option override.
13157 (struct cl_option_handlers): Likewise.
13158 (set_default_handlers): Add new argument.
13159 (decode_options): Likewise.
13160 (common_handle_option): Likewise.
13161 (target_handle_option): Likewise.
13162 * toplev.c (toplev::main): Pass targetm.target_option.override
13165 2017-08-29 Richard Biener <rguenther@suse.de>
13166 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13168 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13169 life to the active subtree.
13171 2017-08-28 Jeff Law <law@redhat.com>
13173 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13174 derive_equivalences.
13175 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13176 Code moved into....
13177 (edge_info::derive_equivalences): New private member function
13179 * tree-ssa-dom.c (class edge_info): Changed from a struct
13180 to a class. Add ctor/dtor, methods and data members.
13181 (edge_info::edge_info): Renamed from allocate_edge_info.
13182 Initialize additional members.
13183 (edge_info::~edge_info): New.
13184 (free_dom_edge_info): Delete the edge info.
13185 (record_edge_info): Use new class & associated member functions.
13186 Tighten forms for testing for edge equivalences.
13187 (record_temporary_equivalences): Iterate over the simple
13188 equivalences rather than assuming there's only one per edge.
13189 (cprop_into_successor_phis): Iterate over the simple
13190 equivalences rather than assuming there's only one per edge.
13191 (optimize_stmt): Use operand_equal_p rather than pointer
13192 equality for mini-DSE code.
13194 2017-08-28 Nathan Sidwell <nathan@acm.org>
13196 * gcc.c (execute): Fold SIGPIPE handling into switch
13197 statement. Adjust internal error message.
13199 2017-08-28 Richard Biener <rguenther@suse.de>
13202 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13203 Do nothing for removed DIEs.
13205 2017-08-28 Richard Biener <rguenther@suse.de>
13207 PR tree-optimization/81977
13208 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13211 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13214 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13215 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13216 using targetm.gen_mem_thread_fence.
13218 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13221 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13222 predicate to register_operand. Reorder operands.
13223 (*btr<mode>): Ditto.
13224 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13225 (*btr<mode>_mask): Ditto.
13227 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13229 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13230 * config/rs6000/xmmintrin.h: New file.
13231 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13233 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13236 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13237 parameter and_insn and return it.
13238 (recombine_lvx_pattern): Insert a copy to ensure availability of
13239 the base register of the copied masking operation at the point of
13240 the instruction replacement.
13241 (recombine_stvx_pattern): Likewise.
13243 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13245 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13246 undocumented switches.
13247 (-mpower9-dform-vector): Likewise.
13248 (-mpower9-dform): Likewise.
13249 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13250 comments to delete references to -mpower9-dform* switches.
13251 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13252 Delete reference to -mpower9-dform* switches, test for
13253 -mpower9-vector instead.
13254 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13255 (OTHER_P9_VECTOR_MASKS): Likewise.
13256 (POWERPC_MASKS): Likewise.
13257 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13258 tests against -mpower9-dform* to -mpower9-vector. Delete code
13259 that checked for -mpower9-dform* consistancy with other options.
13260 Add test for -mpower9-misc to enable other power9 switches.
13261 (rs6000_init_hard_regno_mode_ok): Likewise.
13262 (rs6000_option_override_internal): Likewise.
13263 (rs6000_emit_prologue): Likewise.
13264 (rs6000_emit_epilogue): Likewise.
13265 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13266 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13267 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13269 (emit_fusion_p9_store): Likewise.
13270 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13271 resetting these macros if the assembler does not support ISA 3.0
13273 (TARGET_P9_DFORM_VECTOR): Likewise.
13274 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13275 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13277 2017-08-25 Alan Modra <amodra@gmail.com>
13279 PR rtl-optimization/81747
13280 * cse.c (cse_extended_basic_block): Don't attempt to record
13281 equivalences for degenerate conditional branches that branch
13282 to their fall-through.
13284 2017-08-24 Martin Sebor <msebor@redhat.com>
13286 PR middle-end/81908
13287 * gimple-fold.c (size_must_be_zero_p): New function.
13288 (gimple_fold_builtin_memory_op): Call it.
13290 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13292 * config/rs6000/mm_malloc.h: New file.
13294 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13296 PR tree-optimization/81913
13297 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13298 analysis when either IVs in condition can wrap.
13300 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13302 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13303 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13305 2017-08-24 Richard Biener <rguenther@suse.de>
13308 * targhooks.c (default_target_can_inline_p): Properly
13309 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13310 is present and always compare.
13311 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13312 infer -mfpmath=sse from TARGET_SSE_P.
13313 (ix86_can_inline_p): Properly use target_option_default_node when
13314 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13316 2017-08-24 Richard Biener <rguenther@suse.de>
13319 * dwarf2out.c (output_die): Handle flag_generate_offload like
13321 (output_comp_unit): Likewise.
13322 (gen_array_type_die): Likewise.
13323 (dwarf2out_early_finish): Likewise.
13324 (note_variable_value_in_expr): Likewise.
13325 (dwarf2out_finish): Likewise. Adjust assert.
13326 * cgraphunit.c (symbol_table::compile): Move setting of
13327 flag_generate_offload earlier ...
13328 (symbol_table::finalize_compilation_unit): ... here, before
13329 early debug finalization.
13331 2017-08-24 Richard Biener <rguenther@suse.de>
13333 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13334 and ipa-fnsummary.h.
13335 (ix86_can_inline_p): When ix86_fpmath flags do not match
13336 check whether the callee uses FP math at all.
13338 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13340 PR middle-end/81931
13341 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13342 instead of TYPE_PRECISION.
13344 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13345 Alan Hayward <alan.hayward@arm.com>
13346 David Sherwood <david.sherwood@arm.com>
13348 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13350 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13352 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13354 (ix86_frame::stack_realign_allocate): New field.
13355 (struct machine_frame_state): Modify comments.
13356 (machine_frame_state::sp_realigned_fp_end): New field.
13357 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13358 layout calculation.
13359 (sp_valid_at): Add assertion to assure no attempt to access invalid
13360 offset of a realigned stack.
13361 (fp_valid_at): Likewise.
13362 (choose_baseaddr): Modify comments.
13363 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
13364 ix86_expand_prologue.
13365 (ix86_expand_prologue): Modify stack realignment and allocation.
13366 (ix86_expand_epilogue): Modify comments.
13367 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
13368 avx2_runtime, avx512f, and avx512f_runtime.
13370 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
13372 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
13373 (mstackrealign): Do not init to -1.
13374 * config/i386/i386.c (ix86_option_override_internal):
13375 Check opts_set, not opts when setting default value of
13376 opts->x_ix86_force_align_arg_pointer.
13378 2017-08-23 Richard Biener <rguenther@suse.de>
13380 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13381 lang_hooks.decl_printable_name.
13382 * print-rtl-function.c (print_rtx_function): Likewise.
13383 * tree-pretty-print.c (dump_function_header): Likewise.
13385 2017-08-23 Richard Biener <rguenther@suse.de>
13388 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
13389 -g0 at compile-time.
13391 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13393 PR middle-end/19706
13394 * doc/sourcebuild.texi (Other hardware attributes):
13397 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13399 PR middle-end/19706
13400 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
13401 Fix single-use check.
13403 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
13405 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
13407 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13409 * doc/install.texi: Modify to add more details on running selected
13412 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13414 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
13415 is combined with -mabi=ms.
13416 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
13419 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13421 PR tree-optimization/81488
13422 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
13423 and cached_basis fields.
13424 (MAX_SPREAD): New constant.
13425 (alloc_cand_and_find_basis): Initialize new fields.
13426 (clear_visited): New function.
13427 (create_phi_basis_1): Rename from create_phi_basis, set visited
13428 and cached_basis fields.
13429 (create_phi_basis): New wrapper function.
13430 (phi_add_costs_1): Rename from phi_add_costs, add spread
13431 parameter, set visited field, short-circuit when limits reached.
13432 (phi_add_costs): New wrapper function.
13433 (record_phi_increments_1): Rename from record_phi_increments, set
13435 (record_phi_increments): New wrapper function.
13436 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
13437 (phi_incr_cost): New wrapper function.
13438 (all_phi_incrs_profitable_1): Rename from
13439 all_phi_incrs_profitable, set visited field.
13440 (all_phi_incrs_profitable): New wrapper function.
13442 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13443 Alan Hayward <alan.hayward@arm.com>
13444 David Sherwood <david.sherwood@arm.com>
13446 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
13447 that takes the outer and inner modes.
13448 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
13449 comparison as the canonical test for a paradoxical subreg.
13450 * combine.c (simplify_set): Use paradoxical_subreg_p.
13451 (make_extraction): Likewise.
13452 (force_to_mode): Likewise.
13453 (rtx_equal_for_field_assignment_p): Likewise.
13454 (gen_lowpart_for_combine): Likewise.
13455 (simplify_comparison): Likewise.
13456 * cse.c (equiv_constant): Likewise.
13457 * expmed.c (store_bit_field_1): Likewise.
13458 * final.c (alter_subreg): Likewise.
13459 * fwprop.c (propagate_rtx): Likewise.
13460 (forward_propagate_subreg): Likewise.
13461 * ira-conflicts.c (ira_build_conflicts): Likewise.
13462 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
13463 * lra-constraints.c (curr_insn_transform): Likewise.
13464 (split_reg): Likewise.
13465 * lra-eliminations.c (move_plus_up): Likewise.
13466 (lra_eliminate_regs_1): Likewise.
13467 * recog.c (general_operand): Likewise.
13468 * ree.c (combine_reaching_defs): Likewise.
13469 * reload.c (push_reload): Likewise.
13470 (find_reloads): Likewise.
13471 * reload1.c (elimination_effects): Likewise.
13472 (compute_reload_subreg_offset): Likewise.
13473 (choose_reload_regs): Likewise.
13474 * rtlanal.c (subreg_lsb_1): Likewise.
13475 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13476 (simplify_subreg): Likewise.
13477 * var-tracking.c (track_loc_p): Likewise.
13478 * emit-rtl.c (byte_lowpart_offset): Likewise.
13479 (paradoxical_subreg_p): Delete out-of-line definition.
13481 2017-08-22 Jeff Law <law@redhat.com>
13483 PR tree-optimization/81741
13484 PR tree-optimization/71947
13485 * tree-ssa-dom.c: Include tree-inline.h.
13486 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
13487 equivalences if one is more expensive to compute than the other.
13488 * tree-ssa-scopedtables.h (class const_or_copies): Make
13489 record_const_or_copy_raw method private.
13490 (class avail_exprs_stack): New method simplify_binary_operation.
13491 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
13492 avail_exprs_stack::simplify_binary_operation as needed.
13493 (avail_exprs_stack::simplify_binary_operation): New function.
13495 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
13497 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
13498 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
13499 (DOT_SYMBOLS): Likewise.
13500 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
13501 (RELOCATABLE_NEEDS_FIXUP): Likewise.
13502 (RS6000_ABI_NAME): Likewise.
13503 (TARGET_CMODEL): Likewise.
13504 (TOC_SECTION_ASM_OP): Likewise.
13505 (SET_CMODEL): New macro.
13506 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
13508 2017-08-22 Richard Biener <rguenther@suse.de>
13510 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
13511 to allow for free-lang-data replacements similar to verify_type_variant.
13513 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13514 Alan Hayward <alan.hayward@arm.com>
13515 David Sherwood <david.sherwood@arm.com>
13517 * config/aarch64/aarch64.md (casesi): Use DImode rather than
13518 VOIDmode for the LABEL_REF.
13520 2017-08-22 Richard Biener <rguenther@suse.de>
13522 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
13524 2017-08-22 Richard Biener <rguenther@suse.de>
13526 * common.opt (feliminate-dwarf2-dups): Ignore.
13527 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
13528 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
13529 same_die_p_wrap, compute_section_prefix,
13530 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
13531 (comdat_symbol_id, comdat_symbol_number): Likewise.
13532 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
13534 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
13535 (output_die): Mark unreachable path unreachable.
13536 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
13537 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
13538 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
13539 (dwarf2out_early_finish): Likewise.
13541 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
13543 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
13545 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
13548 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
13549 not VAR_P. Filter attribute warnings with OPT_Wattributes.
13550 (avr_attribute_table) <io, io_low, address>: Initialize
13551 .decl_required with true.
13553 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
13555 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
13556 undocumented debugging options.
13557 (-mvsx-scalar-double): Likewise.
13558 (-mallow-df-permute): Likewise.
13559 (-mvectorize-builtins): Likewise.
13560 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13561 (rs6000_builtin_vectorized_function): Likewise.
13562 (rs6000_builtin_md_vectorized_function): Likewise.
13563 (rs6000_opt_vars): Likewise.
13565 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
13568 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
13569 (*btrq_imm): Rename from *btrq.
13570 (*btcq_imm): Rename from *btcq.
13571 (btsc): New code attribute.
13572 (*<btsc><mode>): New insn pattern.
13573 (*btr<mode>): Ditto.
13574 (*<btsc><mode>_mask): New insn_and_split pattern.
13575 (*btr<mode>_mask): Ditto.
13577 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13578 Alan Hayward <alan.hayward@arm.com>
13579 David Sherwood <david.sherwood@arm.com>
13581 * function.c (pad_below): Simplify padding calculation.
13583 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13584 Alan Hayward <alan.hayward@arm.com>
13585 David Sherwood <david.sherwood@arm.com>
13587 * target.def (function_prologue): Remove frame size argument.
13588 (function_epilogue): Likewise.
13589 * doc/tm.texi: Regenerate.
13590 * final.c (final_start_function): Update call to function_prologue.
13591 (final_end_function): Update call to function_epilogue.
13592 (default_function_pro_epilogue): Remove frame size argument.
13593 * output.h (default_function_pro_epilogue): Likewise.
13594 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
13595 (arm_output_function_prologue): Likewise.
13596 * config/frv/frv.c (frv_function_prologue): Likewise.
13597 (frv_function_epilogue): Likewise.
13598 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
13599 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
13600 (ia64_output_function_epilogue): Likewise.
13601 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
13602 (m32r_output_function_epilogue): Likewise.
13603 * config/microblaze/microblaze.c (microblaze_function_prologue)
13604 (microblaze_function_epilogue): Likewise.
13605 * config/mips/mips.c (mips_output_function_prologue): Likewise.
13606 (mips_output_function_epilogue): Likewise.
13607 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
13608 (mmix_target_asm_function_epilogue): Likewise.
13609 * config/msp430/msp430.c (msp430_start_function): Likewise.
13610 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
13611 (nds32_asm_function_epilogue): Likewise.
13612 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
13613 * config/pa/pa.c (pa_output_function_prologue): Likewise.
13614 (pa_output_function_epilogue): Likewise.
13615 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
13616 (rs6000_output_function_epilogue): Likewise.
13617 * config/rl78/rl78.c (rl78_start_function): Likewise.
13618 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
13619 (rs6000_output_function_epilogue): Likewise.
13620 * config/rx/rx.c (rx_output_function_prologue): Likewise.
13621 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
13622 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
13623 (sparc_asm_function_epilogue): Likewise.
13625 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13627 * tree.h (type_has_mode_precision_p): New function.
13628 * convert.c (convert_to_integer_1): Use it.
13629 * expr.c (expand_expr_real_2): Likewise.
13630 (expand_expr_real_1): Likewise.
13631 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
13632 * match.pd: Likewise.
13633 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
13634 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
13635 * tree-tailcall.c (process_assignment): Likewise.
13636 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13637 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
13638 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
13639 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13640 (vectorizable_assignment): Likewise.
13641 (vectorizable_shift): Likewise.
13642 (vectorizable_operation): Likewise.
13643 * tree-vrp.c (register_edge_assert_for_2): Likewise.
13645 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
13647 * match.pd: Add pow (C, x) simplification.
13649 2017-08-21 Richard Biener <rguenther@suse.de>
13651 PR tree-optimization/81900
13652 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
13653 for blocks with abnormal predecessors.
13654 (compute_antic): Do not set visited flag prematurely.
13656 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
13659 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
13661 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13663 * stor-layout.h (vector_type_mode): Move to...
13664 * tree.h (vector_type_mode): ...here.
13665 * stor-layout.c (vector_type_mode): Move to...
13666 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
13668 2017-08-21 Richard Biener <rguenther@suse.de>
13670 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
13671 register_external_die hooks.
13672 (debug_false_tree_charstarstar_uhwistar): Declare.
13673 (debug_nothing_tree_charstar_uhwi): Likewise.
13674 * debug.c (do_nothing_debug_hooks): Adjust.
13675 (debug_false_tree_charstarstar_uhwistar): New do nothing.
13676 (debug_nothing_tree_charstar_uhwi): Likewise.
13677 * dbxout.c (dbx_debug_hooks): Adjust.
13678 (xcoff_debug_hooks): Likewise.
13679 * sdbout.c (sdb_debug_hooks): Likewise.
13680 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13681 * dwarf2out.c (macinfo_label_base): New global.
13682 (dwarf2out_register_external_die): New function for the
13683 register_external_die hook.
13684 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
13685 (dwarf2_debug_hooks): Use them.
13686 (dwarf2_lineno_debug_hooks): Adjust.
13687 (struct die_struct): Add with_offset flag.
13688 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
13689 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
13690 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
13691 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
13692 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
13693 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
13694 defining section names for the early LTO debug variants.
13695 (reset_indirect_string): New helper.
13696 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
13697 (print_dw_val): Add support for offsetted symbol references.
13698 (get_ultimate_context): Split out from is_cxx.
13699 (is_cxx): Use get_ultimate_context.
13700 (is_fortran): Add decl overload.
13701 (compute_comp_unit_symbol): Split out worker from
13702 compute_section_prefix.
13703 (compute_section_prefix): Call compute_comp_unit_symbol and
13704 set comdat_type_p here.
13705 (output_die): Skip DIE symbol output for the LTO added one.
13706 Handle DIE symbol references with offset.
13707 (output_comp_unit): Guard section name mangling properly.
13708 For LTO debug sections emit a symbol at the section beginning
13709 which we use to refer to its DIEs.
13710 (add_abstract_origin_attribute): For DIEs registered via
13711 dwarf2out_register_external_die directly refer to the early
13712 DIE rather than indirectly through the shadow one we created.
13713 Remove obsolete call to dwarf2out_abstract_function for
13714 non-function/block origins.
13715 (gen_array_type_die): When generating early LTO debug do
13716 not emit DW_AT_string_length.
13717 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
13718 late when in LTO. As suggested place a gcc_unreachable for
13719 the DECL_ABSTRACT_P case.
13720 (gen_subprogram_die): Avoid another specification DIE
13721 for early built declarations/definitions for the late LTO case.
13722 (gen_variable_die): Add type references for late duplicated VLA dies
13724 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
13725 we have the abstract instance already.
13726 (process_scope_var): Adjust decl DIE contexts in LTO which
13727 first puts them in limbo.
13728 (gen_decl_die): Do not generate type DIEs late apart from
13729 types for VLAs or for decls we do not yet have a DIE. Do not
13730 call dwarf2out_abstract_function late.
13731 (dwarf2out_early_global_decl): Make sure to create DIEs
13732 for abstract instances of a decl first.
13733 (dwarf2out_late_global_decl): Adjust comment.
13734 (output_macinfo_op): With multiple macro sections use
13735 macinfo_label_base to distinguish labels.
13736 (output_macinfo): Likewise. Update macinfo_label_base.
13737 Pass in the line info label.
13738 (note_variable_value_in_expr): When generating LTO resolve
13739 all variable values here by generating DIEs as needed.
13740 (init_sections_and_labels): Add early LTO debug flag parameter
13741 and generate different sections and names if set. Add generation
13742 counter for the labels so we can have multiple of them.
13743 (reset_dies): Helper to allow DIEs to be output multiple times.
13744 (dwarf2out_finish): When outputting DIEs to the fat part of an
13745 LTO object first reset DIEs.
13746 (dwarf2out_early_finish): Output early DIEs when generating LTO.
13747 (modified_type_die): Check for decl_ultimate_origin being self
13749 (gen_type_die_with_usage): Likewise.
13750 (gen_typedef_die): Allow decl_ultimate_origin being self.
13751 (set_decl_abstract_flags): Remove.
13752 (set_block_abstract_flags): Likewise.
13753 (dwarf2out_abstract_function): Treat the early generated DIEs
13754 as the abstract copy and only add DW_AT_inline and
13755 DW_AT_artificial here and call set_decl_origin_self.
13756 If the DIE has an abstract origin don't do anything.
13757 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
13758 if we have none yet (Go fails to build one, PR78628).
13759 (variably_modified_type_p): Prevent endless recursion for Ada
13760 cyclic pointer types.
13761 * lto-streamer-in.c: Include debug.h.
13762 (dref_queue): New global.
13763 (lto_read_tree_1): Stream in DIE references.
13764 (lto_input_tree): Register DIE references.
13765 (input_function): Stream DECL_DEBUG_ARGS.
13766 * lto-streamer-out.c: Include debug.h.
13767 (lto_write_tree_1): Output DIE references.
13768 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
13769 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
13770 (output_function): Stream DECL_DEBUG_ARGS.
13771 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
13772 Stream DECL_ABSTRACT_ORIGIN.
13773 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
13774 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
13775 DECL_CONTEXT for file-scope decls.
13776 * lto-streamer.h (struct dref_entry): Declare.
13777 (dref_queue): Likewise.
13778 * cfgexpand.c (pass_expand::execute): Do not call the
13779 outlining_inline_function hook here.
13780 * lto-wrapper.c (debug_obj): New global.
13781 (tool_cleanup): Unlink it if required.
13782 (debug_objcopy): New function.
13783 (run_gcc): Handle early debug sections in the IL files by
13784 extracting them to separate files, partially linkin them and
13785 feeding the result back as result to the linker.
13786 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
13787 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
13788 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
13789 sections into a separate segment.
13790 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
13792 (darwin_asm_dwarf_section): Likewise.
13793 (darwin_asm_output_dwarf_offset): Likewise.
13794 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
13796 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13797 Alan Hayward <alan.hayward@arm.com>
13798 David Sherwood <david.sherwood@arm.com>
13800 * read-md.h (md_reader::record_potential_iterator_use): Replace
13801 pointer argument with an rtx and an index.
13802 * read-rtl.c (iterator_group::apply_iterator): Likewise.
13803 (apply_mode_iterator): Likewise.
13804 (apply_code_iterator): Likewise.
13805 (apply_int_iterator): Likewise.
13806 (apply_subst_iterator): Likewise.
13807 (record_iterator_use): Likewise.
13808 (record_attribute_use): Likewise.
13809 (md_reader::record_potential_iterator_use): Likewise. Update calls
13810 to record_iterator_use and apply_iterator.
13811 (iterator_use): Replace ptr with x and index.
13812 (attribute_use): Likewise.
13813 (apply_attribute_uses): Update calls to apply_iterator.
13814 (apply_iterators): Likewise. Update initialization of iterator_use.
13815 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
13816 and record_potential_iterator_use.
13817 (rtx_reader::read_rtx_operand): Likewise.
13819 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13820 Alan Hayward <alan.hayward@arm.com>
13821 David Sherwood <david.sherwood@arm.com>
13823 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
13826 2017-08-21 Richard Biener <rguenther@suse.de>
13828 PR middle-end/81884
13829 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
13830 at struct end conservatively when comparing common bases.
13832 2017-08-21 Richard Biener <rguenther@suse.de>
13834 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
13835 (mem_ref_in_stmt): Remove.
13836 (determine_max_movement): Use ref index to get at the reference.
13837 (invariantness_dom_walker::before_dom_children): Deal with
13838 lim data already initialized.
13839 (gather_mem_refs_stmt): Initialize lim data and record ref index.
13841 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13843 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
13844 (TARGET_ISA_ROUND): Ditto.
13845 (TARGET_ROUND): Ditto.
13846 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
13847 * config/i386/i386.md: Ditto.
13848 * config/i386/sse.md: Ditto.
13849 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
13850 with OPTION_MASK_ISA_SSE4_1.
13852 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13855 * doc/extend.texi (x86 Built-in Functions): Correct the name of
13856 __builtin_ia32_lzcnt_u16.
13858 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
13861 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
13862 (rs6000_set_current_function): Rewrite function to use it.
13864 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
13867 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
13868 and TYPE_WARN_IF_NOT_ALIGN.
13869 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
13870 (handle_warn_if_not_align): New.
13871 (place_union_field): Call handle_warn_if_not_align.
13872 (place_field): Call handle_warn_if_not_align.
13873 Copy TYPE_WARN_IF_NOT_ALIGN.
13874 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
13875 (layout_type): Likewise.
13876 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
13878 (tree_decl_common): Add warn_if_not_align.
13879 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
13880 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
13881 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
13882 (DECL_WARN_IF_NOT_ALIGN): Likewise.
13883 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
13884 * doc/extend.texi: Document warn_if_not_aligned attribute.
13885 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
13887 2017-08-17 Martin Liska <mliska@suse.cz>
13890 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
13891 (get_data_dependence): Use it as pointer type.
13892 (distribute_loop): Likewise.
13894 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13896 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
13897 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
13898 (p8_vmrgow_<mode>_direct): New define_insn.
13899 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
13900 handle endianness for vmrgew and vmrgow permute patterns.
13902 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13904 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
13905 * config/rs6000/rs6000-cpus.def: Remove comment.
13906 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
13907 (POWERPC_MASKS): Likewise.
13908 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
13909 use of TARGET_VSX_TIMODE.
13910 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
13911 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
13912 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
13913 (rs6000_option_override_internal): Remove dead code.
13914 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
13915 (rs6000_legitimize_reload_address): Likewise.
13916 (rs6000_legitimate_address_p): Likewise.
13917 (rs6000_opt_masks): Delete "vsx-timode".
13918 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
13919 from function comment.
13920 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
13921 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
13922 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
13924 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
13925 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
13926 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
13927 (VSX_TI): Likewise.
13929 (define_peephole2): Likewise.
13931 2017-08-17 Martin Sebor <msebor@redhat.com>
13934 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
13935 past the end of an array.
13936 (test_pp_format): Add test cases.
13938 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
13940 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
13941 missing ECF_NOTHROW flags.
13943 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13946 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
13947 operands residing in integer registers.
13948 (*vsx_le_perm_load_<mode>): Likewise.
13949 (*vsx_le_perm_store_<mode>): Likewise.
13950 (define_peephole2): Add peepholes to optimize the above.
13952 2017-08-17 Marek Polacek <polacek@redhat.com>
13954 PR middle-end/81814
13955 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
13956 to mimic what shorten_compare did. Change the return type to bool.
13957 (fold_cond_expr_with_comparison): Update call to
13958 operand_equal_for_comparison_p.
13959 (fold_ternary_loc): Likewise.
13961 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
13963 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
13965 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
13967 2017-08-17 Richard Biener <rguenther@suse.de>
13969 * tree-ssa-structalias.c (solve_graph): When propagating
13970 to successors update the graphs succ edges and avoid duplicate work.
13972 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
13975 * config/i386/i386.c (ix86_option_override_internal): Save target
13976 specific options after ix86_stack_protector_guard_reg was changed.
13978 2017-08-17 Richard Biener <rguenther@suse.de>
13980 PR tree-optimization/81827
13981 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
13982 (new_var_info): Initialize it conservatively.
13983 (get_call_vi): Mark register vars.
13984 (new_scalar_tmp_constraint_exp): Likewise.
13985 (handle_rhs_call): Likewise.
13986 (handle_const_call): Likewise.
13987 (create_function_info_for): Likewise.
13988 (solve_constraints): Sort varinfos to separate register from
13989 non-register vars to pack points-to solution bitmaps during
13992 2017-08-17 Marek Polacek <polacek@redhat.com>
13994 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
13996 2017-08-17 Richard Biener <rguenther@suse.de>
13998 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
13999 to true when overflow is undefined and we saturated the result.
14001 2017-08-17 Alan Modra <amodra@gmail.com>
14004 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
14005 Don't use store multiple if only one reg needs saving.
14006 (interesting_frame_related_regno): New function.
14007 (rs6000_frame_related): Don't emit frame info for regs that
14009 (rs6000_emit_epilogue): Likewise.
14011 2017-08-16 Nathan Sidwell <nathan@acm.org>
14013 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
14014 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
14015 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14016 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14017 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14018 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14019 (verify_type): Adjust for TYPE_BINFO move.
14020 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14021 process TYPE_BINFO directly.
14022 (hash_tree): Likewise.
14023 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14025 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14028 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14030 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14032 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14035 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14036 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14037 Add ix86_binary_operator_ok to insn constraint.
14038 (*iordi_1_bts): Ditto.
14039 (*xordi_1_btc): Ditto.
14040 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14041 Update corresponding peephole2 pattern.
14045 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14047 PR tree-optimization/81832
14048 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14049 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14051 2017-08-16 Marek Polacek <polacek@redhat.com>
14054 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14055 regarding size_zero_node.
14057 2017-08-16 Martin Liska <mliska@suse.cz>
14060 * config.gcc: Respect previously set extra_objs in case
14063 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14065 PR tree-optimization/81835
14066 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14067 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14068 not depend on the phi.
14070 2017-08-16 Alan Modra <amodra@gmail.com>
14072 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14075 2017-08-16 Alan Modra <amodra@gmail.com>
14077 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14078 (save_reg_p): ..into this. Update all callers.
14079 (first_reg_to_save): Simplify.
14081 2017-08-16 Alan Modra <amodra@gmail.com>
14083 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14086 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14090 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14091 constant count if that count is less than 32.
14093 2017-08-15 Nathan Sidwell <nathan@acm.org>
14095 * gcc.c (execute): Emit friendlier message if inferior is killed
14096 by an external cause.
14098 2017-08-15 Richard Biener <rguenther@suse.de>
14100 PR tree-optimization/81790
14101 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14102 CONSTRUCTORs from simplifying and VN.
14104 2017-08-14 Martin Sebor <msebor@redhat.com>
14106 * builtin-attrs.def: Add comments.
14108 2017-08-14 Martin Sebor <msebor@redhat.com>
14111 * doc/extend.texi (attribute nonstring): Document new attribute.
14113 2017-08-14 Martin Sebor <msebor@redhat.com>
14116 * tree-diagnostic.c (default_tree_printer): Handle %G.
14117 * gimple-pretty-print.h (percent_G_format): Declare new function.
14118 * gimple-pretty-print.c (percent_G_format): Define.
14119 * tree-pretty-print.c (percent_K_format): Add argument.
14121 2017-08-14 Martin Sebor <msebor@redhat.com>
14123 PR translation/79998
14124 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14125 Remove a stray space.
14127 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14130 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14131 (*iordi_1_bts): Ditto.
14132 (*xordi_1_btc): Ditto.
14134 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14137 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14138 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14140 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14141 quoted strings, and make more translator-friendly.
14142 (darwin_rs6000_override_options): Likewise.
14143 (rs6000_option_override_internal): Likewise.
14144 (rs6000_return_in_memory): Fix overlong line.
14145 (init_cmulative_args): Use quoted strings, and make more
14146 translator-friendly.
14147 (rs6000_pass_by_reference): Fix overlong line.
14148 (def_builtin): Use quoted strings.
14149 (altivec_expand_predicate_builtin): Use quoted strings, and make
14150 more translator-friendly.
14151 (htm_expand_builtin): Use quoted strings.
14152 (cpu_expand_builtin): Use quoted strings, and make more
14153 translator-friendly.
14154 (altivec_expand_builtin): Likewise.
14155 (paired_expand_predicate_builtin): Likewise.
14156 (rs6000_invalid_builtin): Likewise.
14157 (builtin_function_type): Use quoted strings.
14158 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14159 more translator-friendly.
14160 (rs6000_trampoline_init): Likewise.
14161 (rs6000_handle_altivec_attribute): Likewise.
14162 (rs6000_inner_target_options): Use quoted strings.
14163 (rs6000_disable_incompatible_switches): Likewise.
14164 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14165 strings, and make more translator-friendly.
14166 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14168 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14170 PR tree-optimization/81799
14171 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14172 cond_expr to simple gimple operand.
14174 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14176 PR middle-end/46932
14177 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14179 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14183 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14184 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14185 TARGET_GASISR_PROLOGUES.
14186 * config/avr/avr.c (avr_option_override): Same.
14187 (avr_pass_pre_proep::execute): Same.
14189 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14192 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14193 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14195 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14197 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14198 define_insn_and_split. Split to a memory load from 0 in
14199 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14200 using PTR mode iterator.
14201 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14202 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14203 (*add_tp_<mode>): Redefine as define_insn_and_split.
14204 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14205 address space. Merge with *add_tp_x32 using PTR mode iterator.
14206 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14207 Split to an add with a memory load from 0 in
14208 DEFAULT_TLS_SEG_REG address space.
14210 2017-08-12 Andrew Pinski <apinski@cavium.com>
14212 * config/aarch64/aarch64-option-extensions.def (rdma):
14213 Fix feature string to what Linux prints out in /proc/cpuinfo.
14215 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14218 * dwarf2out.c (modified_type_die): For C typedef types that have
14219 an ultimate origin, process the ultimate origin instead of the
14221 (gen_typedef_die): Assert that input DECLs have no ultimate
14223 (gen_type_die_with_usage): For typedef variants that have an
14224 ultimate origin, just call gen_decl_die on the original DECL.
14225 (process_scope_var): Avoid creating DIEs for local typedefs and
14226 concrete static variables.
14228 2017-08-12 Alan Modra <amodra@gmail.com>
14232 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14233 match gnu-user.h startfile.
14234 (ENDFILE_LINUX_SPEC): Similarly.
14236 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14239 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14241 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14243 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14244 * config/aarch64/aarch64.md (mov<mode>): Change.
14245 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14246 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14247 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14249 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14251 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14252 for storage order barriers.
14254 2017-08-11 Martin Liska <mliska@suse.cz>
14257 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14258 variables of void type.
14260 2017-08-11 Martin Liska <mliska@suse.cz>
14262 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14263 TARGET_SUPPORTS_ALIASES.
14264 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14265 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14266 (optimize_weakref): Likewise.
14267 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14268 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14269 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14271 2017-08-11 Martin Liska <mliska@suse.cz>
14274 * config/i386/i386.c (make_resolver_func): Do complete
14275 refactoring of the function.
14277 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14280 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14281 * config/i386/i386.c (ix86_stack_protect_guard): Use
14282 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14283 * doc/invoke.texi (x86 Options): Document
14284 -mstack-protector-guard-symbol= option.
14286 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14288 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14289 * config/i386/i386.c (ix86_split_stack_guard): New function.
14290 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14291 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14292 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14293 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14294 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14295 (split_stack_space_check): Call ix86_split_stack_guard.
14297 2017-08-10 Martin Sebor <msebor@redhat.com>
14299 * print-tree.c (print_node): Print location using the established
14301 Replace spaces with colons.
14302 (debug_raw, debug): Ditto.
14304 2017-08-10 Martin Sebor <msebor@redhat.com>
14307 * pretty-print.c (pp_format): Correct the handling of %s precision.
14309 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14312 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14314 (ix86_finalize_stack_frame_flags): This. Also clear
14315 frame_pointer_needed if -fno-omit-frame-pointer is used without
14317 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14318 with ix86_finalize_stack_frame_flags.
14319 (ix86_expand_epilogue): Likewise.
14320 (ix86_expand_split_stack_prologue): Likewise.
14321 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14323 2017-08-10 Martin Liska <mliska@suse.cz>
14326 * c-attribs.c (handle_target_attribute):
14327 Report warning for an empty string argument of target attribute.
14329 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14332 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14334 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14335 or DECL_NONLOCAL labels.
14336 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14337 or DECL_NONLOCAL labels here.
14339 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14341 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14342 to indicate when early gimple folding has been disabled.
14343 (rs6000_gimple_fold_builtin): Add debug content.
14344 (rs6000_invalid_builtin): Fix whitespace.
14345 (rs6000_expand_builtin): Fix whitespace.
14346 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14348 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14351 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14352 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14354 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14356 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14357 (qdf24xx): Likewise.
14358 * config/aarch64/aarch64.md: Include falkor.md.
14359 * config/aarch64/falkor.md: New.
14361 2017-08-09 Marek Polacek <polacek@redhat.com>
14364 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
14365 * diagnostic.c (emit_diagnostic): Add a comment.
14366 (emit_diagnostic_valist): New function.
14368 2017-08-09 Marek Polacek <polacek@redhat.com>
14371 * input.c (make_location): New overload.
14372 * input.h (make_location): Declare.
14374 2017-08-08 Alan Modra <amodra@gmail.com>
14375 H.J. Lu <hongjiu.lu@intel.com>
14378 * gcc.c (NO_PIE_SPEC): Delete.
14379 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14381 (LINK_PIE_SPEC): ..to here.
14382 (LINK_COMMAND_SPEC): Support -no-pie.
14383 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14384 chain of crtbegin*.o selection, update for PIE_SPEC changes and
14386 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
14387 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
14388 (ENDFILE_CRTEND_SPEC): Similarly.
14390 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
14393 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
14394 (mstack-protector-guard-offset=): Ditto.
14395 * config/i386/i386.c (ix86_option_override): Handle
14396 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
14398 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
14399 ix86_stack_protect_guard_offset variables.
14400 (TARGET_STACK_PROTECT_GUARD): Always define.
14401 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
14402 and -mstack-protector-guard-offset= options.
14404 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14406 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
14407 boundary case for the last candidate.
14409 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14411 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
14412 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
14414 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14416 PR middle-end/19706
14417 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
14418 * config/aarch64/aarch64-builtins.c
14419 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
14420 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
14421 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
14423 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14424 Andrew Pinski <pinskia@gmail.com>
14426 PR middle-end/19706
14427 * internal-fn.def (XORSIGN): New.
14428 * optabs.def (xorsign_optab): New.
14429 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
14430 (convert_expand_mult_copysign): New.
14431 (pass_optimize_widening_mul::execute): Call
14432 convert_expand_mult_copysign.
14434 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14436 PR tree-optimization/81354
14437 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
14438 Insert on edges rather than explicitly creating landing pads.
14439 (analyze_candidates_and_replace): Commit edge inserts.
14441 2017-08-08 Richard Biener <rguenther@suse.de>
14443 PR middle-end/81719
14444 * tree-ssa-loop-niter.c: Include tree-dfa.h.
14445 (expand_simple_operations): Also look through ADDR_EXPRs with
14446 MEM_REF bases treating them as POINTER_PLUS_EXPR.
14448 2017-08-08 Richard Biener <rguenther@suse.de>
14450 PR tree-optimization/81723
14451 * tree-vect-slp.c (struct bst_traits): New hash traits.
14452 (bst_fail): New global.
14453 (vect_build_slp_tree_2): New worker, split out from ...
14454 (vect_build_slp_tree): ... this now wrapping it with using
14455 bst_fail set to cache SLP tree build fails. Properly handle
14457 (vect_analyze_slp_instance): Allocate and free bst_fail.
14459 2017-08-08 Martin Liska <mliska@suse.cz>
14462 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
14463 LABEL_DECLs that can be from a different function.
14465 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14467 PR tree-optimization/81744
14468 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
14469 loop's number of iterations.
14471 2017-08-08 Martin Liska <mliska@suse.cz>
14473 * asan.c: Include header files.
14474 * attribs.c (build_decl_attribute_variant): New function moved
14476 (build_type_attribute_qual_variant): Likewise.
14477 (cmp_attrib_identifiers): Likewise.
14478 (simple_cst_list_equal): Likewise.
14479 (omp_declare_simd_clauses_equal): Likewise.
14480 (attribute_value_equal): Likewise.
14481 (comp_type_attributes): Likewise.
14482 (build_type_attribute_variant): Likewise.
14483 (lookup_ident_attribute): Likewise.
14484 (remove_attribute): Likewise.
14485 (merge_attributes): Likewise.
14486 (merge_type_attributes): Likewise.
14487 (merge_decl_attributes): Likewise.
14488 (merge_dllimport_decl_attributes): Likewise.
14489 (handle_dll_attribute): Likewise.
14490 (attribute_list_equal): Likewise.
14491 (attribute_list_contained): Likewise.
14492 * attribs.h (lookup_attribute): New function moved from tree.[ch].
14493 (lookup_attribute_by_prefix): Likewise.
14494 * bb-reorder.c: Include header files.
14495 * builtins.c: Likewise.
14496 * calls.c: Likewise.
14497 * cfgexpand.c: Likewise.
14498 * cgraph.c: Likewise.
14499 * cgraphunit.c: Likewise.
14500 * convert.c: Likewise.
14501 * dwarf2out.c: Likewise.
14502 * final.c: Likewise.
14503 * fold-const.c: Likewise.
14504 * function.c: Likewise.
14505 * gimple-expr.c: Likewise.
14506 * gimple-fold.c: Likewise.
14507 * gimple-pretty-print.c: Likewise.
14508 * gimple.c: Likewise.
14509 * gimplify.c: Likewise.
14510 * hsa-common.c: Likewise.
14511 * hsa-gen.c: Likewise.
14512 * internal-fn.c: Likewise.
14513 * ipa-chkp.c: Likewise.
14514 * ipa-cp.c: Likewise.
14515 * ipa-devirt.c: Likewise.
14516 * ipa-fnsummary.c: Likewise.
14517 * ipa-inline.c: Likewise.
14518 * ipa-visibility.c: Likewise.
14520 * lto-cgraph.c: Likewise.
14521 * omp-expand.c: Likewise.
14522 * omp-general.c: Likewise.
14523 * omp-low.c: Likewise.
14524 * omp-offload.c: Likewise.
14525 * omp-simd-clone.c: Likewise.
14526 * opts-global.c: Likewise.
14527 * passes.c: Likewise.
14528 * predict.c: Likewise.
14529 * sancov.c: Likewise.
14530 * sanopt.c: Likewise.
14531 * symtab.c: Likewise.
14532 * toplev.c: Likewise.
14533 * trans-mem.c: Likewise.
14534 * tree-chkp.c: Likewise.
14535 * tree-eh.c: Likewise.
14536 * tree-into-ssa.c: Likewise.
14537 * tree-object-size.c: Likewise.
14538 * tree-parloops.c: Likewise.
14539 * tree-profile.c: Likewise.
14540 * tree-ssa-ccp.c: Likewise.
14541 * tree-ssa-live.c: Likewise.
14542 * tree-ssa-loop.c: Likewise.
14543 * tree-ssa-sccvn.c: Likewise.
14544 * tree-ssa-structalias.c: Likewise.
14545 * tree-ssa.c: Likewise.
14546 * tree-streamer-in.c: Likewise.
14547 * tree-vectorizer.c: Likewise.
14548 * tree-vrp.c: Likewise.
14549 * tsan.c: Likewise.
14550 * ubsan.c: Likewise.
14551 * varasm.c: Likewise.
14552 * varpool.c: Likewise.
14553 * tree.c: Remove functions moved to attribs.[ch].
14554 * tree.h: Likewise.
14555 * config/aarch64/aarch64.c: Add attrs.h header file.
14556 * config/alpha/alpha.c: Likewise.
14557 * config/arc/arc.c: Likewise.
14558 * config/arm/arm.c: Likewise.
14559 * config/avr/avr.c: Likewise.
14560 * config/bfin/bfin.c: Likewise.
14561 * config/c6x/c6x.c: Likewise.
14562 * config/cr16/cr16.c: Likewise.
14563 * config/cris/cris.c: Likewise.
14564 * config/darwin.c: Likewise.
14565 * config/epiphany/epiphany.c: Likewise.
14566 * config/fr30/fr30.c: Likewise.
14567 * config/frv/frv.c: Likewise.
14568 * config/ft32/ft32.c: Likewise.
14569 * config/h8300/h8300.c: Likewise.
14570 * config/i386/winnt.c: Likewise.
14571 * config/ia64/ia64.c: Likewise.
14572 * config/iq2000/iq2000.c: Likewise.
14573 * config/lm32/lm32.c: Likewise.
14574 * config/m32c/m32c.c: Likewise.
14575 * config/m32r/m32r.c: Likewise.
14576 * config/m68k/m68k.c: Likewise.
14577 * config/mcore/mcore.c: Likewise.
14578 * config/microblaze/microblaze.c: Likewise.
14579 * config/mips/mips.c: Likewise.
14580 * config/mmix/mmix.c: Likewise.
14581 * config/mn10300/mn10300.c: Likewise.
14582 * config/moxie/moxie.c: Likewise.
14583 * config/msp430/msp430.c: Likewise.
14584 * config/nds32/nds32-isr.c: Likewise.
14585 * config/nds32/nds32.c: Likewise.
14586 * config/nios2/nios2.c: Likewise.
14587 * config/nvptx/nvptx.c: Likewise.
14588 * config/pa/pa.c: Likewise.
14589 * config/pdp11/pdp11.c: Likewise.
14590 * config/powerpcspe/powerpcspe.c: Likewise.
14591 * config/riscv/riscv.c: Likewise.
14592 * config/rl78/rl78.c: Likewise.
14593 * config/rx/rx.c: Likewise.
14594 * config/s390/s390.c: Likewise.
14595 * config/sh/sh.c: Likewise.
14596 * config/sol2.c: Likewise.
14597 * config/sparc/sparc.c: Likewise.
14598 * config/spu/spu.c: Likewise.
14599 * config/stormy16/stormy16.c: Likewise.
14600 * config/tilegx/tilegx.c: Likewise.
14601 * config/tilepro/tilepro.c: Likewise.
14602 * config/v850/v850.c: Likewise.
14603 * config/vax/vax.c: Likewise.
14604 * config/visium/visium.c: Likewise.
14605 * config/xtensa/xtensa.c: Likewise.
14607 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14610 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
14611 constraints since the -mupper-regs-* switches have been
14613 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
14614 into a vector from a double word element that was extracted from
14615 another vector, and eliminate extra XXPERMDI instructions.
14616 (vsx_concat_<mode>_2): Likewise.
14617 (vsx_concat_<mode>_3): Likewise.
14618 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
14619 concat to allow optimizing inserts from previous extracts.
14621 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
14623 * config/i386/i386.c (ix86_stack_protect_guard): Generate
14624 memory reference to a SSP offset in TLS address space.
14625 (ix86_print_operand) <case '@'>: Remove.
14626 (ix86_print_operand_punct_valid_p): Remove '@' code.
14627 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
14628 UNSPEC_SP_TLS_TEST.
14629 (stack_tls_protect_set_<mode>): Remove.
14630 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
14631 (stack_tls_protect_test_<mode>): Remove.
14632 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
14634 2017-08-07 Olivier Hainque <hainque@adacore.com>
14637 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
14639 2017-08-07 Douglas Rupp <rupp@adacore.com>
14641 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
14642 variable was referenced as multidir in command.
14644 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14647 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
14650 2017-08-07 Martin Liska <mliska@suse.cz>
14652 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
14653 * config/rl78/rl78.c: Add include of attribs.h.
14654 * config/sh/sh.c: Likewise.
14655 * config/v850/v850.c: Likewise.
14657 2017-08-07 Tom de Vries <tom@codesourcery.com>
14659 PR middle-end/78266
14660 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
14662 2017-08-07 Martin Liska <mliska@suse.cz>
14664 * config/mips/mips.c: Include attribs.h.
14666 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
14669 * doc/invoke.texi: Document change in behvaior for -Ofast for
14672 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
14674 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
14676 (aarch64_pop_regs): Likewise.
14677 (aarch64_gen_load_pair): Likewise.
14678 (aarch64_save_callee_saves): Likewise.
14679 (aarch64_restore_callee_saves): Likewise.
14681 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14683 * config/i386/i386.c: Revert the last change.
14685 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14688 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14690 (ix86_finalize_stack_frame_flags): This. Also clear
14691 frame_pointer_needed if -fno-omit-frame-pointer is used without
14693 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14694 with ix86_finalize_stack_frame_flags.
14695 (ix86_expand_epilogue): Likewise.
14696 (ix86_expand_split_stack_prologue): Likewise.
14698 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14701 * config/i386/i386.c (get_builtin_code_for_version): Set priority
14702 to P_AES for Westmere.
14704 2017-08-07 Jonathan Yong <10walls@gmail.com>
14706 * config/i386/mingw.opt (fset-stack-executable): Removed.
14707 * config/i386/cygming.opt (fset-stack-executable): Moved
14709 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
14711 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
14713 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
14715 2017-08-07 Marek Polacek <polacek@redhat.com>
14717 PR middle-end/81737
14718 * fold-const.c (fold_indirect_ref_1): Check type_domain.
14720 2017-08-07 Martin Liska <mliska@suse.cz>
14722 * attribs.h (canonicalize_attr_name): New function.
14723 (cmp_attribs): Move from c-format.c and adjusted.
14724 (is_attribute_p): Moved from tree.h.
14725 * tree-inline.c: Add new includes.
14726 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
14727 (private_is_attribute_p): Remove.
14728 (private_lookup_attribute): Likewise.
14729 (private_lookup_attribute_by_prefix): Simplify.
14730 (remove_attribute): Use is_attribute_p.
14731 * tree.h: Remove removed declarations.
14733 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14735 PR middle-end/81698
14736 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
14737 instead of computing it in the function. Formatting fix.
14738 (expand_case): Don't rely on default_edge being the first edge,
14739 clear it if removing it, pass default_edge to
14740 emit_case_dispatch_table.
14741 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
14744 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
14746 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
14747 insn in the function, emit NOP after the insn.
14749 2017-08-06 Tom de Vries <tom@codesourcery.com>
14751 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
14754 2017-08-06 Tom de Vries <tom@codesourcery.com>
14756 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
14759 2017-08-04 Yury Gribov <tetra2005@gmail.com>
14761 PR tree-optimization/57371
14762 * match.pd: New pattern.
14764 2017-08-04 Marek Polacek <polacek@redhat.com>
14766 PR middle-end/81695
14767 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
14768 perform the computation in offset_int.
14770 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14772 PR tree-optimization/81136
14773 * tree-vectorizer.h: Include tree-hash-traits.h.
14774 (vec_base_alignments): New typedef.
14775 (vec_info): Add a base_alignments field.
14776 (vect_record_base_alignments): Declare.
14777 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
14779 (DR_IS_CONDITIONAL_IN_STMT): New macro.
14780 (create_data_ref): Add an is_conditional_in_stmt argument.
14781 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
14782 the is_conditional_in_stmt field.
14783 (data_ref_loc): Add an is_conditional_in_stmt field.
14784 (get_references_in_stmt): Set the is_conditional_in_stmt field.
14785 (find_data_references_in_stmt): Update call to create_data_ref.
14786 (graphite_find_data_references_in_stmt): Likewise.
14787 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
14788 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
14789 (vect_record_base_alignment): New function.
14790 (vect_record_base_alignments): Likewise.
14791 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
14792 for nested statements even if we fail to compute a misalignment.
14793 Use pooled base alignments for unconditional references.
14794 (vect_find_same_alignment_drs): Compare base addresses instead
14796 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
14797 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
14799 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14801 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
14802 Add an explicit name for the enum. Use auto_vec for slp_instances
14803 and grouped_stores.
14804 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
14806 (_bb_vec_info): Add a constructor and destructor.
14807 (vinfo_for_stmt): Return NULL for uids of -1 as well.
14808 (destroy_loop_vec_info): Delete.
14809 (vect_destroy_datarefs): Likewise.
14810 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
14811 (vec_info::vec_info): New function.
14812 (vec_info::~vec_info): Likewise.
14813 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
14814 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
14815 destroy_loop_vec_info.
14816 * tree-vect-loop.c (new_loop_vec_info): Replace with...
14817 (_loop_vec_info::_loop_vec_info): ...this.
14818 (destroy_loop_vec_info): Replace with...
14819 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
14820 the stmt_vec_infos. Leave handling of vec_info information to its
14821 destructor. Remove explicit vector releases.
14822 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
14823 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
14824 * tree-vect-slp.c (new_bb_vec_info): Replace with...
14825 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
14826 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
14827 (destroy_bb_vec_info): Replace with...
14828 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
14829 information to its destructor.
14830 (vect_slp_analyze_bb_1): Use new and delete instead of
14831 new_bb_vec_info and destroy_bb_vec_info.
14832 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
14835 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14837 * tree-data-ref.h (subscript): Add access_fn field.
14838 (data_dependence_relation): Add could_be_independent_p.
14839 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
14840 (same_access_functions): Move to tree-data-ref.c.
14841 * tree-data-ref.c (ref_contains_union_access_p): New function.
14842 (access_fn_component_p): Likewise.
14843 (access_fn_components_comparable_p): Likewise.
14844 (dr_analyze_indices): Add a reference to access_fn_component_p.
14845 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
14847 (constant_access_functions): Likewise.
14848 (add_other_self_distances): Likewise.
14849 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
14850 (initialize_data_dependence_relation): Use XCNEW and remove
14851 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
14852 of access functions that have the same type. Allow the
14853 subsequence to end with different bases in some circumstances.
14854 Record the chosen access functions in SUB_ACCESS_FN.
14855 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
14856 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
14857 (subscript_dependence_tester_1): Likewise dra and drb.
14858 (build_classic_dist_vector): Update calls accordingly.
14859 (subscript_dependence_tester): Likewise.
14860 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
14861 DDR_COULD_BE_INDEPENDENT_P.
14862 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
14863 comp_alias_ddrs instead of may_alias_ddrs.
14864 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
14866 (vect_analyze_data_ref_dependence): Use it if
14867 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
14868 distance vectors if that fails.
14869 (dependence_distance_ge_vf): New function.
14870 (vect_prune_runtime_alias_test_list): Use it. Don't clear
14871 LOOP_VINFO_MAY_ALIAS_DDRS.
14873 2017-08-04 Richard Biener <rguenther@suse.de>
14875 PR middle-end/81705
14876 * fold-const.c (fold_binary_loc): Properly restrict
14877 minus_var0 && minus_var1 case when associating undefined overflow
14880 2017-08-04 Tom de Vries <tom@codesourcery.com>
14882 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
14884 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14886 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14887 Don't start diagnostic messages with a capital letter.
14888 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14890 (rs6000_invalid_builtin): Likewise.
14891 (rs6000_trampoline_init): Likewise.
14893 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14896 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
14897 after setting changeable df flags.
14899 2017-08-03 Richard Biener <rguenther@suse.de>
14901 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
14902 up if the use is in USE - X.
14904 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
14906 * toplev.c (dumpfile.h): New include.
14907 (internal_error_reentered): New static function. Use it...
14908 (internal_error_function): ...here to handle reentered internal_error.
14910 2017-08-03 Richard Biener <rguenther@suse.de>
14912 PR middle-end/81148
14913 * fold-const.c (split_tree): Add minus_var and minus_con
14914 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
14915 here but always use minus_*.
14916 (associate_trees): Assert we never associate with MINUS_EXPR
14917 and NULL first operand. Do not recurse for PLUS_EXPR operands
14918 when associating as MINUS_EXPR either.
14919 (fold_binary_loc): Track minus_var and minus_con.
14921 2017-08-03 Tom de Vries <tom@codesourcery.com>
14924 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
14925 ACCEL_COMPILER, apply finish_options on
14926 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
14928 2017-08-03 Tom de Vries <tom@codesourcery.com>
14931 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
14932 function_entry_patch_area_size > 0.
14934 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14937 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
14938 instead of 10??LU, perform unit multiplication in wide_int,
14939 don't change alloc_object_size_limit if the limit is larger
14942 PR tree-optimization/81655
14943 PR tree-optimization/81588
14944 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
14945 the case when ranges[i].low and high are 1 for unsigned type with
14948 PR middle-end/81052
14949 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
14950 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
14952 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14954 * tree-vrp.h: Add include guard.
14956 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
14959 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
14960 (ud2): New insn pattern.
14961 * config/i386/i386.c (ix86_expand_epilogue):
14962 For naked functions, generate ud2 instead of trap insn.
14964 2017-08-02 Marek Polacek <polacek@redhat.com>
14967 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
14969 2017-08-02 Tom de Vries <tom@codesourcery.com>
14970 Cesar Philippidis <cesar@codesourcery.com>
14972 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
14973 Add missing edge probabilities.
14975 2017-08-02 Tamar Christina <tamar.christina@arm.com>
14977 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
14978 Correct endianness.
14980 2017-08-02 Jakub Jelinek <jakub@redhat.com>
14982 PR middle-end/79499
14983 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
14984 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
14985 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
14986 prologue_seq sequences - if any.
14988 2017-08-02 Richard Biener <rguenther@suse.de>
14990 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
14991 via vectors if supported, integer extracts via punning if supported
14992 or otherwise vector extracts.
14994 2017-08-02 Richard Biener <rguenther@suse.de>
14996 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
14998 (bitmap_insert_into_set): ... this.
15000 2017-08-02 Richard Biener <rguenther@suse.de>
15002 PR tree-optimization/81633
15004 2015-08-17 Alan Hayward <alan.hayward@arm.com>
15006 PR tree-optimization/71752
15007 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
15009 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
15011 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
15012 (machine_function::call_ms2sysv_pad_out): Remove field.
15013 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
15014 (ix86_compute_frame_layout): Likewise.
15016 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15019 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15020 attribute with interrupt attribute.
15022 2017-08-01 Andrew Pinski <apinski@cavium.com>
15024 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15025 BIT_INSERT_EXPR's operand 1
15026 to see if the types precision matches.
15028 2017-08-01 Martin Liska <mliska@suse.cz>
15030 PR middle-end/70140
15031 * builtins.c (expand_builtin_memcpy_args): Remove.
15032 (expand_builtin_memcpy): Call newly added function
15033 expand_builtin_memory_copy_args.
15034 (expand_builtin_memcpy_with_bounds): Likewise.
15035 (expand_builtin_mempcpy): Remove last argument.
15036 (expand_builtin_mempcpy_with_bounds): Likewise.
15037 (expand_builtin_memory_copy_args): New function created from
15038 expand_builtin_mempcpy_args with small modifications.
15039 (expand_builtin_mempcpy_args): Remove.
15040 (expand_builtin_stpcpy): Remove unused argument.
15041 (expand_builtin): Likewise.
15042 (expand_builtin_with_bounds): Likewise.
15044 2017-08-01 Martin Liska <mliska@suse.cz>
15047 Make mempcpy more optimal (PR middle-end/70140).
15049 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15052 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15053 __builtin_vec_cmpne verify both arguments are compatible vectors
15054 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15055 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15056 move computation of aligned to after checking the argument types.
15060 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15061 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15064 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15065 Doug Rupp <rupp@adacore.com>
15066 Olivier Hainque <hainque@adacore.com>
15068 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15069 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15071 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15072 for TARGET_OS_CPP_BUILTIN.
15073 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15074 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15076 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15077 passing required abi options to the assembler for EABI configurations.
15078 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15079 of .text.hot and .text.unlikely sections for kernel modules when
15080 using ARM style exceptions.
15081 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15082 options. Add EXTRA_CC1_SPEC.
15083 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15085 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15087 (ARM_TARGET2_DWARF_FORMAT): Define.
15088 * config/arm/t-vxworks: Adjust multilib control to removal of the
15089 Diab command line options.
15091 2017-08-01 Martin Liska <mliska@suse.cz>
15093 PR gcov-profile/81561
15094 * gcov.c (unblock): Make unblocking safe as we need to preserve
15095 index correspondence of blocks and block_lists.
15097 2017-08-01 Richard Biener <rguenther@suse.de>
15099 PR tree-optimization/81181
15100 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15101 (compute_antic): ... end of iteration here.
15103 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15105 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15106 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15107 (ftree-slp-vectorize): Likewise.
15108 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15109 can no longer be set independent of flag_tree_loop_vectorize.
15110 * omp-general.c (emp_max_vf): Likewise.
15111 * opts.c (enable_fdo_optimizations): Remove references to
15112 flag_tree_vectorize, these are now implicit.
15113 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15114 and leave it for the options machinery.
15116 2017-08-01 Martin Liska <mliska@suse.cz>
15118 PR middle-end/70140
15119 * builtins.c (expand_builtin_memcpy_args): Remove.
15120 (expand_builtin_memcpy): Call newly added function
15121 expand_builtin_memory_copy_args.
15122 (expand_builtin_memcpy_with_bounds): Likewise.
15123 (expand_builtin_mempcpy): Remove last argument.
15124 (expand_builtin_mempcpy_with_bounds): Likewise.
15125 (expand_builtin_memory_copy_args): New function created from
15126 expand_builtin_mempcpy_args with small modifications.
15127 (expand_builtin_mempcpy_args): Remove.
15128 (expand_builtin_stpcpy): Remove unused argument.
15129 (expand_builtin): Likewise.
15130 (expand_builtin_with_bounds): Likewise.
15132 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15135 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15136 print "ds:" only for immediates in generic address space.
15138 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15141 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15142 (ix86_function_ok_for_sibcall): Return false for naked functions.
15144 2017-08-01 Richard Biener <rguenther@suse.de>
15146 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15147 (compute_antic): Seed worklist with exit block predecessors.
15148 * cfganal.c (dfs_find_deadend): For a cycle return the source
15149 of the edge closing it.
15151 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15153 * config/aarch64/aarch64.c
15154 (aarch64_can_const_movi_rtx_p): Move 0 check.
15156 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15158 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15159 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15161 * match.pd: Ditto in address comparison pattern.
15163 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15165 PR tree-optimization/81627
15166 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15167 closed ssa form for store-store chain.
15169 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15171 PR tree-optimization/81620
15172 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15173 for store-store chain.
15175 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15177 PR tree-optimization/81588
15178 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15179 ranges[i].in_p, invert comparison code ccode. For >/>=,
15180 swap rhs1 and rhs2 and comparison code unconditionally,
15181 for </<= don't do that. Don't swap rhs1/rhs2 again if
15182 ranges[i].in_p, instead invert comparison code ccode if
15183 opcode or oe->rank is BIT_IOR_EXPR.
15186 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15187 a direct optab to conversion optab.
15188 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15189 with GET_MODE_INNER as last argument instead of optab_handler.
15190 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15191 vector extraction if possible and optab is available.
15192 * expr.c (store_constructor): Use convert_optab_handler instead
15193 of optab_handler. Use vector initialization from smaller
15194 vectors if possible and optab is available.
15195 * tree-vect-stmts.c (vectorizable_load): Likewise.
15196 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15197 now have two modes.
15198 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15199 of vec_init from half-sized vectors with the same element mode.
15200 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15201 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15202 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15203 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15204 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15205 after mode in gen_vec_extract* calls.
15206 (vec_extract<mode>): Renamed to ...
15207 (vec_extract<mode><ssescalarmodelower>): ... this.
15208 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15209 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15210 element mode after mode in gen_vec_init* calls.
15211 (VEC_INIT_HALF_MODE): New mode iterator.
15212 (vec_init<mode>): Renamed to ...
15213 (vec_init<mode><ssescalarmodelower>): ... this.
15214 (vec_init<mode><ssehalfvecmodelower>): New expander.
15215 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15216 (vec_extractv2sfsf): ... this.
15217 (vec_initv2sf): Renamed to ...
15218 (vec_initv2sfsf): ... this.
15219 (vec_extractv2si): Renamed to ...
15220 (vec_extractv2sisi): ... this.
15221 (vec_initv2si): Renamed to ...
15222 (vec_initv2sisi): ... this.
15223 (vec_extractv4hi): Renamed to ...
15224 (vec_extractv4hihi): ... this.
15225 (vec_initv4hi): Renamed to ...
15226 (vec_initv4hihi): ... this.
15227 (vec_extractv8qi): Renamed to ...
15228 (vec_extractv8qiqi): ... this.
15229 (vec_initv8qi): Renamed to ...
15230 (vec_initv8qiqi): ... this.
15231 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15232 (vec_init<mode>): Renamed to ...
15233 (vec_init<mode><VEC_base_l>): ... this.
15234 (vec_extract<mode>): Renamed to ...
15235 (vec_extract<mode><VEC_base_l>): ... this.
15236 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15237 (vec_initv2sfsf): ... this.
15238 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15239 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15240 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15241 element mode after mode in gen_vec_init* calls.
15242 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15243 (vec_init<mode><Vel>): ... this.
15244 (vec_extract<mode>): Renamed to ...
15245 (vec_extract<mode><Vel>): ... this.
15246 * config/aarch64/iterators.md (Vel): New mode attribute.
15247 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15248 Add element mode after mode in gen_vec_extract* calls.
15249 * config/s390/vector.md (non_vec_l): New mode attribute.
15250 (vec_extract<mode>): Renamed to ...
15251 (vec_extract<mode><non_vec_l>): ... this.
15252 (vec_init<mode>): Renamed to ...
15253 (vec_init<mode><non_vec_l>): ... this.
15254 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15255 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15257 * config/arm/iterators.md (V_elem_l): New mode attribute.
15258 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15259 (vec_extract<mode><V_elem_l>): ... this.
15260 (vec_extractv2di): Renamed to ...
15261 (vec_extractv2didi): ... this.
15262 (vec_init<mode>): Renamed to ...
15263 (vec_init<mode><V_elem_l>): ... this.
15264 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15265 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15266 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15267 Add element mode after gen_vec_extract* calls.
15268 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15269 (vec_init<mode><unitmode>): ... this.
15270 (vec_extract<mode>): Renamed to ...
15271 (vec_extract<mode><unitmode>): ... this.
15272 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15273 (vec_init<mode><unitmode>): ... this.
15274 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15275 (vec_initv2sfsf): ... this.
15276 (vec_extractv2sf): Renamed to ...
15277 (vec_extractv2sfsf): ... this.
15278 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15279 Add element mode after gen_vec_extract* calls.
15280 * config/mips/mips.md (unitmode): New mode iterator.
15281 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15282 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15283 * config/spu/spu.md (inner_l): New mode attribute.
15284 (vec_init<mode>): Renamed to ...
15285 (vec_init<mode><inner_l>): ... this.
15286 (vec_extract<mode>): Renamed to ...
15287 (vec_extract<mode><inner_l>): ... this.
15288 * config/sparc/sparc.md (veltmode): New mode iterator.
15289 (vec_init<VMALL:mode>): Renamed to ...
15290 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15291 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15292 (vec_initv2sisi): ... this.
15293 (vec_initv2sf): Renamed to ...
15294 (vec_initv2sfsf): ... this.
15295 (vec_extractv2sf): Renamed to ...
15296 (vec_extractv2sfsf): ... this.
15297 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15298 (vec_init<mode>): Renamed to ...
15299 (vec_init<mode><VEC_base_l>): ... this.
15300 (vec_extract<mode>): Renamed to ...
15301 (vec_extract<mode><VEC_base_l>): ... this.
15302 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15303 (vec_initv2sfsf): ... this.
15304 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15305 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15306 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15307 gen_vec_init* calls.
15309 2017-08-01 Richard Biener <rguenther@suse.de>
15311 PR tree-optimization/81297
15312 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15313 TREE_OVERFLOW from INTEGER_CSTs.
15315 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15317 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15319 2017-07-31 Carl Love <cel@us.ibm.com>
15321 * config/rs6000/rs6000-c: Add support for built-in functions
15322 vector signed char vec_xl_be (signed long long, signed char *);
15323 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15324 vector signed int vec_xl_be (signed long long, signed int *);
15325 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15326 vector signed long long vec_xl_be (signed long long, signed long long *);
15327 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15328 vector signed short vec_xl_be (signed long long, signed short *);
15329 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15330 vector double vec_xl_be (signed long long, double *);
15331 vector float vec_xl_be (signed long long, float *);
15332 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15333 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15334 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15336 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15337 (altivec_expand_builtin): Add switch statement to call
15338 altivec_expand_xl_be for each builtin.
15339 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15340 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15341 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15342 __builtin_vsx_le_be_v16qi.
15343 * doc/extend.texi: Update the built-in documentation file for the
15344 new built-in functions.
15346 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15349 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15351 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15353 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15355 * config.gcc: Add z14.
15356 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15357 CPU model numbers for z13s and z14.
15358 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
15360 * config/s390/s390-opts.h (enum processor_type): Rename
15361 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15362 * config/s390/s390.c (processor_table): Add field for CPU name to
15363 be passed to Binutils.
15364 (s390_asm_output_machine_for_arch): Use the new field in
15365 processor_table for Binutils.
15366 (s390_expand_builtin): Replace arch12 with z14.
15367 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15368 (s390_get_sched_attrmask): Likewise.
15369 (s390_get_unit_mask): Likewise.
15370 * config/s390/s390.opt: Add z14 to processor_type enum.
15372 2017-07-31 Martin Jambor <mjambor@suse.cz>
15375 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
15376 regardless of optimization level.
15378 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
15379 Martin Liska <mliska@suse.cz>
15381 * predict.def: Remove old comment and adjust probability.
15382 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15383 PREDICT statements.
15385 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15388 * config/i386/i386.c (ix86_function_naked): New function.
15389 (ix86_can_use_return_insn_p): Return false for naked functions.
15390 (ix86_expand_prologue): Skip prologue for naked functions.
15391 (ix86_expand_epilogue): Skip epilogue for naked functions
15392 and emit trap instruction.
15393 (ix86_warn_func_return): New function.
15394 (ix86_attribute_table): Add "naked" attribute specification.
15395 (TARGET_WARN_FUNC_RETURN): Define.
15396 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
15398 2017-07-31 Martin Liska <mliska@suse.cz>
15400 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
15401 (dump_gimple_bb_header): Always dump BB info.
15402 (pp_cfg_jump): Do not append info about BB when dumping a jump.
15404 2017-07-31 Martin Liska <mliska@suse.cz>
15407 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
15408 also with current_function_decl non-null equality.
15410 2017-07-31 Jakub Jelinek <jakub@redhat.com>
15413 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
15414 change type to the element type, instead add eltype variable and
15415 use it where we are interested in the element type.
15417 PR tree-optimization/81603
15418 * ipa-polymorphic-call.c
15419 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
15420 offset arithmetic in offset_int, bail out if the resulting bit offset
15421 doesn't fit into shwi.
15423 2017-07-31 Martin Liska <mliska@suse.cz>
15425 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
15426 (gimplify_save_expr): Fix comment.
15428 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
15431 * config/i386/i386.c (ix86_function_arg): Update arguments for
15433 (ix86_compute_frame_layout): Set the initial stack offset to
15434 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
15435 INCOMING_FRAME_SP_OFFSET.
15436 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
15437 stack before exception handler returns.
15438 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
15439 the 'ERROR_CODE' for exception handler.
15441 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
15443 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
15444 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
15445 (ASM_OUTPUT_REG_POP): Ditto.
15446 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
15447 instead of asm_fprintf to output pure string.
15449 2017-07-29 Jakub Jelinek <jakub@redhat.com>
15451 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
15452 to imported_module_or_decl hook.
15453 (debug_nothing_tree_tree_tree_bool): Remove.
15454 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
15455 * debug.c (do_nothing_debug_hooks): Use
15456 debug_nothing_tree_tree_tree_bool_bool instead of
15457 debug_nothing_tree_tree_tree_bool.
15458 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15459 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
15460 * sdbout.c (sdb_debug_hooks): Likewise.
15461 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
15462 (gen_namespace_die): Add DW_AT_export_symbols attribute if
15464 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
15465 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
15466 attribute, don't add anything.
15468 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15470 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
15471 (fold_build2_stat_loc): Likewise.
15472 (fold_build3_stat_loc): Likewise.
15473 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
15474 (fold_build1_loc): Remove macro.
15475 (fold_build2_loc): Likewise.
15476 (fold_build3_loc): Likewise.
15478 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15480 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
15481 (gimple_build_debug_bind_source_stat): Likewise.
15482 * gimple.h (gimple_build_debug_bind): Remove macro.
15483 (gimple_build_debug_bind_source): Likewise.
15485 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15487 * bitmap.c (bitmap_alloc): Adjust.
15488 (bitmap_gc_alloc): Likewise.
15489 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
15491 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15493 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
15494 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
15495 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
15496 (bitmap_gc_alloc_stat): Likewise.
15497 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
15499 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15501 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
15502 * rtl.h (shallow_copy_rtx): Remove macro.
15504 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15506 * emit-rtl.c (gen_raw_REG): Adjust.
15507 * gengenrtl.c (gendef): Likewise.
15508 * rtl.c (rtx_alloc_stat): Remove _stat from name.
15509 * rtl.h (rtx_alloc): Remove macro.
15511 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15513 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
15514 (build_tree_list_stat): Likewise.
15515 * tree.h (build_tree_list): Remove macro.
15516 (build_tree_list_vec): Likewise.
15518 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15520 * tree.c (make_vector_stat): Remove _stat from name.
15521 (build_vector_stat): Likewise.
15522 * tree.h (make_vector_stat): Remove macro.
15523 (build_vector_stat): Likewise.
15525 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15527 * tree.h (build_var_debug_value): Remove prototype.
15529 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15531 * tree.c (tree_cons_stat): Remove _stat from name.
15532 * tree.h (tree_cons): Remove macro.
15534 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15536 * tree.c (build_vl_exp_stat): Remove _stat from name.
15537 * tree.h (build_vl_exp): Remove macro.
15539 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15541 * tree.c (build_decl_stat): Remove _stat from name.
15542 * tree.h (build_decl): Remove macro.
15544 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15546 * gimple.c (gimple_build_with_ops_stat): Adjust.
15547 (gimple_alloc_stat): Remove _stat from name.
15548 * gimple.h (gimple_alloc): Remove macro.
15550 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15552 * tree.c (make_tree_vec_stat): Remove _stat from name.
15553 (grow_tree_vec_stat): Likewise.
15554 * tree.h (make_tree_vec_stat): Adjust prototype.
15555 (grow_tree_vec_stat): Likewise.
15556 (make_tree_vec): Remove macro.
15557 (grow_tree_vec): Likewise.
15559 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15561 * fold-const.c (fold_build1_stat_loc): Adjust.
15562 (fold_build2_stat_loc): Likewise.
15563 (fold_build3_stat_loc): Likewise.
15564 * tree.c (build0_stat): Remove _stat from name.
15565 (build1_stat): Likewise.
15566 (build2_stat): Likewise.
15567 (build3_stat): Likewise.
15568 (build4_stat): Likewise.
15569 (build5_stat): Likewise.
15570 * tree.h (build1_loc): Remove macro, and rename _stat function
15572 (build2_loc): Likewise.
15573 (build3_loc): Likewise.
15574 (build4_loc): Likewise.
15575 (build5_loc): Likewise.
15577 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15579 * tree.c (make_int_cst_stat): Remove _stat from name.
15580 * tree.h (make_int_cst_stat): Adjust prototype.
15581 (make_int_cst): Remove macro.
15583 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15585 * tree.c (make_tre_binfo_stat): Remove _stat from name.
15586 * tree.h (make_tree_binfo_stat): Adjust prototype.
15587 (make_tree_binfo): Remove.
15589 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15591 * tree.c (copy_node_stat): Rename to copy_node.
15592 (build_distinct_type_copy): Adjust.
15593 * tree.h (copy_node_stat): Adjust prototype.
15594 (copy_node): Remove macro.
15596 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15598 * tree.c (make_node_stat): rename to make_node.
15599 (build_tree_list_stat): Adjust.
15600 (build0_stat): Likewise.
15601 (build2_stat): Likewise.
15602 (build3_stat): Likewise.
15603 (build4_stat): Likewise.
15604 (build5_stat): Likewise.
15605 (build_decl_stat): Likewise.
15606 * tree.h (make_node_stat): Adjust prototype.
15607 (make_node): remove macro.
15609 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
15611 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
15612 (PPC_FEATURE2_SCV): Likewise.
15613 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
15615 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15617 * config/aarch64/aarch64.c
15618 (aarch64_internal_mov_immediate): Add new special pattern.
15619 * config/aarch64/aarch64.md (*movdi_aarch64):
15620 Add reg/32bit const mov case.
15622 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15623 Richard Sandiford <richard.sandiford@linaro.org>
15625 * config/aarch64/aarch64.md (mov<mode>): Generalize.
15626 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15627 Add integer and movi cases.
15628 (movi-split-hf-df-sf split, fp16): New.
15629 (enabled): Added TARGET_FP_F16INST.
15630 * config/aarch64/iterators.md (GPF_HF): New.
15631 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
15633 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15635 * config/aarch64/aarch64.c
15636 (aarch64_simd_container_mode): Add prototype.
15637 (aarch64_expand_mov_immediate): Add HI support.
15638 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
15639 (aarch64_can_const_movi_rtx_p): New.
15640 (aarch64_preferred_reload_class):
15641 Remove restrictions of using FP registers for certain SIMD operations.
15642 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
15643 (aarch64_valid_floating_const): Add integer move validation.
15644 (aarch64_simd_imm_scalar_p): Remove.
15645 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
15646 (aarch64_legitimate_constant_p): Expand list of supported cases.
15647 * config/aarch64/aarch64-protos.h
15648 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
15649 (aarch64_reinterpret_float_as_int): New.
15650 (aarch64_simd_imm_scalar_p): Remove.
15651 * config/aarch64/constraints.md (Uvi): New.
15652 (Dd): Split into Ds and new Dd.
15653 * config/aarch64/aarch64.md (*movsi_aarch64):
15655 (*movdi_aarch64): Add SIMD mov case.
15657 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15659 * tree-predcom.c: (struct chain): Handle store-store chain in which
15660 stores for elimination only store loop invariant values.
15661 (execute_pred_commoning_chain): Ditto.
15662 (prepare_initializers_chain_store_elim): Ditto.
15663 (prepare_finalizers): Ditto.
15664 (is_inv_store_elimination_chain): New function.
15665 (initialize_root_vars_store_elim_1): New function.
15667 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15669 * tree-predcom.c: Revise general description of the pass.
15670 (enum chain_type): New enum type for store elimination.
15671 (struct chain): New field supporting store elimination.
15672 (struct component): Ditto.
15673 (dump_chain): Dump store-stores chain.
15674 (release_chain): Release resources.
15675 (split_data_refs_to_components): Compute and create component
15676 contains only stores for elimination.
15677 (get_chain_last_ref_at): New function.
15678 (make_invariant_chain): Initialization.
15679 (make_rooted_chain): Specify chain type in parameter and record it.
15680 (add_looparound_copies): Skip for store-stores chain.
15681 (determine_roots_comp): Compute type of chain and pass it to
15683 (initialize_root_vars_store_elim_2): New function.
15684 (finalize_eliminated_stores): New function.
15685 (remove_stmt): Handle store for elimination.
15686 (execute_pred_commoning_chain): Execute predictive commoning on
15687 store-store chains.
15688 (determine_unroll_factor): Skip unroll for store-stores chain.
15689 (prepare_initializers_chain_store_elim): New function.
15690 (prepare_initializers_chain): Hanlde store-store chain.
15691 (prepare_finalizers_chain, prepare_finalizers): New function.
15692 (tree_predictive_commoning_loop): Return integer value indicating
15693 if loop is unrolled or lcssa form is corrupted.
15694 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
15696 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15698 * tree-predcom.c (initialize_root): Delete.
15699 (execute_pred_commoning_chain): Initialize root vars and replace
15700 reference of non-combined chain directly, rather than call above
15703 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15705 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
15706 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
15708 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15710 * tree-predcom.c (struct chain): New field init_seq.
15711 (release_chain): Release init_seq.
15712 (prepare_initializers_chain): Record intialization stmts in above
15714 (insert_init_seqs): New function.
15715 (tree_predictive_commoning_loop): Call insert_init_seqs.
15717 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15719 * tree-predcom.c (determine_roots_comp): Skip trivial components.
15721 2017-07-28 Richard Biener <rguenther@suse.de>
15723 * match.pd: Remove superfluous :c.
15724 * genmatch.c (simplify::id): Add member.
15725 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
15727 (current_id): New global.
15728 (dt_node::parent): Move from ...
15729 (dt_operand::parent): ... here. Add for_id member.
15730 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
15731 (decision_tree::find_node): Relax order requirement when
15732 merging DT_TRUE nodes to ones inbetween the current simplify
15733 and the one we try to merge with. Add diagnostic whenever
15734 we need to enforce pattern order by not merging.
15735 (decision_tree::insert): Set current_id.
15736 (decision_tree::print_node): Dump parent node and for_id.
15737 (parser::last_id): Add member.
15738 (parser::push_simplify): Assign unique id.
15739 (parser::parser): Initialize last_id.
15741 2017-07-28 Martin Liska <mliska@suse.cz>
15744 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
15745 gimple_build_debug_bind.
15747 2017-07-28 Richard Biener <rguenther@suse.de>
15749 PR tree-optimization/81502
15750 * match.pd: Add pattern combining BIT_INSERT_EXPR with
15752 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
15754 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
15755 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
15756 for BIT_FIELD_REF args.
15757 * fold-const.c (make_bit_field_ref): Likewise.
15758 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
15760 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15763 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
15764 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
15765 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
15766 Or it into SANITIZER_UNDEFINED.
15767 * ubsan.c: Include gimple-fold.h and varasm.h.
15768 (ubsan_expand_ptr_ifn): New function.
15769 (instrument_pointer_overflow): New function.
15770 (maybe_instrument_pointer_overflow): New function.
15771 (instrument_object_size): Formatting fix.
15772 (pass_ubsan::execute): Call instrument_pointer_overflow
15773 and maybe_instrument_pointer_overflow.
15774 * internal-fn.c (expand_UBSAN_PTR): New function.
15775 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
15776 * sanitizer.def (__ubsan_handle_pointer_overflow,
15777 __ubsan_handle_pointer_overflow_abort): New builtins.
15778 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
15779 * internal-fn.def (UBSAN_PTR): New internal function.
15780 * opts.c (sanitizer_opts): Add pointer-overflow.
15781 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
15782 * fold-const.c (build_range_check): Compute pointer range check in
15783 integral type if pointer arithmetics would be needed. Formatting
15786 2017-07-28 Martin Liska <mliska@suse.cz>
15789 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
15790 parameters that are of a variable-length.
15792 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
15794 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
15796 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
15797 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
15798 (CRT_CALL_STATIC_FUNCTION): Likewise.
15799 (ASM_DEFAULT_SPEC): New define.
15800 (ASM_SPEC32): Likewise.
15801 (ASM_SPEC64): Likewise.
15802 (ASM_SPEC_COMMON): Likewise.
15803 (ASM_SPEC): Likewise.
15804 (INVALID_64BIT): Likewise.
15805 (LINK_OS_DEFAULT_SPEC): Likewise.
15806 (LINK_OS_SPEC32): Likewise.
15807 (LINK_OS_SPEC64): Likewise.
15808 (POWERPC_LINUX): Likewise.
15809 (PTRDIFF_TYPE): Likewise.
15810 (RESTORE_FP_PREFIX): Likewise.
15811 (RESTORE_FP_SUFFIX): Likewise.
15812 (SAVE_FP_PREFIX): Likewise.
15813 (SAVE_FP_SUFFIX): Likewise.
15814 (SIZE_TYPE): Likewise.
15815 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
15816 (TARGET_64BIT): Likewise.
15817 (TARGET_64BIT): Likewise.
15818 (TARGET_AIX): Likewise.
15819 (WCHAR_TYPE_SIZE): Likewise.
15820 (WCHAR_TYPE): Undefine.
15821 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
15822 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
15823 (CPP_OS_RTEMS_SPEC): Delete.
15824 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
15825 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
15827 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
15829 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15831 PR tree-optimization/81578
15832 * tree-parloops.c (build_new_reduction): Bail out if
15833 reduction_code isn't one of the standard OpenMP reductions.
15834 Move the details printing after that decision.
15836 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15838 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
15839 related to reload_in_progress.
15840 (splat_input_operand): Likewise.
15841 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
15843 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
15845 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
15846 (TARGET_INSTANTIATE_DECLS): Likewise.
15847 (legitimate_indexed_address_p): Delete reload_in_progress code.
15848 (rs6000_debug_legitimate_address_p): Likewise.
15849 (rs6000_eliminate_indexed_memrefs): Likewise.
15850 (rs6000_emit_le_vsx_store): Likewise.
15851 (rs6000_emit_move_si_sf_subreg): Likewise.
15852 (rs6000_emit_move): Likewise.
15853 (register_to_reg_type): Likewise.
15854 (rs6000_pre_atomic_barrier): Likewise.
15855 (rs6000_machopic_legitimize_pic_address): Likewise.
15856 (rs6000_allocate_stack_temp): Likewise.
15857 (rs6000_address_for_fpconvert): Likewise.
15858 (rs6000_address_for_altivec): Likewise.
15859 (rs6000_secondary_memory_needed_rtx): Delete function.
15860 (rs6000_check_sdmode): Likewise.
15861 (rs6000_alloc_sdmode_stack_slot): Likewise.
15862 (rs6000_instantiate_decls): Likewise.
15863 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
15864 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
15865 Delete reload_in_progress.
15866 (*vec_reload_and_plus_<mptrsize>): Likewise.
15867 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
15868 (vsx_div_v2di): Likewise.
15869 (vsx_udiv_v2di): Likewise.
15871 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15873 * config/rs6000/rs6000.opt (mlra): Replace with stub.
15874 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
15875 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
15876 (rs6000_debug_reg_global): Delete print of LRA status.
15877 (rs6000_option_override_internal): Delete dead LRA related code.
15878 (rs6000_lra_p): Delete function.
15879 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
15881 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
15883 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
15884 * config/riscv/rtems.h: New file.
15886 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15887 Sudakshina Das <sudi.das@arm.com>
15889 * config/aarch64/aarch64.md
15890 (define_split for and<mode>3nr_compare): Move
15891 non aarch64_logical_operand to a register.
15892 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
15893 register immediate operand to a register.
15894 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
15896 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15898 PR middle-end/81564
15899 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
15901 2017-07-27 Richard Biener <rguenther@suse.de>
15903 PR tree-optimization/81573
15904 PR tree-optimization/81494
15905 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
15906 multi defuse cycle case.
15908 2017-07-27 Richard Biener <rguenther@suse.de>
15910 PR tree-optimization/81571
15911 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
15914 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
15916 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
15917 earlier and only if MASK_FPU is set. Adjust formatting.
15919 2017-07-27 Martin Liska <mliska@suse.cz>
15921 * opt-functions.awk: Add validation of value of Init.
15922 * optc-gen.awk: Pass new argument.
15924 2017-07-27 Martin Liska <mliska@suse.cz>
15926 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
15927 Fix wrong condition.
15929 2017-07-27 Martin Liska <mliska@suse.cz>
15931 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
15932 BBs and edges seen by autoFDO.
15934 2017-07-27 Richard Biener <rguenther@suse.de>
15936 PR tree-optimization/81502
15937 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
15938 with incompatible but same sized type.
15939 (execute_update_addresses_taken): Likewise.
15941 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
15943 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
15944 flag_tree_loop_vectorize rather than flag_tree_vectorize.
15946 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15949 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
15950 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
15951 Change s_operand to memory_operand.
15953 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
15955 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
15956 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
15957 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
15958 Emit instructions rather than returning an expression. Handle TFmode
15959 and KFmode by casting to TImode.
15960 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
15961 (rs6000_emit_le_vsx_store): Likewise.
15962 * config/rs6000/vsx.md (VSX_TI): New iterator.
15963 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
15964 (*vsx_le_undo_permute_<mode>): Likewise.
15965 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
15966 emit the split sequence.
15967 (*vsx_le_perm_store_<mode>): Likewise.
15969 2017-07-27 Jakub Jelinek <jakub@redhat.com>
15971 PR tree-optimization/81555
15972 PR tree-optimization/81556
15973 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
15974 if true, force CHANGED for the recursive invocation.
15975 (reassociate_bb): Remember original length of ops array, pass
15976 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
15978 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
15979 attributes for noipa attribute. For naked attribute use
15980 lookup_attribute first before lookup_attribute_spec.
15981 * final.c (rest_of_handle_final): Disable IPA RA for functions with
15983 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
15984 for functions with noipa attribute.
15985 (cgraph_externally_visible_p): Return true for functions with noipa
15987 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
15988 for functions with noipa attribute.
15989 * doc/extend.texi: Document noipa function attribute.
15990 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
15991 also for functions with noipa attribute.
15992 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
15994 2017-07-26 Andrew Pinski <apinski@cavium.com>
15996 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
15997 vec_unalign_load_cost and vec_unalign_store_cost.
15999 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
16001 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
16002 -mvsx-small-integer option.
16003 (ISA_3_0_MASKS_IEEE): Likewise.
16004 (OTHER_VSX_VECTOR_MASKS): Likewise.
16005 (POWERPC_MASKS): Likewise.
16006 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
16007 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
16008 code, only testing for DImode being allowed in non-VSX floating
16010 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
16011 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
16013 (rs6000_option_override_internal): Delete -mvsx-small-integer.
16014 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
16015 TARGET_P8_VECTOR test.
16016 (rs6000_secondary_reload_simple_move): Likewise.
16017 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16018 since TARGET_P9_VECTOR was already tested.
16019 (rs6000_opt_masks): Remove -mvsx-small-integer.
16020 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16021 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16023 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16024 test for TARGET_VEXTRACTUB was used, and that uses
16026 (p9 extract splitter): Likewise.
16027 (vsx_extract_<mode>_di_p9): Likewise.
16028 (vsx_extract_<mode>_store_p9): Likewise.
16029 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16030 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16031 the elimination of TARGET_VSX_SMALL_INTEGER.
16032 (vsx_extract_<mode>_p8): Likewise.
16033 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16034 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16035 (vsx_set_<mode>_p9): Likewise.
16036 (vsx_set_v4sf_p9): Likewise.
16037 (vsx_set_v4sf_p9_zero): Likewise.
16038 (vsx_insert_extract_v4sf_p9): Likewise.
16039 (vsx_insert_extract_v4sf_p9_2): Likewise.
16040 * config/rs6000/rs6000.md (sign extend splitter): Change
16041 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16042 (floatsi<mode>2_lfiwax_mem): Likewise.
16043 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16044 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16045 since a test for TARGET_P9_VECTOR was used.
16046 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16047 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16048 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16049 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16050 TARGET_P8_VECTOR test.
16051 (fix_trunc<mode>si2_stfiwx): Likewise.
16052 (fix_trunc<mode>si2_internal): Likewise.
16053 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16054 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16056 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16057 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16058 TARGET_P8_VECTOR test.
16059 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16060 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16061 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16063 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16064 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16065 since a test for TARGET_P9_VECTOR was used.
16066 (splitter for loading small constants): Likewise.
16068 2017-07-26 Andrew Pinski <apinski@cavium.com>
16070 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16073 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16076 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16077 (fp_valid_at): Likewise.
16079 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16081 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16082 (qdf24xx_addrcost_table): Likewise.
16083 (cortexa57_tunings): Update to use generic_branch_cost.
16084 (cortexa72_tunings): Likewise.
16085 (cortexa73_tunings): Likewise.
16086 (qdf24xx_tunings): Likewise.
16088 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16090 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16091 (thunderx2t99_branch_cost): Likewise.
16092 (cortexa35_tunings): Update to use generic_branch_cost.
16093 (cortexa53_tunings): Likewise.
16094 (cortexa57_tunings): Likewise.
16095 (cortexa72_tunings): Likewise.
16096 (cortexa73_tunings): Likewise.
16097 (thunderx2t99_tunings): Likewise.
16099 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16101 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16102 (sparc_option_override): Honour MASK_FSMULD.
16103 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16104 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16105 * config/sparc/sparc.opt (mfsmuld): New option.
16106 * doc/invoke.texi (mfsmuld): Document option.
16108 2017-07-26 Marek Polacek <polacek@redhat.com>
16110 PR middle-end/70992
16111 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16113 2017-07-26 Richard Biener <rguenther@suse.de>
16115 * gimple-match-head.c (do_valueize): Return OP if valueize
16117 (get_def): New helper to get at the def stmt of a SSA name
16118 if valueize allows.
16119 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16120 do_valueize to get at the def stmt.
16121 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16123 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16125 PR middle-end/46932
16126 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16128 2017-07-26 Martin Liska <mliska@suse.cz>
16131 * function.c (expand_function_start): Make expansion of
16132 nonlocal_goto_save_area after parm_birth_insn.
16134 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16136 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16137 from all CPU target flags enable members.
16139 2017-07-26 Richard Biener <rguenther@suse.de>
16141 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16142 (decision_tree::gen): Make 'type' const.
16143 (write_predicate): Likewise.
16145 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16147 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16148 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16149 (rs6000_option_override_internal): Likewise.
16150 (rs6000_expand_vector_set): Likewise.
16151 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16152 (TARGET_UPPER_REGS_SF): Likewise.
16153 (TARGET_UPPER_REGS_DI): Likewise.
16154 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16155 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16156 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16157 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16158 (Splitters for DI constants in Altivec registers): Likewise.
16159 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16160 (vsx_set_v4sf_p9): Likewise.
16161 (vsx_set_v4sf_p9_zero): Likewise.
16162 (vsx_insert_extract_v4sf_p9): Likewise.
16163 (vsx_insert_extract_v4sf_p9_2): Likewise.
16165 2017-07-25 Carl Love <cel@us.ibm.com>
16167 * doc/extend.texi: Update the built-in documentation file for the
16168 existing built-in functions
16169 vector signed char vec_cnttz (vector signed char);
16170 vector unsigned char vec_cnttz (vector unsigned char);
16171 vector signed short vec_cnttz (vector signed short);
16172 vector unsigned short vec_cnttz (vector unsigned short);
16173 vector signed int vec_cnttz (vector signed int);
16174 vector unsigned int vec_cnttz (vector unsigned int);
16175 vector signed long long vec_cnttz (vector signed long long);
16176 vector unsigned long long vec_cnttz (vector unsigned long long);
16178 2017-07-25 Andrew Pinski <apinski@cavium.com>
16180 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16181 accesses where the use is for the first operand of a BIT_INSERT.
16183 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16186 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16187 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16189 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16191 * config/i386/gstabs.h: Delete.
16192 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16194 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16196 * config/i386/i386.c (ix86_decompose_address): Do not check for
16197 register RTX when looking at index_reg or base_reg.
16198 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16200 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16202 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16203 to update EH info here.
16205 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16207 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16209 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16211 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16213 2017-07-25 Torsten Duwe <duwe@suse.de>
16215 * common.opt: Introduce -fpatchable-function-entry
16216 command line option, and its variables function_entry_patch_area_size
16217 and function_entry_patch_area_start.
16218 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16219 including a two-value parser.
16220 * target.def (print_patchable_function_entry): New target hook.
16221 * targhooks.h (default_print_patchable_function_entry): New function.
16222 * targhooks.c (default_print_patchable_function_entry): Likewise.
16223 * toplev.c (process_options): Switch off IPA-RA if
16224 patchable function entries are being generated.
16225 * varasm.c (assemble_start_function): Look at the
16226 patchable-function-entry command line switch and current
16227 function attributes and maybe generate NOP instructions by
16228 calling the print_patchable_function_entry hook.
16229 * doc/extend.texi: Document patchable_function_entry attribute.
16230 * doc/invoke.texi: Document -fpatchable_function_entry
16231 command line option.
16232 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16234 * doc/tm.texi: Re-generate.
16236 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16239 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16240 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16242 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16244 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16246 2017-07-25 Richard Biener <rguenther@suse.de>
16248 PR tree-optimization/81455
16249 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16250 not walk in cycles when looking for guards.
16252 2017-07-25 Richard Biener <rguenther@suse.de>
16254 PR tree-optimization/81529
16255 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16256 when optimizing backedge uses.
16258 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16260 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16262 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16263 to dl_section_ref. On AIX, append an expression to subtract
16264 the size of the section length to dl_section_ref.
16266 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16268 * configure.ac: If any of the config.* scripts fail, exit 1.
16269 * configure: Regenerate.
16271 2017-07-25 Richard Biener <rguenther@suse.de>
16273 PR middle-end/81546
16274 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16275 of immediate uses, be more verbose on errors.
16277 2017-07-25 Richard Biener <rguenther@suse.de>
16279 PR tree-optimization/81510
16280 * tree-vect-loop.c (vect_is_simple_reduction): When the
16281 reduction stmt is not inside the loop bail out.
16283 2017-07-25 Richard Biener <rguenther@suse.de>
16285 PR tree-optimization/81303
16286 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16287 profitability check against LOOP_VINFO_NITERSM1.
16289 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16291 * domwalk.c (cmp_bb_postorder): Simplify.
16292 (sort_bbs_postorder): New function. Use it...
16293 (dom_walker::walk): ...here to optimize common cases.
16295 2017-07-25 Martin Liska <mliska@suse.cz>
16298 * ipa-visibility.c (function_and_variable_visibility): Make the
16299 redirection just on target that supports aliasing.
16300 Fix GNU coding style.
16302 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16305 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16307 * config/arm/rtems.h: Likewise.
16308 * config/bfin/rtems.h: Likewise.
16309 * config/i386/rtemself.h: Likewise.
16310 * config/lm32/rtems.h: Likewise.
16311 * config/m32c/rtems.h: Likewise.
16312 * config/m68k/rtemself.h: Likewise.
16313 * config/microblaze/rtems.h: Likewise.
16314 * config/mips/rtems.h: Likewise.
16315 * config/moxie/rtems.h: Likewise.
16316 * config/nios2/rtems.h: Likewise.
16317 * config/powerpcspe/rtems.h: Likewise.
16318 * config/rs6000/rtems.h: Likewise.
16319 * config/rtems.h: Likewise.
16320 * config/sh/rtems.h: Likewise.
16321 * config/sh/rtemself.h: Likewise.
16322 * config/sparc/rtemself.h: Likewise.
16324 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16327 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16328 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16329 * tree-ssa-structalias.c (alias_get_name): Same.
16331 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16334 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16335 instructions if no du chain is found.
16337 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16339 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16341 2017-07-25 Richard Biener <rguenther@suse.de>
16343 PR middle-end/81505
16344 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16347 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16349 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16350 upper-regs options.
16351 (ISA_2_7_MASKS_SERVER): Likewise.
16352 (ISA_3_0_MASKS_IEEE): Likewise.
16353 (OTHER_P8_VECTOR_MASKS): Likewise.
16354 (OTHER_VSX_VECTOR_MASKS): Likewise.
16355 (POWERPC_MASKS): Likewise.
16356 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16357 duplicate list of options.
16358 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
16359 explicit -mupper-regs options.
16360 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
16361 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
16362 alias for -mupper-regs-df.
16363 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16364 (rs6000_init_hard_regno_mode_ok): Likewise.
16365 (rs6000_option_override_internal): Likewise.
16366 (rs6000_opt_masks): Likewise.
16367 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
16368 options in terms of whether -mvsx or -mpower8-vector was used.
16369 (TARGET_UPPER_REGS_DI): Likewise.
16370 (TARGET_UPPER_REGS_SF): Likewise.
16371 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
16372 -mupper-regs-* options.
16374 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16376 * passes.c (emergency_dump_function): Print some empty lines and a
16377 header before the RTL dump.
16379 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16381 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16383 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
16386 * config/aarch64/aarch64.c (aarch64_classify_symbol):
16387 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
16389 2017-07-24 Carl Love <cel@us.ibm.com>
16391 * config/rs6000/rs6000-c.c: Add support for built-in functions
16392 vector float vec_extract_fp32_from_shorth (vector unsigned short);
16393 vector float vec_extract_fp32_from_shortl (vector unsigned short);
16394 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
16395 vec_extract_fp_from_shortl): Add defines for the two builtins.
16396 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
16397 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
16399 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
16400 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
16401 * doc/extend.texi: Update the built-in documentation file for the
16402 new built-in function.
16404 2017-07-24 Jakub Jelinek <jakub@redhat.com>
16407 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
16409 * doc/generic.texi: Likewise.
16410 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16411 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16413 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
16415 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
16416 (aarch64_mls_elt_merge<mode>): Likewise.
16418 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
16420 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16421 having __cxa_atexit.
16423 2017-07-23 Michael Collison <michael.collison@arm.com>
16425 * config/arm/arm.c (arm_option_override): Deprecate
16426 use of -mstructure-size-boundary.
16427 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
16428 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
16430 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16433 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16434 Reduce cost estimate for direct moves.
16436 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
16439 * config/i386/i386.c (ix86_option_override_internal): Disable
16440 BMI, BMI2 and TBM instructions for -m16.
16442 2017-07-21 Carl Love <cel@us.ibm.com>
16444 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16445 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16446 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16447 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16448 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16449 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16450 VMULOSW): New enum "unspec" values.
16451 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16452 altivec_vmulosw): New patterns.
16453 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16454 VMULOSW): Add definitions.
16456 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
16458 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
16459 (qdf24xx): Likewise.
16460 * config/aarch64/aarch64-options-extensions.def (rdma); New.
16461 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
16462 (AARCH64_FL_V8_1): Renumber.
16463 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
16464 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
16465 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
16466 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
16467 rdma to feature modifiers list.
16469 2017-07-21 Yury Gribov <tetra2005@gmail.com>
16471 PR middle-end/56727
16472 * ipa-visibility (function_and_variable_visibility): Convert
16473 recursive PLT call to direct call if appropriate.
16475 2017-07-21 Andrew Pinski <apinski@cavium.com>
16477 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
16478 operand 1 to see if the types precision matches.
16479 * fold-const.c (operand_equal_p): Likewise.
16481 2017-07-21 Richard Biener <rguenther@suse.de>
16483 PR tree-optimization/81303
16484 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
16485 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
16486 (vect_peeling_hash_get_lowest_cost): Adjust.
16487 (vect_enhance_data_refs_alignment): Likewise. Use
16488 vect_get_peeling_costs_all_drs to compute the penalty for no
16489 peeling to match up costs.
16491 2017-07-21 Richard Biener <rguenther@suse.de>
16493 PR tree-optimization/81500
16494 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
16495 we didn't identify a reduction path.
16497 2017-07-21 Tom de Vries <tom@codesourcery.com>
16498 Cesar Philippidis <cesar@codesourcery.com>
16500 PR gcov-profile/81442
16501 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
16504 2017-07-21 Tom de Vries <tom@codesourcery.com>
16507 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
16509 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
16510 nvptx_override_options_after_change.
16512 2017-07-21 Ulrich Drepper <drepper@redhat.com>
16514 * dwarf2out.c (output_file_names): Avoid double testing for
16515 dwarf_version >= 5.
16517 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
16519 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
16521 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16523 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
16525 (try_crossjump_to_edge): Do not punt on partitioned functions.
16527 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16529 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
16530 Put all BBs reachable only via paths crossing cold region to cold
16532 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
16534 2016-07-21 Richard Biener <rguenther@suse.de>
16536 PR tree-optimization/81303
16537 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
16538 into account prologue and epilogue iterations when raising
16539 min_profitable_iters to sth at least covering one vector iteration.
16541 2017-07-21 Tamar Christina <tamar.christina@arm.com>
16543 * config/arm/arm.c (arm_test_cpu_arch_dat):
16546 2017-07-20 Nathan Sidwell <nathan@acm.org>
16548 Remove TYPE_METHODS.
16549 * tree.h (TYPE_METHODS): Delete.
16550 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
16551 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
16552 (dbxout_type_methods): Scan TYPE_FIELDS.
16553 (dbxout_type): Don't check TYPE_METHODS here.
16554 * function.c (use_register_for_decl): Always ignore register for
16555 class types when not optimizing.
16556 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
16557 * tree.c (free_lang_data_in_type): Stitch out member functions and
16558 templates from TYPE_FIELDS.
16559 (build_distinct_type_copy, verify_type_variant,
16560 verify_type): Member fns are on TYPE_FIELDS.
16561 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
16562 * tree-pretty-print.c (dump_generic_node): Likewise.
16564 2017-07-20 Jakub Jelinek <jakub@redhat.com>
16567 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
16568 V2TImode and V4TImode.
16569 (ix86_expand_vector_extract): Likewise.
16570 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
16571 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
16572 (ssescalarmode): Handle V4TImode and V2TImode.
16573 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
16574 (*vec_extractv2ti, *vec_extractv4ti): New insns.
16575 (VEXTRACTI128_MODE): New mode iterator.
16576 (splitter for *vec_extractv?ti first element): New.
16577 (VEC_INIT_MODE): New mode iterator.
16578 (vec_init<mode>): Consolidate 3 expanders into one using
16579 VEC_INIT_MODE mode iterator.
16581 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16583 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
16584 non_spilled_static_chain_regno_p.
16586 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16588 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
16590 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
16592 * bb-reorder.c (connect_traces): Allow copying of blocks within
16595 2017-07-20 Richard Biener <rguenther@suse.de>
16597 * gimple.h (gimple_phi_result): Add gphi * overload.
16598 (gimple_phi_result_ptr): Likewise.
16599 (gimple_phi_arg): Likewise. Adjust index assert to only
16600 allow actual argument accesses rather than all slots available
16602 (gimple_phi_arg_def): Add gphi * overload.
16603 * tree-phinodes.c (make_phi_node): Initialize only actual
16605 (resize_phi_node): Clear memory not covered by old node,
16606 do not initialize excess argument slots.
16607 (reserve_phi_args_for_new_edge): Initialize new argument slot
16610 2017-07-20 Bin Cheng <bin.cheng@arm.com>
16612 PR tree-optimization/81388
16614 2016-07-21 Bin Cheng <bin.cheng@arm.com>
16616 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
16617 by removing computation of may_be_zero.
16619 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16620 Tom de Vries <tom@codesourcery.com>
16622 PR middle-end/81030
16623 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
16624 when gimple level profile disagrees with what RTL expander did.
16626 2017-07-20 Richard Biener <rguenther@suse.de>
16628 PR tree-optimization/61171
16629 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
16630 (vect_analyze_stmt): Add slp instance parameter.
16631 (vectorizable_reduction): Likewise.
16632 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
16633 (vect_is_simple_reduction): Deal with chains not detected
16634 as SLP reduction chain, specifically not properly associated
16635 chains containing a mix of plus/minus.
16636 (get_reduction_op): Remove.
16637 (get_initial_defs_for_reduction): Simplify, pass in whether
16638 this is a reduction chain, pass in the SLP node for the PHIs.
16639 (vect_create_epilog_for_reduction): Get the SLP instance as
16641 (vectorizable_reduction): Get the SLP instance as arg.
16642 During analysis remember the SLP node with the PHIs in the
16643 instance. Simplify getting at the vectorized reduction PHIs.
16644 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
16645 through SLP instance.
16646 (vect_slp_analyze_operations): Likewise.
16647 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
16648 (vect_transform_stmt): Likewise.
16650 2017-07-20 Tom de Vries <tom@codesourcery.com>
16652 PR tree-optimization/81489
16653 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
16654 read of phi arg location to before loop that modifies phi.
16656 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16658 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
16661 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16663 PR middle-end/81331
16664 * except.c (execute): Fix ordering issue.
16666 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16668 PR rtl-optimization/81423
16669 * combine.c (make_compound_operation_int): Don't try to optimize
16670 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
16672 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16674 PR rtl-optimization/81423
16675 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
16676 with a constant that is -1 in the truncated to mode.
16678 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16680 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
16681 (determine_unlikely_bbs): ... here.
16682 * predict.h (propagate_unlikely_bbs_forward): Declare.
16683 * cfgexpand.c (pass_expand::execute): Use it.
16684 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
16686 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
16687 propagate_unlikely_bbs_forward.
16689 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16691 PR middle-end/81331
16692 * except.c (maybe_add_nop_after_section_switch): New function.
16695 2017-07-19 Tom de Vries <tom@codesourcery.com>
16697 * gimple.h (gimple_phi_set_arg): Make assert more strict.
16699 2017-07-19 Tom de Vries <tom@codesourcery.com>
16701 * gimple.h (gimple_phi_arg): Make assert more strict.
16703 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
16705 * config.gcc (powerpc*-*-*): Add mmintrin.h.
16706 * config/rs6000/mmintrin.h: New file.
16707 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
16709 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16711 PR tree-optimization/81346
16712 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
16714 2017-07-19 Tom de Vries <tom@codesourcery.com>
16716 * config/nvptx/nvptx.md (VECIM): Add V2DI.
16718 2017-07-19 Tom de Vries <tom@codesourcery.com>
16720 * config/nvptx/nvptx-modes.def: Add V2DImode.
16721 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
16722 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
16723 (nvptx_output_mov_insn): Handle lack of mov.b128.
16724 (nvptx_print_operand): Handle 'H' and 'L' codes.
16725 (nvptx_vector_mode_supported): Allow V2DImode.
16726 (nvptx_preferred_simd_mode): New function.
16727 (nvptx_data_alignment): New function.
16728 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
16729 nvptx_preferred_simd_mode.
16730 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
16732 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
16734 2017-07-19 Tom de Vries <tom@codesourcery.com>
16736 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
16737 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
16738 (nvptx_vector_mode_supported): New function. Allow V2SImode.
16739 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
16740 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
16741 (mov<VECIM>_insn): New define_insn.
16742 (define_expand "mov<VECIM>): New define_expand.
16744 2017-07-19 Tom de Vries <tom@codesourcery.com>
16746 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
16748 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16750 PR tree-optimization/81346
16751 * fold-const.h (fold_div_compare, range_check_type): Declare.
16752 * fold-const.c (range_check_type): New function.
16753 (build_range_check): Use range_check_type.
16754 (fold_div_compare): No longer static, rewritten into
16755 a match.pd helper function.
16756 (fold_comparison): Don't call fold_div_compare here.
16757 * match.pd (X / C1 op C2): New optimization using fold_div_compare
16758 as helper function.
16760 2017-07-19 Nathan Sidwell <nathan@acm.org>
16762 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
16763 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
16764 * tree.c (find_decls_types_r, verify_type): Use
16765 TYPE_{MIN,MAX}_VALUE_RAW.
16766 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
16767 (hash_tree): Likewise.
16768 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16770 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16773 2017-07-18 Tom de Vries <tom@codesourcery.com>
16775 PR middle-end/81464
16776 * omp-expand.c (expand_omp_for_static_chunk): Handle
16777 equal-argument loop exit phi.
16779 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
16782 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
16783 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
16784 operand 2 predicate.
16785 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
16786 operand 2 predicate.
16787 (ror,rol -> rorx splitters): Use const_int_operand as
16788 operand 2 predicate.
16790 2017-06-18 Richard Biener <rguenther@suse.de>
16792 PR tree-optimization/81410
16793 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
16794 the gap in the ! slp_perm SLP case after each group.
16796 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16798 PR middle-end/81463
16799 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
16802 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16804 PR middle-end/81462
16805 * predict.c (set_even_probabilities): Cleanup; do not affect
16806 probabilities that are already known.
16807 (combine_predictions_for_bb): Call even when count is set.
16809 2017-07-18 Nathan Sidwell <nathan@acm.org>
16811 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
16814 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16817 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
16818 optimization for loop niter analysis.
16820 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
16823 * config/avr/avr.c (avr_optimize_casesi): Don't use
16824 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
16826 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
16828 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
16829 body_cost_vec from _vect_peel_extended_info.
16830 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
16831 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
16834 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16836 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
16838 2017-07-18 Richard Biener <rguenther@suse.de>
16840 PR tree-optimization/80620
16841 PR tree-optimization/81403
16842 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
16843 info when re-using a VN table entry.
16845 2017-07-18 Richard Biener <rguenther@suse.de>
16847 PR tree-optimization/81418
16848 * tree-vect-loop.c (vectorizable_reduction): Properly compute
16849 vectype_in. Verify that with lane-reducing reduction operations
16850 we have a single def-use cycle.
16852 2017-07-17 Carl Love <cel@us.ibm.com>
16854 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
16856 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16857 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16858 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16859 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16860 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16861 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16862 VMULOSW): New enum "unspec" values.
16863 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16864 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16865 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16866 altivec_vmulosw): New patterns.
16867 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16868 VMULOSW): Add definitions.
16869 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
16871 * config/alpha/alpha.c: Include predict.h.
16873 2017-07-17 Yury Gribov <tetra2005@gmail.com>
16875 * tree-vrp.c (compare_assert_loc): Fix comparison function
16876 to return predictable results.
16878 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16880 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
16882 (subdi3): Likewise.
16883 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
16884 * doc/invoke.texi (mexpand-adddi): Update text.
16886 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16888 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
16889 that also clobbers the CC register. The old expand code is moved
16891 (*arc_clzsi2): ... here.
16892 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
16893 the CC register. The old expand code is moved to ...
16894 (arc_ctzsi2): ... here.
16896 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16898 * config/arc/arc.opt (mindexed-loads): Use initial value
16899 TARGET_INDEXED_LOADS_DEFAULT.
16900 (mauto-modify-reg): Use initial value
16901 TARGET_AUTO_MODIFY_REG_DEFAULT.
16902 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16903 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16904 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16905 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16907 2017-07-17 Martin Liska <mliska@suse.cz>
16910 * opts.c (finish_options): Do not allow -fgnu-tm
16911 w/ -fsanitize={kernel-,}address. Say sorry.
16913 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16916 * tree-loop-distribution.c (classify_partition): Only assert on
16917 numer of iterations.
16918 (merge_dep_scc_partitions): Delete prameter. Update function call.
16919 (distribute_loop): Remove code handling loop with unknown niters.
16920 (pass_loop_distribution::execute): Skip loop with unknown niters.
16922 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16925 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
16926 function sort_partitions_by_post_order.
16928 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16930 PR tree-optimization/81374
16931 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
16932 the max index of basic blocks, rather than number of basic blocks.
16934 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16936 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
16938 (arc_legitimate_pic_operand_p): Likewise.
16939 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
16941 (arc_needs_pcl_p): Likewise.
16942 (arc_legitimate_pc_offset_p): Likewise.
16943 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
16944 function is also used in constrains.md.
16945 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
16946 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
16947 PLUS. Only return true/false in known cases, otherwise assert.
16948 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
16949 is already called in arc_legitimate_constant_p.
16950 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
16952 (LEGITIMATE_PIC_OPERAND_P): Use
16953 arc_raw_symbolic_reference_mentioned_p function.
16954 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
16959 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16960 Andrew Burgess <andrew.burgess@embecosm.com>
16962 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
16963 (arc_return_address_register): New function.
16964 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
16965 (arc_handle_fndecl_attribute): Add naked attribute.
16966 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16967 (TARGET_WARN_FUNC_RETURN): Likewise.
16968 (arc_allocate_stack_slots_for_args): New function.
16969 (arc_warn_func_return): Likewise.
16970 (machine_function): Change type fn_type.
16971 (arc_compute_function_type): Consider new naked function type,
16972 change function return type.
16973 (arc_must_save_register): Adapt to handle new
16974 arc_compute_function_type's return type.
16975 (arc_expand_prologue): Likewise.
16976 (arc_expand_epilogue): Likewise.
16977 (arc_return_address_regs): Delete.
16978 (arc_return_address_register): New function.
16979 (arc_epilogue_uses): Use above function.
16980 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
16981 (arc_function_type): Change encoding, add naked type.
16982 (ARC_INTERRUPT_P): Change to handle the new encoding.
16983 (ARC_FAST_INTERRUPT_P): Likewise.
16984 (ARC_NORMAL_P): Define.
16985 (ARC_NAKED_P): Likewise.
16986 (arc_compute_function_type): Delete prototype.
16987 * config/arc/arc.md (in_ret_delay_slot): Use
16988 arc_return_address_register function.
16989 (simple_return): Likewise.
16990 (p_return_i): Likewise.
16992 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16994 PR tree-optimization/81428
16995 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
16996 can't be built for those types.
16998 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17000 Remove stuff dead since r239246.
17002 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
17003 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
17004 (avr_inform_devices): Remove dead stuff.
17006 2017-07-17 Tamar Christina <tamar.christina@arm.com>
17008 * config/arm/arm_neon.h: Fix softp typo.
17010 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17012 PR tree-optimization/81365
17013 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
17014 aggregate moves onto bb predecessor edges, make sure there are no
17015 loads that could alias the lhs in between the start of bb and the
17018 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17021 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17022 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17023 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17025 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17027 PR tree-optimization/81396
17028 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17029 (init_symbolic_number): Initialize it to 1.
17030 (perform_symbolic_merge): Add n_ops from both operands into the new
17032 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17033 without base_addr as useless if they need more than one operation.
17034 (bswap_replace): Handle !bswap case for NULL base_addr.
17036 2017-07-17 Tom de Vries <tom@codesourcery.com>
17039 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17042 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17044 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17045 conditional builtin define __FIX_LEON3FT_B2BST.
17047 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17049 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17050 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17053 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17055 PR rtl-optimization/81424
17056 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17057 to remove potential trapping from operands if -fnon-call-exceptions.
17059 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17061 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17062 profile_proability for scalling.
17063 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17065 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17067 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17069 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17071 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17072 fixpoint arithmetics.
17074 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17076 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17077 fixpoint arithmetics.
17079 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17081 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17082 fixpoint arithmetics.
17084 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17086 * profile-count.h (profile_probability::from_reg_br_prob_note,
17087 profile_probability::to_reg_br_prob_note): New functions.
17088 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17089 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17090 * predict.c (probability_reliable_p): Update.
17091 (edge_probability_reliable_p): Update.
17092 (br_prob_note_reliable_p): Update.
17093 (invert_br_probabilities): Update.
17094 (add_reg_br_prob_note): New function.
17095 (combine_predictions_for_insn): Update.
17096 * asan.c (asan_clear_shadow): Update.
17097 * cfgbuild.c (compute_outgoing_frequencies): Update.
17098 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17099 (update_br_prob_note): Update.
17100 (rtl_verify_edges): Update.
17101 (purge_dead_edges): Update.
17102 (fixup_reorder_chain): Update.
17103 * emit-rtl.c (try_split): Update.
17104 * ifcvt.c (cond_exec_process_insns): Update.
17105 (cond_exec_process_if_block): Update.
17106 (dead_or_predicable): Update.
17107 * internal-fn.c (expand_addsub_overflow): Update.
17108 (expand_neg_overflow): Update.
17109 (expand_mul_overflow): Update.
17110 * loop-doloop.c (doloop_modify): Update.
17111 * loop-unroll.c (compare_and_jump_seq): Update.
17112 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17113 * predict.h: Update.
17114 * reorg.c (mostly_true_jump): Update.
17116 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17117 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17118 * config/arc/arc.c: (emit_unlikely_jump): Update.
17119 * config/arm/arm.c: (emit_unlikely_jump): Update.
17120 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17121 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17122 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17123 (ix86_print_operand): Update.
17124 (ix86_split_fp_branch): Update.
17125 (predict_jump): Update.
17126 * config/ia64/ia64.c (ia64_print_operand): Update.
17127 * config/mmix/mmix.c (mmix_print_operand): Update.
17128 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17129 (rs6000_expand_split_stack_prologue): Update.
17130 * config/rs6000/rs6000.c: Update.
17131 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17132 (s390_expand_vec_movstr): Update.
17133 (s390_expand_cs_tdsi): Update.
17134 (s390_expand_split_stack_prologue): Update.
17135 * config/sh/sh.c (sh_print_operand): Update.
17136 (expand_cbranchsi4): Update.
17137 (expand_cbranchdi4): Update.
17138 * config/sparc/sparc.c (output_v9branch): Update.
17139 * config/spu/spu.c (get_branch_target): Update.
17140 (ea_load_store_inline): Update.
17141 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17142 * config/tilepro/tilepro.c: Update.
17144 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17146 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17147 (gimplify_save_expr): Likewise.
17149 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17151 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17153 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17155 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17157 * timevar.def (TV_IPA_FNSUMMARY): Define.
17159 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17161 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17162 to back store errata sensitive sequence from being generated.
17163 (sqrtdf2_fix): Likewise.
17165 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17167 * tree-ssa-threadupdate.c (compute_path_counts,
17168 update_joiner_offpath_counts): Use profile_probability.
17170 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17173 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17175 * config/arm/arm-c.c (arm_cpu_builtins): Define
17176 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17178 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17180 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17181 array entries to represent __ieee128 versions of the
17182 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17183 scalar_extract_sig, and scalar_insert_exp built-in functions.
17184 (altivec_resolve_overloaded_builtin): Add special case handling
17185 for the __builtin_scalar_insert_exp function, as represented by
17186 the P9V_BUILTIN_VEC_VSIEDP constant.
17187 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17188 exponent support for __ieee128 argument.
17189 (VSESQP): Add scalar extract signature support for __ieee128
17191 (VSTDCNQP): Add scalar test negative support for __ieee128
17193 (VSIEQP): Add scalar insert exponent support for __int128 argument
17194 with __ieee128 result.
17195 (VSIEQPF): Add scalar insert exponent support for __ieee128
17196 argument with __ieee128 result.
17197 (VSTDCQP): Add scalar test data class support for __ieee128
17199 (VSTDCNQP): Add overload support for scalar test negative with
17200 __ieee128 argument.
17201 (VSTDCQP): Add overload support for scalar test data class
17202 __ieee128 argument.
17203 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17204 UNSPEC_VSX_SXSIGDP.
17205 (UNSPEC_VSX_SIEXPQP): New constant.
17206 (xsxexpqp): New insn for VSX scalar extract exponent quad
17208 (xsxsigqp): New insn for VSX scalar extract significand quad
17210 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17211 precision with floating point argument.
17212 (xststdcqp): New expand for VSX scalar test data class quad
17214 (xststdcnegqp): New expand for VSX scalar test negative quad
17216 (xststdcqp): New insn to match expansions for VSX scalar test data
17217 class quad precision and VSX scalar test negative quad precision.
17218 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17219 special case operand checking to enforce that second operand of
17220 VSX scalar test data class with quad precision argument is a 7-bit
17222 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17223 prototypes and descriptions of __ieee128 versions of
17224 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17225 scalar_test_data_class, and scalar_test_neg built-in functions.
17227 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17229 PR tree-optimization/81162
17230 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17231 replace a negate with an add.
17233 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17235 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17237 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17239 * config/arm/arm-c.c (arm_cpu_builtins): Define
17240 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17242 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17244 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17245 (armv8-r): Set ARM Cortex-R52 as default CPU.
17246 * config/arm/arm-tables.opt: Regenerate.
17247 * config/arm/arm-tune.md: Regenerate.
17248 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17250 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17251 extension for -mcpu=cortex-r52.
17253 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17255 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17256 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17257 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17258 (fp-armv8): Define it as FP_ARMv8 only.
17259 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17260 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17262 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17263 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17264 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17265 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17266 than TARGET_FPU_ARMV8.
17267 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17268 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17269 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17271 * config/arm/neon.md (neon_vrint): Likewise.
17272 (neon_vcvt): Likewise.
17273 (neon_<fmaxmin_op><mode>): Likewise.
17274 (<fmaxmin><mode>3): Likewise.
17275 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17276 * config/arm/predicates.md (arm_cond_move_operator): Check against
17277 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17279 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17281 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17282 to top of function.
17284 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17286 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17287 loop in comment with memset.
17289 2017-07-14 Martin Liska <mliska@suse.cz>
17291 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17292 * dwarf2out.c (is_java): Remove the function.
17293 (output_pubname): Remove usage of the function.
17294 (lower_bound_default): Remove usage of DW_LANG_Java.
17295 (gen_compile_unit_die): Likewise.
17296 * gcc.c: Remove compiler defaults for .java and .zip files.
17297 * gimple-expr.c (remove_suffix): Change as there's no longer
17298 extension than 4-letter one.
17299 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17300 (gimplify_save_expr): Likewise.
17301 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17302 as it's possible even for other languages than Java.
17303 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17304 * lto-opts.c (lto_write_options): Remove reference to Java.
17305 * opts.c (strip_off_ending): Update file extension handling.
17306 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17307 * tree-eh.c (lower_resx): Likewise.
17308 * tree.c (free_lang_data_in_type): Remove dead code.
17309 (find_decls_types_r): Likewise.
17310 (build_common_builtin_nodes): Remove Java from a comment.
17311 (verify_type): Remove dead code.
17312 * varasm.c (assemble_external): Remove Java from a comment.
17314 2017-07-14 Martin Liska <mliska@suse.cz>
17316 * opts.c (finish_options): Add quotes.
17317 (common_handle_option): Likewise.
17319 2017-07-14 Martin Liska <mliska@suse.cz>
17321 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17322 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17323 Remove N_SO_PASCAL.
17324 * dwarf2out.c (lower_bound_default): Do not handle
17326 (gen_compile_unit_die): Likewise.
17327 * gcc.c: Remove default extension binding for GNU Pascal.
17328 * stmt.c: Remove Pascal language from a comment.
17329 * xcoffout.c: Likewise.
17331 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17334 * diagnostic-show-locus.c (fixit_cmp): New function.
17335 (layout::layout): Sort m_fixit_hints.
17336 (column_range::column_range): Assert that the values are valid.
17337 (struct char_span): New struct.
17338 (correction::overwrite): New method.
17339 (struct source_line): New struct.
17340 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17341 calls in terms of classes source_line and char_span, and
17342 correction::overwrite.
17343 (selftest::test_overlapped_fixit_printing_2): New function.
17344 (selftest::diagnostic_show_locus_c_tests): Call it.
17346 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17348 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17349 early if there is no lhs.
17351 2017-07-13 Martin Liska <mliska@suse.cz>
17353 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17354 (gen_reference_type_die): Likewise.
17355 * stor-layout.c: Remove Pascal-related comment.
17357 2017-07-13 Martin Liska <mliska@suse.cz>
17359 * opts.c (finish_options): Add quotes to error messages.
17360 (parse_sanitizer_options): Likewise.
17362 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
17364 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
17366 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
17368 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
17370 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17372 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
17374 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
17376 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17379 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17380 provides the hardware capability bits, define the macro
17381 __BUILTIN_CPU_SUPPORTS__.
17382 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17383 if GLIBC does not provide the hardware capability bits. Add a
17384 gcc_unreachable call if the built-in cpu function is neither
17385 __builtin_cpu_is nor __builtin_cpu_supports.
17386 (rs6000_get_function_versions_dispatcher): Change the warning
17387 that an old GLIBC is used which does not export the capability
17388 bits to be an error.
17389 * doc/extend.texi (target_clones attribute): Document the
17390 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
17391 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
17392 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
17393 the macros defined by GCC if the newer GLIBC is available.
17395 2017-07-12 Jeff Law <law@redhat.com>
17397 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
17398 remaining includes slightly.
17399 * config/riscv/riscv-builtins.c: Include profile-count.h.
17401 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17404 * config/avr/avr.c (avr_set_current_function): In diagnostic
17405 messages: Quote keywords and (parts of) identifiers.
17406 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
17409 2017-07-12 Carl Love <cel@us.ibm.com>
17411 * config/rs6000/rs6000-c.c: Add support for built-in functions
17412 vector bool char vec_revb (vector bool char);
17413 vector bool short vec_revb (vector short char);
17414 vector bool int vec_revb (vector bool int);
17415 vector bool long long vec_revb (vector bool long long);
17416 * doc/extend.texi: Update the built-in documentation file for the
17417 new built-in functions.
17419 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17421 * config/s390/s390.md: Remove movcc splitter.
17423 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17425 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
17426 load/store on condition.
17428 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17431 * config/avr/avr.c (avr_encode_section_info)
17432 [progmem && !TREE_READONLY]: Error if progmem object needs
17435 2017-07-11 Michael Collison <michael.collison@arm.com>
17437 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
17440 2017-07-11 Carl Love <cel@us.ibm.com>
17442 * config/rs6000/rs6000-c.c: Add support for builtins
17443 vector unsigned int vec_parity_lsbb (vector signed int);
17444 vector unsigned int vec_parity_lsbb (vector unsigned int);
17445 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
17446 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
17447 vector unsigned long long vec_parity_lsbb (vector signed long long);
17448 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
17449 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
17450 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
17451 * doc/extend.texi: Update the built-in documentation file for the
17452 new built-in functions.
17454 2017-07-11 David Malcolm <dmalcolm@redhat.com>
17456 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
17457 (layout::m_primary_loc): New field.
17458 (layout::layout): Initialize new field. Move location filtering
17459 logic from here to...
17460 (layout::maybe_add_location_range): ...this new method. Add
17461 support for filtering to just the lines already specified by other
17463 (layout::will_show_line_p): New method.
17464 (gcc_rich_location::add_location_if_nearby): New method.
17465 (selftest::test_add_location_if_nearby): New test function.
17466 (selftest::diagnostic_show_locus_c_tests): Call it.
17467 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
17470 2017-07-11 Tom de Vries <tom@codesourcery.com>
17472 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
17473 (bb_first_real_insn): New function.
17474 (nvptx_single): Add extra initialization of broadcasted condition
17477 2017-07-11 Nathan Sidwell <nathan@acm.org>
17479 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
17481 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
17483 * doc/extend.texi (AVR Function Attributes): Remove weblink to
17484 Binutils doc as TEXI will mess them up.
17485 * doc/invoke.texi (AVR Options): Same here.
17487 2017-07-11 Daniel Cederman <cederman@gaisler.com>
17489 * config/sparc/sparc.opt (mfix-ut700): New option.
17490 (mfix-gr712rc): Likewise.
17491 (sparc_fix_b2bst): New variable.
17492 * doc/invoke.texi (SPARC options): Document them.
17493 (ARM options): Fix warnings.
17494 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
17495 instructions to prevent sequences that can trigger the store-store
17496 errata for certain LEON3FT processors.
17497 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
17498 (sparc_option_override): Set sparc_fix_b2bst appropriately.
17499 * config/sparc/sparc.md (fix_b2bst): New attribute.
17500 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
17502 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
17505 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
17507 (*rsqrtsf2_sse): Ditto.
17509 (div<mode>3): Macroize insn from divdf3 and divsf3
17510 using MODEF mode iterator.
17512 2017-07-10 Martin Sebor <msebor@redhat.com>
17514 PR tree-optimization/80397
17515 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
17516 instead of testing for equality to INTEGER_TYPE.
17518 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
17520 * config.gcc: Remove uclibc from arc target spec.
17522 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
17524 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
17526 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17529 * lto-wrapper.c (remove_option): New function.
17530 (merge_and_complain): Merge PIC/PIE options more realistically.
17532 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17534 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
17538 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
17539 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
17540 * config.in: Regenerate.
17541 * configure: Regenerate.
17542 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
17543 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
17544 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
17545 (TARGET_GASISR_PROLOGUES): ...target mask.
17546 * common/config/avr/avr-common.c
17547 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
17548 Set -mgas-isr-prologues.
17549 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
17550 INSERT_PASS_BEFORE for it.
17551 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
17552 * config/avr/avr.c (avr_option_override)
17553 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
17554 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
17555 (avr_attribute_table) <no_gccisr>: Add new function attribute.
17556 (avr_set_current_function) <is_no_gccisr>: Init machine field.
17557 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
17559 (make_avr_pass_pre_proep): New function.
17560 (emit_push_sfr) <treg>: Add argument to function and use it
17561 instead of TMP_REG.
17562 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
17563 and set machine->gasisr.yes.
17564 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
17565 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
17566 __gcc_isr.n_pushed to .L__stack_usage.
17567 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
17568 (avr_asm_final_postscan_insn): ...this new static function.
17569 * config/avr/avr.h (machine_function)
17570 <is_no_gccisr, use_L__stack_usage>: New fields.
17571 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
17572 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
17573 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
17574 (gasisr, *gasisr): New expander and insn.
17575 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17576 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
17577 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
17579 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
17581 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
17584 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17586 Move jump-tables out of .text again.
17589 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
17590 (ASM_OUTPUT_ADDR_VEC): New function.
17591 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
17592 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
17593 INSN_ADDRESSes as asm comment.
17594 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
17595 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
17596 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
17597 * config/avr/avr.md (*tablejump): Adjust comment.
17598 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
17599 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
17601 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
17602 (avr_output_addr_vec): New proto.
17603 (avr_log_t) <insn_addresses>: New field.
17605 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
17608 * config/i386/i386.c (ix86_function_arg_advance): Set
17609 outgoing_args_on_stack to true if there are outgoing arguments
17611 (ix86_function_arg): Likewise.
17612 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
17613 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
17614 * config/i386/i386.h (machine_function): Add
17615 outgoing_args_on_stack.
17617 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
17619 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17620 supporting pthreds.
17621 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
17623 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
17625 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
17626 (REAL_H): Remove $(MACHMODE_H).
17627 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
17629 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
17630 $(MACHMODE_H) and double-int.h.
17631 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
17633 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
17636 2017-07-07 Andrew Pinski <apinski@cavium.com>
17638 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
17639 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
17641 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17643 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
17644 Add warning if GCC was not configured to link against a GLIBC that
17645 exports the hardware capability bits.
17646 (make_resolver_func): Make resolver function private and not a
17647 COMDAT function. Create the name with clone_function_name instead
17648 of make_unique_name.
17651 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
17652 correct operand in doing the split.
17654 2017-07-07 Carl Love <cel@us.ibm.com>
17656 * config/rs6000/rs6000-c: Add support for built-in function
17657 vector unsigned short vec_pack_to_short_fp32 (vector float,
17659 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
17660 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
17661 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
17662 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
17663 (convert_4f32_8i16): Add define_expand.
17664 * doc/extend.texi: Update the built-in documentation file for the
17665 new built-in function.
17667 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17669 * config/sparc/m8.md: New file.
17670 * config/sparc/sparc.md: Include m8.md.
17672 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17674 * config/sparc/sparc.opt: New option -mvis4b.
17675 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
17676 (sparc_option_override): Handle VIS4B.
17677 (enum sparc_builtins): Define
17678 SPARC_BUILTIN_DICTUNPACK{8,16,32},
17679 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
17680 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
17681 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
17682 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
17683 (check_constant_argument): New function.
17684 (sparc_vis_init_builtins): Define builtins
17685 __builtin_vis_dictunpack{8,16,32},
17686 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
17687 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
17688 __builtin_vis_fpcmpde{8,16,32}shl and
17689 __builtin_vis_fpcmpur{8,16,32}shl.
17690 (sparc_expand_builtin): Check that the constant operands to
17691 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
17692 constant and in range.
17693 * config/sparc/sparc-c.c (sparc_target_macros): Handle
17695 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
17696 (SPARC_IMM5_P): Likewise.
17697 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
17698 (enabled): Handle vis4b.
17699 (UNSPEC_DICTUNPACK): New unspec.
17700 (UNSPEC_FPCMPSHL): Likewise.
17701 (UNSPEC_FPUCMPSHL): Likewise.
17702 (UNSPEC_FPCMPDESHL): Likewise.
17703 (UNSPEC_FPCMPURSHL): Likewise.
17704 (cpu_feature): New CPU feature `vis4b'.
17705 (dictunpack{8,16,32}): New insns.
17706 (FPCSMODE): New mode iterator.
17707 (fpcscond): New code iterator.
17708 (fpcsucond): Likewise.
17709 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
17710 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
17711 (fpcmpde{8,16,32}{si,di}shl): Likewise.
17712 (fpcmpur{8,16,32}{si,di}shl): Likewise.
17713 * config/sparc/constraints.md: Define constraints `q' for unsigned
17714 2-bit integer constants and `t' for unsigned 5-bit integer
17716 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
17718 (imm5_operand_dictunpack16): Likewise.
17719 (imm5_operand_dictunpack32): Likewise.
17720 (imm2_operand): Likewise.
17721 * doc/invoke.texi (SPARC Options): Document -mvis4b.
17722 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
17723 ditunpack* and fpcmp*shl builtins.
17725 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17727 * config.gcc: Handle m8 in --with-{cpu,tune} options.
17728 * config.in: Add HAVE_AS_SPARC6 define.
17729 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
17731 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
17733 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
17734 (CPP_CPU_SPEC): Handle m8.
17735 (ASM_CPU_SPEC): Likewise.
17736 * config/sparc/sparc-opts.h (enum processor_type): Add
17738 * config/sparc/sparc.c (m8_costs): New struct.
17739 (sparc_option_override): Handle TARGET_CPU_m8.
17740 (sparc32_initialize_trampoline): Likewise.
17741 (sparc64_initialize_trampoline): Likewise.
17742 (sparc_issue_rate): Likewise.
17743 (sparc_register_move_cost): Likewise.
17744 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
17745 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
17746 (ASM_CPU64_DEFAULT_SPEC): Likewise.
17747 (CPP_CPU_SPEC): Handle M8.
17748 (ASM_CPU_SPEC): Likewise.
17749 (AS_M8_FLAG): Define.
17750 * config/sparc/sparc.md: Add m8 to the cpu attribute.
17751 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
17752 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
17754 * configure: Regenerate.
17755 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
17758 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17760 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
17762 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
17763 ("*movdi_insn_sp32"): Do not set v3pipe.
17764 ("*movsi_insn"): Likewise.
17765 ("*movdi_insn_sp64"): Likewise.
17766 ("*movsf_insn"): Likewise.
17767 ("*movdf_insn_sp32"): Likewise.
17768 ("*movdf_insn_sp64"): Likewise.
17769 ("*zero_extendsidi2_insn_sp64"): Likewise.
17770 ("*sign_extendsidi2_insn"): Likewise.
17771 ("*mov<VM32:mode>_insn"): Likewise.
17772 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17773 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17774 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17775 ("<vlop:code><VL:mode>3"): Likewise.
17776 ("*not_<vlop:code><VL:mode>3"): Likewise.
17777 ("*nand<VL:mode>_vis"): Likewise.
17778 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
17779 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
17780 ("one_cmpl<VL:mode>2"): Likewise.
17781 ("faligndata<VM64:mode>_vis"): Likewise.
17782 ("alignaddrsi_vis"): Likewise.
17783 ("alignaddrdi_vis"): Likweise.
17784 ("alignaddrlsi_vis"): Likewise.
17785 ("alignaddrldi_vis"): Likewise.
17786 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17787 ("bmaskdi_vis"): Likewise.
17788 ("bmasksi_vis"): Likewise.
17789 ("bshuffle<VM64:mode>_vis"): Likewise.
17790 ("cmask8<P:mode>_vis"): Likewise.
17791 ("cmask16<P:mode>_vis"): Likewise.
17792 ("cmask32<P:mode>_vis"): Likewise.
17793 ("pdistn<P:mode>_vis"): Likewise.
17794 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17796 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17798 * config/sparc/sparc.md ("subtype"): New insn attribute.
17799 ("*wrgsr_sp64"): Set insn subtype.
17800 ("*rdgsr_sp64"): Likewise.
17801 ("alignaddrsi_vis"): Likewise.
17802 ("alignaddrdi_vis"): Likewise.
17803 ("alignaddrlsi_vis"): Likewise.
17804 ("alignaddrldi_vis"): Likewise.
17805 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17806 ("fexpand_vis"): Likewise.
17807 ("fpmerge_vis"): Likewise.
17808 ("faligndata<VM64:mode>_vis"): Likewise.
17809 ("bshuffle<VM64:mode>_vis"): Likewise.
17810 ("cmask8<P:mode>_vis"): Likewise.
17811 ("cmask16<P:mode>_vis"): Likewise.
17812 ("cmask32<P:mode>_vis"): Likewise.
17813 ("fchksm16_vis"): Likewise.
17814 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
17815 ("fmean16_vis"): Likewise.
17816 ("fp<plusminus_insn>64_vis"): Likewise.
17817 ("<plusminus_insn>v8qi3"): Likewise.
17818 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17819 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
17820 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
17821 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
17822 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
17823 ("*movqi_insn"): Likewise.
17824 ("*movhi_insn"): Likewise.
17825 ("*movsi_insn"): Likewise.
17826 ("movsi_pic_gotdata_op"): Likewise.
17827 ("*movdi_insn_sp32"): Likewise.
17828 ("*movdi_insn_sp64"): Likewise.
17829 ("movdi_pic_gotdata_op"): Likewise.
17830 ("*movsf_insn"): Likewise.
17831 ("*movdf_insn_sp32"): Likewise.
17832 ("*movdf_insn_sp64"): Likewise.
17833 ("*zero_extendhisi2_insn"): Likewise.
17834 ("*zero_extendqihi2_insn"): Likewise.
17835 ("*zero_extendqisi2_insn"): Likewise.
17836 ("*zero_extendqidi2_insn"): Likewise.
17837 ("*zero_extendhidi2_insn"): Likewise.
17838 ("*zero_extendsidi2_insn_sp64"): Likewise.
17839 ("ldfsr"): Likewise.
17840 ("prefetch_64"): Likewise.
17841 ("prefetch_32"): Likewise.
17842 ("tie_ld32"): Likewise.
17843 ("tie_ld64"): Likewise.
17844 ("*tldo_ldub_sp32"): Likewise.
17845 ("*tldo_ldub1_sp32"): Likewise.
17846 ("*tldo_ldub2_sp32"): Likewise.
17847 ("*tldo_ldub_sp64"): Likewise.
17848 ("*tldo_ldub1_sp64"): Likewise.
17849 ("*tldo_ldub2_sp64"): Likewise.
17850 ("*tldo_ldub3_sp64"): Likewise.
17851 ("*tldo_lduh_sp32"): Likewise.
17852 ("*tldo_lduh1_sp32"): Likewise.
17853 ("*tldo_lduh_sp64"): Likewise.
17854 ("*tldo_lduh1_sp64"): Likewise.
17855 ("*tldo_lduh2_sp64"): Likewise.
17856 ("*tldo_lduw_sp32"): Likewise.
17857 ("*tldo_lduw_sp64"): Likewise.
17858 ("*tldo_lduw1_sp64"): Likewise.
17859 ("*tldo_ldx_sp64"): Likewise.
17860 ("*mov<VM32:mode>_insn"): Likewise.
17861 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17862 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17864 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17866 * config/sparc/sparc.md ("type"): New insn type viscmp.
17867 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
17869 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
17870 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
17871 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17872 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
17874 ("n7_vis_logical_11cycle"): Likewise.
17875 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
17876 * config/sparc/niagara2.md ("niag3_vis": Likewise.
17877 * config/sparc/niagara.md ("niag_vis"): Likewise.
17878 * config/sparc/ultra3.md ("us3_fga"): Likewise.
17879 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
17881 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17883 * config/sparc/sparc.md: New instruction type `bmask'.
17884 (bmaskdi_vis): Use the `bmask' type.
17885 (bmasksi_vis): Likewise.
17886 * config/sparc/ultra3.md (us3_array): Likewise.
17887 * config/sparc/niagara7.md (n7_array): Likewise.
17888 * config/sparc/niagara4.md (n4_array): Likewise.
17889 * config/sparc/niagara2.md (niag2_vis): Likewise.
17890 (niag3_vis): Likewise.
17891 * config/sparc/niagara.md (niag_vis): Likewise.
17893 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17895 * ipa-comdats.c: Remove optimize check from gate.
17896 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
17897 for functions not optimized.
17898 (ipa_fn_summary_read): Skip optimize check.
17899 (ipa_fn_summary_write): Likewise.
17900 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
17902 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
17904 (can_inline_edge_p): Check flag_pcc_struct_return for match.
17905 (check_callers): Give up on caller which is not optimized.
17906 (inline_small_functions): Likewise.
17907 (ipa_inline): Do not give up when not optimizing.
17908 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
17909 away unoptimizes cdtors.
17910 (whole_program_function_and_variable_visibility): Do
17911 ipa_discover_readonly_nonaddressable_vars in LTO mode.
17912 * ipa.c (process_references): Do not check optimize.
17913 (symbol_table::remove_unreachable_nodes): Update optimize check.
17914 (set_writeonly_bit): Update optimize check.
17915 (pass_ipa_cdtor_merge::gate): Do not check optimize.
17916 (pass_ipa_single_use::gate): Remove.
17918 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17920 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
17921 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
17922 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
17923 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
17924 permute_load, permute_store, adjust_extract, adjust_splat,
17925 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
17926 replace_swap_with_copy, dump_swap_insn_table,
17927 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
17928 recombine_lvx_pattern, recombine_stvx_pattern,
17929 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
17930 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
17931 to file rs6000-p8swap.c.
17932 * config/rs6000/rs6000-p8swap.c: New file.
17933 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
17934 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
17935 and rs6000*-*-* targets.
17937 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17939 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
17941 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17943 * lto-wrapper.c (merge_and_complain): Do not merge
17944 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
17945 fsigned_zeros, ftrapping_math, fwrapv.
17946 (append_compiler_options): Do not track these options.
17947 (append_linker_options): Likewie
17949 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17951 * cgraphunit.c (cgraph_node::finalize_function): When
17952 !flag_toplevel_reorde set no_reorder flag.
17953 (varpool_node::finalize_decl): Likewise.
17954 (symbol_table::compile): Drop no toplevel reorder path.
17956 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17958 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
17959 edges; zero probability is not better than uninitialized.
17961 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17963 * asan.h (asan_sanitize_allocas_p): Declare.
17964 * asan.c (asan_sanitize_allocas_p): New function.
17965 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
17966 (handle_builtin_alloca): Likewise.
17967 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
17968 if !asan_sanitize_allocas_p.
17969 * params.def (asan-instrument-allocas): Add new option.
17970 * params.h (ASAN_PROTECT_ALLOCAS): Define.
17971 * opts.c (common_handle_option): Disable allocas sanitization for
17974 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17976 * asan.c: Include gimple-fold.h.
17977 (get_last_alloca_addr): New function.
17978 (handle_builtin_stackrestore): Likewise.
17979 (handle_builtin_alloca): Likewise.
17980 (asan_emit_allocas_unpoison): Likewise.
17981 (get_mem_refs_of_builtin_call): Add new parameter, remove const
17982 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
17983 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
17984 (instrument_builtin_call): Pass gimple iterator to
17985 get_mem_refs_of_builtin_call.
17986 (last_alloca_addr): New global.
17987 * asan.h (asan_emit_allocas_unpoison): Declare.
17988 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
17989 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
17990 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
17991 if function calls alloca.
17992 * gimple-fold.c (replace_call_with_value): Remove static keyword.
17993 * gimple-fold.h (replace_call_with_value): Declare.
17994 * internal-fn.c: Include asan.h.
17995 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
17996 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
17998 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18000 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
18001 (C_SELFTEST_FLAGS): New.
18002 (CPP_SELFTEST_FLAGS): New.
18003 (SELFTEST_DEPS): New, from deps of s-selftest.
18004 (C_SELFTEST_DEPS): New, from deps of s-selftest.
18005 (CPP_SELFTEST_DEPS): New.
18006 (selftest): Add dependency on s-selftest-c++.
18007 (s-selftest): Rename to...
18008 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
18009 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
18010 than SELFTEST_FLAGS.
18011 (selftest-gdb): Rename to...
18012 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
18014 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
18015 (selftest-valgrind): Rename to...
18016 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18018 (selftest-valgrind): Reintroduce as an alias for
18019 selftest-c-valgrind.
18020 (s-selftest-c++): New.
18021 (selftest-c++-gdb): New.
18022 (selftest-c++-valgrind): New.
18024 2017-07-06 Olivier Hainque <hainque@adacore.com>
18026 * gcc.c (process_command): When deciding if undefined variables
18027 should be ignored when processing specs, accept "gcc -v" as well.
18029 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18031 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18032 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18034 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18036 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18037 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18038 * config/arm/arm-tables.opt: Regenerate.
18039 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18041 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18043 2017-07-06 Carl Love <cel@us.ibm.com>
18045 * ChangeLog: Clean up from mid air collision
18047 2017-07-06 Carl Love <cel@us.ibm.com>
18049 * config/rs6000/rs6000-c.c: Add support for built-in functions
18050 vector signed int vec_subc (vector signed int, vector signed int);
18051 vector signed __int128 vec_subc (vector signed __int128,
18052 vector signed __int128);
18053 vector unsigned __int128 vec_subc (vector unsigned __int128,
18054 vector unsigned __int128);
18055 vector signed int vec_sube (vector signed int, vector signed int,
18056 vector signed int);
18057 vector unsigned int vec_sube (vector unsigned int,
18058 vector unsigned int,
18059 vector unsigned int);
18060 vector signed __int128 vec_sube (vector signed __int128,
18061 vector signed __int128,
18062 vector signed__int128);
18063 vector unsigned __int128 vec_sube (vector unsigned __int128,
18064 vector unsigned __int128,
18065 vector unsigned __int128);
18066 vector signed int vec_subec (vector signed int, vector signed int,
18067 vector signed int);
18068 vector unsigned int vec_subec (vector unsigned int,
18069 vector unsigned int,
18070 vector unsigned int);
18071 vector signed __int128 vec_subec (vector signed __int128,
18072 vector signed __int128,
18073 vector signed__int128);
18074 vector unsigned __int128 vec_subec (vector unsigned __int128,
18075 vector unsigned __int128,
18076 vector unsigned __int128);
18077 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18078 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18079 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18080 BU_ALTIVEC_OVERLOAD_X definitions.
18081 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18082 * doc/extend.texi: Update the built-in documentation file for the new
18083 built-in functions.
18085 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18088 * diagnostic-show-locus.c (layout::layout): Use start and finish
18089 spelling location for the start and finish of each range.
18090 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18091 Add unused aspect param.
18092 * input.c (expand_location_1): Add "aspect" param, and use it
18093 to access the correct part of the location.
18094 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18096 (expand_location_to_spelling_point): Likewise.
18097 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18098 param, and pass it to expand_location_1.
18100 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18102 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18103 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18104 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18105 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18106 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18107 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18108 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18109 _mm_maskz_getmant_ss): New intrinsics.
18110 (__builtin_ia32_getexpss128_mask): Changed to ...
18111 __builtin_ia32_getexpss128_round ... this.
18112 (__builtin_ia32_getexpsd128_mask): Changed to ...
18113 __builtin_ia32_getexpsd128_round ... this.
18114 * config/i386/i386-builtin-types.def
18115 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18116 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18117 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18118 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18119 __builtin_ia32_getmantss_mask_round): New builtins.
18120 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18121 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18122 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18123 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18124 * config/i386/sse.md
18125 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18126 avx512f_sgetexp<mode><mask_scalar_name>
18127 <round_saeonly_scalar_name> ... this.
18128 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18129 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18130 vgetexp<ssescalarmodesuffix>
18131 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18132 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18133 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18134 avx512f_vgetmant<mode><mask_scalar_name>
18135 <round_saeonly_scalar_name> ... this.
18136 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18137 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18138 vgetmant<ssescalarmodesuffix>
18139 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18140 %0<mask_scalar_operand4>, %1, %2
18141 <round_saeonly_scalar_mask_op4>, %3} ... this.
18142 * config/i386/subst.md (mask_scalar_operand4,
18143 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18144 round_saeonly_scalar_nimm_predicate): New subst attributes.
18146 2017-07-06 Julia Koval <julia.koval@intel.com>
18148 * config/i386/i386.c (ix86_erase_embedded_rounding):
18149 Remove code for old rounding pattern.
18151 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18153 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18155 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18157 * doc/sourcebuild.texi (Test Directives, Variants of
18158 dg-require-support): Add documentation for dg-require-stack-check.
18160 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18162 * config/i386/subst.md (mask_scalar, round_scalar,
18163 round_saeonly_scalar): New meta-templates.
18164 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18165 round_scalar_mask_operand3, round_scalar_mask_op3,
18166 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18167 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18168 round_saeonly_scalar_constraint,
18169 round_saeonly_scalar_prefix): New subst attribute.
18170 * config/i386/sse.md
18171 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18172 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18173 <round_scalar_name> ... this.
18174 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18175 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18176 <round_scalar_name> ... this.
18177 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18178 <sse>_vm<code><mode>3<mask_scalar_name>
18179 <round_saeonly_scalar_name> ... this.
18180 (v<plusminus_mnemonic><ssescalarmodesuffix>
18181 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18182 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18183 v<plusminus_mnemonic><ssescalarmodesuffix>
18184 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18185 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18186 (v<multdiv_mnemonic><ssescalarmodesuffix>
18187 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18188 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18189 v<multdiv_mnemonic><ssescalarmodesuffix>
18190 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18191 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18192 (v<maxmin_float><ssescalarmodesuffix>
18193 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18194 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18195 v<maxmin_float><ssescalarmodesuffix>
18196 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18197 %0<mask_scalar_operand3>, %1, %<iptr>2
18198 <round_saeonly_scalar_mask_op3>} ... this.
18200 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18202 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18203 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18205 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18206 Alan Hayward <alan.hayward@arm.com>
18207 David Sherwood <david.sherwood@arm.com>
18209 * combine.c (simplify_if_then_else): Remove "enum" before
18211 * compare-elim.c (can_eliminate_compare): Likewise.
18212 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18214 (aarch64_lookup_simd_builtin_type): Likewise.
18215 (aarch64_simd_builtin_type): Likewise.
18216 (aarch64_init_simd_builtin_types): Likewise.
18217 (aarch64_simd_expand_args): Likewise.
18218 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18220 (aarch64_reverse_mask): Likewise.
18221 (aarch64_simd_emit_reg_reg_move): Likewise.
18222 (aarch64_gen_adjusted_ldpstp): Likewise.
18223 (aarch64_ccmp_mode_to_code): Likewise.
18224 (aarch64_operands_ok_for_ldpstp): Likewise.
18225 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18226 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18228 (aarch64_min_divisions_for_recip_mul): Likewise.
18229 (aarch64_reassociation_width): Likewise.
18230 (aarch64_get_condition_code_1): Likewise.
18231 (aarch64_simd_emit_reg_reg_move): Likewise.
18232 (aarch64_simd_attr_length_rglist): Likewise.
18233 (aarch64_reverse_mask): Likewise.
18234 (aarch64_operands_ok_for_ldpstp): Likewise.
18235 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18236 (aarch64_gen_adjusted_ldpstp): Likewise.
18237 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18239 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18240 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18241 (arm_lookup_simd_builtin_type): Likewise.
18242 (arm_simd_builtin_type): Likewise.
18243 (arm_init_simd_builtin_types): Likewise.
18244 (arm_expand_builtin_args): Likewise.
18245 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18246 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18247 (ft32_setup_incoming_varargs): Likewise.
18248 (ft32_function_arg): Likewise.
18249 (ft32_function_arg_advance): Likewise.
18250 (ft32_pass_by_reference): Likewise.
18251 (ft32_arg_partial_bytes): Likewise.
18252 (ft32_valid_pointer_mode): Likewise.
18253 (ft32_addr_space_pointer_mode): Likewise.
18254 (ft32_addr_space_legitimate_address_p): Likewise.
18255 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18257 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18258 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18259 (iamcu_alignment): Likewise.
18260 (canonicalize_vector_int_perm): Likewise.
18261 (ix86_noce_conversion_profitable_p): Likewise.
18262 (ix86_mpx_bound_mode): Likewise.
18263 (ix86_operands_ok_for_move_multiple): Likewise.
18264 * config/microblaze/microblaze-protos.h
18265 (microblaze_expand_conditional_branch_reg): Likewise.
18266 * config/microblaze/microblaze.c
18267 (microblaze_expand_conditional_branch_reg): Likewise.
18268 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18270 (rs6000_reassociation_width): Likewise.
18271 (rs6000_invalid_binary_op): Likewise.
18272 (fusion_p9_p): Likewise.
18273 (emit_fusion_p9_load): Likewise.
18274 (emit_fusion_p9_store): Likewise.
18275 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18277 (riscv_hard_regno_mode_ok_p): Likewise.
18278 (riscv_address_insns): Likewise.
18279 (riscv_split_symbol): Likewise.
18280 (riscv_legitimize_move): Likewise.
18281 (riscv_function_value): Likewise.
18282 (riscv_hard_regno_nregs): Likewise.
18283 (riscv_expand_builtin): Likewise.
18284 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18285 (riscv_build_integer): Likewise.
18286 (riscv_split_integer): Likewise.
18287 (riscv_legitimate_constant_p): Likewise.
18288 (riscv_cannot_force_const_mem): Likewise.
18289 (riscv_regno_mode_ok_for_base_p): Likewise.
18290 (riscv_valid_base_register_p): Likewise.
18291 (riscv_valid_offset_p): Likewise.
18292 (riscv_valid_lo_sum_p): Likewise.
18293 (riscv_classify_address): Likewise.
18294 (riscv_legitimate_address_p): Likewise.
18295 (riscv_address_insns): Likewise.
18296 (riscv_load_store_insns): Likewise.
18297 (riscv_force_binary): Likewise.
18298 (riscv_split_symbol): Likewise.
18299 (riscv_force_address): Likewise.
18300 (riscv_legitimize_address): Likewise.
18301 (riscv_move_integer): Likewise.
18302 (riscv_legitimize_const_move): Likewise.
18303 (riscv_legitimize_move): Likewise.
18304 (riscv_address_cost): Likewise.
18305 (riscv_subword): Likewise.
18306 (riscv_output_move): Likewise.
18307 (riscv_canonicalize_int_order_test): Likewise.
18308 (riscv_emit_int_order_test): Likewise.
18309 (riscv_function_arg_boundary): Likewise.
18310 (riscv_pass_mode_in_fpr_p): Likewise.
18311 (riscv_pass_fpr_single): Likewise.
18312 (riscv_pass_fpr_pair): Likewise.
18313 (riscv_get_arg_info): Likewise.
18314 (riscv_function_arg): Likewise.
18315 (riscv_function_arg_advance): Likewise.
18316 (riscv_arg_partial_bytes): Likewise.
18317 (riscv_function_value): Likewise.
18318 (riscv_pass_by_reference): Likewise.
18319 (riscv_setup_incoming_varargs): Likewise.
18320 (riscv_print_operand): Likewise.
18321 (riscv_elf_select_rtx_section): Likewise.
18322 (riscv_save_restore_reg): Likewise.
18323 (riscv_for_each_saved_reg): Likewise.
18324 (riscv_register_move_cost): Likewise.
18325 (riscv_hard_regno_mode_ok_p): Likewise.
18326 (riscv_hard_regno_nregs): Likewise.
18327 (riscv_class_max_nregs): Likewise.
18328 (riscv_memory_move_cost): Likewise.
18329 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18330 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18331 (rl78_addr_space_address_mode): Likewise.
18332 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18334 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18335 (rs6000_reassociation_width): Likewise.
18336 (rs6000_invalid_binary_op): Likewise.
18337 (fusion_p9_p): Likewise.
18338 (emit_fusion_p9_load): Likewise.
18339 (emit_fusion_p9_store): Likewise.
18340 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18341 (ok_for_simple_move_operands): Likewise.
18342 (ok_for_simple_move_strict_operands): Likewise.
18343 (ok_for_simple_arith_logic_operands): Likewise.
18344 (visium_legitimize_reload_address): Likewise.
18345 (visium_select_cc_mode): Likewise.
18346 (output_cbranch): Likewise.
18347 (visium_split_double_move): Likewise.
18348 (visium_expand_copysign): Likewise.
18349 (visium_expand_int_cstore): Likewise.
18350 (visium_expand_fp_cstore): Likewise.
18351 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18352 (visium_function_arg): Likewise.
18353 (visium_function_arg_advance): Likewise.
18354 (visium_libcall_value): Likewise.
18355 (visium_setup_incoming_varargs): Likewise.
18356 (visium_legitimate_constant_p): Likewise.
18357 (visium_legitimate_address_p): Likewise.
18358 (visium_legitimize_address): Likewise.
18359 (visium_secondary_reload): Likewise.
18360 (visium_register_move_cost): Likewise.
18361 (visium_memory_move_cost): Likewise.
18362 (prepare_move_operands): Likewise.
18363 (ok_for_simple_move_operands): Likewise.
18364 (ok_for_simple_move_strict_operands): Likewise.
18365 (ok_for_simple_arith_logic_operands): Likewise.
18366 (visium_function_value_1): Likewise.
18367 (rtx_ok_for_offset_p): Likewise.
18368 (visium_legitimize_reload_address): Likewise.
18369 (visium_split_double_move): Likewise.
18370 (visium_expand_copysign): Likewise.
18371 (visium_expand_int_cstore): Likewise.
18372 (visium_expand_fp_cstore): Likewise.
18373 (visium_split_cstore): Likewise.
18374 (visium_select_cc_mode): Likewise.
18375 (visium_split_cbranch): Likewise.
18376 (output_cbranch): Likewise.
18377 (visium_print_operand_address): Likewise.
18378 * expmed.c (flip_storage_order): Likewise.
18379 * expmed.h (emit_cstore): Likewise.
18380 (flip_storage_order): Likewise.
18381 * genrecog.c (validate_pattern): Likewise.
18382 * hsa-gen.c (gen_hsa_addr): Likewise.
18383 * internal-fn.c (expand_arith_overflow): Likewise.
18384 * ira-color.c (allocno_copy_cost_saving): Likewise.
18385 * lra-assigns.c (find_hard_regno_for_1): Likewise.
18386 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
18387 (process_invariant_for_inheritance): Likewise.
18388 * lra-eliminations.c (move_plus_up): Likewise.
18389 * omp-low.c (lower_oacc_reductions): Likewise.
18390 * simplify-rtx.c (simplify_subreg): Likewise.
18391 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
18392 (TARGET_CHKP_BOUND_MODE): Likewise..
18393 * targhooks.c (default_chkp_bound_mode): Likewise.
18394 (default_setup_incoming_vararg_bounds): Likewise.
18395 * targhooks.h (default_chkp_bound_mode): Likewise.
18396 (default_setup_incoming_vararg_bounds): Likewise.
18397 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
18398 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
18399 (have_whole_vector_shift): Likewise.
18400 * tree-vect-stmts.c (vectorizable_load): Likewise.
18401 * doc/tm.texi: Regenerate.
18403 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18405 Graceful degrade if Binutils PR21472 is not available.
18408 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
18409 .rodata in flash test fails.
18410 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
18411 * confgure: Regenerate.
18412 * config.in: Regenerate.
18413 * config/avr/avr.c (avr_asm_named_section)
18414 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
18415 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
18416 (avr_asm_init_sections): Same.
18418 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18420 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
18421 (fma<VH:mode>4_intrinsic): Likewise.
18422 (*fmsub<VCVTF:mode>4): Likewise.
18423 (*fmsub<VH:mode>4_intrinsic): Likewise.
18425 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18428 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
18429 Don't depend on "optimize > 0".
18430 (out_movhi_r_mr, out_movqi_mr_r): Same.
18431 (out_movhi_mr_r, out_movqi_r_mr): Same.
18432 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
18433 io_address_operand on "optimize > 0".
18435 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18437 * tree-loop-distribution.c: Add general explanantion on the pass.
18438 (generate_loops_for_partition): Mark distributed loop.
18439 (pg_add_dependence_edges): New parameter. Handle alias data
18440 dependence specially and record it in the parameter if asked.
18441 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
18442 (init_partition_graph_vertices, add_partition_graph_edge): New.
18443 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
18444 (free_partition_graph_vdata, build_partition_graph): New.
18445 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
18446 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
18447 (data_ref_segment_size, latch_dominated_by_data_ref): New.
18448 (compute_alias_check_pairs, version_loop_by_alias_check): New.
18449 (version_for_distribution_p, finalize_partitions): New.
18450 (distribute_loop): Handle alias data dependence specially. Factor
18451 out loop fusion code as functions and call these functions.
18453 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18455 * tree-loop-distribution.c (classify_partition): New parameter and
18456 better handle reduction statement.
18457 (rdg_build_partitions): Revise comment.
18458 (distribute_loop): Compute statements in all partitions and pass it
18459 to classify_partition.
18461 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18463 * tree-loop-distribution.c (enum partition_type): New.
18464 (struct partition): New field type.
18465 (partition_merge_into): Add parameter. Update partition type.
18466 (data_dep_in_cycle_p, update_type_for_merge): New functions.
18467 (build_rdg_partition_for_vertex): Compute partition type.
18468 (rdg_build_partitions): Dump partition type.
18469 (distribute_loop): Update calls to partition_merge_into.
18471 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18473 * tree-loop-distribution.c (struct ddr_hasher): New.
18474 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
18476 (classify_partition): Call get_data_dependence.
18477 (pg_add_dependence_edges): Ditto.
18478 (distribute_loop): Release data dependence hash table.
18480 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18482 * tree-loop-distribution.c (ref_base_address): Delete.
18483 (similar_memory_accesses): Rename ...
18484 (share_memory_accesses): ... to this. Check if partitions access
18485 the same memory reference.
18486 (distribute_loop): Call share_memory_accesses.
18488 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18490 * tree-loop-distribution.c (struct partition): New field recording
18491 its data reference.
18492 (partition_alloc, partition_free): Init and release data refs.
18493 (partition_merge_into): Merge data refs.
18494 (build_rdg_partition_for_vertex): Collect data refs for partition.
18495 (pg_add_dependence_edges): Change parameters from vector to bitmap.
18497 (distribute_loop): Remve data refs from vertice data of partition
18500 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18502 * tree-loop-distribution.c (params.h): Include header file.
18503 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
18504 (datarefs_vec): New global var.
18505 (create_rdg_vertices): Use datarefs_vec directly.
18506 (free_rdg): Don't free data references.
18507 (build_rdg): Update use. Don't free data references.
18508 (distribute_loop): Compute global variable for data references.
18509 Bail out if there are too many data references.
18511 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18513 * tree-loop-distribution.c (loop_nest): New global var.
18514 (build_rdg): Use loop directly, rather than loop nest.
18515 (pg_add_dependence_edges): Remove loop nest parameter. Use global
18517 (distribute_loop): Compute global variable loop nest. Update use.
18519 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18521 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
18522 (partition_merge_into): New parameter. Dump reason for fusion.
18523 (distribute_loop): Update use of partition_merge_into.
18525 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18527 * tree-loop-distribution.c (bb_top_order_index): New.
18528 (bb_top_order_index_size, bb_top_order_cmp): New.
18529 (stmts_from_loop): Use topological order.
18530 (pass_loop_distribution::execute): Compute and release topological
18531 order for basic blocks.
18533 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18535 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
18538 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18540 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
18541 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
18542 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
18543 * internal-fn.def (LOOP_DIST_ALIAS): New.
18544 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
18545 (fold_loop_internal_call): ... this.
18546 (vect_loop_dist_alias_call): New function.
18547 (set_uid_loop_bbs): Call fold_loop_internal_call.
18548 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
18551 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18554 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
18555 Require dead FLAGS_REG at the beginning of a peephole.
18557 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18560 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
18561 arguments in the call to __builtin_ia32_sbb_u32.
18562 (_subborrow_u64): Swap _X and _Y arguments in the call to
18563 __builtin_ia32_sbb_u64.
18565 2017-07-04 Jakub Jelinek <jakub@redhat.com>
18568 * tree-vrp.c (compare_assert_loc): Turn into a function template
18569 with stable template parameter. Only test if a->e is NULL,
18570 !a->e == !b->e has been verified already. Use e == NULL or
18571 e != NULL instead of e or ! e tests. If stable is true, don't use
18572 iterative_hash_expr, on the other side allow a or b or both NULL
18573 and sort the NULLs last.
18574 (process_assert_insertions): Sort using compare_assert_loc<false>
18575 instead of compare_assert_loc, later sort using
18576 compare_assert_loc<true> before calling process_assert_insertions_for
18577 in a loop. Use break instead of continue once seen NULL pointer.
18579 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18581 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18582 Cortex-R7 and Cortex-R8 processors.
18584 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18586 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
18587 uninitialized while src is not.
18589 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
18591 * common/config/arm/arm-common.c: Adjust include path for
18593 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
18594 (arm-cpu.h): Create in build directory. Adjust dependency rules.
18595 (arm-cpu-data.h): Likewise.
18596 (arm-cpu-cdata.h): Likewise.
18597 * config/arm/arm-cpu.h: Delete.
18598 * config/arm/arm-cpu-cdata.h: Delete.
18599 * config/arm/arm-cpu-data.h: Delete.
18601 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
18603 * config/arm/arm-cpus.in (cortex-a55): New.
18604 (cortex-a75): Likewise.
18605 (cortex-a75.cortex-a55): Likewise.
18606 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
18608 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
18609 * config/arm/arm-cpu-cdata.h: Regenerate.
18610 * config/arm/arm-cpu-data.h: Regenerate.
18611 * config/arm/arm-cpu.h: Regenerate.
18612 * config/arm/arm-tables.opt: Regenerate.
18613 * config/arm/arm-tune.md: Regenerate.
18615 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18617 * haifa-sched.c (sched_create_recovery_edges): Update profile.
18619 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18621 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
18624 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
18626 PR tree-optimization/81292
18627 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
18628 full_string_p, also call adjust_related_strinfos if the adjustment
18629 is simple, otherwise invalidate related strinfos.
18631 2017-07-04 Martin Liska <mliska@suse.cz>
18634 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
18635 newly created variable as DECL_IGNORED_P.
18637 2017-07-04 Martin Liska <mliska@suse.cz>
18640 * ipa-inline.c (inline_small_functions):
18641 Use xstrdup_for_dump.
18643 2017-07-04 Tom de Vries <tom@codesourcery.com>
18645 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
18647 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18650 * config/darwin.c (darwin_function_switched_text_sections):
18653 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
18655 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
18657 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18659 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
18661 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18663 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
18664 min_profitable_iters, and th as inclusive lower bounds.
18665 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
18666 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
18667 for min_profitable_iters and min_profitable_estimate.
18668 (vect_transform_loop): Treat th as an inclusive lower bound.
18669 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
18671 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18674 * config/darwin.c (darwin_function_switched_text_sections):
18675 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
18676 in two pieces, and suppress the use of buf.
18678 2017-07-03 Nathan Sidwell <nathan@acm.org>
18680 * hash-table.h (hash_table_mod1): Fix indentation.
18682 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18684 PR middle-end/81290
18685 * predict.c (force_edge_cold): Be more careful about propagation
18687 * profile-count.h (profile_probability::guessed,
18688 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
18690 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
18692 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
18694 * doc/invoke.texi (rcpc architecture extension): Document it.
18696 2017-07-03 Richard Biener <rguenther@suse.de>
18698 PR tree-optimization/60510
18699 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
18700 the scalar reduction PHI and use it.
18701 (vectorizable_reduction): Properly guard the single_defuse_cycle
18702 path for non-SLP reduction chains where we cannot use it.
18703 Rework reduc_def/index and vector type deduction. Rework
18704 vector operand gathering during reduction op code-gen.
18705 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
18706 chains dissolve the chain and leave it to non-SLP reduction
18709 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18711 * tree-data-ref.h (dr_alignment): Declare.
18712 * tree-data-ref.c (dr_alignment): New function.
18713 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
18714 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
18716 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
18718 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18720 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
18721 and base_misalignment fields.
18722 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
18723 * tree-data-ref.c: Include builtins.h.
18724 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
18725 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
18726 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
18727 * tree-vect-data-refs.c: Include tree-cfg.h.
18728 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
18729 fields instead of calculating an alignment here.
18730 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
18731 innermost_loop_behavior fields.
18733 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18735 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
18737 (DR_STEP_ALIGNMENT): New macro.
18738 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
18739 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
18740 (create_data_ref): Print it.
18741 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
18742 to tell whether the step preserves vector (mis)alignment.
18743 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18744 Move the check for an integer step and generalise to all INTEGER_CST.
18745 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
18746 Print the outer step alignment.
18748 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18750 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
18751 with offset_alignment.
18752 (DR_ALIGNED_TO): Delete.
18753 (DR_OFFSET_ALIGNMENT): New macro.
18754 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
18755 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
18756 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
18757 (create_data_ref): Likewise.
18758 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18759 (vect_analyze_data_refs): Likewise.
18760 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
18761 creating dummy innermost behavior.
18763 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18765 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
18766 with a "innermost_loop_behavior *" and refeence tree.
18767 * tree-data-ref.c (dr_analyze_innermost): Likewise.
18768 (create_data_ref): Update call accordingly.
18769 * tree-predcom.c (find_looparound_phi): Likewise.
18771 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18773 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
18774 fields with dr_wrt_vec_loop.
18775 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
18776 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
18777 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
18778 (vect_dr_behavior): New function.
18779 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18780 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
18781 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
18782 track whether the step preserves the misalignment.
18783 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18784 Use vect_dr_behavior.
18785 (vect_setup_realignment): Update call accordingly.
18786 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
18787 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
18788 call to vect_create_addr_base_for_vector_ref.
18789 (vect_create_cond_for_align_checks): Likewise.
18790 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
18791 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
18792 (vect_recog_mask_conversion_pattern): Likewise.
18793 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
18794 (new_stmt_vec_info): Remove redundant zeroing.
18796 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18798 * common/config/arm/arm-common.c (arm_be8_option): New function.
18799 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
18800 (ISA_ARMv6): Add isa_bit_be8.
18801 * config/arm/arm.h (arm_be8_option): Add prototype.
18802 (BE8_SPEC_FUNCTION): New define.
18803 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
18804 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
18805 (mlittle-endian): Similarly.
18806 (mbe8, mbe32): New options.
18807 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
18808 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
18810 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18812 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
18814 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18816 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
18817 (cleanup_tree_cfg_bb): Use it.
18818 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
18820 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
18822 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18825 * loop-doloop.c (add_test): Update profile.
18827 2017-07-03 Martin Liska <mliska@suse.cz>
18830 * sanopt.c (rewrite_usage_of_param): New function.
18831 (sanitize_rewrite_addressable_params): Likewise.
18832 (pass_sanopt::execute): Call rewrite_usage_of_param.
18834 2017-07-03 Richard Biener <rguenther@suse.de>
18836 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
18837 back to using VIEW_CONVERT_EXPR.
18839 2017-07-03 Martin Liska <mliska@suse.cz>
18842 * doc/extend.texi: Document when a resolver function is
18843 generated for target_clones.
18845 2017-07-03 Martin Liska <mliska@suse.cz>
18847 * asan.c (asan_emit_stack_protection): Unpoison just red zones
18848 and shadow memory of auto variables which are subject of
18849 use-after-scope sanitization.
18850 (asan_expand_mark_ifn): Add do set only when is_poison.
18852 2016-07-03 Richard Biener <rguenther@suse.de>
18854 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
18856 (vect_force_simple_reduction): Record reduction def -> phi mapping.
18857 (vectorizable_reduction): Perform reduction PHI creation when
18858 visiting a reduction PHI and adjust and simplify code generation
18859 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
18860 (vect_transform_loop): Visit reduction PHIs.
18861 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
18862 defs into the SLP tree.
18863 (vect_build_slp_tree): Reduction defs terminate the recursion.
18864 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
18866 (vect_get_vec_defs_for_stmt_copy): Export.
18867 (vect_get_vec_defs): Likewise.
18868 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
18870 (vect_get_vec_defs_for_stmt_copy): Declare.
18871 (vect_get_vec_defs): Likewise.
18873 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18875 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
18876 parameter with a "loop" parameter and use it instead of the
18877 loop containing DR_STMT. Don't check simple_iv when doing
18878 BB analysis. Describe the two analysis modes in the comment.
18880 2017-07-03 Tom de Vries <tom@codesourcery.com>
18882 PR tree-optimization/69468
18883 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
18884 (find_same_succ_bb): Handle ignore_edge_flags.
18886 2017-07-03 Tom de Vries <tom@codesourcery.com>
18888 PR tree-optimization/81192
18889 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
18891 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
18893 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
18895 2017-07-03 Tom de Vries <tom@codesourcery.com>
18897 PR tree-optimization/81192
18898 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
18899 BB_SAME_SUCC (bb) == NULL.
18901 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18903 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
18906 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18908 * dumpfile.c: Include profile-count.h
18909 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
18911 (insert_cond_bb): Update profile.
18912 * tree-cfg.h (insert_cond_bb): Update prototype.
18913 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
18914 * tree-dump.c: Do not include tree-cfg.
18916 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18918 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
18920 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18922 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
18925 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18927 * tree-complex.c (expand_complex_div_wide): update profile.
18929 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18930 Alan Hayward <alan.hayward@arm.com>
18931 David Sherwood <david.sherwood@arm.com>
18933 * Makefile.in (MACHMODE_H): Remove insn-modes.h
18934 (CORETYPES_H): New define.
18935 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
18936 (insn-modes-inline.h, s-modes-inline-h): New rules.
18937 (generated_files): Add insn-modes-inline.h.
18938 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
18939 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
18940 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
18941 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
18942 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
18943 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
18944 (build/gencodes.o, build/genconditions.o): Likewise.
18945 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
18946 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
18947 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
18948 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
18949 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
18950 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
18951 * coretypes.h: Include everything up to real.h for generators.
18952 Include insn-modes.h first. Include wide-int-print.h after
18953 wide-int.h. Include insn-modes-inline.h and then machmode.h.
18954 * machmode.h: Don't include insn-modes.h here.
18955 * function-tests.c: Remove includes of signop.h, machmode.h,
18956 double-int.h and wide-int.h.
18958 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
18960 * optc-save-gen.awk: Likewise.
18961 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
18962 * godump.c: Remove include of wide-int-print.h.
18963 * pretty-print.h: Likewise.
18964 * wide-int-print.cc: Likewise.
18965 * wide-int.cc: Likewise.
18966 * hash-map-tests.c: Remove include of signop.h.
18967 * hash-set-tests.c: Likewise.
18968 * rtl-tests.c: Likewise.
18969 * mkconfig.sh: Remove include of machmode.h.
18970 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
18972 (emit_insn_modes_inline_h): ...this new function. Emit the code
18973 into an insn-modes-inline.h header file, adding appropriate
18974 include guards and end comments.
18975 (emit_insn_modes_c_header): Remove include of machmode.h.
18976 (emit_min_insn_modes_c_header): Include coretypes.h rather than
18978 (main): Handle -i flag and call emit_insn_modes_inline_h when
18981 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18983 * tree-ssa-strlen.c (strinfo): Rename the length field to
18984 nonzero_chars. Add a full_string_p field.
18985 (compare_nonzero_chars, zero_length_string_p): New functions.
18986 (get_addr_stridx): Add an offset_out parameter.
18987 Use compare_nonzero_chars.
18988 (get_stridx): Update accordingly. Use compare_nonzero_chars.
18989 (new_strinfo): Update after above changes to strinfo.
18990 (set_endptr_and_length): Set full_string_p.
18991 (get_string_length): Update after above changes to strinfo.
18992 (unshare_strinfo): Update call to new_strinfo.
18993 (maybe_invalidate): Likewise.
18994 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
18995 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
18996 as a uhwi instead of an shwi. Update after above changes to
18997 strinfo and new_strinfo.
18998 (zero_length_string): Assert that chainsi contains full strings.
18999 Use zero_length_string_p. Update call to new_strinfo.
19000 (adjust_related_strinfos): Update after above changes to strinfo.
19001 Copy full_string_p from origsi.
19002 (adjust_last_stmt): Use zero_length_string_p.
19003 (handle_builtin_strlen): Update after above changes to strinfo and
19004 new_strinfo. Install the lhs as the string length if the previous
19005 entry didn't describe a full string.
19006 (handle_builtin_strchr): Update after above changes to strinfo
19008 (handle_builtin_strcpy): Likewise.
19009 (handle_builtin_strcat): Likewise.
19010 (handle_builtin_malloc): Likewise.
19011 (handle_pointer_plus): Likewise.
19012 (handle_builtin_memcpy): Likewise. Track nonzero characters
19013 that aren't necessarily followed by a nul terminator.
19014 (handle_char_store): Likewise.
19016 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19018 PR tree-optimization/80769
19019 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19020 for malloc and calloc. Document the new invariant that all related
19021 strinfos have delayed lengths or none do.
19022 (verify_related_strinfos): Move earlier in file.
19023 (set_endptr_and_length): New function, split out from...
19024 (get_string_length): ...here. Also set the lengths of related
19026 (zero_length_string): Assert that chainsi has known (rather than
19028 (adjust_related_strinfos): Likewise.
19030 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19032 PR tree-optimization/81136
19033 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19034 assert that two references with the same misalignment have the same
19035 compile-time misalignment if those compile-time misalignments
19038 2017-07-01 Andi Kleen <ak@linux.intel.com>
19040 * print-tree.c (print_node): Print all attributes.
19042 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19044 * cfg.c (scale_bbs_frequencies): New function.
19045 * cfg.h (scale_bbs_frequencies): Declare it.
19046 * cfgloopanal.c (single_likely_exit): Cleanup.
19047 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19049 (scale_loop_profile): Likewise.
19050 (loop_version): Likewise.
19051 (create_empty_loop_on_edge): Update.
19052 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19053 scale_loop_frequencies, scale_loop_profile, loopify,
19054 loop_version): Update prototypes.
19055 * modulo-sched.c (sms_schedule): Update.
19056 * predict.c (unlikely_executed_edge_p): Also check probability.
19057 (probably_never_executed_edge_p): Fix typo.
19058 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19059 * tree-parloops.c (gen_parallel_loop): Update.
19060 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19061 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19062 * tree-ssa-loop-split.c (split_loop): Update.
19063 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19064 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19065 (vect_loop_versioning): Update.
19066 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19068 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19070 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19072 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19074 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19075 to keep profile consistent.
19077 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19079 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19080 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19081 * profile-count.h (max_safe_multiplier): Make unsigned.
19082 (profile_count::guessed_zero): New.
19084 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19086 * bb-reorder.c (fix_up_crossing_landing_pad,
19087 fix_crossing_conditional_branches): Use make_single_succ_edge
19088 to keep profile consistent.
19090 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19092 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19095 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19098 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19099 the right scopes, make sure cond_jump isn't preserved between multiple
19100 iterations. Search for fallthru edge whenever there are 3+ edges and
19101 use find_fallthru_edge for it.
19103 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19105 Patch by Alexander Monakov <amonakov@ispras.ru>
19106 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19107 probabilities consistently.
19109 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19111 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19112 emit_cmp_and_jump_insns.
19114 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19117 * tree-inline.c (expand_call_inline): Combine profile statuses.
19119 2017-06-30 Andrew Pinski <apinski@cavium.com>
19121 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19122 fold_stmt returned true.
19124 2017-06-30 Nathan Sidwell <nathan@acm.org>
19126 * ggc.h (empty_string): Delete.
19127 * cfgexpand.c (expand_asm_stmt): Use plain "".
19128 * optabs.c (expand_asm_memory_barrier): Likewise.
19129 * stringpool.c (empty_string): Delete.
19130 (digit_vector, digit_string): Delete.
19131 (ggc_alloc_string): Use plain "", don't optimize single digit
19132 strings. Use ggc_alloc_atomic.
19134 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19136 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19137 comparison set and one other set, use the cost of the non-comparison
19140 2017-06-30 Nathan Sidwell <nathan@acm.org>
19142 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19145 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19147 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19148 loops. Remove now unneeded calls to gimple_switch_set_label() that
19149 just set removed labels to NULL_TREE.
19151 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19153 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19154 (set_range_info): ...here. Only call set_range_info_raw if domain
19156 (set_nonzero_bits): Call set_range_info_raw.
19157 * tree-ssanames.h (set_range_info_raw): New.
19159 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19162 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19163 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19164 of nonimmediate_operand and <store_mask_constraint> instead of m
19165 for the input operand. For V8FI iterator, always split if input
19166 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19167 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19168 <store_mask_predicate> instead of register_operand and
19169 <store_mask_constraint> instead of v for the input operand. Make
19170 sure both operands aren't MEMs for if not <mask_applied>.
19172 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19174 * lto-wrapper.c (copy_file) Close both file descriptors before
19177 2017-06-30 Martin Liska <mliska@suse.cz>
19180 * multiple_target.c (create_dispatcher_calls): Make ifunc
19181 also for function that don't have calls or are not referenced.
19183 2017-06-30 Richard Biener <rguenther@suse.de>
19185 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19186 analyze the first scalar stmt. Move vector type computation
19187 for the BB case here from ...
19188 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19189 live operation processing in the SLP case properly.
19191 2017-06-30 Richard Biener <rguenther@suse.de>
19193 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19195 2017-06-30 Martin Liska <mliska@suse.cz>
19198 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19199 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19201 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19203 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19206 2017-06-30 Richard Biener <rguenther@suse.de>
19208 PR tree-optimization/81249
19209 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19210 condition reduction result to original scalar type.
19212 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19214 * profile-count.h (enum profile_quality): Fix typos and whitespace
19217 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19219 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19220 type for branch probabilities.
19222 2017-06-29 Julian Brown <julian@codesourcery.com>
19223 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19225 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19226 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19227 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19228 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19230 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19232 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19233 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19234 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19235 CC usage from generic code to here.
19236 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19237 CC usage into the target macros.
19239 2017-06-29 Maya Rashish <coypu@sdf.org>
19241 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19244 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19246 * arm/arm-builtins.c: Include profile-count.h
19247 * except.c (sjlj_emit_function_enter): Use
19248 profile_probability::unlikely.
19250 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19252 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19253 and tocrel_offset be pointer args rather than implicitly using
19255 (legitimate_constant_pool_address_p, rs6000_emit_move,
19256 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19257 tocrel_offset and use in toc_relative_expr_p call.
19258 (print_operand, print_operand_address): Use static tocrel_base_oac
19259 and tocrel_offset_oac.
19260 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19263 2017-06-29 Maya Rashish <coypu@sdf.org>
19265 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19267 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19269 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19270 objects, take into account only the alignment of 'op0' and 'mode1' if
19273 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19275 * ccmp.c (ccmp_tree_comparison_p): New function.
19276 (ccmp_candidate_p): Update to use above function.
19277 (get_compare_parts): New function.
19278 (expand_ccmp_next): Update to use new functions.
19279 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19281 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19282 take mode as argument.
19283 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19284 * expr.c (expand_expr_real_1): Pass mode as argument.
19286 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19288 * combine.c (combine_instructions): Print insns to dump_file, together
19291 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19293 * asan.c (asan_emit_stack_protection): Update.
19294 (create_cond_insert_point): Update.
19295 * auto-profile.c (afdo_propagate_circuit): Update.
19296 * basic-block.h (struct edge_def): Turn probability to
19297 profile_probability.
19298 (EDGE_FREQUENCY): Update.
19299 * bb-reorder.c (find_traces_1_round): Update.
19300 (better_edge_p): Update.
19301 (sanitize_hot_paths): Update.
19302 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19303 (make_single_succ_edge): Update.
19304 (check_bb_profile): Update.
19305 (dump_edge_info): Update.
19306 (update_bb_profile_for_threading): Update.
19307 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19308 probabilitycount to 0.
19309 * cfgbuild.c (compute_outgoing_frequencies): Update.
19310 * cfgcleanup.c (try_forward_edges): Update.
19311 (outgoing_edges_match): Update.
19312 (try_crossjump_to_edge): Update.
19313 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19314 (expand_gimple_tailcall): Update.
19315 (construct_init_block): Use make_single_succ_edge.
19316 (construct_exit_block): Use make_single_succ_edge.
19317 * cfghooks.c (verify_flow_info): Update.
19318 (redirect_edge_succ_nodup): Update.
19319 (split_edge): Update.
19320 (account_profile_record): Update.
19321 * cfgloopanal.c (single_likely_exit): Update.
19322 * cfgloopmanip.c (scale_loop_profile): Update.
19323 (set_zero_probability): Remove.
19324 (duplicate_loop_to_header_edge): Update.
19325 * cfgloopmanip.h (loop_version): Update prototype.
19326 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19327 (force_nonfallthru_and_redirect): Update.
19328 (update_br_prob_note): Update.
19329 (rtl_verify_edges): Update.
19330 (purge_dead_edges): Update.
19331 (rtl_lv_add_condition_to_bb): Update.
19332 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19333 * cgraphunit.c (init_lowered_empty_function): Update.
19334 (cgraph_node::expand_thunk): Update.
19335 * cilk-common.c: Include profile-count.h
19336 * dojump.c (inv): Remove.
19337 (jumpifnot): Update.
19338 (jumpifnot_1): Update.
19339 (do_jump_1): Update.
19341 (do_jump_by_parts_greater_rtx): Update.
19342 (do_compare_rtx_and_jump): Update.
19343 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19344 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19345 * dwarf2cfi.c: Include profile-count.h
19346 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19347 (sjlj_emit_dispatch_table): Likewise.
19348 * explow.c: Include profile-count.h
19349 * expmed.c (emit_store_flag_force): Update.
19350 (do_cmp_and_jump): Update.
19351 * expr.c (compare_by_pieces_d::generate): Update.
19352 (compare_by_pieces_d::finish_mode): Update.
19353 (emit_block_move_via_loop): Update.
19354 (store_expr_with_bounds): Update.
19355 (store_constructor): Update.
19356 (expand_expr_real_2): Update.
19357 (expand_expr_real_1): Update.
19358 * expr.h (try_casesi, try_tablejump): Update prototypes.
19359 * gimple-pretty-print.c (dump_probability): Update.
19360 (dump_profile): New.
19361 (dump_gimple_label): Update.
19362 (dump_gimple_bb_header): Update.
19363 * graph.c (draw_cfg_node_succ_edges): Update.
19364 * hsa-gen.c (convert_switch_statements): Update.
19365 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
19366 (find_if_case_1): Update.
19367 (find_if_case_2): Update.
19368 * internal-fn.c (expand_arith_overflow_result_store): Update.
19369 (expand_addsub_overflow): Update.
19370 (expand_neg_overflow): Update.
19371 (expand_mul_overflow): Update.
19372 (expand_vector_ubsan_overflow): Update.
19373 * ipa-cp.c (good_cloning_opportunity_p): Update.
19374 * ipa-split.c (split_function): Use make_single_succ_edge.
19375 * ipa-utils.c (ipa_merge_profiles): Update.
19376 * loop-doloop.c (add_test): Update.
19377 (doloop_modify): Update.
19378 * loop-unroll.c (compare_and_jump_seq): Update.
19379 (unroll_loop_runtime_iterations): Update.
19380 * lra-constraints.c (lra_inheritance): Update.
19381 * lto-streamer-in.c (input_cfg): Update.
19382 * lto-streamer-out.c (output_cfg): Update.
19383 * mcf.c (adjust_cfg_counts): Update.
19384 * modulo-sched.c (sms_schedule): Update.
19385 * omp-expand.c (expand_omp_for_init_counts): Update.
19386 (extract_omp_for_update_vars): Update.
19387 (expand_omp_ordered_sink): Update.
19388 (expand_omp_for_ordered_loops): Update.
19389 (expand_omp_for_generic): Update.
19390 (expand_omp_for_static_nochunk): Update.
19391 (expand_omp_for_static_chunk): Update.
19392 (expand_cilk_for): Update.
19393 (expand_omp_simd): Update.
19394 (expand_omp_taskloop_for_outer): Update.
19395 (expand_omp_taskloop_for_inner): Update.
19396 * omp-simd-clone.c (simd_clone_adjust): Update.
19397 * optabs.c (expand_doubleword_shift): Update.
19398 (expand_abs): Update.
19399 (emit_cmp_and_jump_insn_1): Update.
19400 (expand_compare_and_swap_loop): Update.
19401 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
19402 * predict.c (predictable_edge_p): Update.
19403 (edge_probability_reliable_p): Update.
19404 (set_even_probabilities): Update.
19405 (combine_predictions_for_insn): Update.
19406 (combine_predictions_for_bb): Update.
19407 (propagate_freq): Update.
19408 (estimate_bb_frequencies): Update.
19409 (force_edge_cold): Update.
19410 * profile-count.c (profile_count::dump): Add missing space into dump.
19411 (profile_count::debug): Add newline.
19412 (profile_count::differs_from_p): Explicitly convert to unsigned.
19413 (profile_count::stream_in): Update.
19414 (profile_probability::dump): New member function.
19415 (profile_probability::debug): New member function.
19416 (profile_probability::differs_from_p): New member function.
19417 (profile_probability::differs_lot_from_p): New member function.
19418 (profile_probability::stream_in): New member function.
19419 (profile_probability::stream_out): New member function.
19420 * profile-count.h (profile_count_quality): Rename to ...
19421 (profile_quality): ... this one.
19422 (profile_probability): New.
19423 (profile_count): Update.
19424 * profile.c (compute_branch_probabilities): Update.
19425 * recog.c (peep2_attempt): Update.
19426 * sched-ebb.c (schedule_ebbs): Update.
19427 * sched-rgn.c (find_single_block_region): Update.
19428 (compute_dom_prob_ps): Update.
19429 (schedule_region): Update.
19430 * sel-sched-ir.c (compute_succs_info): Update.
19431 * stmt.c (struct case_node): Update.
19432 (do_jump_if_equal): Update.
19433 (get_outgoing_edge_probs): Update.
19434 (conditional_probability): Update.
19435 (emit_case_dispatch_table): Update.
19436 (expand_case): Update.
19437 (expand_sjlj_dispatch_table): Update.
19438 (emit_case_nodes): Update.
19439 * targhooks.c: Update.
19440 * tracer.c (better_p): Update.
19441 (find_best_successor): Update.
19442 * trans-mem.c (expand_transaction): Update.
19443 * tree-call-cdce.c: Update.
19444 * tree-cfg.c (gimple_split_edge): Upate.
19445 (move_sese_region_to_fn): Upate.
19446 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
19447 * tree-eh.c (lower_resx): Upate.
19448 (cleanup_empty_eh_move_lp): Upate.
19449 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19450 * tree-inline.c (copy_edges_for_bb): Update.
19451 (copy_cfg_body): Update.
19452 * tree-parloops.c (gen_parallel_loop): Update.
19453 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
19454 (gimple_gen_time_profiler): Update.
19455 * tree-ssa-dce.c (remove_dead_stmt): Update.
19456 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19457 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
19458 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
19459 (unloop_loops): Update.
19460 (try_peel_loop): Update.
19461 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19462 * tree-ssa-loop-split.c (connect_loops): Update.
19463 (split_loop): Update.
19464 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19465 (hoist_guard): Update.
19466 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
19467 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
19468 (value_replacement): Update.
19469 * tree-ssa-reassoc.c (branch_fixup): Update.
19470 * tree-ssa-tail-merge.c (replace_block_by): Update.
19471 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
19472 (create_edge_and_update_destination_phis): Update.
19473 (compute_path_counts): Update.
19474 (recompute_probabilities): Update.
19475 (update_joiner_offpath_counts): Update.
19476 (freqs_to_counts_path): Update.
19477 (duplicate_thread_path): Update.
19478 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
19479 (struct switch_conv_info): Update.
19480 (gen_inbound_check): Update.
19481 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19482 (vect_do_peeling): Update.
19483 (vect_loop_versioning): Update.
19484 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19485 (optimize_mask_stores): Update.
19486 * ubsan.c (ubsan_expand_null_ifn): Update.
19487 * value-prof.c (gimple_divmod_fixed_value): Update.
19488 (gimple_divmod_fixed_value_transform): Update.
19489 (gimple_mod_pow2): Update.
19490 (gimple_mod_pow2_value_transform): Update.
19491 (gimple_mod_subtract): Update.
19492 (gimple_mod_subtract_transform): Update.
19493 (gimple_ic): Update.
19494 (gimple_stringop_fixed_value): Update.
19495 (gimple_stringops_transform): Update.
19496 * value-prof.h: Update.
19498 2017-06-29 Carl Love <cel@us.ibm.com>
19500 * config/rs6000/rs6000-c.c: Add support for built-in functions
19501 vector signed int vec_signed (vector float);
19502 vector signed long long vec_signed (vector double);
19503 vector signed int vec_signed2 (vector double, vector double);
19504 vector signed int vec_signede (vector double);
19505 vector signed int vec_signedo (vector double);
19506 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
19507 instruction generator.
19508 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19509 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
19510 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
19512 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
19513 vunsignede_v2df): Add define_expands.
19514 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
19515 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
19516 VEC_UNSIGNEDO): Add definitions.
19517 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19518 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
19519 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
19520 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
19521 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
19522 * config/rs6000/altivec.h (vec_signed, vec_signed2,
19523 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
19524 vec_unsignede, vec_unsignedo): Add builtin defines.
19525 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
19527 * doc/extend.texi: Update the built-in documentation file for the
19528 new built-in functions.
19530 2017-06-29 Richard Biener <rguenther@suse.de>
19532 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
19533 reduction chains to LOOP_VINFO_REDUCTIONS.
19534 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
19535 SLP reductions after processing reduction chains.
19537 2017-06-29 Nathan Sidwell <nathan@acm.org>
19539 * builtins.c (fold_builtin_FUNCTION): Use
19540 lang_hooks.decl_printable_name.
19542 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
19544 PR middle-end/81194
19545 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
19546 with only one label.
19547 * stmt.c (expand_case): Assert NCASES is greater than one.
19549 2017-06-29 Richard Biener <rguenther@suse.de>
19551 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
19553 (group_case_labels): Likewise.
19554 (find_taken_edge): Push sanity checking on val to workers...
19555 (find_taken_edge_cond_expr): ... here
19556 (find_taken_edge_switch_expr): ... and here, handle cases
19557 with just a default label.
19558 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
19559 (group_case_labels): Likewise.
19560 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
19561 group_case_labels does anything cleanup the CFG again.
19563 2017-06-29 Bin Cheng <bin.cheng@arm.com>
19565 PR tree-optimization/81196
19566 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
19567 exit condition comparing two IVs.
19569 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
19571 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
19572 profile to the dummy entry at the end of the list of architectures.
19573 * config/arm/arm-cpu-cdata.h: Regenerated.
19575 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19576 Michael Collison <michael.collison@arm.com>
19579 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
19581 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
19582 (*aarch64_reg_<mode>3_minus_mask): New pattern.
19583 (*aarch64_<optab>_reg_di3_mask2): New pattern.
19584 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
19585 of shift when the shift amount is masked with constant equal to
19586 the size of the mode.
19587 * config/aarch64/predicates.md (subreg_lowpart_operator): New
19590 2017-06-29 Martin Liska <mliska@suse.cz>
19592 * config/i386/i386.opt: Change range from [1,5] to [0,5].
19594 2017-06-29 Yury Gribov <tetra2005@gmail.com>
19597 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
19600 (edge_growth_cache_entry::edge_growth_cache_entry): New
19602 (reset_edge_growth_cache): Update to use constructor.
19604 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19606 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
19607 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19608 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19610 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
19612 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
19613 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
19615 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
19617 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
19618 (*-linux-uclibc*): Add t-uclibc tmake_file.
19619 * config/t-musl: New.
19620 * config/t-uclibc: New.
19622 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
19624 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
19626 (gen_comm_data): Emit architectural setting of arch_prof.
19627 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
19629 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
19630 (armv8-m.base, armv8-m.main): Likewise.
19631 * arm-protos.h (arm_build_target): Add profile field.
19632 (arch_option): Likewise.
19633 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
19635 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
19636 arm_active_target.profile.
19638 2017-06-28 Richard Biener <rguenther@suse.de>
19640 PR middle-end/81227
19641 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
19642 TYPE_OVERFLOW_WRAPS.
19643 * match.pd (negate_expr_p): Likewise.
19644 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
19645 fold_build2, not fold_binary.
19647 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19649 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
19650 Convert memory address to Pmode.
19651 (aarch64_print_operand): Assert MEM operands are always Pmode.
19653 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19656 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
19657 Remove redundant if.
19658 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
19659 * config/arm/aarch-common-protos.h
19660 (aarch_forward_to_shift_is_not_shifted_re): Remove.
19661 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
19663 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
19666 * multiple_target.c (create_dispatcher_calls): Set the default
19667 clone to be static, not public.
19669 2017-06-28 Richard Biener <rguenther@suse.de>
19671 * tree-vect-loop.c (vectorizable_reduction): Move special
19672 cond reduction IV var creation ...
19673 (vect_create_epilog_for_reduction): ... here. Remove induction_index
19674 parameter. Use STMT_VINFO_VECTYPE.
19675 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
19678 2017-06-28 Martin Liska <mliska@suse.cz>
19681 * ipa-visibility.c (non_local_p): Handle visibility.
19683 2017-06-28 Martin Liska <mliska@suse.cz>
19686 * common.opt: Add IntegerRange to various options.
19687 * opt-functions.awk (integer_range_info): New function.
19688 * optc-gen.awk: Add integer_range_info to cl_options struct.
19689 * opts-common.c (decode_cmdline_option): Handle
19690 CL_ERR_INT_RANGE_ARG.
19691 (cmdline_handle_error): Likewise.
19692 * opts.c (print_filtered_help): Show valid interval in
19693 when --help is provided.
19694 * opts.h (struct cl_option): Add range_min and range_max fields.
19695 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
19697 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
19699 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
19700 (x * C EQ/NE y * C): New transformation.
19702 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
19704 * genmultilib (combination_space): Accept '+' in option names.
19706 2017-06-28 Martin Liska <mliska@suse.cz>
19709 * asan.c (instrument_derefs): Bail out inner references
19710 that are hard register variables.
19712 2017-06-28 Jakub Jelinek <jakub@redhat.com>
19715 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
19716 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
19718 2017-06-28 Richard Biener <rguenther@suse.de>
19720 * tree-vectorizer.h (vect_get_vec_defs): Remove.
19721 (vect_get_slp_defs): Adjust.
19722 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
19724 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
19726 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
19727 get_initial_defs_for_reduction instead of vect_get_vec_defs.
19728 (vectorizable_reduction): Adjust.
19729 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
19731 (vect_get_slp_defs): Likewise.
19732 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
19733 (vectorizable_bswap): Adjust.
19734 (vectorizable_call): Likewise.
19735 (vectorizable_conversion): Likewise.
19736 (vectorizable_assignment): Likewise.
19737 (vectorizable_shift): Likewise.
19738 (vectorizable_operation): Likewise.
19739 (vectorizable_store): Likewise.
19740 (vectorizable_condition): Likewise.
19741 (vectorizable_comparison): Likewise.
19743 2017-06-28 Michael Collison <michael.collison@arm.com>
19746 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
19748 (arm_gen_movmemqi): Removed unused variable 'i'.
19749 Convert 'for' loop into 'while' loop.
19750 (arm_expand_prologue): Remove last unnecessary set of insn.
19751 (thumb_pop): Remove unused variable 'pushed_words'.
19752 (thumb_exit): Remove last unnecessary set of regs_to_pop.
19754 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19756 * config/s390/predicates.md: Use s390_rel_address_ok_p.
19757 * config/s390/s390-protos.h: Add prototype of
19758 s390_rel_address_ok_p.
19759 * config/s390/s390.c (s390_got_symbol): New function.
19760 (s390_rel_address_ok_p): New function.
19761 (legitimize_pic_address): Use s390_rel_address_ok_p.
19762 (s390_load_got): Use s390_got_symbol.
19763 (s390_option_override): Issue error if
19764 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
19765 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
19767 * config/s390/s390.opt: New option mpic-data-is-text-relative.
19769 2017-06-27 Andrew Pinski <apinski@cavium.com>
19771 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
19772 (X * copysign (1.0, X)): New pattern.
19773 (X * copysign (1.0, -X)): New pattern.
19774 (copysign (-1.0, CST)): New pattern.
19776 2017-06-27 Joseph Myers <joseph@codesourcery.com>
19778 * genmultilib (combination_space): Remove variable.
19779 Validate reuse rules against regular expression for any sequence
19780 of multilib options in any order.
19782 2017-06-27 Michael Collison <michael.collison@arm.com>
19784 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
19785 call aarch64_split_simd_combine.
19786 * (aarch64_combine_internal<mode>): Delete pattern.
19787 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
19788 Allow register and subreg operands.
19790 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19792 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
19793 specific need, just fallback on defaults.
19794 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
19796 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19797 Olivier Hainque <hainque@adacore.com>
19799 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
19801 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
19802 targets. Pick a default if no particular attempt applied.
19803 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
19806 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19808 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
19809 (x86_64-wrs-vxworks7): Likewise.
19811 2017-06-27 Marek Polacek <polacek@redhat.com>
19814 * ubsan.c (instrument_null): Check get_base_address's result for null.
19816 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19818 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
19820 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19822 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
19823 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
19824 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
19825 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
19826 New function types.
19827 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
19828 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
19829 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
19830 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
19831 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
19832 BUILT_IN_FEUPDATEENV): New builtins.
19833 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
19834 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
19835 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
19836 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
19838 (builtin_structptr_types): Adjust size.
19839 * tree.c (builtin_structptr_types): Add four entries.
19841 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19842 Olivier Hainque <hainque@adacore.com>
19844 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
19845 (TLS_SYM): New local macro, forcing reference to __tls__ on
19846 link command lines for VxWorks 7 RTPs, triggering initialization
19848 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
19849 OS features TLS support, true for RTPs on VxWorks 7.
19850 * config/vxworks.c (vxworks_override_options): Setup emutls
19853 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19855 * predict.c (test_prediction_value_range): Use -1U instead of -1
19856 to avoid narrowing conversion warning.
19857 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
19858 to avoid narrowing conversion warning.
19859 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
19861 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
19863 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19865 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
19866 64bit configurations.
19867 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
19868 (SIZE_TYPE): Likewise.
19869 * config/vxworks.c (vxworks_emutls_var_fields): Use
19870 long_unsigned_type_node instead of unsigned_type_node as the offset
19871 field type, which is "pointer" mode in emutls.c.
19873 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19876 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
19878 PR middle-end/81207
19879 * gimple-fold.c (replace_call_with_call_and_fold): Handle
19880 gimple_vuse copying separately from gimple_vdef copying.
19882 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19884 * value-prof.c (free_hist): Remove call to memset and the enclosing if
19887 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19888 Olivier Hainque <hainque@adacore.com>
19890 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
19891 for all vxworks7 targets.
19892 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
19893 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
19894 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
19895 variations for VX6/VX7 and 32/64bits later on in ...
19896 (VXWORKS_LIB_SPEC): Leverage new macros.
19897 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
19898 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
19900 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19902 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
19903 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
19905 2017-06-26 Carl Love <cel@us.ibm.com>
19907 * config/rs6000/rs6000-c.c: Add support for built-in functions
19908 vector bool char vec_reve (vector bool char);
19909 vector signed char vec_reve (vector signed char);
19910 vector unsigned char vec_reve (vector unsigned char);
19911 vector bool int vec_reve (vector bool int);
19912 vector signed int vec_reve (vector signed int);
19913 vector unsigned int vec_reve (vector unsigned int);
19914 vector bool long long vec_reve (vector bool long long);
19915 vector signed long long vec_reve (vector signed long long);
19916 vector unsigned long long vec_reve (vector unsigned long long);
19917 vector bool short vec_reve (vector bool short);
19918 vector signed short vec_reve (vector signed short);
19919 vector double vec_reve (vector double);
19920 vector float vec_reve (vector float);
19921 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
19922 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
19923 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
19924 (altivec_vreve): New pattern.
19925 * config/rs6000/altivec.h (vec_reve): New define.
19926 * doc/extend.texi (vec_rev): Update the built-in documentation file
19927 for the new built-in functions.
19929 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19931 PR tree-optimization/71815
19932 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
19934 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
19936 (slsr_process_phi): Likewise.
19937 (replace_uncond_cands_and_profitable_phis): Don't replace a
19938 multiply candidate with a stride of 1 (copy or cast).
19939 (phi_incr_cost): Call uses_consumed_by_stmt rather than
19941 (lowest_cost_path): Likewise.
19942 (total_savings): Likewise.
19944 2017-06-26 Richard Biener <rguenther@suse.de>
19947 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
19948 Use def_builtin_pure for all gather builtins.
19950 2017-06-26 Richard Biener <rguenther@suse.de>
19952 PR tree-optimization/81203
19953 * tree-tailcall.c (find_tail_calls): Do not move stmts into
19954 non-dominating BBs.
19956 2017-06-26 Marek Polacek <polacek@redhat.com>
19959 * doc/invoke.texi: Document -Wmultistatement-macros.
19961 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
19963 * doc/sourcebuild.texi (ARM-specific attributes): Document new
19964 arm_neon_ok_no_float_abi effective target.
19966 2017-06-26 Richard Biener <rguenther@suse.de>
19968 PR tree-optimization/80928
19969 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
19970 (copy_bbs): Set BB_DUPLICATED flag early.
19971 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
19973 (execute_on_shrinking_pred): Likewise.
19974 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
19975 BB_DUPLICATED blocks.
19976 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
19977 iterate over all PHIs considering removal of *gsi.
19979 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
19981 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
19984 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19986 * config/rs6000/rs6000-string.c: (expand_block_clear,
19987 do_load_for_compare, select_block_compare_mode,
19988 compute_current_alignment, expand_block_compare,
19989 expand_strncmp_align_check, expand_strn_compare,
19990 expand_block_move, rs6000_output_load_multiple)
19991 Move functions related to string/block move/compare
19992 to a separate file.
19993 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
19994 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
19995 for this function which is now used in two files.
19996 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
19997 * config.gcc: Add rs6000-string.o to extra_objs for
19998 targets powerpc*-*-* and rs6000*-*-*.
20000 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
20003 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
20004 32-bit, since indexed is not valid for DImode.
20005 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
20006 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
20007 (define_peephole2 for Altivec d-form load): Add 32-bit support.
20008 (define_peephole2 for Altivec d-form store): Likewise.
20011 * multiple_target.c (create_dispatcher_calls): Only create the
20012 dispatcher call if the function is the default clone of a
20013 versioned function.
20015 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20017 PR middle-end/80902
20018 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20019 a call, force the call to not be a tail call.
20021 2017-06-23 Jeff Law <law@redhat.com>
20023 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20026 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20028 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20029 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20030 handling for early expansion of vector shifts (sl,sr,sra,rl).
20031 (builtin_function_type): Add vector shift right instructions
20032 to the unsigned argument list.
20034 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20036 rtl-optimizatoin/79286
20037 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20038 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20039 trap. PIC register plus a const unspec without offset can never trap.
20041 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20043 * tree.h (builtin_structptr_type): New type.
20044 (builtin_structptr_types): Declare new array.
20045 * tree.c (builtin_structptr_types): New array.
20046 (free_lang_data, build_common_tree_nodes): Use it.
20048 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20051 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20054 2017-06-22 Matt Turner <mattst88@gmail.com>
20056 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20057 Lake models to skylake case. Assume skylake for unknown
20058 models with clflushopt.
20060 2017-06-22 Jeff Law <law@redhat.com>
20062 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20063 frame sizes that do not satisfy aarch64_uimm12_shift.
20065 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20067 * profile-count.h (apply_probability,
20068 apply_scale, probability_in): Fix checks for zero.
20070 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20072 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20073 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20075 2016-06-22 Richard Biener <rguenther@suse.de>
20077 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20078 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20079 REDUC_MAX_EXPR support.
20080 (vectorizable_reduction): Likewise.
20081 (vect_create_epilog_for_reduction): Likewise.
20083 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20085 * match.pd (A / (1 << B) -> A >> B): New.
20086 * generic-match-head.c: Include optabs-tree.h.
20087 * gimple-match-head.c: Likewise.
20088 * optabs-tree.h (target_supports_op_p): New.
20089 * optabs-tree.c (target_supports_op_p): New.
20091 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20093 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20094 $gcc_cv_ld --help output.
20095 (gcc_cv_ld_demangle): Likewise.
20096 (gcc_cv_ld_eh_frame_hdr): Likewise.
20097 (gcc_cv_ld_pie): Likewise.
20098 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20099 (gcc_cv_ld_buildid): Likewise.
20100 (gcc_cv_ld_sysroot): Likewise.
20101 (ld_bndplt_support): Likewise.
20102 (ld_pushpopstate_support): Likewise.
20103 * configure: Regenerate.
20104 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20106 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20109 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20110 operands indexes to avoid gap between operands and match_dups.
20112 2017-06-21 Andrew Pinski <apinski@cavium.com>
20114 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20115 Increment Arith_shift and Arith_shift_reg by 1.
20116 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20118 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20119 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20120 (aarch64_strip_extend): Add new argument and test for it.
20121 (aarch64_cheap_mult_shift_p): New function.
20122 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20123 add a cost if it is true.
20124 Update calls to aarch64_strip_extend.
20125 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20127 2017-06-21 Andrew Pinski <apinski@cavium.com>
20129 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20131 (thunderxt88): Likewise.
20132 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20133 (thunderx_prefetch_tune): New variable.
20134 (thunderx2t99_prefetch_tune): Update for the correct values.
20135 (thunderxt88_tunings): New variable.
20136 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20137 generic_prefetch_tune.
20138 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20140 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20142 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20143 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20144 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20145 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20146 (aarch64_atomic_cas<mode>, GPI): Likewise.
20148 2017-06-21 Martin Liska <mliska@suse.cz>
20150 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20151 statements on cold and hot labels.
20152 * predict.c (tree_estimate_probability_bb): Remove the
20153 prediction from this place.
20155 2017-06-21 Martin Liska <mliska@suse.cz>
20157 PR tree-optimization/79489
20158 * gimplify.c (maybe_add_early_return_predict_stmt): New
20160 (gimplify_return_expr): Call the function.
20161 * predict.c (tree_estimate_probability_bb): Remove handling
20163 * predict.def: Update comment about early return predictor.
20164 * gimple-predict.h (is_gimple_predict): New function.
20165 * predict.def: Change default value of early return to 66.
20166 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20168 * passes.def: Put pass_strip_predict_hints to the beginning of
20171 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20173 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20174 FUNCTION_DECL declarations.
20175 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20177 (dwaf2out_decl): Likewise.
20178 * godump.c (go_early_global_decl): Skip call to the real debug hook
20179 for FUNCTION_DECL declarations.
20180 * passes.c (rest_of_decl_compilation): Skip call to the
20181 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20182 -fdump-go-spec is passed.
20184 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20186 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20187 Reorder for compactness.
20188 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20189 (def_builtin_pure, def_builtin_pure2): New functions.
20190 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20192 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20194 * match.pd (nop_convert): New predicate.
20195 ((A +- CST1) +- CST2): Allow some NOP conversions.
20197 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20200 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20201 with ctors/dtors if GOVD_SHARED is set.
20203 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20205 * config/aarch64/aarch64.md (movti_aarch64):
20206 Emit mov rather than orr.
20207 (movtf_aarch64): Likewise.
20208 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20209 Emit mov rather than orr.
20211 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20213 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20214 Swap alternatives, make integer dup more expensive.
20216 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20218 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20219 Return true for non-tls symbols.
20221 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20223 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20224 (cortex-a75): Likewise.
20225 (cortex-a75.cortex-a55): Likewise.
20226 * config/aarch64/aarch64-tune.md: Regenerate.
20227 * doc/invoke.texi (-mtune): Document new values for -mtune.
20229 2017-06-21 Tom de Vries <tom@codesourcery.com>
20231 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20232 stack_size feature.
20233 (Effective-Target Keywords, Other attributes): Suggest using
20234 dg-add-options stack_size feature to get stack limit in stack_size
20235 effective target documentation.
20237 2017-06-21 Julian Brown <julian@codesourcery.com>
20238 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20240 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20241 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20242 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20244 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20245 attribute type list for neon_multiply.
20246 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20247 attribute type list for neon_multiply.
20248 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20249 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20250 attribute type list for neon_multiply.
20251 * config/arm/types.md (crypto_pmull): Add.
20252 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20253 attribute type list.
20255 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20257 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20260 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20262 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20263 to make sure not to dereference a NULL cost_classes_ptr pointer.
20265 2017-06-20 Carl Love <cel@us.ibm.com>
20267 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20268 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20269 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20270 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20271 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20272 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20273 VMULOSW): New enum "unspec" values.
20274 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20275 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20276 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20277 altivec_vmulosw): New patterns.
20278 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20279 VMULOSW): Add definitions.
20281 2017-06-20 Julia Koval <julia.koval@intel.com>
20283 * config/i386/i386.c: Fix rounding expand for new pattern.
20284 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20286 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20288 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20289 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20291 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20293 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20296 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20298 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20299 architecture, then by implementer ID.
20300 * config/aarch64/aarch64-tune.md: Regenerate.
20302 2017-06-20 Richard Biener <rguenther@suse.de>
20304 PR middle-end/81097
20305 * fold-const.c (split_tree): Fold to type before negating.
20307 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20309 * diagnostic-show-locus.c
20310 (selftest::test_fixit_deletion_affecting_newline): New function.
20311 (selftest::diagnostic_show_locus_c_tests): Call it.
20313 2017-06-20 Andreas Schwab <schwab@suse.de>
20316 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20319 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20321 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20322 __ARM_FEATURE_COPROC according to support.
20324 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20326 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20327 Rewritten to avoid overflow for > 32-bit pointers.
20330 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20331 by removing enum keyword.
20332 (ubsan_type_descriptor): Likewise. Formatting fix.
20335 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20336 splitter): Require TARGET_SSE2 in the condition.
20338 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20341 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20342 for doing vector set of SFmode on ISA 3.0.
20343 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20344 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20346 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20347 SFmode value into a V4SF variable that was extracted from another
20348 V4SF variable without converting the element to double precision
20349 and back to single precision vector format.
20350 (vsx_insert_extract_v4sf_p9_2): Likewise.
20352 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20354 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20355 in UWHI to avoid undefined overflow.
20358 * ubsan.h (enum ubsan_encode_value_phase): New.
20359 (ubsan_encode_value): Change second argument to
20360 enum ubsan_encode_value_phase with default value of
20361 UBSAN_ENCODE_VALUE_GENERIC.
20362 * ubsan.c (ubsan_encode_value): Change second argument to
20363 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
20364 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
20365 create_tmp_var_raw instead of create_tmp_var and use a
20367 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
20368 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
20369 ubsan_encode_value callers.
20372 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
20373 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
20374 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
20376 2017-06-19 Richard Biener <rguenther@suse.de>
20378 PR middle-end/81118
20379 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20380 estimates if we changed anything.
20382 2017-06-19 Richard Biener <rguenther@suse.de>
20384 PR tree-optimization/80887
20385 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
20386 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
20387 simplified lookups, then reset mprts_hook.
20388 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
20390 (try_to_simplify): Likewise.
20392 2017-06-19 Martin Liska <mliska@suse.cz>
20395 * gimplify.c (gimplify_switch_expr):
20396 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
20398 2017-06-19 Martin Liska <mliska@suse.cz>
20400 * doc/install.texi: Document that PGO runs in 4 stages.
20402 2017-06-19 Martin Liska <mliska@suse.cz>
20405 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
20406 to dispatcher function name.
20407 * multiple_target.c (replace_function_decl): New function.
20408 (create_dispatcher_calls): Redirect both edges and references.
20410 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
20412 * profile-count.c (profile_count::dump): Dump quality.
20413 (profile_count::differs_from_p): Update for unsigned val.
20414 * profile-count.h (profile_count_quality): New enum.
20415 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
20417 2017-06-19 Richard Biener <rguenther@suse.de>
20419 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
20420 struct function as arg.
20421 (estimate_numbers_of_iterations): Export overload with loop arg.
20422 (free_numbers_of_iterations_estimates_loop): Use an overload of
20423 free_numbers_of_iterations_estimates instead.
20424 * tree-cfg.c (remove_bb): Adjust.
20425 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
20426 * tree-parloops.c (gen_parallel_loop): Likewise.
20427 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
20429 (tree_unroll_loops_completely): Likewise.
20430 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
20431 Use an overload instead and export.
20432 (estimated_loop_iterations): Adjust.
20433 (max_loop_iterations): Likewise.
20434 (likely_max_loop_iterations): Likewise.
20435 (estimate_numbers_of_iterations): Take struct function as arg
20437 (loop_exits_before_overflow): Adjust.
20438 (free_numbers_of_iterations_estimates_loop): Use an overload.
20439 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
20440 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
20442 2017-06-19 Richard Biener <rguenther@suse.de>
20445 * ipa-prop.c (find_constructor_constant_at_offset): Handle
20446 RANGE_EXPR conservatively.
20448 2017-06-16 Carl Love <cel@us.ibm.com>
20450 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20451 definitions for vec_float, vec_float2, vec_floato,
20452 vec_floate built-ins.
20453 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
20454 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
20456 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
20457 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
20458 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
20459 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
20460 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
20461 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
20462 vec_floato): Add builtin defines.
20463 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
20464 Update the built-in documentation file for the new built-in
20467 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20469 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
20470 (mthumb): Mark as the negative of -marm.
20472 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20474 * doc/invoke.texi (ARM Options, -mcpu): Document supported
20476 (ARM Options, -mtune): Document that this accepts the same
20477 extension options as -mcpu.
20478 (ARM Options, -mfpu): Document addition of -mfpu=auto.
20480 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20482 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
20483 permitted extensions.
20485 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20487 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
20488 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
20489 (armv8-m.main): Add option +nodsp.
20490 * config/arm/arm-cpu-cdata.h: Regenerated.
20492 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20494 * config/arm/t-fuchsia: New file.
20495 * config.gcc (arm*-*-fuchsia*): Use it.
20497 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20499 * config/arm/t-symbian: Rewrite for new option infrastructure.
20501 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20503 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
20504 (MULTILIB_REQUIRED): Likewise.
20506 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20508 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
20509 (MULTILIB_RESUE): Likewise.
20510 (MULTILIB_MATCHES): Likewise.
20511 (MULTLIB_REQUIRED): Likewise.
20513 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20515 * config/arm/t-rtems: Rewrite for new option framework.
20517 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20519 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
20520 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
20521 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
20522 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
20523 * config/arm/t-multilib: ... here.
20524 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
20525 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
20526 armv7-a and armv8*-a when A-profile libraries have not been built.
20527 * config/arm/t-rmprofile: Rewrite.
20529 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20531 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
20532 with a backslash. Remove the backslash after substituting unescaped
20534 * doc/fragments.texi (MULTILIB_REUSE): Document it.
20536 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20538 * config.gcc: (arm*-*-*): When building a-profile libraries, force
20539 the driver to pass through the default setting of -mfloat-abi.
20540 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
20542 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
20543 (all_feat_combs): New rule.
20544 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
20546 * config/arm/t-aprofile: Rewrite.
20548 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20550 * config/arm/arm.h (FPUTYPE_AUTO): Define.
20551 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
20552 fpu is not specified by the user/command-line.
20553 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
20554 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
20555 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
20556 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
20557 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
20558 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
20560 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20562 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
20563 * config/arm/t-arm-elf: Rewritten.
20565 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20567 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
20568 have some floating-point instructions.
20569 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
20570 (TARGET_MAYBE_HARD_FLOAT): New macro.
20571 * config/arm/arm-builtins.c (arm_init_builtins): Use
20572 TARGET_MAYBE_HARD_FLOAT.
20573 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
20575 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20577 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
20578 (configargs.h): Include it.
20579 (arm_print_hint_for_fpu_option): New function.
20580 (arm_parse_fpu_option): New function.
20581 (candidate_extension): New class.
20582 (arm_canon_for_multilib): New function.
20583 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
20584 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
20585 (ARCH_CANONICAL_SPECS): New macro.
20586 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
20588 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20590 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
20591 are set after handling multilib fragments. Set target_cpu_default2
20594 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20596 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
20598 (arm*-*-*): Set target_cpu_default2 to a quoted string.
20599 * config/arm/parsecpu.awk (check_cpu): Validate any extension
20601 (check_arch): Likewise.
20602 * config/arm/arm.c (arm_configure_build_target): Handle
20603 TARGET_CPU_DEFAULT being a string constant. Scan any feature
20604 options in the default.
20606 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20608 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
20609 when an option is an alias of another.
20610 * config/arm/parsecpu.awk (optalias): New parser token.
20611 (gen_comm_data): Mark non-alias options as such. Emit entries
20612 for extension aliases.
20613 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
20614 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
20615 (armv6kz, armv6zk, armv6t2): Likewise.
20616 (armv7): Make vfpv3-d16 an alias.
20617 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
20619 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
20620 Sort in canonical order.
20621 (armv8-a): Sort in canonical order.
20622 (armv8.1-a, armv8.2-a): Likewise.
20623 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
20625 (cortex-a9): Sort in canonical order.
20626 * config/arm/arm.c (selftests.h): Include it.
20627 (arm_test_cpu_arch_data): New function.
20628 (arm_run_self_tests): New function.
20629 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
20630 (targetm): Move declaration to the end of the file.
20631 * arm-cpu-cdata.h: Regenerated.
20633 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20635 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
20636 call to target_mode_check describing the type of option passed.
20637 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
20638 (arm_target_thumb_only): Use arm_parse_arch_option_name or
20639 arm_parse_cpu_option_name to match parameters against list of
20641 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
20642 arm_arch_core_flags data structure.
20643 * config/arm/arm-cpu_cdata.h: Regenerated.
20645 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20647 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
20649 (arm_print_hint_for_cpu_option): Likewise.
20650 (arm_print_hint_for_arch_option): Likewise.
20651 (arm_parse_cpu_option_name): Likewise.
20652 (arm_parse_arch_option_name): Likewise.
20653 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
20654 of entries in the all_fpus list.
20655 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
20656 (arm_parse_cpu_option_name): Declare.
20657 (arm_parse_arch_option_name): Declare.
20658 (arm_parse_option_features): Declare.
20659 (arm_intialize_isa): Declare.
20660 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
20662 (gen_comm_data): ... here. Make definitions non-static.
20663 * config/arm/arm-cpu-data.h: Regenerated.
20664 * config/arm/arm-cpu-cdata.h: Regenerated.
20666 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20668 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
20669 (cpu_arch_extension): New structure.
20670 (cpu_arch_option, arch_option, cpu_option): New structures.
20671 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
20672 architecture types.
20673 (gen_data): Generate new format data tables.
20674 * config/arm/arm.c (cpu_tune): New structure.
20675 (cpu_option, processors): Delete.
20676 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
20677 (arm_print_hint_for_cpu_option): ... this and ...
20678 (arm_print_hint_for_arch_option): ... this.
20679 (arm_parse_arch_cpu_name): Delete. Replace with ...
20680 (arm_parse_cpu_option_name): ... this and ...
20681 (arm_parse_arch_option_name): ... this.
20682 (arm_unrecognized_feature): Change type of target parameter to
20684 (arm_parse_arch_cpu_features): Delete. Replace with ...
20685 (arm_parse_option_features): ... this.
20686 (arm_configure_build_target): Rework to use new configuration data
20688 (arm_print_tune_info): Rework for new configuration data tables.
20689 * config/arm/arm-cpu-data.h: Regenerated.
20690 * config/arm/arm-cpu.h: Regenerated.
20692 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20694 * Makefile.in (OBJS): Move sbitmap.o from here ...
20695 (OBJS-libcommon): ... to here.
20697 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20699 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
20700 (ISA_ALL_CRYPTO): New macro.
20701 (ISA_ALL_SIMD): New macro
20702 (ISA_ALL_FP): New macro.
20703 * config/arm/arm.c (fpu_bitlist): Update initializer.
20704 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
20706 (arm9e): Add fpu. Add option for nofp
20707 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
20708 (arm926ej-s, arm1026ej-s): Likewise.
20709 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
20710 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
20711 neon-fp16, neon-vfpv4, nofp and nosimd.
20712 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
20713 (cortex-a8): Add fpu. Add option for nofp.
20714 (cortex-a9): Add fpu. Add options for nosimd and nofp.
20715 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
20716 (cortex-r4f): Add fpu.
20717 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
20718 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
20720 (cortex-r8): Likewise.
20721 (cortex-m4): Add fpu. Add option for nofp.
20722 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
20723 (cortex-a17.cortex-a7): Likewise.
20724 (cortex-a32): Add fpu. Add options for crypto and nofp.
20725 (cortex-a35, cortex-a53): Likewise.
20726 (cortex-a57): Add fpu. Add option for crypto.
20727 (cortex-a72, cortex-a73): Likewise.
20728 (exynos-m1): Likewise.
20729 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
20730 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
20731 (cortex-m33): Add fpu. Add option for nofp.
20732 * config/arm/arm-cpu-cdata.h: Regenerated
20733 * config/arm/arm-cpu-data.h: Regenerated.
20735 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20737 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
20738 (armv5te, armv5tej): Likewise.
20739 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
20740 (armv7): Add options fp and vfpv3-d16.
20741 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
20742 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
20744 (armv7ve): Likewise.
20745 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
20746 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
20747 (armv8-a): Add nocrypto option.
20748 (armv8.1-a, armv8.2-a): Likewise.
20749 (armv8-m.main): add options fp, fp.dp and nofp.
20751 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20753 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
20755 (armv8-a+crc): Delete.
20756 (armv8.1-a): Add options simd, crypto and nofp.
20757 (armv8.2-a): Add options fp16, simd, crypto and nofp.
20758 (armv8.2-a+fp16): Delete.
20759 (armv8-m.main): Add option dsp.
20760 (armv8-m.main+dsp): Delete.
20761 (cortex-a8): Add fpu. Add nofp option.
20762 (cortex-a9): Add fpu. Add nofp and nosimd options.
20763 * config/arm/parsecpu.awk (gen_data): Generate option tables and
20764 link to main cpu and architecture data structures.
20765 (gen_comm_data): Only put isa attributes from the main architecture
20767 (option): New statement for architecture and CPU entries.
20768 * arm.c (struct cpu_option): New structure.
20769 (struct processors): Add entry for options.
20770 (arm_unrecognized_feature): New function.
20771 (arm_parse_arch_cpu_name): Ignore any characters after the first
20773 (arm_parse_arch_cpu_feature): New function.
20774 (arm_configure_build_target): Separate out any CPU and architecture
20775 features and parse separately. Don't error out if -mfpu=auto is
20776 used with only an architecture string.
20777 (arm_print_asm_arch_directives): New function.
20778 (arm_file_start): Call it.
20779 * config/arm/arm-cpu-cdata.h: Regenerated.
20780 * config/arm/arm-cpu-data.h: Likewise.
20781 * config/arm/arm-tables.opt: Likewise.
20783 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20785 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
20786 assembler when it is not -mfpu=auto.
20788 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20790 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
20791 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
20792 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
20793 (ASM_CPU_SPEC): Rewrite.
20794 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
20795 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
20796 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
20797 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
20798 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
20799 copied string is NUL-terminated. Also strip any characters prefixed
20801 (arm_rewrite_selected_arch): New function.
20802 (arm_rewrite_march): New function.
20804 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20806 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
20807 (x_arm_cpu_string, x_arm_tune_string): Likewise.
20808 (march, mcpu, mtune): Convert to string-based options.
20809 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
20810 (arm_parse_arch_cpu_name): New function.
20811 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
20812 identify selected architecture or CPU.
20813 (arm_option_save): New function.
20814 (TARGET_OPTION_SAVE): Redefine.
20815 (arm_option_restore): Restore string options.
20816 (arm_option_print): Print string options.
20818 2017-06-16 Martin Sebor <msebor@redhat.com>
20820 PR tree-optimization/80933
20821 PR tree-optimization/80934
20822 * builtins.c (fold_builtin_3): Do not handle bcmp here.
20823 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
20824 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
20825 (gimple_fold_builtin): Call them.
20827 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20829 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
20830 as unlikely; update profile.
20832 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20834 * predict.c (force_edge_cold): Handle declaring edges impossible
20837 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20839 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
20841 (try_unroll_loop_completely): Fix reporting.
20843 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20845 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
20847 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
20850 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
20851 if given a non-constant argument for an intrinsic which requires a
20854 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20856 * profile.c (compare_freqs): New function.
20857 (branch_prob): Sort edge list.
20858 (find_spanning_tree): Assume that the list is priority sorted.
20860 2017-06-16 Richard Biener <rguenther@suse.de>
20862 PR tree-optimization/81090
20863 * passes.def (pass_record_bounds): Remove.
20864 * tree-pass.h (make_pass_record_bounds): Likewise.
20865 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
20866 make_pass_record_bounds): Likewise.
20867 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
20868 not free niter estimates at the beginning but at the end.
20869 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
20871 2017-06-16 Richard Biener <rguenther@suse.de>
20873 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
20874 initializer to workaround ICE in host GCC 4.8.
20876 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20878 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
20880 (clone_inlined_nodes): Update.
20882 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20884 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
20885 prefetch settings, and enable prefetching by default at -O3.
20887 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20889 * config/aarch64/aarch64.c (aarch64_override_options_internal):
20890 Set flag_prefetch_loop_arrays according to tuning data.
20892 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20894 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
20895 New tune structure.
20896 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
20897 [Unrelated to main purpose of the patch] Place the pointer field last
20898 to enable type checking errors when tune structure are wrongly merged.
20899 * config/aarch64/aarch64.c (generic_prefetch_tune,)
20900 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
20901 (thunderx2t99_prefetch_tune): New tune constants.
20902 (tune_params *_tunings): Update all tunings (no functional change).
20903 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
20904 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
20905 from tunings structures.
20907 2017-06-16 Jakub Jelinek <jakub@redhat.com>
20910 * ubsan.c (instrument_null): Add T argument, use it instead
20911 of computing it based on IS_LHS.
20912 (instrument_object_size): Likewise.
20913 (pass_ubsan::execute): Adjust instrument_null and
20914 instrument_object_size callers to pass gimple_get_lhs or
20915 gimple_assign_rhs1 result to it. Use instrument_null instead of
20916 calling get_base_address and instrument_mem_ref. Handle
20917 aggregate call arguments for object-size sanitization.
20919 2017-06-16 Yury Gribov <tetra2005@gmail.com>
20921 PR tree-optimization/81089
20922 * tree-vrp.c (is_masked_range_test): Validate operands of
20925 2017-06-15 Martin Sebor <msebor@redhat.com>
20928 * dumpfile.c (dump_register): Avoid calling memset to initialize
20929 a class with a default ctor.
20930 * gcc.c (struct compiler): Remove const qualification.
20931 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
20932 * hash-table.h: Ditto.
20933 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
20935 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
20936 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
20938 * params.h (struct param_info): Make struct members non-const.
20939 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
20940 with default initialization.
20941 * vec.h (vec_copy_construct, vec_default_construct): New helper
20943 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
20944 with vec_copy_construct.
20945 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
20946 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
20947 * doc/invoke.texi (-Wclass-memaccess): Document.
20949 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20951 * emit-rtl.h (is_leaf): Update comment about local
20952 register allocator.
20954 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20957 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
20958 for a variable to have a section before checking if the section has a
20960 Set section to.persistent if persistent attribute is set.
20961 Warn if .persistent attribute is used on an automatic variable.
20963 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
20965 PR rtl-optimization/80474
20966 * reorg.c (update_block): Do not ignore instructions in a delay slot.
20968 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
20970 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
20973 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
20975 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
20976 (casesi): Emit bounds checking as RTL.
20977 (casesi_internal_mips16_<mode>): Remove bounds checking.
20979 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
20981 * config/xtensa/xtensa.c (xtensa_option_override): Append
20982 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
20983 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
20984 xtensa_doloop_hooks): Define unconditionally.
20985 (xtensa_reorg_loops): Only call reorg_loops in the presence of
20987 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
20988 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
20989 for it in xtensa_option_override.
20990 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
20991 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
20993 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
20995 * doc/cppopts.texi: Document '-' special value to -MF.
20997 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
20999 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
21000 (cortex_a53_fconst): Likewise.
21001 (cortex_a53_fpmul): Likewise.
21002 (cortex_a53_f_load_64): Likewise.
21003 (cortex_a53_f_load_many): Likewise.
21004 (cortex_a53_advsimd_alu): Likewise.
21005 (cortex_a53_advsimd_alu_q): Likewise.
21006 (cortex_a53_advsimd_mul): Likewise.
21007 (cortex_a53_advsimd_mul_q): Likewise.
21008 (fpmac bypass): Add new bypass for fpmac-fpmac case.
21009 Add missing fmul, r2f_cvt and fconst cases.
21011 2017-06-14 Richard Biener <rguenther@suse.de>
21013 PR middle-end/81088
21014 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21016 (fold_binary_loc): When associating do not treat pre-existing
21017 TREE_OVERFLOW on literal constants as a reason to allow
21018 TREE_OVERFLOW on associated literal constants.
21020 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21022 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21023 (MASK_FEATURES): New macro.
21024 * config/sparc/sparc.c (sparc_option_override): Remove the special
21025 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21027 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21029 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21030 a division of 0 if non-call exceptions are enabled.
21032 2017-06-14 Andrew Pinski <apinski@cavium.com>
21033 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21036 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21037 Improve vector initialization code gen for only variable case.
21039 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21041 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21043 2017-06-14 Richard Biener <rguenther@suse.de>
21045 PR tree-optimization/81083
21046 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21049 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21051 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21052 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21053 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21054 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21055 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21056 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21058 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21060 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21061 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21063 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21065 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21067 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21069 * config/rs6000/t-rtems: Don't handle SPE.
21071 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21073 * config/rs6000/t-linux: Don't handle SPE.
21075 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21077 * config/rs6000/eabispe.h: Delete file.
21079 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21081 * config/rs6000/t-spe: Delete file.
21083 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21085 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21086 (rs6000_legitimate_offset_address_p): Return false for anything in
21087 V2SImode or V2SFmode.
21089 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21091 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21092 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21093 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21095 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21096 (rs6000_legitimate_offset_address_p): Ditto.
21097 (rs6000_emit_move): Ditto.
21098 (rs6000_init_builtins): Remove V4HI_type_node.
21100 2017-06-13 Martin Liska <mliska@suse.cz>
21103 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21104 (gate_asan): Likewise.
21105 * asan.h (asan_no_sanitize_address_p): Remove the function.
21106 (sanitize_flags_p): New function.
21107 * builtins.def: Fix coding style.
21108 * common.opt: Use renamed enum value.
21109 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21110 * doc/extend.texi: Document no_sanitize attribute.
21111 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21112 to SANITIZE_UNDEFINED_NONDEFAULT.
21113 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21114 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21115 Use sanitize_flags_p.
21116 * gimplify.c (gimplify_function_tree): Likewise.
21117 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21118 * opts.c (parse_no_sanitize_attribute): New function.
21119 (common_handle_option): Use renamed enum value.
21120 * opts.h (parse_no_sanitize_attribute): Declare.
21121 * tree.c (sanitize_flags_p): New function.
21122 * tree.h: Declared here.
21123 * tsan.c: Use sanitize_flags_p.
21124 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21125 (instrument_mem_ref): Likewise.
21126 (instrument_bool_enum_load): Likewise.
21127 (do_ubsan_in_current_function): Remove the function.
21128 (pass_ubsan::execute): Use sanitize_flags_p.
21129 * ubsan.h: Remove do_ubsan_in_current_function
21130 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21131 (dump_function_to_file): Use it here.
21133 2017-06-13 Martin Jambor <mjambor@suse.cz>
21135 PR tree-optimization/80803
21136 PR tree-optimization/81063
21137 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21138 (propagate_subaccesses_across_link): Enqueue subtree whenever
21139 necessary instead of relying on the caller.
21141 2017-06-13 Martin Jambor <mjambor@suse.cz>
21143 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21144 that have a first_link.
21145 (sort_and_splice_var_accesses): Do not check first_link before
21147 (subtree_mark_written_and_enqueue): Likewise.
21148 (propagate_all_subaccesses): Likewise and do not stop at first
21149 parent with a first_link.
21151 2017-06-13 Martin Jambor <mjambor@suse.cz>
21153 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21156 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21158 * match.pd: New pattern.
21160 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21162 * tree-vrp.c (is_masked_range_test): New function.
21163 (register_edge_assert_for): Determine ranges for
21166 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21168 PR tree-optimization/67328
21169 * fold-const.c (maskable_range_p): New function.
21170 (build_range_check): Generate bittests if possible.
21172 2017-06-13 Martin Liska <mliska@suse.cz>
21174 * gimple-pretty-print.c (dump_probability): Add new argument.
21175 (dump_edge_probability): Dump both probability and count.
21176 (dump_gimple_label): Likewise.
21177 (dump_gimple_bb_header): Likewise.
21179 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21182 * config/avr/avr-devices.c: Fix indentation.
21183 * config/avr/gen-avr-mmcu-specs.c: Dito.
21185 2017-06-13 Richard Biener <rguenther@suse.de>
21187 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21188 instead get vector type from stmt_info.
21189 (vectorizable_reduction): Adjust. Remove dead code.
21191 2017-06-13 Richard Biener <rguenther@suse.de>
21193 PR middle-end/81065
21194 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21195 case of C * (x * C2 + C3).
21196 (fold_addr_of_array_ref_difference): Properly fold index difference.
21198 2017-06-12 David S. Miller <davem@davemloft.net>
21201 * config/sparc/sparc.md (return expander): Emit frame blockage if
21202 function uses alloca.
21204 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21206 * combine.c (make_field_assignment): Check len rather than the mode
21207 precision when calling force_to_mode.
21209 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21211 Support multilibs and devices that see flash in RAM address range.
21214 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21215 (avr_mcu_t) <flash_pm_offset>: New field.
21216 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21217 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21218 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21219 (AVR_TINY_PM_OFFSET): Remove macro.
21220 * config/avr/avr.opt (-mshort-calls): New option.
21221 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21222 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21223 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21224 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21225 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21226 instead of avr_arch->have_jmp_call.
21227 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21228 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21229 avr_arch->flash_pm_offset to define.
21230 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21231 new field flash_pm_offset. Add entry for avrxmega3.
21232 (avr_texinfo): Add entry for avrxmega3.
21233 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21234 attiny212, attiny214,
21235 attiny412, attiny414, attiny416, attiny417,
21236 attiny814, attiny816, attiny817,
21237 attiny1614, attiny1616, attiny1617,
21238 attiny3214, attiny3216, attiny3217.
21239 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21240 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21241 (avr_print_operand_address) [AVR_TINY]: Same.
21242 (avr_asm_init_sections) <readonly_data_section>: Only patch
21243 callback if avr_arch->flash_pm_offset = 0.
21244 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21245 for rodata if avr_arch->flash_pm_offset != 0.
21246 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21247 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21248 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21249 (m_options): Append opt_rcall.
21250 (m_dirnames): Append dir_rcall.
21251 * config/avr/t-multilib: Regenerate.
21253 * configure.ac [target=avr]: Check whether avrxmega3 default
21254 linker description file works as needed.
21255 * configure: Regenerate.
21256 * doc/avr-mmcu.texi: Regenerate.
21257 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21258 <__AVR_ARCH__>: Document avrxmega3 and 103.
21259 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21260 <__AVR_SHORT_CALLS__>: Document it.
21261 <__AVR_PM_BASE_ADDRESS__>: Document it.
21262 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21263 (AVR Variable Attributes) <progmem>: Document this is
21264 not needed for avrxmega3.
21265 (AVR Named Address Spaces) <__flash>: Dito.
21267 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21269 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21271 2017-06-12 Doug Rupp <rupp@adacore.com>
21273 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21274 Append vxworks-stdint.h to the tm_file list.
21275 * config/vxworks-stdint.h: New file.
21277 2017-06-12 Martin Liska <mliska@suse.cz>
21279 PR tree-optimization/81041
21280 * tree-profile.c (gimple_gen_ic_func_profiler):
21281 Create an extra BB in profile-generate
21282 (gimple_gen_time_profiler): Likewise.
21284 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21286 PR tree-optimization/81003
21287 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21288 (update_range_test): Use it instead of force_gimple_operand_gsi.
21290 2017-06-12 Richard Biener <rguenther@suse.de>
21292 PR tree-optimization/81053
21293 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21294 with backedge value not defined in loop. Simplify def stmt
21297 2017-06-11 Tom de Vries <tom@codesourcery.com>
21300 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21302 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21303 nvptx_cannot_force_const_mem.
21305 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21307 * opts.c (finish_options): Move test for flag_split_stack after
21308 it has been initialized.
21310 2017-06-11 Jason Merrill <jason@redhat.com>
21312 * tree.h (id_equal): New.
21313 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21314 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21315 instead of strcmp of IDENTIFIER_POINTER.
21317 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21319 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21320 (mark_all_inlined_calls_cdtor): Fix formating.
21321 (inline_transform): Rescale profile before inlining.
21323 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21325 * cgraph.h (cgraph_edge::clone): Update prototype.
21326 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21327 (cgraph_node::create_clone): Update.
21328 (cgraph_node::create_version_clone): Update.
21329 * tree-inline.c (copy_bb): Update.
21330 (expand_call_inline): Update.
21332 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21334 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21335 factored out from ...
21336 (rs6000_emit_prologue): ... here.
21338 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21340 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21341 factored out from ...
21342 (rs6000_emit_prologue): ... here.
21344 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21346 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21348 (handle_missing_profiles): Fix computation of tp_first_run.
21349 (counts_to_freqs): Do not touch freqs when count is 0.
21351 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21353 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21356 2017-06-10 Tom de Vries <tom@codesourcery.com>
21358 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21359 attributes): Document signal effective target.
21361 2017-06-10 Tom de Vries <tom@codesourcery.com>
21363 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21364 Document effective target stack_size.
21366 2017-06-09 David Malcolm <dmalcolm@redhat.com>
21368 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
21369 to the edit_context if they can be auto-applied.
21371 2017-06-9 Ian Lance Taylor <iant@golang.org>
21373 * opts.c (finish_options): If -fsplit-stack, disable implicit
21374 -forder-blocks-and-partition.
21375 * doc/invoke.texi (Optimize Options): Document that when using
21376 -fsplit-stack -forder-blocks-and-partition is not implicitly
21379 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21381 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21382 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21383 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21384 * builtins.def (abort, trap, unreachable): Declare cold.
21385 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
21386 * tree-core.h (ECF_COLD): New.
21387 * tree.c (set_call_expr_flags): Handle ECF_COLD.
21388 (build_common_builtin_nodes): Mark unreachable and abort as cold.
21390 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21392 * predict.c (unlikely_executed_stmt_p): Cleanup.
21394 2017-06-09 Richard Biener <rguenther@suse.de>
21396 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
21397 model if the ref is always written to.
21399 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21401 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
21403 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21405 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
21408 2017-06-09 Tom de Vries <tom@codesourcery.com>
21411 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
21412 "target cannot support label values" when encountering LABEL_REF.
21414 2017-06-09 Martin Liska <mliska@suse.cz>
21416 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
21417 (gimple_gen_ic_func_profiler): Emit direct comparison
21418 of __gcov_indirect_call_callee with NULL.
21419 (gimple_gen_time_profiler): Change probability from
21420 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
21422 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21424 * profile.c (edge_gcov_counts): Turn to pointer.
21425 (compute_branch_probabilities, compute_branch_probabilities): Update.
21426 (branch_prob): Do not clear edge_gcov_count.
21427 * profile.h (edge_gcov_counts): Turn to pointer.
21428 (edge_gcov_count): Update.
21430 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21432 * gimple.h (gimple_check_failed): Mark cold.
21434 2017-06-09 Richard Biener <rguenther@suse.de>
21436 PR tree-optimization/66623
21437 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
21438 refactor check_reduction into two parts, properly computing
21439 whether we have to check reduction validity for outer loop
21442 2017-06-09 Richard Biener <rguenther@suse.de>
21444 PR tree-optimization/79483
21445 * graphite-scop-detection.c (order): New global.
21446 (get_order): Compute bb to order mapping that satisfies code
21447 generation constraints.
21448 (cmp_pbbs): New helper.
21449 (build_scops): Start domwalk at entry block, sort generated
21452 2017-06-09 Richard Biener <rguenther@suse.de>
21454 PR middle-end/81007
21455 * ipa-polymorphic-call.c
21456 (ipa_polymorphic_call_context::restrict_to_inner_class):
21457 Skip FIELD_DECLs with error_mark_node type.
21458 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
21461 2017-06-09 Martin Liska <mliska@suse.cz>
21463 * predict.c (struct branch_predictor): New struct.
21464 (test_prediction_value_range): New test.
21465 (predict_c_tests): New function.
21466 * selftest-run-tests.c (selftest::run_tests): Run the function.
21467 * selftest.h: Declare new tests.
21469 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
21472 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
21473 gen_add3_insn did not fail.
21474 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
21475 r0, construct that number in a temporary reg and add that reg to r0.
21476 If asked to put the result in r0 as well, fail.
21478 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
21480 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21481 for early expansion of vec_eqv.
21483 2017-06-08 Jakub Jelinek <jakub@redhat.com>
21485 PR middle-end/81005
21486 * ubsan.c (instrument_null): Avoid pointless code temporary.
21487 (pass_ubsan::execute): Instrument aggregate arguments of calls.
21489 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
21493 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
21496 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
21499 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21501 * predict.c (unlikely_executed_edge_p): Move ahead.
21502 (probably_never_executed_edge_p): Use it.
21504 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21506 PR middle-end/79988
21507 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
21508 gimple_call_builtin_p call.
21510 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21512 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
21513 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
21514 rtl_check_failed_type2, rtl_check_failed_code1,
21515 rtl_check_failed_code2, rtl_check_failed_code_mode,
21516 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
21517 rtvec_check_failed_bounds, rtl_check_failed_flag,
21518 _fatal_insn_not_found, _fatal_insn): Likewise.
21519 * tree.h (tree_contains_struct_check_failed,
21520 tree_check_failed, tree_not_check_failed,
21521 tree_class_check_failed, tree_range_check_failed,
21522 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
21523 tree_vec_elt_check_failed, phi_node_elt_check_failed,
21524 tree_operand_check_failed, omp_clause_check_failed,
21525 omp_clause_operand_check_failed, omp_clause_range_check_failed):
21528 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21530 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
21531 flag_branch_probabilities.
21532 * ipa-inline.c (edge_badness): Likewise.
21533 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
21534 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
21535 * predict.c (maybe_hot_frequency_p): Likewise.
21536 (probably_never_executed): Likewise.
21537 * sched-ebb.c (schedule_ebbs): Likewise.
21538 * sched-rgn.c (find_single_block_region): Likewise.
21539 * tracer.c (tail_duplicate): Likewise.
21541 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21543 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
21544 longer requires x_flag_profile_use.
21546 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21548 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
21549 instead of flag_reorder_blocks_and_partition.
21550 * dbxout.c (dbxout_function_end): Likewise.
21551 * dwarf2out.c (gen_subprogram_die): Likewise.
21552 * haifa-sched.c (sched_create_recovery_edges): Likewise.
21553 * hw-doloop.c (reorg_loops): Likewise.
21554 * varasm.c (assemble_start_function,
21555 assemble_end_function): Likewise.
21556 (decide_function_section): Do not check for
21557 flag_reorder_blocks_and_partition.
21559 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21561 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21563 (chkp_get_hard_register_fake_addr_expr): Ditto.
21564 (chkp_build_addr_expr): Add check for hard reg case.
21565 (chkp_parse_array_and_component_ref): Ditto.
21566 (chkp_find_bounds_1): Ditto.
21567 (chkp_process_stmt): Don't generate bounds store for
21570 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21572 * predict.c (maybe_hot_bb_p): Do not check profile status.
21573 (maybe_hot_edge_p): Likewise.
21574 (probably_never_executed): Check for zero counts even if profile
21576 (unlikely_executed_edge_p): New function.
21577 (unlikely_executed_stmt_p): New function.
21578 (unlikely_executed_bb_p): New function.
21579 (set_even_probabilities): Use unlikely predicates.
21580 (combine_predictions_for_bb): Likewise.
21581 (predict_paths_for_bb): Likewise.
21582 (predict_paths_leading_to_edge): Likewise.
21583 (determine_unlikely_bbs): New function.
21584 (estimate_bb_frequencies): Use it.
21585 (compute_function_frequency): Use zero counts even if profile is
21587 * profile-count.h: Fix typo.
21589 2017-08-08 Julia Koval <julia.koval@intel.com>
21591 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
21592 _mm512_mask_cvtsepi16_storeu_epi8,
21593 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
21594 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
21595 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
21596 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
21597 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
21598 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
21599 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
21600 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
21601 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
21602 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
21603 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
21604 __builtin_ia32_pmovuswb256mem_mask,
21605 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
21606 __builtin_ia32_pmovwb512mem_mask): New builtins.
21608 2017-08-08 Julia Koval <julia.koval@intel.com>
21610 PR target/73350,80862
21611 * config/i386/subst.md (round): Fix round pattern.
21612 * config/i386/i386.c (ix86_erase_embedded_rounding):
21613 Fix erasing rounding for the fixed pattern.
21615 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21617 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
21619 2017-06-08 Martin Liska <mliska@suse.cz>
21621 PR gcov-profile/80911
21622 * gcov.c (block_info::block_info): New constructor.
21624 2017-06-07 Carl Love <cel@us.ibm.com>
21626 * config/rs6000/rs6000-c: The return type of the following
21627 built-in functions was implemented as int not long long. Fix sign
21628 of return value for the unsigned version of vec_mulo and vec_mule.
21629 vector unsigned long long vec_bperm (vector unsigned long long,
21630 vector unsigned char)
21631 vector signed long long vec_mule (vector signed int,
21633 vector unsigned long long vec_mule (vector unsigned int,
21634 vector unsigned int)
21635 vector signed long long vec_mulo (vector signed int,
21637 vector unsigned long long vec_mulo (vector unsigned int,
21638 vector unsigned int)
21639 * doc/extend.texi: Fix the documentation for the built-in
21642 2017-06-07 Carl Love <cel@us.ibm.com>
21645 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
21648 2017-06-07 Carl Love <cel@us.ibm.com>
21650 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
21651 support, Generate doublehv for signed int/float for BE case only.
21653 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
21655 * doc/invoke.texi (mcx16): Rewrite.
21657 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21659 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
21660 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
21661 *mov<mode>_softfloat, and an anonymous splitter): Use
21662 nonimmediate_operand instead of rs6000_nonimmediate_operand.
21664 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21666 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
21668 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21669 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
21670 (rs6000_debug_reg_global): Adjust.
21671 (rs6000_init_hard_regno_mode_ok): Adjust.
21672 (rs6000_dbx_register_number): Adjust.
21673 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
21674 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21675 Remove SPE_ACC and SPEFSCR.
21676 (REG_ALLOC_ORDER): Ditto.
21677 (FRAME_POINTER_REGNUM): Change to 111.
21678 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
21679 (REG_CLASS_NAMES): Ditto.
21680 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
21681 (REGISTER_NAMES): Ditto.
21682 (ADDITIONAL_REG_NAMES): Ditto.
21683 (rs6000_reg_names): Ditto.
21684 * config/rs6000/rs6000.md: Renumber some register number
21687 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21689 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
21691 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21692 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
21694 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
21695 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21696 Delete the SPE high registers.
21697 (REG_ALLOC_ORDER): Ditto.
21698 (enum reg_class): Remove SPE_HIGH_REGS.
21699 (REG_CLASS_NAMES): Ditto.
21700 (REG_CLASS_CONTENTS): Delete the SPE high registers.
21701 (REGISTER_NAMES): Ditto.
21702 (rs6000_reg_names): Ditto.
21703 * doc/tm.texi.in: Remove SPE as example.
21704 * doc/tm.texi: Regenerate.
21706 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21708 * config/rs6000/8540.md (ppc8540_brinc): Delete.
21709 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
21710 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
21711 * config/rs6000/rs6000.md (type): Remove "brinc".
21713 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21715 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
21716 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
21717 * config/rs6000/linuxspe.h: Delete file.
21718 * config/rs6000/rs6000.md: Don't include spe.md.
21719 * config/rs6000/spe.h: Delete file.
21720 * config/rs6000/spe.md: Delete file.
21721 * config/rs6000/t-rs6000: Remove spe.md.
21723 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21725 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
21726 (reg_or_none500mem_operand): Delete.
21727 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
21728 instead of reg_or_none500mem_operand.
21730 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21732 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21733 handling of SPE flags.
21734 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
21736 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21738 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
21740 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
21741 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
21742 paired_divv2sf3): Similar.
21743 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
21744 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
21745 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
21747 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
21748 Rename the paired_* instruction patterns.
21749 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
21751 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
21752 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
21753 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
21754 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
21755 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
21756 PAIRED_VECTOR_MODE.
21757 (struct machine_function): Delete field spe_insn_chain_scanned_p.
21758 (spe_func_has_64bit_regs_p): Delete.
21759 (spe_expand_predicate_builtin): Delete.
21760 (spe_expand_evsel_builtin): Delete.
21761 (TARGET_DWARF_REGISTER_SPAN): Do not define.
21762 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
21763 (invalid_e500_subreg): Delete.
21764 (rs6000_legitimize_address): Always force_reg op2 as well, for
21765 paired single memory accesses.
21766 (rs6000_member_type_forces_blk): Delete.
21767 (rs6000_spe_function_arg): Delete.
21768 (rs6000_expand_unop_builtin): Delete SPE handling.
21769 (rs6000_expand_binop_builtin): Ditto.
21770 (spe_expand_stv_builtin): Delete.
21771 (bdesc_2arg_spe): Delete.
21772 (spe_expand_builtin): Delete.
21773 (spe_expand_predicate_builtin): Delete.
21774 (spe_expand_evsel_builtin): Delete.
21775 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
21776 (spe_init_builtins): Delete.
21777 (spe_func_has_64bit_regs_p): Delete.
21778 (savres_routine_name): Delete "info" parameter. Adjust callers.
21779 (rs6000_emit_stack_reset): Ditto.
21780 (rs6000_dwarf_register_span): Delete.
21781 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
21782 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
21783 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
21785 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
21787 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
21788 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
21789 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
21790 mulv2sf3, divv2sf3): Delete expanders.
21792 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21794 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
21796 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21798 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
21799 * config/rs6000/rs6000.c: Ditto.
21801 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21803 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
21804 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
21805 comparison_operator.
21807 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21809 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
21810 * config/rs6000/rs6000.opt: Ditto.
21811 * config/rs6000/t-rtems: Ditto.
21813 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21815 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
21816 TARGET_E500_SINGLE by 0, simplify.
21817 * config/rs6000/rs6000.c: Ditto.
21818 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
21819 (spe_build_register_parallel): Delete.
21820 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
21821 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
21822 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
21823 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
21824 (E500_CONVERT): Delete.
21825 * config/rs6000/spe.md: Remove many patterns and all define_constants.
21827 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21829 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
21830 * config/rs6000/dfp.md: Ditto.
21831 (negdd2, *negdd2_fpr): Merge.
21832 (absdd2, *absdd2_fpr): Merge.
21833 (negtd2, *negtd2_fpr): Merge.
21834 (abstd2, *abstd2_fpr): Merge.
21835 * config/rs6000/e500.h: Delete file.
21836 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
21837 TARGET_FPRS by 1 and simplify.
21838 * config/rs6000/rs6000-c.c: Ditto.
21839 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
21840 TARGET_DF_SPE by 0.
21841 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
21843 * config/rs6000/rs6000.md: Ditto.
21844 (floatdidf2, *floatdidf2_fpr): Merge.
21845 (move_from_CR_gt_bit): Delete.
21846 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
21847 (E500_CR_IOR_COMPARE): Delete.
21848 (All patterns that require !TARGET_FPRS): Delete.
21849 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
21851 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21853 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
21855 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21857 * graphds.c (add_edge): Intitialize edge's attached data.
21858 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
21859 pointer parameter. Call pointed function on each edge during
21860 graph traversing. Skip traversing the edge when the function
21862 (graphds_dfs, graphds_scc): Ditto.
21863 (for_each_edge): New parameter. Pass the new parameter to callback
21865 * graphds.h (skip_edge_callback): New function pointer type.
21866 (graphds_dfs, graphds_scc): New function pointer parameter.
21867 (graphds_edge_callback, for_each_edge): New parameter.
21869 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21871 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
21872 out code checking if runtime alias check is possible to below ...
21873 Call the new function.
21874 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
21875 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
21877 2017-06-07 Marek Polacek <polacek@redhat.com>
21880 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
21881 TYPE_OVERFLOW_WRAPS check.
21883 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21885 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
21886 if versioning is required.
21887 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
21888 peeling with the check for versioning.
21890 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21892 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
21893 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
21894 Set true to new parameter if new ssa variable is defined.
21895 (vect_gen_vector_loop_niters): Refactor. Set range information
21896 for the new vector loop bound variable.
21897 (vect_do_peeling): Ditto.
21899 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21901 * tree-affine.c (ssa.h): Include header file.
21902 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
21903 has wrapping overflow behavior.
21905 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21907 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
21909 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21911 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
21912 (tree_to_aff_combination): ... here.
21914 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21916 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
21917 reg_pressure model function.
21918 (ivopts_global_cost_for_size): Delete.
21919 (determine_set_costs, iv_ca_recount_cost): Call new model function
21920 ivopts_estimate_reg_pressure.
21922 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21924 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
21925 expensive than udiv. Remove floating point cases from mod.
21927 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21929 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
21930 Increase idiv cost.
21932 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21934 * config/aarch64/aarch64.md
21935 (copysignsf3): Fix mask generation.
21937 2017-06-07 Jakub Jelinek <jakub@redhat.com>
21939 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
21941 (class dump_manager): Add register_dumps method.
21942 * dumpfile.c: Include langhooks.h.
21943 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
21944 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
21945 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
21946 (dump_manager::dump_register): Start with 512 entries instead of 32.
21947 (dump_manager::register_dumps): New method.
21948 * toplev.c (general_init): Instead of invoking register_dumps
21949 langhook, invoke register_dumps method on the dump manager.
21950 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
21953 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
21955 * doc/md.texi: Clarify the restrictions on a define_insn condition.
21956 Say that # requires an associated define_split to exist, and that
21957 the define_split must be suitable for use after register allocation.
21959 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21961 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
21962 (compute_outgoing_frequencies): Also initialize zero counts.
21963 (find_many_sub_basic_blocks): Do not produce uninitialized profile
21964 around loops; preserve more of profile when nothing changes.
21966 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
21968 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
21970 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
21971 * config/arm/arm-cpu-cdata.h: Regenerate.
21972 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
21973 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
21974 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
21975 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
21976 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
21978 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
21980 * config/arm/t-rmprofile: Likewise.
21981 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
21983 2017-06-06 David S. Miller <davem@davemloft.net>
21986 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
21987 blockage if function uses alloca.
21989 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21991 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
21992 New "uid" fields to hold pretty-print IDs of group and ref.
21993 Memory references are now identified as <group_id>:<ref_id>
21994 instead of using [random] addresses.
21995 (dump_mem_details): Simplify, no functional change.
21996 (dump_mem_ref): Simplify and make output more concise.
21997 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
21998 (find_or_create_group): Initialize group uid.
21999 (record_ref): Initialize ref uid. Improve debug output.
22000 (prune_group_by_reuse, should_issue_prefetch_p,)
22001 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
22002 (mark_nontemporal_store, determine_loop_nest_reuse):
22003 Improve debug output.
22005 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22007 * dbgcnt.def (prefetch): New debug counter.
22008 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
22009 (schedule_prefetches): Stop issueing prefetches if debug counter
22012 2017-06-06 Tom de Vries <tom@codesourcery.com>
22014 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
22015 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22017 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22019 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22020 Use aarch64_reg_or_zero predicate for operand 4.
22021 (aarch64_compare_and_swap<mode> define_insn_and_split):
22022 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22023 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22025 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22027 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22028 (arm_compute_save_core_reg_mask): This.
22029 (thumb1_compute_save_reg_mask): Rename into ...
22030 (thumb1_compute_save_core_reg_mask): This.
22031 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22032 (arm_compute_frame_layout): Likewise.
22034 2017-06-06 Richard Biener <rguenther@suse.de>
22036 PR tree-optimization/80974
22037 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22038 keep or clear leaders SSA info.
22040 2017-06-06 Tom de Vries <tom@codesourcery.com>
22042 * config/nvptx/nvptx.c (split_mode_p): New function.
22043 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22045 2017-06-06 Tom de Vries <tom@codesourcery.com>
22047 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22049 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22052 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22055 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22057 * shrink-wrap.c (handle_simple_exit): Update profile.
22058 (try_shrink_wrapping): Upate profile.
22060 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22062 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22063 (tree_guess_outgoing_edge_probabilities): New.
22064 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22065 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22067 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22069 * ipa-split.c (split_function): Initialize return bb profile.
22071 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22073 * profile.c (compute_branch_probabilities): Also initialize
22074 EXIT_BLOCK profile.
22076 2017-06-06 Richard Biener <rguenther@suse.de>
22078 PR tree-optimization/80928
22079 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22080 (vect_analyze_loop_operations): Properly guard analysis for
22082 (vect_transform_loop): Likewise.
22083 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22084 (vect_model_induction_cost): Do not cost for pure SLP.
22085 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22086 of induction in inner loop vectorization.
22087 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22088 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22089 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22091 (vect_analyze_slp_cost_1): Cost induction.
22092 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22093 (vect_get_slp_vect_defs): Likewise.
22094 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22095 (vect_transform_stmt): Handle SLP reductions.
22096 * tree-vectorizer.h (vectorizable_induction): Adjust.
22098 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22100 * config/rs6000/rs6000.c (make_resolver_func): Update
22101 init_lowered_empty_function call.
22103 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22105 * doc/invoke.texi: Document the -fprofile-abs-path option.
22106 * common.opt (fprofile-abs-path): New option.
22107 * gcov-io.h (gcov_write_filename): Declare.
22108 * gcov-io.c (gcov_write_filename): New function.
22109 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22110 * profile.c (output_location): Likewise.
22112 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22114 * shring-wrap.c: Revert accidental commit.
22116 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22118 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22120 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22122 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22124 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22125 profile in callgraph edge.
22126 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22127 (apply_scale): Likewise.
22128 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22129 Also scale profile when inlining function with zero profile.
22130 (initialize_cfun): Update exit block profile even when it is zero.
22131 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22132 when profile is read.
22134 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22136 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22137 (CLONE_*): New constants to define the processors we can generate
22138 code for with the target_clone attribute.
22139 (rs6000_clone_map): New array to identify which clone processors
22140 the current program is running on.
22141 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22142 target_clone attribute.
22143 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22144 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22145 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22146 (cpu_expand_builtin): Add support for target_clone attribute.
22147 (rs6000_valid_attribute_p): Allow "default" attribute.
22148 (get_decl_name): New debug function to simplify printing the
22149 current function name in debugging statements.
22150 (rs6000_clone_priority): New functions to support the target_clone
22151 attribute, and be able to generate code to switch between ISA 2.05
22152 through ISA 3.0 (power6 through power9).
22153 (rs6000_compare_version_priority): Likewise.
22154 (rs6000_get_function_versions_dispatcher): Likewise.
22155 (make_resolver_func): Likewise.
22156 (add_condition_to_bb): Likewise.
22157 (dispatch_function_versions): Likewise.
22158 (rs6000_generate_version_dispatcher_body): Likewise.
22159 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22160 (fusion_gpr_load_p): Fix a spacing issue.
22161 * doc/extend.texi (Common Function Attributes): Document that the
22162 PowerPC supports the target_clone attribute.
22164 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22166 * config/arm/arm.h: explain F symbol found in description of ARM
22167 register allocation in its legend.
22169 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22171 * config/mips/frame-header-opt.c: Include profile-count.h.
22172 * config/riscv/riscv.c: Include profile-count.h
22174 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22176 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22178 (sm_set_flag_if_changed): Add bbs field.
22179 (execute_sm_if_changed_flag_set): Pass BBS.
22180 (execute_sm): Update.
22182 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22184 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22187 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22189 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22190 (peephole2): New peephole2 to emit the above.
22191 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22193 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22195 * config/aarch64/aarch64.c (define_peephole2 above
22196 *sub_<shift>_<mode>): New peephole.
22198 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22200 * config/i386/i386.c (make_resolver_func): Update.
22201 * Makefile.in: Add profile-count.h and profile-count.o
22202 * auto-profile.c (afdo_indirect_call): Update to new API.
22203 (afdo_set_bb_count): Update.
22204 (afdo_propagate_edge): Update.
22205 (afdo_propagate_circuit): Update.
22206 (afdo_calculate_branch_prob): Update.
22207 (afdo_annotate_cfg): Update.
22208 * basic-block.h: Include profile-count.h
22209 (struct edge_def): Turn count to profile_count.
22210 (struct basic_block_def): Likewie.
22211 (REG_BR_PROB_BASE): Move to profile-count.h
22212 (RDIV): Move to profile-count.h
22213 * bb-reorder.c (max_entry_count): Turn to profile_count.
22214 (find_traces): Update.
22215 (rotate_loop):Update.
22216 (connect_traces):Update.
22217 (sanitize_hot_paths):Update.
22218 * bt-load.c (migrate_btr_defs): Update.
22219 * cfg.c (RDIV): Remove.
22220 (init_flow): Use alloc_block.
22221 (alloc_block): Uninitialize count.
22222 (unchecked_make_edge): Uninitialize count.
22223 (check_bb_profile): Update.
22224 (dump_edge_info): Update.
22225 (dump_bb_info): Update.
22226 (update_bb_profile_for_threading): Update.
22227 (scale_bbs_frequencies_int): Update.
22228 (scale_bbs_frequencies_gcov_type): Update.
22229 (scale_bbs_frequencies_profile_count): New.
22230 * cfg.h (update_bb_profile_for_threading): Update.
22231 (scale_bbs_frequencies_profile_count): Declare.
22232 * cfgbuild.c (compute_outgoing_frequencies): Update.
22233 (find_many_sub_basic_blocks): Update.
22234 * cfgcleanup.c (try_forward_edges): Update.
22235 (try_crossjump_to_edge): Update.
22236 * cfgexpand.c (expand_gimple_tailcall): Update.
22237 (construct_exit_block): Update.
22238 * cfghooks.c (verify_flow_info): Update.
22239 (dump_bb_for_graph): Update.
22240 (split_edge): Update.
22241 (make_forwarder_block): Update.
22242 (duplicate_block): Update.
22243 (account_profile_record): Update.
22244 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22245 (get_estimated_loop_iterations): Update.
22246 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22247 (single_likely_exit): Update.
22248 * cfgloopmanip.c (scale_loop_profile): Update.
22250 (set_zero_probability): Update.
22251 (lv_adjust_loop_entry_edge): Update.
22252 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22253 (purge_dead_edges): Update.
22254 (rtl_account_profile_record): Update.
22255 * cgraph.c (cgraph_node::create): Uninitialize count.
22256 (symbol_table::create_edge): Uninitialize count.
22257 (cgraph_update_edges_for_call_stmt_node): Update.
22258 (cgraph_edge::dump_edge_flags): Update.
22259 (cgraph_node::dump): Update.
22260 (cgraph_edge::maybe_hot_p): Update.
22261 * cgraph.h: Include profile-count.h
22262 (create_clone), create_edge, create_indirect_edge): Update.
22263 (cgraph_node): Turn count to profile_count.
22264 (cgraph_edge0: Likewise.
22265 (make_speculative, clone): Update.
22266 (create_edge): Update.
22267 (init_lowered_empty_function): Update.
22268 * cgraphclones.c (cgraph_edge::clone): Update.
22269 (duplicate_thunk_for_node): Update.
22270 (cgraph_node::create_clone): Update.
22271 * cgraphunit.c (cgraph_node::analyze): Update.
22272 (cgraph_node::expand_thunk): Update.
22273 * final.c (dump_basic_block_info): Update.
22274 * gimple-streamer-in.c (input_bb): Update.
22275 * gimple-streamer-out.c (output_bb): Update.
22276 * graphite.c (print_global_statistics): Update.
22277 (print_graphite_scop_statistics): Update.
22278 * hsa-brig.c: Include basic-block.h.
22279 * hsa-dump.c: Include basic-block.h.
22280 * hsa-gen.c (T sum_slice): Update.
22281 (convert_switch_statements):Update.
22282 * hsa-regalloc.c: Include basic-block.h.
22283 * ipa-chkp.c (chkp_produce_thunks): Update.
22284 * ipa-cp.c (struct caller_statistics): Update.
22285 (init_caller_stats): Update.
22286 (gather_caller_stats): Update.
22287 (ipcp_cloning_candidate_p): Update.
22288 (good_cloning_opportunity_p): Update.
22289 (get_info_about_necessary_edges): Update.
22290 (dump_profile_updates): Update.
22291 (update_profiling_info): Update.
22292 (update_specialized_profile): Update.
22293 (perhaps_add_new_callers): Update.
22294 (decide_about_value): Update.
22295 (ipa_cp_c_finalize): Update.
22296 * ipa-devirt.c (struct odr_type_warn_count): Update.
22297 (struct decl_warn_count): Update.
22298 (struct final_warning_record): Update.
22299 (possible_polymorphic_call_targets): Update.
22300 (ipa_devirt): Update.
22301 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22302 * ipa-icf.c (sem_function::merge): Update.
22303 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22304 * ipa-inline.c (compute_uninlined_call_time): Update.
22305 (compute_inlined_call_time): Update.
22306 (want_inline_small_function_p): Update.
22307 (want_inline_self_recursive_call_p): Update.
22308 (edge_badness): Update.
22309 (lookup_recursive_calls): Update.
22310 (recursive_inlining): Update.
22311 (inline_small_functions): Update.
22312 (dump_overall_stats): Update.
22313 (dump_inline_stats): Update.
22314 * ipa-profile.c (ipa_profile_generate_summary): Update.
22315 (ipa_propagate_frequency): Update.
22316 (ipa_profile): Update.
22317 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22318 * ipa-utils.c (ipa_merge_profiles): Update.
22319 * loop-doloop.c (doloop_modify): Update.
22320 * loop-unroll.c (report_unroll): Update.
22321 (unroll_loop_runtime_iterations): Update.
22322 * lto-cgraph.c (lto_output_edge): Update.
22323 (lto_output_node): Update.
22324 (input_node): Update.
22325 (input_edge): Update.
22326 (merge_profile_summaries): Update.
22327 * lto-streamer-in.c (input_cfg): Update.
22328 * lto-streamer-out.c (output_cfg): Update.
22329 * mcf.c (create_fixup_graph): Update.
22330 (adjust_cfg_counts): Update.
22331 (sum_edge_counts): Update.
22332 * modulo-sched.c (sms_schedule): Update.
22333 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22334 * predict.c (maybe_hot_count_p): Update.
22335 (probably_never_executed): Update.
22336 (dump_prediction): Update.
22337 (combine_predictions_for_bb): Update.
22338 (propagate_freq): Update.
22339 (handle_missing_profiles): Update.
22340 (counts_to_freqs): Update.
22341 (rebuild_frequencies): Update.
22342 (force_edge_cold): Update.
22343 * predict.h: Include profile-count.h
22344 (maybe_hot_count_p, counts_to_freqs): UPdate.
22345 * print-rtl-function.c: Do not include cfg.h
22346 * print-rtl.c: Include basic-block.h
22347 * profile-count.c: New file.
22348 * profile-count.h: New file.
22349 * profile.c (is_edge_inconsistent): Update.
22350 (correct_negative_edge_counts): Update.
22351 (is_inconsistent): Update.
22352 (set_bb_counts): Update.
22353 (read_profile_edge_counts): Update.
22354 (compute_frequency_overlap): Update.
22355 (compute_branch_probabilities): Update; Initialize and deinitialize
22357 (branch_prob): Update.
22358 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
22359 (edge_gcov_count): New.
22360 (bb_gcov_count): New.
22361 * shrink-wrap.c (try_shrink_wrapping): Update.
22362 * tracer.c (better_p): Update.
22363 * trans-mem.c (expand_transaction): Update.
22364 (ipa_tm_insert_irr_call): Update.
22365 (ipa_tm_insert_gettmclone_call): Update.
22366 * tree-call-cdce.c: Update.
22367 * tree-cfg.c (gimple_duplicate_sese_region): Update.
22368 (gimple_duplicate_sese_tail): Update.
22369 (gimple_account_profile_record): Update.
22370 (execute_fixup_cfg): Update.
22371 * tree-inline.c (copy_bb): Update.
22372 (copy_edges_for_bb): Update.
22373 (initialize_cfun): Update.
22374 (freqs_to_counts): Update.
22375 (copy_cfg_body): Update.
22376 (expand_call_inline): Update.
22377 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22378 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
22379 (try_unroll_loop_completely): Update.
22380 (try_peel_loop): Update.
22381 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22382 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22383 * tree-ssa-loop-split.c (connect_loops): Update.
22384 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22385 * tree-ssa-reassoc.c (branch_fixup): Update.
22386 * tree-ssa-tail-merge.c (replace_block_by): Update.
22387 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
22388 (compute_path_counts): Update.
22389 (update_profile): Update.
22390 (recompute_probabilities): Update.
22391 (update_joiner_offpath_counts): Update.
22392 (estimated_freqs_path): Update.
22393 (freqs_to_counts_path): Update.
22394 (clear_counts_path): Update.
22395 (ssa_fix_duplicate_block_edges): Update.
22396 (duplicate_thread_path): Update.
22397 * tree-switch-conversion.c (case_bit_test_cmp): Update.
22398 (struct switch_conv_info): Update.
22399 * tree-tailcall.c (decrease_profile): Update.
22400 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22401 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22402 * value-prof.c (check_counter): Update.
22403 (gimple_divmod_fixed_value): Update.
22404 (gimple_mod_pow2): Update.
22405 (gimple_mod_subtract): Update.
22406 (gimple_ic_transform): Update.
22407 (gimple_stringop_fixed_value): Update.
22408 * value-prof.h (gimple_ic): Update.
22410 2017-06-02 Carl Love <cel@us.ibm.com>
22412 * config/rs6000/rs6000-c: Add support for built-in functions
22413 vector double vec_doublee (vector signed int);
22414 vector double vec_doublee (vector unsigned int);
22415 vector double vec_doublee (vector float);
22416 vector double vec_doubleh (vector signed int);
22417 vector double vec_doubleh (vector unsigned int);
22418 vector double vec_doubleh (vector float);
22419 vector double vec_doublel (vector signed int);
22420 vector double vec_doublel (vector unsigned int);
22421 vector double vec_doublel (vector float);
22422 vector double vec_doubleo (vector signed int);
22423 vector double vec_doubleo (vector unsigned int);
22424 vector double vec_doubleo (vector float);.
22425 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
22426 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
22428 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
22429 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
22430 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
22432 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
22433 vec_doublel, vec_doubleh.
22434 * doc/extend.texi: Update the built-in documentation file for the
22435 new built-in functions.
22437 2017-06-02 David Malcolm <dmalcolm@redhat.com>
22440 * ipa-inline-analysis.c (free_growth_caches): Set
22441 edge_removal_hook_holder to NULL after removing it.
22443 2017-06-02 Sudakshina Das <sudi.das@arm.com>
22445 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
22446 comparision with zero.
22448 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22449 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22450 for early expansion of vec_min and vec_max builtins.
22451 (builtin_function_type): Add min/max unsigned variants to those
22452 identified as having unsigned arguments.
22454 2017-06-02 Olivier Hainque <hainque@adacore.com>
22456 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
22458 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22460 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
22461 Use VALL_F16 iterator rather than VALL.
22463 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22465 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
22466 Emit CBNZ inside loop when doing a strong exchange and comparing
22467 against zero. Generate the CC flags after the loop.
22469 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
22471 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
22472 (dl_section_ref): New.
22473 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
22474 On AIX, append an expression to subtract the size of the
22475 section length to dl_section_ref.
22477 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22479 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22480 for early expansion of vector absolute builtins.
22482 2017-06-02 Richard Biener <rguenther@suse.de>
22484 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
22485 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
22487 2017-06-02 Richard Biener <rguenther@suse.de>
22489 PR tree-optimization/80948
22490 * tree-tailcall.c (find_tail_calls): Track stmts to move in
22491 stmt order as well.
22493 2017-06-02 Richard Biener <rguenther@suse.de>
22495 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
22497 * tree-vect-stmts.c (process_use): Do not mark backedge defs
22498 for inductions as relevant.
22500 2017-06-02 Richard Biener <rguenther@suse.de>
22502 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
22503 (vectorizable_induction): ... this. Remove dead code.
22505 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
22507 * builtins. (expand_builtin_alloca): Remove second parameter and
22508 infer its value from the first parameter instead.
22509 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
22511 2017-06-02 Jakub Jelinek <jakub@redhat.com>
22513 PR rtl-optimization/80903
22514 * loop-doloop.c (add_test): Unshare sequence.
22516 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
22518 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
22520 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
22522 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
22524 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
22525 xlogue_layout::get_instance, logue_layout::xlogue_layout,
22526 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
22527 (xlogue_layout::get_stub_rtx): Make static.
22528 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
22529 (xlogue_layout::compute_stub_managed_regs): Rename to...
22530 (xlogue_layout::count_stub_managed_regs): ...this.
22531 (xlogue_layout::is_stub_managed_reg): New function.
22532 (xlogue_layout::m_stub_names): Rename to...
22533 (xlogue_layout::s_stub_names): ...this, make static.
22534 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
22535 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
22536 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
22537 xlogue_layout::s_stub_names): Instantiate statics.
22538 (stub_managed_regs): Remove.
22539 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
22540 (disable_call_ms2sysv_xlogues): Rename to...
22541 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
22542 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
22544 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
22545 change after reload_completed.
22546 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
22548 (ix86_expand_prologue): Likewise.
22549 (ix86_expand_epilogue): Likewise.
22550 (ix86_expand_split_stack_prologue): Likewise.
22551 (ix86_compute_frame_layout): Remove frame parameter ...
22552 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
22553 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
22555 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
22556 (ix86_frame): Move from here ...
22557 * config/i386/i386.h (ix86_frame): ... to here.
22558 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
22559 complete ix86_frame data structure instead. Remove some_ld_name.
22561 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22563 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
22564 symbols that hold a DECL_VALUE_EXPR.
22566 2017-06-01 Martin Jambor <mjambor@suse.cz>
22568 PR tree-optimization/80898
22569 * tree-sra.c (process_subtree_disqualification): Removed.
22570 (disqualify_candidate): Do not acll
22571 process_subtree_disqualification.
22572 (subtree_mark_written_and_enqueue): New function.
22573 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
22574 RHS has been disqualified and re-queue LHS if necessary. Apart
22575 from that, ignore disqualified RHS.
22577 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22579 * config/s390/s390.c (s390_emit_epilogue): Disable early return
22580 address fetch for z10 or later.
22582 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22584 * config/arc/arc.md (tst_movb): Add guard when splitting.
22586 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22588 * config/arc/arc.c (arc_can_eliminate): Test against
22589 arc_frame_pointer_needed.
22591 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22593 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
22594 to prevent store reordering.
22595 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
22596 (type): Add block type.
22597 (stack_tie): Define special instruction to be used in
22600 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22602 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
22603 constraint. It is not valid for the pattern.
22604 (noncommutative_binary_comparison): Likewise.
22606 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22608 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
22611 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22613 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
22614 be used by the reg-alloc.
22616 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22618 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
22619 reg-alloc when having mul64 or mul32x16 instructions.
22620 (mulsidi3): Likewise.
22621 (umulsidi3): Likewise.
22622 (mulsi32x16): New pattern.
22623 (mulsi64): Likewise.
22624 (mulsidi64): Likewise.
22625 (umulsidi64): Likewise.
22626 (MUL32x16_REG): Define.
22627 (mul64_600): Use MUL32x16_REG.
22628 (mac64_600): Likewise.
22629 (umul64_600): Likewise.
22630 (umac64_600): Likewise.
22632 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22634 * config/arc/arc.md (mulsi3_700): Make it commutative.
22636 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
22638 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
22640 (*sign_extendsidi2_insn): Likewise for movstosw.
22642 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22644 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
22645 the type of the input discriminant value. Convert the
22646 discriminant value of signedness vary.
22648 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
22650 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
22651 Add to -Wall section.
22653 2017-06-01 Richard Biener <rguenther@suse.de>
22655 PR middle-end/66313
22656 * fold-const.c (fold_plusminus_mult_expr): If the factored
22657 factor may be zero use a wrapping type for the inner operation.
22658 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
22659 and handle moved defs.
22660 (process_assignment): Properly guard the unary op case. Return a
22661 tri-state indicating that moving the stmt before the call may allow
22662 to continue. Pass through to_move.
22663 (find_tail_calls): Handle moving unrelated defs before
22666 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
22669 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
22670 splitter result in the canonical way.
22672 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
22674 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
22675 also for 32bit target. Update insn attributes.
22676 (zero-extendsidi2 splitter): Allow all registers for operand 1.
22678 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
22680 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
22681 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
22682 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
22683 (_mm_maskz_min_ss): New intrinsics.
22685 2017-05-31 Martin Liska <mliska@suse.cz>
22687 * tree-vect-loop.c (vect_create_epilog_for_reduction):
22688 Change comment style to one we normally use.
22689 (vectorizable_reduction): Likewise.
22690 (vectorizable_induction): Likewise.
22691 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
22692 (vectorizable_call): Likewise.
22693 (vectorizable_simd_clone_call): Likewise.
22694 (vectorizable_conversion): Likewise.
22695 (vectorizable_assignment): Likewise.
22696 (vectorizable_shift): Likewise.
22697 (vectorizable_operation): Likewise.
22698 (vectorizable_store): Likewise.
22699 (vectorizable_load): Likewise.
22700 * tree-vectorizer.h: Likewise.
22702 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
22704 * passes.c (emergency_dump_function): New.
22705 * tree-pass.h (emergency_dump_function): Declare.
22706 * plugin.c (plugins_internal_error_function): Remove.
22707 * plugin.h (plugins_internal_error_function): Remove declaration.
22708 * toplev.c (internal_error_function): New static function. Use it...
22709 (general_init): ...here.
22711 2017-05-31 Graham Markall <graham.markall@embecosm.com>
22713 * config/arc/arc.c (arc_print_operand): Handle constant operands.
22714 (arc_rtx_costs): Add costs for new patterns.
22715 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
22716 * config/arc/predicates.md: Add _1_2_3_operand predicate.
22718 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22720 * tree-ssa-strlen.c (get_next_strinfo): New function.
22721 (get_stridx_plus_constant): Use it.
22722 (zero_length_string): Likewise.
22723 (adjust_related_strinfos): Likewise.
22724 (adjust_last_stmt): Likewise.
22726 2017-05-31 Richard Biener <rguenther@suse.de>
22729 * config/i386/i386.c (ix86_expand_builtin): Remove assert
22730 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
22732 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22734 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
22735 loop_vinfo argument and use of dependence distance vectors.
22736 Check instead whether the two references differ only in their
22737 initial value and assume that they have the same alignment if the
22738 difference is a multiple of the vector alignment.
22739 (vect_analyze_data_refs_alignment): Update call accordingly.
22741 2017-05-31 Martin Liska <mliska@suse.cz>
22744 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
22746 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22748 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
22749 (create_intersect_range_checks): Move from ...
22750 * tree-data-ref.c (create_intersect_range_checks_index)
22751 (create_intersect_range_checks): ... to here.
22752 (create_runtime_alias_checks): New function factored from ...
22753 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
22754 here. Call above function.
22755 * tree-data-ref.h (create_runtime_alias_checks): New function.
22757 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22759 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
22760 segment length for dr_b and compute it in wide_int.
22762 2017-05-31 Richard Biener <rguenther@suse.de>
22764 PR tree-optimization/80906
22765 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
22766 and pass through iv_map.
22767 (copy_bb_and_scalar_dependences): Adjust.
22768 (translate_pending_phi_nodes): Likewise.
22769 (copy_loop_close_phi_args): Handle code-generating IVs instead
22772 2017-05-30 David Malcolm <dmalcolm@redhat.com>
22774 * diagnostic-color.c (color_dict): Add "type-diff".
22775 (parse_gcc_colors): Update comment.
22776 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
22777 -fdiagnostics-show-template-tree and -fno-elide-type.
22778 (GCC_COLORS): Add type-diff to example.
22780 (-fdiagnostics-show-template-tree): New.
22781 (-fno-elide-type): New.
22782 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
22783 the pp_format_decoder callback. Call any m_format_postprocessor's
22785 (pretty_printer::pretty_printer): Initialize
22786 m_format_postprocessor.
22787 (pretty_printer::~pretty_printer): Delete any
22788 m_format_postprocessor.
22789 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
22790 (class format_postprocessor): New class.
22791 (struct pretty_printer::format_decoder): Document the new parameters.
22792 (struct pretty_printer::m_format_postprocessor): New field.
22793 * tree-diagnostic.c (default_tree_printer): Update for new
22794 bool and const char ** params.
22795 * tree-diagnostic.h (default_tree_printer): Likewise.
22797 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
22799 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
22800 (lwa_operand): Delete rs6000_gen_cell_microcode test.
22801 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22802 rs6000_gen_cell_microcode code.
22803 (rs6000_final_prescan_insn): Delete.
22804 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
22805 "warn-cell-microcode" entries.
22806 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
22807 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
22808 throughout. Change cc_reg_not_micro_cr0_operand to
22809 cc_reg_not_cr0_operand throughout.
22810 (*extendhi<mode>2_noload): Delete.
22811 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
22812 (mwarn-cell-microcode): Delete.
22813 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
22814 -mgen-cell-microcode and -mwarn-cell-microcode.
22816 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
22819 * config/i386/constraints.md (Yd): New constraint.
22821 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
22822 and (?Yd, r) alternatives. Update insn attributes.
22823 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
22824 and (?*Yd, r) alternatives. Update insn attributes.
22825 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
22827 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
22829 * gimplify.c (gimplify_modify_expr): Don't create a
22830 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
22833 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
22835 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
22837 2017-05-30 Richard Biener <rguenther@suse.de>
22839 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
22840 and reduc_def fields.
22841 (STMT_VINFO_REDUC_TYPE): New define.
22842 (STMT_VINFO_REDUC_DEF): Likewise.
22843 (vect_force_simple_reduction): Adjust prototype.
22844 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
22845 (vect_is_simple_reduction): Remove check_reduction argument.
22846 (vect_force_simple_reduction): Adjust and set
22847 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22848 (vectorizable_reduction): Do not re-do reduction analysis
22849 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22850 * tree-parloops.c (gather_scalar_reductions): Adjust.
22852 2017-05-30 Richard Biener <rguenther@suse.de>
22854 PR middle-end/80901
22855 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
22858 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
22860 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
22861 Introduce unknown_misalignment parameter and remove vf.
22862 (vect_peeling_hash_get_lowest_cost):
22863 Pass unknown_misalignment parameter.
22864 (vect_enhance_data_refs_alignment):
22865 Fix unsupportable data ref treatment.
22867 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22869 * tree-vect-data-refs.c (vect_get_data_access_cost):
22870 Workaround for SLP handling.
22871 (vect_enhance_data_refs_alignment):
22872 Compute costs for doing no peeling at all, compare to the best
22873 peeling costs so far and avoid peeling if cheaper.
22875 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22877 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
22878 Return peeling info and set costs to zero for unlimited cost
22880 (vect_enhance_data_refs_alignment): Also inspect all datarefs
22881 with unknown misalignment. Compute and costs for unknown
22882 misalignment, compare them to the costs for known misalignment
22883 and choose the cheapest for peeling.
22885 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22887 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
22888 (vect_get_peeling_costs_all_drs): Create function.
22889 (vect_peeling_hash_get_lowest_cost):
22890 Use vect_get_peeling_costs_all_drs.
22891 (vect_peeling_supportable): Create function.
22892 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
22894 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22896 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
22897 DR_HAS_NEGATIVE_STEP.
22898 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
22899 (vect_enhance_data_refs_alignment): Use.
22900 (vect_duplicate_ssa_name_ptr_info): Use.
22901 * tree-vectorizer.h (dr_misalignment): Use.
22902 (known_alignment_for_access_p): Use.
22904 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22907 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
22909 (has_section_name): New function.
22911 2017-05-30 Martin Liska <mliska@suse.cz>
22914 * auto-profile.c (get_function_decl_from_block): Fix
22917 2017-05-30 Richard Biener <rguenther@suse.de>
22919 PR middle-end/80876
22920 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
22922 2017-05-30 Martin Liska <mliska@suse.cz>
22924 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
22925 * dumpfile.h (struct dump_file_info): Remove ctors.
22927 2017-05-30 Martin Liska <mliska@suse.cz>
22929 * predict.def: Fix GNU coding style.
22931 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22933 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22934 Mark 'to' argument with ATTRIBUTE_UNUSED.
22936 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22938 * config/xtensa/xtensa.c (xtensa_emit_call): Use
22939 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
22940 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
22943 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
22945 * doc/install.texi (Options specification): Restore entry of
22946 --enable-sjlj-exceptions.
22948 2017-05-27 Michael Eager <eager@eagercon.com>
22951 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
22953 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
22955 * config/microblaze/microblaze.h
22956 (FIXED_REGISTERS): Update in macro.
22957 (CALL_USED_REGISTERS): Update in macro.
22959 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
22961 * doc/install.texi: Add links to macOS binary distributions.
22963 2017-05-27 Jakub Jelinek <jakub@redhat.com>
22967 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22969 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22971 2017-05-26 Martin Liska <mliska@suse.cz>
22973 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
22975 2017-05-26 Martin Liska <mliska@suse.cz>
22977 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
22978 always leading ';; '.
22979 (dump_bb_info): Likewise.
22980 (brief_dump_cfg): Likewise.
22981 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
22982 * dumpfile.c: Remove usage of TDF_VERBOSE.
22983 * dumpfile.h (enum dump_kind): Likewise.
22984 (dump_gimple_bb_header): Do not use TDF_COMMENT.
22985 * print-tree.c (debug_verbose): Remove.
22986 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
22987 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
22988 * tree-diagnostic.c (default_tree_printer): Replace
22989 TDF_DIAGNOSTIC with TDF_SLIM.
22991 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22993 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
22994 in parameter loop, rather than loop_vinfo.
22995 (create_intersect_range_checks): Ditto.
22996 (vect_create_cond_for_alias_checks): Update call to above functions.
22998 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23000 PR tree-optimization/80815
23001 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
23002 for merging runtime alias checks. Handle negative DR_STEPs.
23004 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23006 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
23008 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
23009 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
23010 out code pruning runtime alias checks.
23011 * tree-data-ref.c (prune_runtime_alias_test_list): New function
23012 factored out from above.
23013 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23015 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23017 (prune_runtime_alias_test_list): New decalaration.
23019 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23021 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23022 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23023 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23024 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23025 (operator==, comp_dr_with_seg_len_pair): Ditto.
23026 (vect_prune_runtime_alias_test_list): Ditto.
23028 2017-05-26 Martin Liska <mliska@suse.cz>
23031 * params.def: Bound partial-inlining-entry-probability param.
23033 2017-05-26 Marek Polacek <polacek@redhat.com>
23036 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23039 2017-05-26 Richard Biener <rguenther@suse.de>
23041 PR tree-optimization/80842
23042 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23045 2017-05-26 Richard Biener <rguenther@suse.de>
23047 PR tree-optimization/80844
23048 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23050 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23052 * doc/md.texi (Machine Constraints): Update x86 family
23053 machine constraints section to match 'config/i386/constraints.md'.
23055 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23057 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23059 2017-05-25 Nathan Sidwell <nathan@acm.org>
23061 * doc/invoke.texi (--enable-languages): Update documentation.
23063 2017-05-25 Martin Liska <mliska@suse.cz>
23065 * dumpfile.c: Add TDF_FOLDING.
23066 * dumpfile.h (enum dump_kind): Likewise.
23067 * genmatch.c (dt_simplify::gen_1): Use it.
23069 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23071 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23073 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23075 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23076 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23078 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23080 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23081 * match.pd (X == C): Rewrite it here.
23082 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23083 with_certain_nonzero_bits2): New predicates.
23084 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23086 2017-05-24 Nathan Sidwell <nathan@acm.org>
23088 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23091 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23094 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23096 * config/powerpcspe: New port. Files are copied from the rs6000
23097 port, with "rs6000" in filenames replaced by "powerpcspe".
23099 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23101 PR rtl-optimization/80754
23102 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23104 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23106 * config/sparc/sparc.md (length): Return the correct value for -mflat
23107 sibcalls to match output_sibcall.
23109 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23113 * config/rs6000/rs6000.c (struct machine_function): Add new field
23115 (rs6000_get_separate_components): Init that field, use it.
23116 (rs6000_components_for_bb): Use the field.
23118 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23120 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23122 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23124 PR middle-end/80823
23125 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23127 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23130 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23131 address against address_operand predicate.
23132 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23134 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23136 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23137 parameters passed indirectly.
23139 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23141 * config/i386/i386.md (*movdi_internal): Remove SSE4
23142 alternative 18 (?r, *v). Update insn attributes.
23143 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23144 Update insn attributes.
23145 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23146 Update insn attributes.
23147 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23148 alternative 1 (r, v). Remove isa attribute.
23149 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23150 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23151 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23153 2017-05-23 Tom de Vries <tom@codesourcery.com>
23155 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23158 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23160 * cgraphunit.c (symbol_table::process_new_functions): Update.
23161 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23162 (inline_generate_summary): Rename to ...
23163 (ipa_fn_summary_generate): ... this one.
23164 (inline_read_summary): Rename to ...
23165 (ipa_fn_summary_read): ... this one.
23166 (inline_write_summary): Rename to ...
23167 (ipa_fn_summary_write): ... this one.
23168 (inline_free_summary): Rename to ...
23169 (ipa_free_fn_summary): ... this one.
23170 (pass_data_local_fn_summary, pass_local_fn_summary,
23171 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23172 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23173 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23174 make_pass_ipa_fn_summary): New.
23175 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23176 inline_write_summary, inline_free_summary): Remove.
23177 (ipa_free_fn_summary) : New.
23178 * ipa-inline.c (ipa_inline): Update.
23179 (pass_ipa_inline): Do not generate summaries.
23180 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23182 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23183 and add pass_ipa_fn_summary.
23184 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23186 (make_pass_inline_parameters): Remove.
23188 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23190 * omp-low.c (struct omp_context): Remove "default_kind" member.
23193 * omp-offload.c (execute_oacc_device_lower): Remove the
23194 parallelism dimensions function attributes for unparallelized
23195 OpenACC kernels constructs.
23197 2017-05-23 Martin Liska <mliska@suse.cz>
23199 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23201 (cgraph_edge::make_speculative): Likewise.
23202 (cgraph_edge::resolve_speculation): Likewise.
23203 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23204 (cgraph_node::dump): Likewise.
23205 * cgraph.h: Likewise.
23206 * cgraphunit.c (analyze_functions): Likewise.
23207 (symbol_table::compile): Likewise.
23208 * ipa-cp.c (print_all_lattices): Likewise.
23209 (determine_versionability): Likewise.
23210 (initialize_node_lattices): Likewise.
23211 (ipcp_verify_propagated_values): Likewise.
23212 (estimate_local_effects): Likewise.
23213 (update_profiling_info): Likewise.
23214 (create_specialized_node): Likewise.
23215 (perhaps_add_new_callers): Likewise.
23216 (decide_about_value): Likewise.
23217 (decide_whether_version_node): Likewise.
23218 (identify_dead_nodes): Likewise.
23219 (ipcp_store_bits_results): Likewise.
23220 * ipa-devirt.c (dump_targets): Likewise.
23221 (ipa_devirt): Likewise.
23222 * ipa-icf.c (sem_item::dump): Likewise.
23223 (sem_function::equals): Likewise.
23224 (sem_variable::equals): Likewise.
23225 (sem_item_optimizer::read_section): Likewise.
23226 (sem_item_optimizer::execute): Likewise.
23227 (congruence_class::dump): Likewise.
23228 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23229 (dump_inline_summary): Likewise.
23230 (estimate_node_size_and_time): Likewise.
23231 (inline_analyze_function): Likewise.
23232 * ipa-inline-transform.c (inline_call): Likewise.
23233 * ipa-inline.c (report_inline_failed_reason): Likewise.
23234 (want_early_inline_function_p): Likewise.
23235 (edge_badness): Likewise.
23236 (update_edge_key): Likewise.
23237 (inline_small_functions): Likewise.
23238 * ipa-profile.c (ipa_profile): Likewise.
23239 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23240 (ipa_make_edge_direct_to_target): Likewise.
23241 (remove_described_reference): Likewise.
23242 (ipa_impossible_devirt_target): Likewise.
23243 (propagate_controlled_uses): Likewise.
23244 (ipa_print_node_params): Likewise.
23245 (ipcp_transform_function): Likewise.
23246 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23247 (propagate_pure_const): Likewise.
23248 * ipa-reference.c (generate_summary): Likewise.
23249 (read_write_all_from_decl): Likewise.
23250 (propagate): Likewise.
23251 (ipa_reference_read_optimization_summary): Likewise.
23252 * ipa-utils.c (ipa_merge_profiles): Likewise.
23253 * ipa.c (walk_polymorphic_call_targets): Likewise.
23254 (symbol_table::remove_unreachable_nodes): Likewise.
23255 (ipa_single_use): Likewise.
23256 * passes.c (execute_todo): Likewise.
23257 * predict.c (drop_profile): Likewise.
23258 * symtab.c (symtab_node::get_dump_name): New function.
23259 (symtab_node::dump_name): Likewise.
23260 (symtab_node::dump_asm_name): Likewise.
23261 (symtab_node::dump_references): Likewise.
23262 (symtab_node::dump_referring): Likewise.
23263 (symtab_node::dump_base): Likewise.
23264 (symtab_node::debug_symtab): Likewise.
23265 * tree-sra.c (convert_callers_for_node): Likewise.
23266 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23267 * value-prof.c (init_node_map): Likewise.
23269 2017-05-23 Martin Liska <mliska@suse.cz>
23271 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23272 and symtab_node::debug_symtab to symbol_table::debug.
23273 * cgraphunit.c (analyze_functions): Use the renamed function.
23274 (symbol_table::compile): Likewise.
23275 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23276 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23277 * passes.c (execute_todo): Likewise.
23278 * symtab.c (symbol_table::dump): New function.
23279 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23281 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23283 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23284 that nonconst implies exec.
23286 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23288 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23289 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23290 (inline_edge_summary_vec): Turn into ...
23291 (ipa_call_summaries): ... this one.
23292 (redirect_to_unreachable, edge_set_predicate,
23293 evaluate_properties_for_edge, inline_summary_alloc,
23294 reset_ipa_call_summary, reset_inline_summary,
23295 inline_summary_t::duplicate): Update.
23296 (inline_edge_duplication_hook): Turn to ...
23297 (ipa_call_summary_t::duplicate): ... this one.
23298 (inline_edge_removal_hook): Turn to ...
23299 (ipa_call_summary_t::remove): ... this one.
23300 (dump_inline_edge_summary): Turn to ...
23301 (dump_ipa_call_summary): ... this one.
23302 (estimate_function_body_sizes): Update.
23303 (inline_update_callee_summaries): Update.
23304 (remap_edge_change_prob): Update.
23305 (remap_edge_summaries): Update.
23306 (inline_merge_summary): Update.
23307 (do_estimate_edge_time): Update.
23308 (inline_generate_summary): Update.
23309 (inline_read_section): Update.
23310 (inline_read_summary): Update.
23311 (inline_free_summary): Update.
23312 * ipa-inline.c (can_inline_edge_p): Update.
23313 (compute_inlined_call_time): Update.
23314 (want_inline_small_function_p): Update.
23315 (edge_badness): Update.
23316 (early_inliner): Update.
23317 * ipa-inline.h (inline_edge_summary): Turn to ...
23318 (ipa_call_summary): ... this one.
23319 (ipa_call_summary_t): New class.
23320 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23321 (ipa_call_summaries): New.
23322 (inline_edge_summary): Remove.
23323 (estimate_edge_growth): Update.
23324 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23325 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23326 * ipa-split.c (execute_split_functions): Update.
23327 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23329 2017-05-23 Tom de Vries <tom@codesourcery.com>
23331 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23332 attributes): Document rdrand effective target.
23334 2017-05-23 Tom de Vries <tom@codesourcery.com>
23336 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23337 attributes): Sort alphabetically.
23339 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23341 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23343 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23346 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23347 V2DF/V2DI splat into two separate patterns, one that handles
23348 registers, and the other that only handles memory. Drop support
23349 for splatting from a GPR on ISA 2.07 and then splitting the
23350 splat into direct move and splat.
23351 (vsx_splat_<mode>_reg): Likewise.
23352 (vsx_splat_<mode>_mem): Likewise.
23354 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23356 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23358 2017-05-22 Jakub Jelinek <jakub@redhat.com>
23360 PR middle-end/80809
23361 * omp-low.c (finish_taskreg_remap): New function.
23362 (finish_taskreg_scan): If unit size of ctx->record_type
23363 is non-constant, unshare the size expression and replace
23364 decls in it with possible outer var refs.
23366 PR middle-end/80809
23367 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
23368 GOVD_SHARED rather than GOVD_PRIVATE with it.
23369 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
23370 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
23372 PR middle-end/80853
23373 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
23374 as last argument to build_outer_var_ref for pointer bases of array
23375 section reductions.
23377 2017-05-19 Martin Sebor <msebor@redhat.com>
23379 * print-tree.c (print_node): Print DECL_READ_P flag.
23381 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23383 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23384 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23385 * cgraph.c: Likewise.
23386 * cgraphunit.c: Likewise.
23387 * gengtype.c: Likewise.
23388 * ipa-cp.c: Likewise.
23389 * ipa-devirt.c: Likewise.
23390 * ipa-icf.c: Likewise.
23391 * ipa-predicate.c: Likewise.
23392 * ipa-profile.c: Likewise.
23393 * ipa-prop.c: Likewise.
23394 * ipa-split.c: Likewise.
23396 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
23397 edge_predicate_pool, dump_inline_hints,
23398 inline_summary::account_size_time, redirect_to_unreachable,
23399 edge_set_predicate, set_hint_predicate,
23400 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
23401 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
23402 inline_summary_t::remove, remap_hint_predicate_after_duplication,
23403 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
23404 ipa_call_summary_t::remove, initialize_growth_caches,
23405 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
23406 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
23407 mark_modified, unmodified_parm_1, unmodified_parm,
23408 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
23409 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
23410 compute_bb_predicates, will_be_nonconstant_expr_predicate,
23411 will_be_nonconstant_predicate, record_modified_bb_info,
23412 get_minimal_bb, record_modified, param_change_prob,
23413 phi_result_unknown_predicate, predicate_for_phi_result,
23414 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
23415 estimate_function_body_sizes, compute_inline_parameters,
23416 compute_inline_parameters_for_curren, pass_data_inline_parameters,
23417 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
23418 inline_update_callee_summaries, remap_edge_change_prob,
23419 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
23420 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
23421 inline_analyze_function, inline_summary_t::insert,
23422 inline_generate_summary, read_ipa_call_summary, inline_read_section,
23423 inline_read_summary, write_ipa_call_summary, inline_write_summary,
23424 inline_free_summary): Move to ipa-fnsummary.h
23425 (predicate_t): Remove.
23426 * ipa-fnsummary.c: New file.
23427 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
23428 (enum inline_hints_vals, inline_hints, agg_position_info,
23429 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
23430 inline_summaries, ipa_call_summary, ipa_call_summary_t,
23431 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
23432 dump_inline_summary, dump_inline_hints, inline_generate_summary,
23433 inline_read_summary, inline_write_summary, inline_free_summary,
23434 inline_analyze_function, initialize_inline_failed,
23435 inline_merge_summary, inline_update_overall_summary,
23436 compute_inline_parameters): Move to ipa-fnsummary.h
23437 * ipa-fnsummary.h: New file.
23438 * ipa-inline-transform.h: Include ipa-inline.h.
23439 * ipa-inline.c: LIkewise.
23441 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23443 * ipa-inline.c (edge_badness): Use inlined_time instead of
23444 inline_summaries->get.
23446 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23448 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
23450 2017-05-22 Nathan Sidwell <nathan@acm.org>
23452 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
23453 (fdump-lang): Document 'raw' option.
23454 * dumpfile.h (TDI_tu): Delete.
23455 * dumpfile.c (dump_files): Remove translation-unit.
23456 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
23458 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
23460 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
23461 command option from $(AWK) call.
23462 * config/avr/genmultilib.awk: Simplify and rewrite so that it
23463 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
23464 [FORMAT]: Remove handling of variable.
23465 * config/avr/t-multilib: Regenerate.
23467 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23469 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
23471 (dump_inline_summary): Do not print self_time.
23472 (estimate_function_body_sizes): Do not set self_time.
23473 (compute_inline_parameters): Likewise.
23474 (inline_read_section, inline_write_summary): Do not stream self_time.
23475 * ipa-inline.h (inline_summary): Drop self_time.
23477 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23479 * ipa-inline-analysis.c (account_size_time): Rename to ...
23480 (inline_summary::account_size_time): ... this one.
23481 (reset_ipa_call_summary): Turn to ...
23482 (ipa_call_summary::reset): ... this one.
23483 (reset_inline_summary): Turn to ...
23484 (inline_summary::reset): ... this one.
23485 (inline_summary_t::remove): Update.
23486 (inline_summary_t::duplicate): Update.
23487 (ipa_call_summary_t::remove): Update.
23488 (dump_inline_summary): Update.
23489 (estimate_function_body_sizes): Update.
23490 (compute_inline_parameters): Update.
23491 (estimate_node_size_and_time): Update.
23492 (inline_merge_summary): Update.
23493 (inline_update_overall_summary): Update.
23494 (inline_read_section): Update.
23495 (inline_write_summary): Update.
23496 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
23497 add account_size_time and reset member functions.
23498 (ipa_call_summary): Add reset function.
23499 * ipa-predicate.h (predicate::operator &): Constify.
23501 2017-05-22 Richard Biener <rguenther@suse.de>
23503 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
23505 2017-05-19 Jason Merrill <jason@redhat.com>
23507 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
23509 2017-05-19 Marek Polacek <polacek@redhat.com>
23512 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
23513 TYPE_OVERFLOW_WRAPS checks.
23515 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
23517 * tree-core.h (enum omp_clause_default_kind): Add
23518 "OMP_CLAUSE_DEFAULT_PRESENT".
23519 * tree-pretty-print.c (dump_omp_clause): Handle it.
23520 * gimplify.c (enum gimplify_omp_var_data): Add
23521 "GOVD_MAP_FORCE_PRESENT".
23522 (gimplify_adjust_omp_clauses_1): Map it to
23523 "GOMP_MAP_FORCE_PRESENT".
23524 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
23526 * gimplify.c (oacc_default_clause): Clarify.
23528 2017-05-19 Nathan Sidwell <nathan@acm.org>
23530 LANG_HOOK_REGISTER_DUMPS
23531 * toplev.c (general_init): Call register dump lang hook.
23532 * doc/invoke.texi: Document -fdump-lang option family.
23533 * dumpfile.c (dump_files): Remove class dump here.
23534 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
23535 * dumpfile.h (tree_dump_index): Remove TDI_class.
23536 * langhooks-def.h (lhd_register_dumps): Declare.
23537 (LANG_HOOKS_REGISTER_DUMPS): Define.
23538 (LANG_HOOKS_INITIALIZER): Add it.
23539 * langhooks.c (lhd_register_dumps): Define.
23540 * langhooks.h (struct lang_hooks): Add register_dumps.
23542 2017-05-19 Nathan Sidwell <nathan@acm.org>
23544 * context.h (context::set_passes): New.
23545 * context.c (context::context): Do not create pass manager.
23546 * toplev.c (general_init): Create pass manager here.
23548 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
23550 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
23551 use this splitter if two add or or instructions would also work for
23552 the constant we want to generate.
23554 2017-05-19 Richard Biener <rguenther@suse.de>
23557 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
23558 predicate evaluation.
23560 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23562 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
23564 * ipa-inline.c (want_inline_small_function_p): Do not cast to
23567 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23569 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23570 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23571 (inline_edge_summary_vec): Turn into ...
23572 (ipa_call_summaries): ... this one.
23573 (redirect_to_unreachable, edge_set_predicate,
23574 evaluate_properties_for_edge, inline_summary_alloc,
23575 reset_ipa_call_summary, reset_inline_summary,
23576 inline_summary_t::duplicate): Update.
23577 (inline_edge_duplication_hook): Turn to ...
23578 (ipa_call_summary_t::duplicate): ... this one.
23579 (inline_edge_removal_hook): Turn to ...
23580 (ipa_call_summary_t::remove): ... this one.
23581 (dump_inline_edge_summary): Turn to ...
23582 (dump_ipa_call_summary): ... this one.
23583 (estimate_function_body_sizes): Update.
23584 (inline_update_callee_summaries): Update.
23585 (remap_edge_change_prob): Update.
23586 (remap_edge_summaries): Update.
23587 (inline_merge_summary): Update.
23588 (do_estimate_edge_time): Update.
23589 (inline_generate_summary): Update.
23590 (inline_read_section): Update.
23591 (inline_read_summary): Update.
23592 (inline_free_summary): Update.
23593 * ipa-inline.c (can_inline_edge_p): Update.
23594 (compute_inlined_call_time): Update.
23595 (want_inline_small_function_p): Update.
23596 (edge_badness): Update.
23597 (early_inliner): Update.
23598 * ipa-inline.h (inline_edge_summary): Turn to ...
23599 (ipa_call_summary): ... this one.
23600 (ipa_call_summary_t): New class.
23601 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23602 (ipa_call_summaries): New.
23603 (inline_edge_summary): Remove.
23604 (estimate_edge_growth): Update.
23605 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23606 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23607 * ipa-split.c (execute_split_functions): Update.
23608 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23610 2017-05-19 Richard Biener <rguenther@suse.de>
23612 PR middle-end/80764
23613 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
23615 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
23617 * config/rs6000/rs6000.c (struct machine_function): Add field
23618 fpr_is_wrapped_separately.
23619 (rs6000_get_separate_components): Use 64 components. Handle the
23620 new FPR components.
23621 (rs6000_components_for_bb): Handle the FPR components.
23622 (rs6000_emit_prologue_components): Handle the FPR components.
23623 (rs6000_emit_epilogue_components): Handle the FPR components.
23624 (rs6000_set_handled_components): Handle the FPR components.
23625 (rs6000_emit_prologue): Don't output prologue code for those FPRs
23626 that are already separately shrink-wrapped.
23627 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
23628 that are already separately shrink-wrapped.
23630 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23633 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
23636 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
23637 (define_peephole2 for Altivec d-form load): Add peepholes to catch
23638 cases where the register allocator uses a move and an offsettable
23639 memory operation to/from a FPR register on ISA 2.06/2.07.
23640 (define_peephole2 for Altivec d-form store): Likewise.
23642 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
23645 * config/i386/mmx.md (*mov<mode>_internal): Enable
23646 alternatives 11, 12, 13 and 14 also for 32bit targets.
23647 Remove alternatives 15, 16, 17 and 18.
23648 * config/i386/sse.md (vec_concatv2di): Change
23649 alternative (!x, *y) to (x, ?!*Yn).
23651 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
23653 * dumpfile.h (enum dump_kind): Remove stray comma.
23655 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23657 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
23658 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
23659 predicate::num_conditions
23660 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
23661 (CHANGED): turn into predicate::changed.
23662 (agg_position_info): Move to ipa-predicate.h
23663 (add_condition, predicate::add_clause, predicate::operator &=,
23664 predicate::or_with, predicate::evaluate, predicate::probability,
23665 dump_condition, dump_clause, predicate::dump,
23666 predicate::remap_after_duplication, predicate::remap_after_inlining,
23667 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
23668 (evaluate_conditions_for_known_args): Update.
23669 (set_cond_stmt_execution_predicate): Update.
23670 * ipa-inline.h: Include ipa-predicate.h
23671 (condition, inline_param_summary, conditions, agg_position_info,
23672 predicate): Move to ipa-predicate.h
23673 * ipa-predicate.c: New file.
23674 * ipa-predicate.h: New file.
23676 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
23678 * final.c (leaf_function_p): Check we are not in a sequence.
23680 2017-05-18 Martin Liska <mliska@suse.cz>
23682 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
23683 * dumpfile.c (dump_register): Use new enum dump_kind.
23684 (get_dump_file_name): Likewise.
23685 (dump_enable_all): Likewise.
23686 (dump_switch_p_1): Likewise.
23687 (enable_rtl_dump_file): Remove usage of TDF_RTL.
23688 * dumpfile.h (enum dump_kind): New enum type.
23689 (struct dump_file_info): Create constructor and
23690 format fields and comments.
23691 * passes.c (pass_manager::register_one_dump_file):
23693 * statistics.c (statistics_early_init): Likewise.
23694 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
23695 TDF_TREE with TDF_SLIM.
23696 (gather_memory_references_ref): Likewise.
23698 2017-05-18 Martin Liska <mliska@suse.cz>
23700 * vec.h (struct vnull): Use it.
23702 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23704 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
23705 (true_predicate, false_predicate, true_predicate_p,
23706 false_predicate_p): Remove.
23707 (single_cond_predicate, not_inlined_predicate): Turn to member function
23709 (add_condition): Update.
23710 (add_clause): Turn to...
23711 (predicate::add_clause): ... this one; update; allow passing NULL
23713 (and_predicates): Turn to ...
23714 (predicate::operator &=): ... this one.
23715 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
23716 (or_predicates): Turn to ...
23717 (predicate::or_with): ... this one.
23718 (evaluate_predicate): Turn to ...
23719 (predicate::evaluate): ... this one.
23720 (predicate_probability): Turn to ...
23721 (predicate::probability): ... this one.
23722 (dump_condition): Update.
23723 (dump_predicate): Turn to ...
23724 (predicate::dump): ... this one.
23725 (account_size_time): Update.
23726 (edge_set_predicate): Update.
23727 (set_hint_predicate): UPdate.
23728 (evaluate_conditions_for_known_args): Update.
23729 (evaluate_properties_for_edge): Update.
23730 (remap_predicate_after_duplication): Turn to...
23731 (predicate::remap_after_duplication): ... this one.
23732 (remap_hint_predicate_after_duplication): Update.
23733 (inline_summary_t::duplicate): UPdate.
23734 (dump_inline_edge_summary): Update.
23735 (dump_inline_summary): Update.
23736 (set_cond_stmt_execution_predicate): Update.
23737 (set_switch_stmt_execution_predicate): Update.
23738 (compute_bb_predicates): Update.
23739 (will_be_nonconstant_expr_predicate): Update.
23740 (will_be_nonconstant_predicate): Update.
23741 (phi_result_unknown_predicate): Update.
23742 (predicate_for_phi_result): Update.
23743 (array_index_predicate): Update.
23744 (estimate_function_body_sizes): Update.
23745 (estimate_node_size_and_time): Update.
23746 (estimate_ipcp_clone_size_and_time): Update.
23747 (remap_predicate): Rename to ...
23748 (predicate::remap_after_inlining): ... this one.
23749 (remap_hint_predicate): Update.
23750 (inline_merge_summary): Update.
23751 (inline_update_overall_summary): Update.
23752 (estimate_size_after_inlining): Update.
23753 (read_predicate): Rename to ...
23754 (predicate::stream_in): ... this one.
23755 (read_inline_edge_summary): Update.
23756 (write_predicate): Rename to ...
23757 (predicate::stream_out): ... this one.
23758 (write_inline_edge_summary): Update.
23759 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
23760 (clause_t): Turn to uint32_t
23761 (predicate): Turn to class; implement constructor and operators
23763 (size_time_entry): Update.
23764 (inline_summary): Update.
23765 (inline_edge_summary): Update.
23767 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
23769 * fold-const.c (fold_binary_loc): Move transformation...
23770 * match.pd (C - X CMP X): ... here.
23772 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
23774 * config/sparc/sparc.c (sparc_option_override): Set function
23775 alignment for -mcpu=niagara7 to 64 to match the I$ line.
23776 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
23778 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
23780 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
23782 2017-05-18 Marek Polacek <polacek@redhat.com>
23785 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
23786 (pass_ubsan::execute): Call gimple_assign_single_p instead of
23787 gimple_assign_load_p.
23789 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
23791 PR middle-end/80692
23792 * real.c (do_compare): Give decimal_do_compare preference over
23793 comparing just the signs.
23795 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
23797 * doc/md.texi (Canonicalization of Instructions): Describe the
23798 canonical form of instructions that inherently set a condition
23801 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
23803 PR middle-end/80775
23804 * tree-cfg.c: Move deletion of unreachable case statements to after
23805 the merging of consecutive case labels.
23807 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
23809 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
23810 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
23811 restoring of callee-saved registers.
23813 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
23815 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
23816 * config/visium/visium.c (single_set_and_flags): Likewise.
23817 * config/visium/visium.md (Substitutions): Likewise.
23819 2017-05-17 Martin Liska <mliska@suse.cz>
23821 * cfg.c: Introduce dump_flags_t type and
23822 use it instead of int type.
23824 * cfghooks.c: Likewise.
23825 * cfghooks.h (struct cfg_hooks): Likewise.
23826 * cfgrtl.c: Likewise.
23827 * cfgrtl.h: Likewise.
23828 * cgraph.c (cgraph_node::get_body): Likewise.
23829 * coretypes.h: Likewise.
23830 * domwalk.c: Likewise.
23831 * domwalk.h: Likewise.
23832 * dumpfile.c (struct dump_option_value_info): Likewise.
23833 (dump_enable_all): Likewise.
23834 (dump_switch_p_1): Likewise.
23835 (opt_info_switch_p): Likewise.
23836 * dumpfile.h (enum tree_dump_index): Likewise.
23837 (struct dump_file_info): Likewise.
23838 * genemit.c: Likewise.
23839 * generic-match-head.c: Likewise.
23840 * gengtype.c (open_base_files): Likewise.
23841 * gimple-pretty-print.c: Likewise.
23842 * gimple-pretty-print.h: Likewise.
23843 * graph.c (print_graph_cfg): Likewise.
23844 * graphite-scop-detection.c (dot_all_sese): Likewise.
23845 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
23846 * loop-unroll.c (report_unroll): Likewise.
23847 * passes.c (pass_manager::register_one_dump_file): Likewise.
23848 * print-tree.c: Likewise.
23849 * statistics.c: Likewise.
23850 * tree-cfg.c: Likewise.
23851 * tree-cfg.h: Likewise.
23852 * tree-dfa.c: Likewise.
23853 * tree-dfa.h: Likewise.
23854 * tree-dump.c (dump_function): Likewise.
23855 * tree-dump.h (struct dump_info): Likewise.
23856 * tree-pretty-print.c: Likewise.
23857 * tree-pretty-print.h: Likewise.
23858 * tree-ssa-live.c: Likewise.
23859 * tree-ssa-live.h: Likewise.
23860 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
23861 * tree-vect-loop.c: Likewise.
23862 * tree-vect-slp.c: Likewise.
23864 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
23865 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23867 PR tree-optimization/80457
23868 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
23869 of all arguments to a statement as scalar_to_vec operations.
23870 (vectorizable_call): Adjust call to vect_model_simple_cost for
23872 (vectorizable_conversion): Likewise.
23873 (vectorizable_assignment): Likewise.
23874 (vectorizable_shift): Likewise.
23875 (vectorizable_operation): Likewise.
23876 (vectorizable_comparison): Likewise.
23877 (vect_is_simple_cond): Record the def types for operands.
23878 (vectorizable_condition): Likewise, call vect_model_simple_cost.
23879 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
23880 for statement argument count.
23882 2017-05-16 Carl Love <cel@us.ibm.com>
23884 * config/rs6000/rs6000-c: Add support for built-in functions
23885 vector unsigned long long vec_bperm (vector unsigned long long,
23886 vector unsigned char)
23887 vector signed long long vec_mule (vector signed int,
23889 vector unsigned long long vec_mule (vector unsigned int,
23890 vector unsigned int)
23891 vector signed long long vec_mulo (vector signed int,
23893 vector unsigned long long vec_mulo (vector unsigned int,
23894 vector unsigned int)
23895 vector signed char vec_sldw (vector signed char,
23896 vector signed char,
23898 vector unsigned char vec_sldw (vector unsigned char,
23899 vector unsigned char,
23901 vector signed short vec_sldw (vector signed short,
23902 vector signed short,
23904 vector unsigned short vec_sldw (vector unsigned short,
23905 vector unsigned short,
23907 vector signed int vec_sldw (vector signed int,
23910 vector unsigned int vec_sldw (vector unsigned int,
23911 vector unsigned int,
23913 vector signed long long vec_sldw (vector signed long long,
23914 vector signed long long,
23916 vector unsigned long long vec_sldw (vector unsigned long long,
23917 vector unsigned long long,
23919 * config/rs6000/rs6000-c: Add support for built-in functions
23920 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
23921 * config/rs6000/altivec.h: Add defintion for vec_sldw.
23922 * doc/extend.texi: Update the built-in documentation for the
23923 new built-in functions.
23925 2017-05-16 Marek Polacek <polacek@redhat.com>
23929 * tree.c (save_expr): Don't fold the expression.
23931 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
23933 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
23934 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
23935 and (?*y,m). Update insn attributes.
23937 2017-05-16 Martin Liska <mliska@suse.cz>
23939 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
23940 flags argument of print_gimple_stmt, print_gimple_expr,
23941 print_generic_stmt and print_generic_expr.
23942 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
23943 * coretypes.h: Likewise.
23944 * except.c (dump_eh_tree): Likewise.
23945 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
23946 * gimple-pretty-print.h: Likewise.
23947 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
23948 (backprop::push_to_worklist): Likewise.
23949 (backprop::pop_from_worklist): Likewise.
23950 (backprop::process_use): Likewise.
23951 (backprop::intersect_uses): Likewise.
23952 (note_replacement): Likewise.
23953 * gimple-ssa-store-merging.c
23954 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
23955 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
23956 (pass_store_merging::execute): Likewise.
23957 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
23958 (ssa_base_cand_dump_callback): Likewise.
23959 (dump_incr_vec): Likewise.
23960 (replace_refs): Likewise.
23961 (replace_mult_candidate): Likewise.
23962 (create_add_on_incoming_edge): Likewise.
23963 (create_phi_basis): Likewise.
23964 (insert_initializers): Likewise.
23965 (all_phi_incrs_profitable): Likewise.
23966 (introduce_cast_before_cand): Likewise.
23967 (replace_one_candidate): Likewise.
23968 * gimplify.c (gimplify_expr): Likewise.
23969 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
23970 (set_rename): Likewise.
23971 (rename_uses): Likewise.
23972 (copy_loop_phi_nodes): Likewise.
23973 (add_close_phis_to_merge_points): Likewise.
23974 (copy_loop_close_phi_args): Likewise.
23975 (copy_cond_phi_args): Likewise.
23976 (graphite_copy_stmts_from_block): Likewise.
23977 (translate_pending_phi_nodes): Likewise.
23978 * graphite-poly.c (print_pdr): Likewise.
23979 (dump_gbb_cases): Likewise.
23980 (dump_gbb_conditions): Likewise.
23981 (print_scop_params): Likewise.
23982 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
23983 (build_cross_bb_scalars_use): Likewise.
23984 (gather_bbs::before_dom_children): Likewise.
23985 * hsa-dump.c (dump_hsa_immed): Likewise.
23986 * ipa-cp.c (print_ipcp_constant_value): Likewise.
23987 (get_replacement_map): Likewise.
23988 * ipa-inline-analysis.c (dump_condition): Likewise.
23989 (estimate_function_body_sizes): Likewise.
23990 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
23991 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
23992 * ipa-prop.c (ipa_dump_param): Likewise.
23993 (ipa_print_node_jump_functions_for_edge): Likewise.
23994 (ipa_modify_call_arguments): Likewise.
23995 (ipa_modify_expr): Likewise.
23996 (ipa_dump_param_adjustments): Likewise.
23997 (ipa_dump_agg_replacement_values): Likewise.
23998 (ipcp_modif_dom_walker::before_dom_children): Likewise.
23999 * ipa-pure-const.c (check_stmt): Likewise.
24000 (pass_nothrow::execute): Likewise.
24001 * ipa-split.c (execute_split_functions): Likewise.
24002 * omp-offload.c (dump_oacc_loop_part): Likewise.
24003 (dump_oacc_loop): Likewise.
24004 * trans-mem.c (tm_log_emit): Likewise.
24005 (tm_memopt_accumulate_memops): Likewise.
24006 (dump_tm_memopt_set): Likewise.
24007 (dump_tm_memopt_transform): Likewise.
24008 * tree-cfg.c (gimple_verify_flow_info): Likewise.
24009 (print_loop): Likewise.
24010 * tree-chkp-opt.c (chkp_print_addr): Likewise.
24011 (chkp_gather_checks_info): Likewise.
24012 (chkp_get_check_result): Likewise.
24013 (chkp_remove_check_if_pass): Likewise.
24014 (chkp_use_outer_bounds_if_possible): Likewise.
24015 (chkp_reduce_bounds_lifetime): Likewise.
24016 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24017 (chkp_mark_completed_bounds): Likewise.
24018 (chkp_register_incomplete_bounds): Likewise.
24019 (chkp_mark_invalid_bounds): Likewise.
24020 (chkp_maybe_copy_and_register_bounds): Likewise.
24021 (chkp_build_returned_bound): Likewise.
24022 (chkp_get_bound_for_parm): Likewise.
24023 (chkp_build_bndldx): Likewise.
24024 (chkp_get_bounds_by_definition): Likewise.
24025 (chkp_generate_extern_var_bounds): Likewise.
24026 (chkp_get_bounds_for_decl_addr): Likewise.
24027 * tree-chrec.c (chrec_apply): Likewise.
24028 * tree-data-ref.c (dump_data_reference): Likewise.
24029 (dump_subscript): Likewise.
24030 (dump_data_dependence_relation): Likewise.
24031 (analyze_overlapping_iterations): Likewise.
24032 * tree-inline.c (expand_call_inline): Likewise.
24033 (tree_function_versioning): Likewise.
24034 * tree-into-ssa.c (dump_defs_stack): Likewise.
24035 (dump_currdefs): Likewise.
24036 (dump_names_replaced_by): Likewise.
24037 (dump_update_ssa): Likewise.
24038 (update_ssa): Likewise.
24039 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24040 * tree-parloops.c (build_new_reduction): Likewise.
24041 (try_create_reduction_list): Likewise.
24042 (ref_conflicts_with_region): Likewise.
24043 (oacc_entry_exit_ok_1): Likewise.
24044 (oacc_entry_exit_single_gang): Likewise.
24045 * tree-pretty-print.h: Likewise.
24046 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24047 (get_scalar_evolution): Likewise.
24048 (add_to_evolution): Likewise.
24049 (get_loop_exit_condition): Likewise.
24050 (analyze_evolution_in_loop): Likewise.
24051 (analyze_initial_condition): Likewise.
24052 (analyze_scalar_evolution): Likewise.
24053 (instantiate_scev): Likewise.
24054 (number_of_latch_executions): Likewise.
24055 (gather_chrec_stats): Likewise.
24056 (final_value_replacement_loop): Likewise.
24057 (scev_const_prop): Likewise.
24058 * tree-sra.c (dump_access): Likewise.
24059 (disqualify_candidate): Likewise.
24060 (create_access): Likewise.
24061 (reject): Likewise.
24062 (maybe_add_sra_candidate): Likewise.
24063 (create_access_replacement): Likewise.
24064 (analyze_access_subtree): Likewise.
24065 (analyze_all_variable_accesses): Likewise.
24066 (sra_modify_assign): Likewise.
24067 (initialize_constant_pool_replacements): Likewise.
24068 (find_param_candidates): Likewise.
24069 (decide_one_param_reduction): Likewise.
24070 (replace_removed_params_ssa_names): Likewise.
24071 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24072 * tree-ssa-copy.c (dump_copy_of): Likewise.
24073 (copy_prop_visit_cond_stmt): Likewise.
24074 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24075 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24076 (record_equivalences_from_stmt): Likewise.
24077 * tree-ssa-dse.c (compute_trims): Likewise.
24078 (delete_dead_call): Likewise.
24079 (delete_dead_assignment): Likewise.
24080 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24081 (forward_propagate_into_cond): Likewise.
24082 (pass_forwprop::execute): Likewise.
24083 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24084 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24086 (move_computations_worker): Likewise.
24087 (execute_sm): Likewise.
24088 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24089 (remove_exits_and_undefined_stmts): Likewise.
24090 (remove_redundant_iv_tests): Likewise.
24091 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24092 (adjust_iv_update_pos): Likewise.
24093 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24094 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24095 (value_replacement): Likewise.
24096 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24097 * tree-ssa-pre.c (print_pre_expr): Likewise.
24098 (get_representative_for): Likewise.
24099 (create_expression_by_pieces): Likewise.
24100 (insert_into_preds_of_block): Likewise.
24101 (eliminate_insert): Likewise.
24102 (eliminate_dom_walker::before_dom_children): Likewise.
24103 (eliminate): Likewise.
24104 (remove_dead_inserted_code): Likewise.
24105 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24106 * tree-ssa-reassoc.c (get_rank): Likewise.
24107 (eliminate_duplicate_pair): Likewise.
24108 (eliminate_plus_minus_pair): Likewise.
24109 (eliminate_not_pairs): Likewise.
24110 (undistribute_ops_list): Likewise.
24111 (eliminate_redundant_comparison): Likewise.
24112 (update_range_test): Likewise.
24113 (optimize_range_tests_var_bound): Likewise.
24114 (optimize_vec_cond_expr): Likewise.
24115 (rewrite_expr_tree): Likewise.
24116 (rewrite_expr_tree_parallel): Likewise.
24117 (linearize_expr): Likewise.
24118 (break_up_subtract): Likewise.
24119 (linearize_expr_tree): Likewise.
24120 (attempt_builtin_powi): Likewise.
24121 (attempt_builtin_copysign): Likewise.
24122 (transform_stmt_to_copy): Likewise.
24123 (transform_stmt_to_multiply): Likewise.
24124 (dump_ops_vector): Likewise.
24125 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24126 (print_scc): Likewise.
24127 (set_ssa_val_to): Likewise.
24128 (visit_reference_op_store): Likewise.
24129 (visit_use): Likewise.
24130 (sccvn_dom_walker::before_dom_children): Likewise.
24131 (run_scc_vn): Likewise.
24132 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24134 (expr_hash_elt::print): Likewise.
24135 (const_and_copies::pop_to_marker): Likewise.
24136 (const_and_copies::record_const_or_copy_raw): Likewise.
24137 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24138 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24139 (dump_predicates): Likewise.
24140 (find_uninit_use): Likewise.
24141 (warn_uninitialized_phi): Likewise.
24142 (pass_late_warn_uninitialized::execute): Likewise.
24143 * tree-ssa.c (verify_vssa): Likewise.
24144 (verify_ssa): Likewise.
24145 (maybe_optimize_var): Likewise.
24146 * tree-vrp.c (dump_value_range): Likewise.
24147 (dump_all_value_ranges): Likewise.
24148 (dump_asserts_for): Likewise.
24149 (register_edge_assert_for_2): Likewise.
24150 (vrp_visit_cond_stmt): Likewise.
24151 (vrp_visit_switch_stmt): Likewise.
24152 (vrp_visit_stmt): Likewise.
24153 (vrp_visit_phi_node): Likewise.
24154 (simplify_cond_using_ranges_1): Likewise.
24155 (fold_predicate_in): Likewise.
24156 (evrp_dom_walker::before_dom_children): Likewise.
24157 (evrp_dom_walker::push_value_range): Likewise.
24158 (evrp_dom_walker::pop_value_range): Likewise.
24159 (execute_early_vrp): Likewise.
24161 2017-05-16 Richard Biener <rguenther@suse.de>
24163 * dwarf2out.c (loc_list_from_tree_1): Do not create
24164 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24166 2017-05-16 Richard Biener <rguenther@suse.de>
24168 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24170 (note_variable_value_in_expr): If we resolved the decl ref
24171 do not push to the stack.
24173 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24175 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24176 operations in fast-math mode.
24177 (vaddq_f16): Likewise.
24178 (vmul_f16): Likewise.
24179 (vmulq_f16): Likewise.
24180 (vsub_f16): Likewise.
24181 (vsubq_f16): Likewise.
24182 * config/arm/neon.md (add<mode>3): New.
24184 (fma:<VH:mode>3): New. Also remove outdated comment.
24187 2017-05-16 Martin Liska <mliska@suse.cz>
24191 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24192 (odr_types_equivalent_p): Likewise.
24194 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24196 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24198 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24201 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24202 non-interunit SSE move alternatives with '?'.
24203 (zero-extendsidi peephole2): New peephole to skip intermediate
24204 general register in SSE zero-extend sequence.
24206 2017-05-15 Jeff Law <law@redhat.com>
24208 * reorg.c (relax_delay_slots): Create a new variable to hold
24209 the temporary target rather than clobbering TARGET_LABEL.
24211 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24212 missing argument to extract_bit_field call.
24213 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24215 2017-05-15 Martin Liska <mliska@suse.cz>
24218 * gcc.c (process_command): Do not allow empty argument of -o option.
24220 2017-05-15 Renlin Li <renlin.li@arm.com>
24222 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24223 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24224 * config/aarch64/constraints.md (Usf): Add long call check.
24225 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24226 (call_value): Likewise.
24227 (sibcall): Likewise.
24228 (sibcall_value): Likewise.
24230 (call_value_insn): New.
24231 (sibcall_insn): Update rtx pattern.
24232 (sibcall_value_insn): Likewise.
24233 (call_internal): Remove.
24234 (call_value_internal): Likewise.
24235 (sibcall_internal): Likewise.
24236 (sibcall_value_internal): Likewise.
24237 (call_reg): Likewise.
24238 (call_symbol): Likewise.
24239 (call_value_reg): Likewise.
24240 (call_value_symbol): Likewise.
24242 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24245 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24247 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24249 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24250 compatible with CCGOCmode and with CCZmode.
24252 2017-05-14 Martin Sebor <msebor@redhat.com>
24254 PR middle-end/77671
24255 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24256 (gimple_fold_builtin_snprintf): Same.
24257 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24258 (gimple_fold_builtin_snprintf): Same.
24259 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24260 of character types.
24261 (is_call_safe): New function.
24262 (try_substitute_return_value): Call it.
24263 (try_simplify_call): New function.
24264 (pass_sprintf_length::handle_gimple_call): Call it.
24266 2017-05-14 Martin Sebor <msebor@redhat.com>
24268 PR middle-end/80669
24269 * builtins.c (expand_builtin_stpncpy): Simplify.
24271 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24273 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24274 * config/i386/i386.h
24275 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24276 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24277 (struct machine_function): Add new members call_ms2sysv,
24278 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24279 (struct machine_frame_state): New fields sp_realigned and
24280 sp_realigned_offset.
24281 * config/i386/i386.c
24282 (enum xlogue_stub): New enum.
24283 (enum xlogue_stub_sets): New enum.
24284 (class xlogue_layout): New class.
24285 (struct ix86_frame): New fields stack_realign_allocate_offset,
24286 stack_realign_offset and outlined_save_offset. Modify comments to
24287 detail stack layout when using out-of-line stubs.
24288 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24289 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24290 -mcall-ms2sysv-xlogues.
24291 (stub_managed_regs): New static variable.
24292 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24293 registers managed by out-of-line stub.
24294 (disable_call_ms2sysv_xlogues): New function.
24295 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24296 m->call_ms2sysv when appropriate and compute frame layout for
24298 (sp_valid_at, fp_valid_at): New inline functions.
24299 (choose_basereg): New function.
24300 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24302 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24303 Use align parameter of choose_baseaddr to generated aligned SSE movs
24305 (pro_epilogue_adjust_stack): Modify to track
24306 machine_frame_state::sp_realigned.
24307 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24308 (ix86_nsaved_sseregs): Likewise.
24309 (ix86_emit_save_regs): Likewise.
24310 (ix86_emit_save_regs_using_mov): Likewise.
24311 (ix86_emit_save_sse_regs_using_mov): Likewise.
24312 (get_scratch_register_on_entry): Likewise.
24313 (gen_frame_set): New function.
24314 (gen_frame_load): Likewise.
24315 (gen_frame_store): Likewise.
24316 (emit_outlined_ms2sysv_save): Likewise.
24317 (emit_outlined_ms2sysv_restore): Likewise.
24318 (ix86_expand_prologue): Modify stack re-alignment code and call
24319 emit_outlined_ms2sysv_save when appropriate.
24320 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24321 parameter rtx_insn *insn, which allows the function to be used to only
24322 generate the notes.
24323 (ix86_expand_epilogue): Modify validity checks of frame and stack
24324 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24325 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24326 * config/i386/predicates.md
24327 (save_multiple): New predicate.
24328 (restore_multiple): Likewise.
24329 * config/i386/sse.md
24330 (save_multiple<mode>): New pattern.
24331 (save_multiple_realign<mode>): Likewise.
24332 (restore_multiple<mode>): Likewise.
24333 (restore_multiple_and_return<mode>): Likewise.
24334 (restore_multiple_leave_return<mode>): Likewise.
24335 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24337 2017-05-14 Julia Koval <julia.koval@intel.com>
24339 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24340 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24341 (__builtin_ia32_xsetbv): New builtins.
24342 * config/i386/i386.c (ix86_expand_special_args_builtin):
24344 (ix86_expand_builtin): Special expand for new intrinsics.
24345 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24346 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24347 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24349 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24351 * cfganal.c (inverted_post_order_compute): Change argument type
24353 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24354 * df-core.c (rest_of_handle_df_initialize): Adjust.
24355 (rest_of_handle_df_finish): Likewise.
24356 (df_analyze_1): Likewise.
24357 (df_analyze): Likewise.
24358 (loop_inverted_post_order_compute): Change argument to be a vec *.
24359 (df_analyze_loop): Adjust.
24360 (df_get_n_blocks): Likewise.
24361 (df_get_postorder): Likewise.
24362 * df.h (struct df_d): Change field to be a vec.
24363 * lcm.c (compute_laterin): Adjust.
24364 (compute_available): Likewise.
24365 * lra-lives.c (lra_create_live_ranges_1): Likewise.
24366 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
24367 * tree-ssa-pre.c (compute_antic): Likewise.
24369 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24371 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
24372 (depth_first_search::depth_first_search): Change structure init
24373 function to this constructor.
24374 (depth_first_search::add_bb): Rename function to this member.
24375 (depth_first_search::execute): Likewise.
24376 (flow_dfs_compute_reverse_finish): Adjust.
24378 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24380 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24381 (longest_simple_path): Likewise.
24382 * shrink-wrap.c (spread_components): Likewise.
24383 (disqualify_problematic_components): Likewise.
24384 (emit_common_heads_for_components): Likewise.
24385 (emit_common_tails_for_components): Likewise.
24386 (insert_prologue_epilogue_for_components): Likewise.
24388 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24390 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
24393 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24395 * df-core.c (df_set_blocks): Start using auto_bitmap.
24396 (df_compact_blocks): Likewise.
24397 * df-problems.c (df_rd_confluence_n): Likewise.
24398 * df-scan.c (df_insn_rescan_all): Likewise.
24399 (df_process_deferred_rescans): Likewise.
24400 (df_update_entry_block_defs): Likewise.
24401 (df_update_exit_block_uses): Likewise.
24402 (df_entry_block_bitmap_verify): Likewise.
24403 (df_exit_block_bitmap_verify): Likewise.
24404 (df_scan_verify): Likewise.
24405 * lra-constraints.c (lra_constraints): Likewise.
24406 (undo_optional_reloads): Likewise.
24407 (lra_undo_inheritance): Likewise.
24408 * lra-remat.c (calculate_gen_cands): Likewise.
24409 (do_remat): Likewise.
24410 * lra-spills.c (assign_spill_hard_regs): Likewise.
24411 (spill_pseudos): Likewise.
24412 * tree-ssa-pre.c (bitmap_set_and): Likewise.
24413 (bitmap_set_subtract_values): Likewise.
24415 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24417 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
24418 management with auto_bitmap.
24419 (fix_inter_tick): Likewise.
24420 (fix_recovery_deps): Likewise.
24421 * ira.c (add_store_equivs): Likewise.
24422 (find_moveable_pseudos): Likewise.
24423 (split_live_ranges_for_shrink_wrap): Likewise.
24424 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
24425 (rtx_reuse_manager::seen_def_p): Likewise.
24426 (rtx_reuse_manager::set_seen_def): Likewise.
24427 * print-rtl.h (class rtx_reuse_manager): Likewise.
24429 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24431 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
24433 (migrate_btr_def): Likewise.
24434 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
24435 * df-core.c (loop_post_order_compute): Likewise.
24436 (loop_inverted_post_order_compute): Likewise.
24437 * hsa-common.h: Likewise.
24438 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
24439 * init-regs.c (initialize_uninitialized_regs): Likewise.
24440 * ipa-inline.c (resolve_noninline_speculation): Likewise.
24441 (inline_small_functions): Likewise.
24442 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
24443 * ira.c (combine_and_move_insns): Likewise.
24444 (build_insn_chain): Likewise.
24445 * loop-invariant.c (find_invariants): Likewise.
24446 * lower-subreg.c (propagate_pseudo_copies): Likewise.
24447 * predict.c (tree_predict_by_opcode): Likewise.
24448 (predict_paths_leading_to): Likewise.
24449 (predict_paths_leading_to_edge): Likewise.
24450 (estimate_loops_at_level): Likewise.
24451 (estimate_loops): Likewise.
24452 * shrink-wrap.c (try_shrink_wrapping): Likewise.
24453 (spread_components): Likewise.
24454 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
24455 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
24456 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
24457 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
24458 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
24459 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
24460 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
24461 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
24462 (mark_threaded_blocks): Likewise.
24463 (thread_through_all_blocks): Likewise.
24464 * tree-ssa.c (verify_ssa): Likewise.
24465 (execute_update_addresses_taken): Likewise.
24466 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
24468 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24470 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
24472 (post_order_compute): Likewise.
24473 (inverted_post_order_compute): Likewise.
24474 (pre_and_rev_post_order_compute_fn): Likewise.
24476 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24478 * genrecog.c (int_set::int_set): Explicitly construct our
24479 auto_vec base class.
24480 * vec.h (auto_vec::auto_vec): New constructor.
24482 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24484 * bitmap.h (class auto_bitmap): New constructor taking
24485 bitmap_obstack * argument.
24487 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24489 * bitmap.h (class auto_bitmap): Change type of m_bits to
24490 bitmap_head, and adjust ctor / dtor and member operators.
24492 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
24494 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
24495 when returned register mode doesn't match original mode.
24497 2017-05-12 Jeff Law <law@redhat.com>
24498 Jakub Jelinek <jakub@redhat.com>
24500 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
24501 we look for cc setter after the compare-elim changes.
24502 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
24503 within the vector to match what compare-elim now expects.
24504 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
24505 (xorsi3_flags, one_cmplsi2_flags): Likewise.
24507 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
24508 after the compare-elim changes.
24509 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
24510 the vector to match what compare-elim now expects.
24511 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
24512 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
24513 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
24514 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
24515 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
24517 * config/visium/visium.c (single_set_and_flags): Fix where
24518 we look for cc setter after the compare-elim changes.
24519 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
24520 with the vector to match what compare-elim now expects.
24521 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
24522 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
24523 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
24524 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
24525 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
24526 (neg<mode>2_insn_set_overflow): Likewise.
24528 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
24530 PR middle-end/79794
24531 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
24532 maybe_expand_insn call, set ops[0].target. If still set after call,
24533 set alt_rtl. Add extra arg to recursive calls.
24534 (extract_bit_field): Add alt_rtl argument. Pass to
24536 * expmed.h (extract_bit_field): Fix prototype.
24537 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
24538 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
24539 to extract_bit_field_calls.
24540 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
24541 Pass alt_rtl to extract_bit_field calls.
24542 * calls.c (store_unaligned_arguments_into_psuedos)
24543 load_register_parameters): Pass extra NULL to extract_bit_field calls.
24544 * optabs.c (maybe_legitimize_operand): Clear op->target when call
24546 * optabs.h (struct expand_operand): Add target bitfield.
24548 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24550 * compare-elim.c (try_eliminate_compare): Canonicalize
24551 operation with embedded compare to
24552 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
24553 (set (reg) (operation)].
24555 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
24557 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24560 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
24561 cost of adding a carry flag for ADC instruction.
24562 [case MINUS]: Ignore the cost of subtracting a carry flag
24563 for SBB instruction.
24565 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
24567 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
24569 * config/rs6000/bmiintrin.h: New file.
24570 * config/rs6000/bmi2intrin.h: New file.
24571 * config/rs6000/x86intrin.h: New file.
24573 2017-05-12 Jeff Law <law@redhat.com>
24575 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
24578 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
24580 PR middle-end/80707
24581 * tree-cfg.c: Remove cfg edges of unreachable case statements.
24583 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24585 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24586 early expansion of vector divide builtins.
24587 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
24588 builtins identified as having unsigned arguments.
24590 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24592 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
24593 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
24594 expansion of vector logical operations (and, andc, or, xor,
24597 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24599 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
24600 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
24602 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24604 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24605 early GIMPLE expansion of vector multiplies.
24607 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24609 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
24610 TARGET_HAVE_MOVT conditional.
24611 (movt splitter): Likewise.
24613 2017-05-12 Richard Biener <rguenther@suse.de>
24615 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
24616 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24617 Fold all stmts not inplace.
24619 2017-05-12 Richard Biener <rguenther@suse.de>
24621 PR tree-optimization/80713
24622 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
24623 inserted_exprs bit for not removed stmts.
24625 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
24627 PR middle-end/69921
24628 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
24629 parallelized" attribute for parallelized OpenACC kernels.
24630 * omp-offload.c (execute_oacc_device_lower): Use it.
24632 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
24633 Set "oacc kernels" attribute.
24634 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
24635 parameter. Adjust all users.
24636 (oacc_fn_attrib_kernels_p): Remove function.
24637 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
24638 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
24639 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
24640 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
24641 assert "oacc kernels" attribute is set.
24643 2017-05-11 Carl Love <cel@us.ibm.com>
24645 * config/rs6000/rs6000-c: Add support for built-in functions
24646 vector unsigned char vec_popcnt (vector signed char)
24647 vector unsigned char vec_popcnt (vector unsigned char)
24648 vector unsigned short vec_popcnt (vector signed short)
24649 vector unsigned short vec_popcnt (vector unsigned short)
24650 vector unsigned int vec_popcnt (vector signed int)
24651 vector unsigned int vec_popcnt (vector unsigned int)
24652 vector unsigned long long vec_popcnt (vector signed long long)
24653 vector unsigned long long vec_popcnt (vector unsigned long long)
24654 vector signed long long vec_slo (vector signed long long,
24655 vector signed char)
24656 vector signed long long vec_slo (vector signed long long,
24657 vector unsigned char)
24658 vector unsigned long long vec_slo (vector unsigned long long,
24659 vector signed char)
24660 vector unsigned long long vec_slo (vector unsigned long long,
24661 vector unsigned char)
24662 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
24663 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
24664 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
24665 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
24666 * doc/extend.texi: Update the built-in documentation file for the
24667 new built-in functions.
24669 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
24671 * attribs.h (sorted_attr_string): Move machine independent
24672 functions for target clone support from the i386 port to common
24673 code. Rename ix86_function_versions to common_function_versions.
24674 Rename make_name to make_unique_name.
24675 (common_function_versions): Likewise.
24676 (make_unique_name): Likewise.
24677 (make_dispatcher_decl): Likewise.
24678 (is_function_default_version): Likewise.
24679 * attribs.c (attr_strcmp): Likewise.
24680 (sorted_attr_string): Likewise.
24681 (common_function_versions): Likewise.
24682 (make_unique_name): Likewise.
24683 (make_dispatcher_decl): Likewise.
24684 (is_function_default_version): Likewise.
24685 * config/i386/i386.c (attr_strcmp): Likewise.
24686 (sorted_attr_string): Likewise.
24687 (ix86_function_versions): Likewise.
24688 (make_name): Likewise.
24689 (make_dispatcher_decl): Likewise.
24690 (is_function_default_version): Likewise.
24691 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
24693 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24696 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
24697 Account for direct move costs for vec_construct of integer
24700 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
24703 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
24704 (UNSPEC_STX_ATOMIC): Ditto.
24705 (loaddi_via_sse): New insn.
24706 (storedi_via_sse): Ditto.
24707 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
24708 Update corresponding peephole2 patterns.
24709 (atomic_storedi_fpu): Ditto.
24711 2017-05-11 Julia Koval <julia.koval@intel.com>
24713 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
24714 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
24716 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
24717 (__builtin_ia32_rsqrt14ss_mask): New builtins.
24718 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
24720 2017-05-11 Nathan Sidwell <nathan@acm.org>
24722 * graphite-poly.c: Include dumpfile.h.
24724 * dumpfle.h (dump_function): Declare here ...
24725 * tree-dump.h (dump_function): ... not here.
24726 * dumpfile.c: #include tree-cfg.h.
24727 (dump_function): Move here from ...
24728 * tree-dump.c (dump_function): ... here.
24729 * gimplify.c: #include splay-tree.h, not tree-dump.h.
24730 * graphite-poly.c: Don't include tree-dump.h.
24731 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
24732 * print-tree.c: Likewise.
24733 * stor-layout.c: Likewise.
24734 * tree-nested.c: Likewise.
24736 * dumpfile.c (dump_start): Use TDF_FLAGS.
24737 (dump_enable_all): Fix TDF_KIND check thinko.
24739 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
24741 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24742 array entries to represent two legal parameterizations of the
24743 overloaded __builtin_cmpb function, as represented by the
24744 P6_OV_BUILTIN_CMPB constant.
24745 (altivec_resolve_overloaded_builtin): Add special case handling
24746 for the __builtin_cmpb function, as represented by the
24747 P6_OV_BUILTIN_CMPB constant.
24748 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
24749 (BU_P6_64BIT_2): New macro.
24750 (BU_P6_OVERLOAD_2): New macro
24751 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
24752 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
24753 (CMPB): Add overload support to represent both 32-bit and 64-bit
24754 compare-bytes function.
24755 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
24756 support for TARGET_CMPB.
24757 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
24758 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24759 documentation of the __builtin_cmpb overloaded built-in function.
24761 2017-05-11 Richard Biener <rguenther@suse.de>
24763 PR tree-optimization/80705
24764 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
24765 bases are not vectorizable.
24767 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24769 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
24770 when counting register pressure.
24772 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24774 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
24775 (iv_ca_more_deps): Renamed to ...
24776 (iv_ca_compare_deps): ... this.
24777 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
24779 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24781 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
24783 (determine_group_iv_costs): ... here.
24784 (find_inv_vars_cb): Record inv var if it's not recorded before.
24786 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24788 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
24789 (get_shiftadd_cost): Ditto.
24791 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24793 * tree-ssa-address.c: Include header file.
24794 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
24796 (add_to_parts): Refactor.
24797 (addr_to_parts): New parameter. Update use of move_hint_to_base.
24798 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
24801 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24803 PR tree-optimization/53090
24804 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
24806 (extract_cond_operands): Detect condition with IV on both sides
24807 and return COMP_IV_EXPR_2.
24808 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
24809 (rewrite_use_compare): Simplify by removing call to function
24810 extract_cond_operands.
24812 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24814 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
24815 (extract_cond_operands): Detect condition comparing against non-
24816 invariant bound and return appropriate enum value.
24817 (find_interesting_uses_cond): Update use of extract_cond_operands.
24818 Handle its return value accordingly.
24819 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
24821 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24823 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
24824 nonlinear iv_use computation in loop invariant sensitive way.
24826 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24828 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
24829 (find_iv_candidates): Call relate_compare_use_with_all_cands.
24831 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24833 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
24834 (dump_cand): Support iv_cand.inv_exprs.
24835 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
24837 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
24840 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24842 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
24844 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
24845 as local function. Include necessary header files.
24846 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
24848 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24850 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
24852 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24854 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
24855 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
24856 RSHIFT_EXPR and BIT_NOT_EXPR.
24858 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24860 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
24861 (adjust_setup_cost): New parameter supporting round up adjustment.
24862 (struct address_cost_data): Delete.
24863 (force_expr_to_var_cost): Don't bound cost with spill_cost.
24864 (split_address_cost, ptr_difference_cost): Delete.
24865 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
24866 (struct ainc_cost_data): New struct.
24867 (get_address_cost_ainc): New function.
24868 (get_address_cost, get_computation_cost): Reimplement.
24869 (determine_group_iv_cost_address): Record inv_expr for all uses of
24871 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
24872 (iv_ca_has_deps): Reimplemented to ...
24873 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
24875 (iv_ca_extend): Call iv_ca_more_deps.
24877 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24879 * tree-ssa-address.c (struct mem_address): Move to header file.
24880 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
24881 * tree-ssa-address.h (struct mem_address): Move from C file.
24882 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
24884 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24886 * tree-affine.h (aff_combination_type): New interface.
24887 (aff_combination_zero_p): Remove static.
24888 (aff_combination_const_p): New interface.
24889 (aff_combination_singleton_var_p): New interfaces.
24891 2017-05-11 Richard Biener <rguenther@suse.de>
24893 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24894 Skip unreachable blocks and destinations.
24895 (eliminate): Move stmt removal and fixup ...
24896 (fini_eliminate): ... here. Skip inserted exprs.
24897 (pass_pre::execute): Move fini_pre after fini_eliminate.
24898 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
24899 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
24900 PRE to get rid of dead code that has invalid SSA form and
24901 split critical edges again.
24903 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24905 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24907 2017-05-11 Richard Biener <rguenther@suse.de>
24909 * passes.c (execute_function_todo): Verify loops if they are
24910 said to be up-to-date.
24911 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
24912 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
24914 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
24917 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
24918 handle calling assemble_external ourself.
24921 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
24922 modes with zero size. Enhance comment.
24924 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24926 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
24927 built-ins for vec_xl and vec_xst with short and char pointer
24930 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24932 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
24933 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
24934 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
24935 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
24936 (_mm_maskz_min_round_ss): New intrinsics.
24937 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
24938 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
24939 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
24940 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
24941 (__builtin_ia32_minss_mask_round): New builtins.
24942 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24943 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24944 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
24946 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
24947 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
24949 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
24952 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24954 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
24955 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
24956 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
24957 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
24958 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
24959 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
24960 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
24961 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24962 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24963 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
24964 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
24965 (__builtin_ia32_mulss_mask_round): New builtins.
24966 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24967 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24968 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
24970 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
24971 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24973 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24976 2017-05-10 Julia Koval <julia.koval@intel.com>
24978 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
24979 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
24980 (_mm256_setr_m128i): New intrinsics.
24982 2017-05-10 Julia Koval <julia.koval@intel.com>
24984 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
24985 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
24986 (_mm_maskz_rcp14_ss): New intrinsics.
24987 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
24988 (__builtin_ia32_rcp14ss_mask): New builtins.
24989 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
24991 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
24993 PR tree-optimization/51513
24994 * tree-cfg.c (gimple_seq_unreachable_p): New function.
24995 (assert_unreachable_fallthru_edge_p): Use it.
24996 (group_case_labels_stmt): Likewise.
24997 * tree-cfg.h: Prototype it.
24998 * stmt.c: Include cfghooks.h and tree-cfg.h.
24999 (emit_case_dispatch_table) <gap_label>: New local variable.
25000 Use it to fill dispatch table gaps.
25001 Test for default_label before updating probabilities.
25002 (expand_case) <default_label>: Remove unneeded initialization.
25003 Test for unreachable default case statement and remove its edge.
25004 Set default_label accordingly.
25005 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
25007 2017-05-10 Carl Love <cel@us.ibm.com>
25009 * config/rs6000/rs6000-c: Add support for built-in functions
25010 vector signed char vec_neg (vector signed char)
25011 vector signed short int vec_neg (vector short int)
25012 vector signed int vec_neg (vector signed int)
25013 vector signed long long vec_neg (vector signed long long)
25014 vector float vec_neg (vector float)
25015 vector double vec_neg (vector double)
25016 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25018 * config/rs6000/altivec.h: Add define for vec_neg
25019 * doc/extend.texi: Update the built-in documentation for the
25020 new built-in functions.
25022 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25024 PR tree-optimization/77644
25025 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25027 2017-05-10 Nathan Sidwell <nathan@acm.org>
25029 * dumpfile.h (TDI_lang_all): New.
25030 (TDF_KIND): New. Renumber others
25031 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25033 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25035 (get_dump_file_name): Adjust suffix generation.
25036 (dump_enable_all): Use TDF_KIND.
25037 * doc/invoke.texi (-fdump-lang-all): Document.
25039 * dumpfile.h: Tabify.
25041 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25044 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25045 Move member access before delete.
25047 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25049 * tree-inline.c (expand_call_inline): Split block at stmt
25052 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25055 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25056 are now unused after splitting mov{sf,sd}_hardfloat.
25057 (f32_lr2): Likewise.
25058 (f32_lm): Likewise.
25059 (f32_lm2): Likewise.
25060 (f32_li): Likewise.
25061 (f32_li2): Likewise.
25062 (f32_lv): Likewise.
25063 (f32_sr): Likewise.
25064 (f32_sr2): Likewise.
25065 (f32_sm): Likewise.
25066 (f32_sm2): Likewise.
25067 (f32_si): Likewise.
25068 (f32_si2): Likewise.
25069 (f32_sv): Likewise.
25070 (f32_dm): Likewise.
25071 (f32_vsx): Likewise.
25072 (f32_av): Likewise.
25073 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25074 For movsf, order stores so the VSX stores occur before the GPR
25075 store which encourages the register allocator to use a traditional
25076 FPR instead of a GPR. For movsd, order the stores so that the GPR
25077 store comes before the VSX stores to allow the power6 to work.
25078 This is due to the power6 not having a 32-bit integer store
25079 instruction from a FPR.
25080 (movsf_hardfloat): Likewise.
25081 (movsd_hardfloat): Likewise.
25083 2017-05-09 Martin Sebor <msebor@redhat.com>
25085 PR translation/80280
25086 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25089 PR translation/80280
25090 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25091 data member added in r247778.
25092 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25094 2017-05-09 Nathan Sidwell <nathan@acm.org>
25096 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25098 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25101 2017-05-09 Marek Polacek <polacek@redhat.com>
25103 * doc/invoke.texi: Fix typo.
25105 2017-05-09 Richard Biener <rguenther@suse.de>
25107 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25108 (vrp_val_is_min): Likewise.
25109 (set_value_range_to_value): Likewise.
25110 (set_value_range_to_nonnegative): Likewise.
25111 (gimple_assign_nonzero_p): Likewise.
25112 (gimple_stmt_nonzero_p): Likewise.
25113 (vrp_int_const_binop): Likewise. Remove unreachable case.
25114 (adjust_range_with_scev): Adjust comments.
25115 (compare_range_with_value): Likewise.
25116 (extract_range_from_phi_node): Likewise.
25117 (test_for_singularity): Likewise.
25119 2017-05-09 Richard Biener <rguenther@suse.de>
25121 * tree-vrp.c (get_single_symbol): Add assert that we don't
25122 get overflowed constants as invariant part.
25123 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25124 checks. Use wi::cmp instead of recursing for integer constants.
25125 (compare_values): Just ignore whether we assumed undefined
25126 overflow instead of failing the compare.
25127 (extract_range_for_var_from_comparison_expr): Add comment before the
25128 TREE_NO_WARNING sets.
25129 (test_for_singularity): Likewise.
25130 (extract_range_from_comparison): Do not disable optimization
25131 when we assumed undefined overflow.
25132 (extract_range_basic): Remove init of unused var.
25134 2017-05-09 Richard Biener <rguenther@suse.de>
25136 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25137 (extract_range_from_multiplicative_op_1): Adjust.
25138 (extract_range_from_binary_expr_1): Use int_const_binop.
25140 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25143 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25144 rs6000_store_data_bypass_p in seven define_bypass directives and
25145 in several comments.
25146 * config/rs6000/rs6000-protos.h: Add prototype for
25147 rs6000_store_data_bypass_p function.
25148 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25149 function implements slightly different (rs6000-specific) semantics
25150 than store_data_bypass_p, returning false rather than aborting
25151 with assertion error when arguments do not satisfy the
25152 requirements of store data bypass.
25153 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25154 rs6000_store_data_bypass_p.
25156 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25158 * config/xtensa/xtensa-protos.h
25159 (xtensa_initial_elimination_offset): New declaration.
25160 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25161 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25162 macro definition, add case for FRAME_POINTER_REGNUM when
25163 FRAME_GROWS_DOWNWARD.
25164 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25165 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25166 xtensa_initial_elimination_offset.
25168 2017-05-08 Nathan Sidwell <nathan@acm.org>
25170 * doc/invoke.texi: Alphabetize -fdump options.
25172 2017-05-08 Martin Sebor <msebor@redhat.com>
25174 PR translation/80280
25175 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25177 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25179 * target.def (compute_frame_layout): New optional target hook.
25180 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25181 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25182 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25184 * reload1.c (verify_initial_elim_offsets): Likewise.
25185 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25186 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25187 (arm_get_frame_offsets): Split up into this ...
25188 (arm_compute_frame_layout): ... and this function.
25190 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25192 * config/aarch64/constraints.md (Usa): New constraint.
25193 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25195 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25197 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25198 with_multilib_list after it has been checked.
25200 2017-05-08 Richard Biener <rguenther@suse.de>
25202 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25203 (bitmap_set_subtract_values): Likewise.
25205 2017-05-08 Richard Biener <rguenther@suse.de>
25207 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25208 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25210 (gimple_stmt_nonzero_warnv_p): Rename to ...
25211 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25213 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25214 (extract_range_basic): Adjust, do not disable propagation on
25215 strict overflow sensitive simplification.
25216 (vrp_visit_cond_stmt): Likewise.
25218 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25220 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25221 body size unconditionally.
25223 2017-05-07 Jeff Law <law@redhat.com>
25226 2017-05-06 Jeff Law <law@redhat.com>
25227 PR tree-optimization/78496
25228 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25231 PR tree-optimization/78496
25232 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25233 (simplify_stmt_using_ranges): Call it.
25234 (vrp_dom_walker::before_dom_children): Extract equivalences
25235 from an ASSERT_EXPR with an equality comparison against a
25238 2017-05-06 Jeff Law <law@redhat.com>
25240 PR tree-optimization/78496
25241 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25244 PR tree-optimization/78496
25245 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25246 (simplify_stmt_using_ranges): Call it.
25247 (vrp_dom_walker::before_dom_children): Extract equivalences
25248 from an ASSERT_EXPR with an equality comparison against a
25251 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25253 * lra-constraints.c (lra_copy_reg_equiv): New function.
25254 (split_reg): Use it to copy equivalence information from the
25255 original register to the spill register.
25257 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25259 PR rtl-optimization/75964
25260 * simplify-rtx.c (simplify_const_relational_operation): Remove
25261 invalid handling of comparisons of integer ABS.
25263 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25265 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25266 initialize to zero.
25267 (init_regs): Remove declaration.
25268 (function_arg_advance_32): Initialize error_p as boolean variable.
25270 2017-05-05 Nathan Sidwell <nathan@acm.org>
25272 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25273 lines. Use for (;;).
25275 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25277 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25278 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25279 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25280 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25281 VF=2 that require versioning.
25283 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25285 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25288 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25290 * diagnostic.h (diagnostic_override_option_index): Convert from
25291 macro to inline function.
25293 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25295 * diagnostic.c (last_module_changed_p): New function.
25296 (set_last_module): New function.
25297 (diagnostic_report_current_module): Convert macro usage to
25298 the above functions.
25299 * diagnostic.h (diagnostic_context::last_module): Strengthen
25300 from const line_map * to const line_map_ordinary *.
25301 (diagnostic_last_module_changed): Delete macro.
25302 (diagnostic_set_last_module): Delete macro.
25304 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25306 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25307 with diagnostic_report_diagnostic.
25308 (diagnostic_n_impl_richloc): Likewise.
25309 * diagnostic.h (report_diagnostic): Delete macro.
25310 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25311 with diagnostic_report_diagnostic.
25312 * substring-locations.c (format_warning_va): Likewise.
25314 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25316 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25317 save/restor of format_spec. Move option-printing code to...
25318 (print_option_information): ...this new function, and
25319 reimplement by simply printing to the pretty_printer,
25320 rather than appending to the format string.
25322 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25324 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25325 handling logic into...
25326 (update_effective_level_from_pragmas): ...this new function.
25328 2017-05-04 Andrew Waterman <andrew@sifive.com>
25330 * config/riscv/riscv.opt (mstrict-align): New option.
25331 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25332 (SLOW_UNALIGNED_ACCESS): Define.
25333 (riscv_slow_unaligned_access): Declare.
25334 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25336 (riscv_slow_unaligned_access): New variable.
25337 (rocket_tune_info): Set slow_unaligned_access to true.
25338 (optimize_size_tune_info): Set slow_unaligned_access to false.
25339 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25340 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25341 (riscv_option_override): Set riscv_slow_unaligned_access.
25342 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25344 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25346 * config/riscv/riscv.md: Unify indentation.
25348 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25353 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25355 (extendsi<mode>2): Add support for doing sign extension via
25356 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25357 don't have ISA 3.0 instructions.
25358 (extendsi<mode>2 splitter): Likewise.
25359 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
25360 generate the normal insns since SImode can now go in vector
25361 registers. Disallow the special UNSPECs needed for previous
25362 machines to hide SImode being used. Add new insns
25363 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
25364 (fix_trunc<mode>si2_stfiwx): Likewise.
25365 (fix_trunc<mode>si2_internal): Likewise.
25366 (fixuns_trunc<mode>si2): Likewise.
25367 (fixuns_trunc<mode>si2_stfiwx): Likewise.
25368 (fctiw<u>z_<mode>_smallint): Likewise.
25369 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
25370 of floating point to 32-bit integer from doing a direct move to
25371 the GPR registers to do a store.
25372 (fctiwz_<mode>): Break long line.
25374 2017-05-05 Bin Cheng <bin.cheng@arm.com>
25376 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
25377 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
25378 (addr_list, addr_offset_valid_p): New.
25379 (split_address_groups): Check offset validity with above function.
25380 (gt-tree-ssa-loop-ivopts.h): Include header file.
25382 2017-05-05 Nathan Sidwell <nathan@acm.org>
25384 * config.gcc (arm*-*-*): Add missing 'fi'.
25386 2017-05-05 Steve Ellcey <sellcey@cavium.com>
25388 * doc/invoke.texi (-fopt-info): Explicitly say order of options
25389 included in -fopt-info does not matter.
25390 * doc/optinfo.texi (-fopt-info): Fix description of default
25391 behavour. Explicitly say order of options included in -fopt-info
25394 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25396 * config.gcc: Allow combinations of aprofile and rmprofile values for
25397 --with-multilib-list.
25398 * config/arm/t-multilib: New file.
25399 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
25400 variables. Remove setting of ISA and floating-point ABI in
25401 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
25402 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
25403 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
25404 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
25406 * config/arm/t-rmprofile: Likewise except for the matches changes.
25407 * doc/install.texi (--with-multilib-list): Document the combination of
25408 aprofile and rmprofile values and warn about pitfalls in doing that.
25410 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25412 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
25413 (movdi_aarch64): Likewise.
25415 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25417 PR tree-optimization/80632
25418 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
25420 (build_arrays): Initialize it for virtual phis.
25421 (fix_phi_nodes): Use it for virtual phis.
25423 PR tree-optimization/80558
25424 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
25425 [x, y] op z into [x op, y op z] for op & or | if conditions
25428 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
25429 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25432 * config/arm/arm.md (use_literal_pool): Remove.
25433 (64-bit immediate split): No longer takes cost into consideration
25434 if arm_disable_literal_pool is enabled.
25435 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
25436 used when arm_disable_literal_pool is enabled.
25437 (arm_max_const_double_inline_cost): Remove use of
25438 arm_disable_literal_pool.
25439 (push_minipool_fix): Add assert.
25440 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
25441 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
25442 (no_literal_pool_sf_immediate): New.
25444 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25446 PR tree-optimization/80613
25447 * tree-ssa-dce.c (propagate_necessity): Remove cases for
25448 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
25450 2017-05-05 Richard Biener <rguenther@suse.de>
25452 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
25454 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
25456 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
25457 of this flag from insn conditions due to removal from r247495.
25459 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25461 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
25463 (arm_early_store_addr_dep_ptr): Likewise.
25464 * config/arm/aarch-common-protos.h
25465 (arm_early_load_addr_dep_ptr): Add prototype.
25466 (arm_early_store_addr_dep_ptr): Likewise.
25467 * config/arm/cortex-a53.md: Add new bypasses.
25469 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25471 * tree.c (next_type_uid): Change type to unsigned.
25472 (type_hash_canon): Decrement back next_type_uid if
25473 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
25474 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
25477 2017-05-04 Martin Sebor <msebor@redhat.com>
25479 * builtins.c: Fix a trivial typo in a comment.
25481 PR middle-end/79234
25482 * builtins.c (check_sizes): Adjust to handle reading past the end.
25483 Avoid printing excessive upper bound of ranges. Use %E to print
25484 tree nodes instead of converting them to %wu.
25485 (expand_builtin_memchr): New function.
25486 (compute_dest_size): Rename...
25487 (compute_objsize): ...to this.
25488 (expand_builtin_memcpy): Adjust.
25489 (expand_builtin_mempcpy): Adjust.
25490 (expand_builtin_strcat): Adjust.
25491 (expand_builtin_strcpy): Adjust.
25492 (check_strncat_sizes): Adjust.
25493 (expand_builtin_strncat): Adjust.
25494 (expand_builtin_strncpy): Adjust and simplify.
25495 (expand_builtin_memset): Adjust.
25496 (expand_builtin_bzero): Adjust.
25497 (expand_builtin_memcmp): Adjust.
25498 (expand_builtin): Handle memcmp.
25499 (maybe_emit_chk_warning): Check strncat just once.
25501 2017-05-04 Martin Sebor <msebor@redhat.com>
25503 PR preprocessor/79214
25504 PR middle-end/79222
25505 PR middle-end/79223
25506 * builtins.c (check_sizes): Add inlining context and issue
25507 warnings even when -Wno-system-headers is set.
25508 (check_strncat_sizes): Same.
25509 (expand_builtin_strncat): Same.
25510 (expand_builtin_memmove): New function.
25511 (expand_builtin_stpncpy): Same.
25512 (expand_builtin): Handle memmove and stpncpy.
25514 2017-05-04 Bin Cheng <bin.cheng@arm.com>
25516 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
25517 which is not used any more.
25519 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25521 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
25523 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25525 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
25526 (cortexa53_tunings): Likewise.
25527 (cortexa57_tunings): Likewise.
25528 (cortexa72_tunings): Likewise.
25529 (cortexa73_tunings): Likewise.
25531 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25533 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
25534 Set loop alignment to 8.
25536 2017-05-04 Martin Sebor <msebor@redhat.com>
25538 PR translation/80280
25539 * builtins.c (expand_builtin_object_size): Add missing quoting to
25540 %D and like directives.
25541 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
25542 (hsa_type_for_tree_type): Same.
25543 (verify_function_arguments): Same.
25544 * symtab.c (symbol_table::change_decl_assembler_name): Same.
25545 * varasm.c (get_section): Same.
25548 2017-05-04 Martin Sebor <msebor@redhat.com>
25550 PR translation/80280
25551 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
25553 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25555 * config/aarch64/aarch64.c (generic_addrcost_table):
25556 Change HI/TI mode setting.
25558 2017-05-04 Martin Jambor <mjambor@suse.cz>
25560 PR tree-optimization/80622
25561 * tree-sra.c (comes_initialized_p): New function.
25562 (build_accesses_from_assign): Only set write lazily when
25563 comes_initialized_p is false.
25564 (analyze_access_subtree): Use comes_initialized_p.
25565 (propagate_subaccesses_across_link): Assert !comes_initialized_p
25566 instead of testing for PARM_DECL.
25568 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25570 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
25571 constraint on operand 0 to allow more general addressing modes.
25572 Adjust output template.
25573 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
25575 * config/aarch64/aarch64-protos.h
25576 (aarch64_address_valid_for_prefetch_p): Declare prototype.
25577 * config/aarch64/constraints.md (Dp): New address constraint.
25578 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
25581 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
25583 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
25584 update use of estimate_ipcp_clone_size_and_time.
25585 (estimate_local_effects): Update use of
25586 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
25587 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
25588 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
25589 Return nonspecialized time.
25591 2017-05-04 Richard Biener <rguenther@suse.de>
25593 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
25594 for the last VUSE which def dominates the PHI. Directly call
25596 (get_continuation_for_phi_1): Remove.
25598 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
25600 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
25601 to explain the use of truncating division. Cap the number of
25602 iterations to the maximum given by nb_iterations_upper_bound,
25605 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25607 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
25608 * configure: Regenerate.
25609 * config.in: Regenerate.
25610 * config/i386/driver-mingw32.c: new file.
25611 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
25612 * config.host: Link driver-mingw32.o on MinGW host.
25613 * doc/install.texi: Document new --enable-mingw-wildcard configure
25616 2017-05-04 Marek Polacek <polacek@redhat.com>
25618 PR tree-optimization/80612
25619 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
25621 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25622 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
25624 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
25625 (movt splitter): Likewise.
25626 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
25627 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
25628 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
25629 block for Thumb-1 with MOVT.
25630 (thumb2_legitimate_address_p): Move code block ...
25631 (can_avoid_literal_pool_for_label_p): ... into this new function.
25632 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
25634 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
25635 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
25636 "M-profile targets with the MOVT instruction".
25638 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25640 * config/arm/arm-builtins.c (arm_init_builtins): Rename
25641 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
25642 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
25644 2017-05-04 Martin Liska <mliska@suse.cz>
25646 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
25647 variable cond_code.
25649 2017-05-04 Richard Biener <rguenther@suse.de>
25651 * tree.c (array_at_struct_end_p): Handle arrays at struct
25652 end with flexarrays more conservatively. Refactor and treat
25653 arrays of arrays or aggregates more strict. Fix
25654 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
25655 * tree.c (array_at_struct_end_p): Adjust prototype.
25656 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
25657 * gimple-fold.c (get_range_strlen): Likewise.
25658 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
25660 2017-05-04 Richard Biener <rguenther@suse.de>
25662 PR tree-optimization/31130
25663 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
25665 (supports_overflow_infinity): Likewise.
25666 (is_negative_overflow_infinity): Likewise.
25667 (is_positive_overflow_infinity): Likewise.
25668 (is_overflow_infinity): Likewise.
25669 (stmt_overflow_infinity): Likewise.
25670 (overflow_infinity_range_p): Likewise.
25671 (usable_range_p): Remove as always returning true.
25672 (make_overflow_infinity): Remove.
25673 (negative_overflow_infinity): Likewise.
25674 (positive_overflow_infinity): Likewise.
25675 (avoid_overflow_infinity): Likewise.
25676 (set_value_range): Adjust accordingly.
25677 (set_value_range_to_nonnegative): Likewise, remove now unused
25678 overflow_infinity arg.
25679 (vrp_operand_equal_p): Adjust.
25680 (update_value_range): Likewise.
25681 (range_int_cst_singleton_p): Likewise.
25682 (operand_less_p): Likewise.
25683 (compare_values_warnv): Likewise.
25684 (extract_range_for_var_from_comparison_expr): Likewise.
25685 (vrp_int_const_binop): Likewise.
25686 (zero_nonzero_bits_from_vr): Likewise.
25687 (extract_range_from_multiplicative_op_1): Likewise.
25688 (extract_range_from_binary_expr_1): Likewise.
25689 (extract_range_from_unary_expr): Likewise.
25690 (extract_range_from_comparison): Likewise.
25691 (extract_range_basic): Likewise.
25692 (adjust_range_with_scev): Likewise.
25693 (compare_ranges): Likewise.
25694 (compare_range_with_value): Likewise.
25695 (dump_value_range): Likewise.
25696 (test_for_singularity): Likewise, remove strict_overflow_p parameter
25698 (simplify_cond_using_ranges): Adjust.
25700 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25702 * brig-builtins.def: Added a builtin for class_f64.
25703 * builtin-types.def: Added a builtin type needed by class_f64.
25705 2017-05-03 Jason Merrill <jason@redhat.com>
25707 * timevar.def: Add TV_CONSTEXPR.
25709 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25711 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
25713 2017-05-03 Martin Jambor <mjambor@suse.cz>
25715 * ipa-prop.c (ipa_update_after_lto_read): Removed.
25716 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
25717 * ipa-cp.c (ipcp_propagate_stage): Do not call
25718 ipa_update_after_lto_read.
25719 * ipa-inline.c (ipa_inline): Likewise.
25721 2017-05-03 Martin Jambor <mjambor@suse.cz>
25723 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
25724 tag. Added a default constructor and a destructor.
25725 (ipa_edge_args_sum_t): New class;
25726 (ipa_edge_args_sum): Declare.
25727 (ipa_edge_args_vector): Remove declaration.
25728 (IPA_EDGE_REF): Use ipa_edge_args_sum.
25729 (ipa_free_edge_args_substructures): Remove declaration.
25730 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
25731 (ipa_edge_args_info_available_for_edge_p): Likewise.
25732 * ipa-prop.c (ipa_edge_args_vector): Removed.
25733 (edge_removal_hook_holder): Likewise.
25734 (edge_duplication_hook_holder): Likewise.
25735 (ipa_edge_args_sum): New variable.
25736 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
25737 ipa_edge_args_vector.
25738 (ipa_free_edge_args_substructures): Likewise.
25739 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
25740 ipa_edge_args_vector.
25741 (ipa_edge_removal_hook): Turned into method
25742 ipa_edge_args_sum_t::remove.
25743 (ipa_edge_duplication_hook): Turned into method
25744 ipa_edge_args_sum_t::duplicate.
25745 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
25746 registering edge hooks.
25747 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
25748 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
25749 ipa_edge_args_sum instead of ipa_edge_args_vector.
25750 * ipa-profile.c (ipa_profile): Likewise.
25752 2017-05-03 Martin Jambor <mjambor@suse.cz>
25754 * symbol-summary.h (function_summary): New method exists.
25755 (function_summary::symtab_removal): Deallocate through release.
25756 (call_summary): New class.
25757 (gt_ggc_mx): New overload.
25758 (gt_pch_nx): Likewise.
25759 (gt_pch_nx): Likewise.
25761 2017-05-03 Jeff Law <law@redhat.com>
25763 PR tree-optimization/78496
25764 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
25765 from simplify_cond_using_ranges. Split off code to walk
25766 backwards through casts into ...
25767 (simplify_cond_using_ranges_2): New function.
25768 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
25769 (execute_vrp): After identifying jump threads, call
25770 simplify_cond_using_ranges_2.
25772 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
25775 * ipa-inline.h (inline_summary): Add ctor.
25776 (create_ggc): Do not use ggc_cleared_alloc.
25778 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
25779 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25781 * gcc.c (handle_braces): Support escaping in switch matching
25783 * doc/invoke.texi (Spec Files): Document it.
25784 Remove superfluous @code markup in items.
25786 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25788 * diagnostic-show-locus.c (struct column_range): New struct.
25789 (get_affected_columns): New function.
25790 (get_printed_columns): New function.
25791 (struct correction): New struct.
25792 (correction::ensure_capacity): New function.
25793 (correction::ensure_terminated): New function.
25794 (struct line_corrections): New struct.
25795 (line_corrections::~line_corrections): New dtor.
25796 (line_corrections::add_hint): New function.
25797 (layout::print_trailing_fixits): Reimplement in terms of the new
25799 (selftest::test_overlapped_fixit_printing): New function.
25800 (selftest::diagnostic_show_locus_c_tests): Call it.
25802 2017-05-03 Nathan Sidwell <nathan@acm.org>
25804 Canonicalize canonical type hashing
25805 * tree.h (type_hash_canon_hash): Declare.
25806 * tree.c (type_hash_list, attribute_hash_list): Move into
25807 type_hash_canon_hash.
25808 (build_type_attribute_qual_variant): Break out hash code calc into
25809 type_hash_canon_hash.
25810 (type_hash_canon_hash): New. Generic type hash computation.
25811 (build_range_type_1, build_array_type_1, build_function_type,
25812 build_method_type_directly, build_offset_type, build_complex_type,
25813 make_vector_type): Call it.
25815 2017-05-03 Richard Biener <rguenther@suse.de>
25817 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
25818 When all DRs have unknown misaligned do not always peel
25819 when there is a store but apply the same costing model as if
25820 there were only loads.
25822 2017-05-03 Richard Biener <rguenther@suse.de>
25825 PR tree-optimization/80492
25826 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25827 compare_base_decls returning dont-know properly.
25829 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
25831 * config/arm/iterators.md (CCSI): New mode iterator.
25832 (arch): New mode attribute.
25833 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
25834 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
25835 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
25836 code iterator for success result mode.
25837 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
25838 the corresponding new insn generators.
25840 2017-05-03 Bin Cheng <bin.cheng@arm.com>
25843 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25844 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25846 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
25848 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
25849 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
25850 (DDR_A): Wrap DDR argument in brackets.
25851 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
25852 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
25853 (DDR_REVERSED_P): Likewise.
25855 2017-05-03 Jakub Jelinek <jakub@redhat.com>
25857 PR tree-optimization/79472
25858 * tree-switch-conversion.c (struct switch_conv_info): Add
25859 contiguous_range and default_case_nonstandard fields.
25860 (collect_switch_conv_info): Compute contiguous_range and
25861 default_case_nonstandard fields, don't clear final_bb if
25862 contiguous_range and only the default case doesn't have the required
25864 (check_all_empty_except_final): Set default_case_nonstandard instead
25865 of failing if contiguous_range and the default case doesn't have empty
25867 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
25868 and only the default case doesn't have the required constants. Skip
25870 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
25871 if default_case_nonstandard.
25872 (build_constructors): Build constant 1 just once. Assert that default
25873 values aren't inserted in between cases if contiguous_range. Skip
25875 (build_arrays): Skip virtual phis.
25876 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
25877 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
25878 Handle virtual phis.
25879 (gen_inbound_check): Handle default_case_nonstandard case.
25880 (process_switch): Adjust check_final_bb caller. Call
25881 gather_default_values with the first non-default case instead of
25882 default case if default_case_nonstandard.
25884 2017-05-02 Nathan Sidwell <nathan@acm.org>
25886 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
25887 check. Fix formatting.
25889 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
25891 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
25892 errors when comparing specialized and unspecialized times.
25894 2017-05-02 David Malcolm <dmalcolm@redhat.com>
25896 * diagnostic-show-locus.c
25897 (layout::should_print_annotation_line_p): Make private.
25898 (layout::print_annotation_line): Make private.
25899 (layout::annotation_line_showed_range_p): Make private.
25900 (layout::show_ruler): Make private.
25901 (layout::print_source_line): Make private. Pass in line and
25902 line_width, rather than calling location_get_source_line. Drop
25904 (layout::print_leading_fixits): New method.
25905 (layout::print_any_fixits): Rename to...
25906 (layout::print_trailing_fixits): ...this, and make private.
25907 Don't print newline fixits.
25908 (diagnostic_show_locus): Move logic for printing one row into...
25909 (layout::print_line): ...this new function. Move the
25910 location_get_source_line call and error-handling from
25911 print_source_line to here. Call print_leading_fixits, and rename
25912 print_any_fixits to print_trailing_fixits.
25913 (selftest::test_fixit_insert_containing_newline): Update now that
25914 newlines are partially supported.
25915 (selftest::test_fixit_insert_containing_newline_2): New test.
25916 (selftest::test_fixit_replace_containing_newline): Update comments.
25917 (selftest::diagnostic_show_locus_c_tests): Call the new test.
25918 * edit-context.c (class added_line): New class.
25919 (class edited_line): Describe newline handling in comment.
25920 (edited_line::actually_edited_p): New method.
25921 (edited_line::print_content): Delete redundant decl.
25922 (edited_line::m_predecessors): New field.
25923 (edited_file::print_content): Call edited_line::print_content.
25924 (edited_file::print_diff): Update to support newlines.
25925 (edited_file::print_diff_hunk): Likewise.
25926 (edited_file::print_run_of_changed_lines): New function.
25927 (edited_file::print_diff_line): Convert to...
25928 (print_diff_line): ...this.
25929 (edited_file::get_effective_line_count): New function.
25930 (edited_line::edited_line): Initialize new field m_predecessors.
25931 (edited_line::~edited_line): Clean up m_predecessors.
25932 (edited_line::apply_fixit): Handle newlines.
25933 (edited_line::get_effective_line_count): New function.
25934 (edited_line::print_content): New function.
25935 (edited_line::print_diff_lines): New function.
25936 (selftest::test_applying_fixits_insert_containing_newline): New
25938 (selftest::test_applying_fixits_replace_containing_newline): New
25940 (selftest::insert_line): New function.
25941 (selftest::test_applying_fixits_multiple_lines): Add example of
25943 (selftest::edit_context_c_tests): Call the new tests.
25945 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25947 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
25948 parameter cand. Update dump information.
25949 (get_computation_cost): Update uses.
25951 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25953 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
25954 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
25955 (get_computation_at, rewrite_use_address): Update use of
25956 get_computation_aff.
25958 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25960 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
25961 (get_computation): Delete.
25962 (get_computation_cost): Implement like get_computation_cost_at.
25963 Use get_computation_at.
25964 (get_computation_cost_at): Delete.
25965 (rewrite_use_nonlinear_expr): Use get_computation_at.
25966 (rewrite_use_compare, remove_unused_ivs): Ditto.
25968 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25970 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
25972 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25974 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
25975 (ivopts_global_cost_for_size): Rename parameter and update uses.
25976 (iv_ca_recount_cost): Update uses.
25977 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
25978 candidates seperately in n_invs and n_cands.
25979 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
25981 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25983 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
25984 (find_inv_vars_cb): New.
25985 (find_depends): Renamed to ...
25986 (find_inv_vars): ... this.
25987 (add_candidate_1, force_var_cost): Call find_inv_vars.
25988 (split_address_cost, determine_group_iv_cost_cond): Ditto.
25990 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25992 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
25993 inv_vars. Add inv_exprs.
25994 (struct iv_cand): Rename depends_on to inv_vars.
25995 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
25996 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
25997 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
25998 (dump_cand): Dump inv_vars.
25999 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
26000 (record_invariant, find_depends, add_candidate_1): Ditto.
26001 (set_group_iv_cost, force_var_cost): Ditto.
26002 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
26003 (get_computation_cost_at, get_computation_cost): Ditto.
26004 (determine_group_iv_cost_generic): Ditto.
26005 (determine_group_iv_cost_address): Ditto.
26006 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
26007 (determine_group_iv_costs): Ditto.
26008 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
26009 (iv_ca_set_remove_invariants): Renamed to ...
26010 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
26011 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
26012 (iv_ca_set_add_invariants): Renamed to ...
26013 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
26014 (iv_ca_set_cp): Use iv_ca_set_add_invs.
26015 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26016 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26017 (create_new_ivs): Remove useless dump.
26019 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26021 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26023 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26024 (iv_ca_set_no_cp, create_new_iv): Ditto.
26026 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26028 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26030 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26032 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26033 function tree_check2.
26035 2017-05-02 Martin Liska <mliska@suse.cz>
26037 * doc/gcov.texi: Add missing preposition.
26038 * gcov.c (function_info::function_info): Properly fill up
26039 all member variables.
26041 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26043 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26045 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26047 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26049 2017-05-02 Martin Liska <mliska@suse.cz>
26052 * lto-streamer-in.c (lto_read_tree_1): Remove
26053 LTO_STREAMER_DEBUG.
26054 * lto-streamer.c (struct tree_hash_entry): Likewise.
26055 (struct tree_entry_hasher): Likewise.
26056 (tree_entry_hasher::hash): Likewise.
26057 (tree_entry_hasher::equal): Likewise.
26058 (lto_streamer_init): Likewise.
26059 (lto_orig_address_map): Likewise.
26060 (lto_orig_address_get): Likewise.
26061 (lto_orig_address_remove): Likewise.
26062 * lto-streamer.h: Likewise.
26063 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26064 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26066 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26068 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26069 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26070 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26071 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26072 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26073 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26074 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26075 (mm_maskz_sub_ss): New intrinsics.
26076 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26077 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26078 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26079 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26080 (__builtin_ia32_subss_mask_round): New builtins.
26081 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26082 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26083 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26085 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26086 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26088 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26091 2017-05-02 Martin Jambor <mjambor@suse.cz>
26093 PR tree-optimization/78687
26094 * tree-sra.c (access): New field parent.
26095 (process_subtree_disqualification): New function.
26096 (disqualify_candidate): Call it.
26097 (build_accesses_from_assign): Reset write flag if creating an
26099 (build_access_subtree): Fill in parent field and also prpagate
26100 down grp_write flag.
26101 (create_artificial_child_access): New parameter set_grp_write, set
26102 grp_write to its value.
26103 (propagate_subaccesses_across_link): Also propagate grp_write flag
26105 (propagate_all_subaccesses): Push the closest parent back to work
26106 queue if add_access_to_work_queue returned true.
26108 2017-05-02 Richard Biener <rguenther@suse.de>
26110 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26111 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26112 -fstrict-overflow documentation.
26113 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26114 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26115 flag_strict_overflow.
26116 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26117 * lto-opts.c (lto_write_options): Do not stream it.
26118 * lto-wrapper.c (merge_and_complain): Do not handle it.
26119 * opts.c (default_options_table): Do not set -fstrict-overflow.
26120 (finish_options): Likewise do not clear it when sanitizing.
26121 * simplify-rtx.c (simplify_const_relational_operation): Do not
26122 test flag_strict_overflow.
26124 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26126 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26127 using enabled attribute.
26128 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26129 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26130 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26131 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26132 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26133 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26134 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26135 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26136 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26137 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26139 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26141 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26143 2017-05-02 Richard Biener <rguenther@suse.de>
26145 PR tree-optimization/80591
26147 2017-04-10 Richard Biener <rguenther@suse.de>
26149 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26152 2017-05-02 Richard Biener <rguenther@suse.de>
26154 PR tree-optimization/80549
26155 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26156 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26157 headers if they do not have a preheader.
26159 2017-05-02 Martin Liska <mliska@suse.cz>
26162 * common.opt: Fix typo.
26163 * doc/invoke.texi: Likewise.
26165 2017-05-01 Jan Beulich <jbeulich@suse.com>
26167 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26168 swapping, add (x,x,m,x,n) alternative.
26170 2017-05-01 Nathan Sidwell <nathan@acm.org>
26172 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26173 unnecessary unadjusted_alignment check.
26175 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26178 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26180 * gimplify.c (gimplify_cilk_detach): New function.
26181 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26182 * tree-core.h: Document EXPR_CILK_SPAWN.
26183 * tree.h (EXPR_CILK_SPAWN): Define.
26185 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26187 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26188 to use new fixit_hint representation, using the "replace" logic.
26189 (get_line_span_for_fixit_hint): Likewise.
26190 (layout::print_any_fixits): Likewise.
26191 (selftest::test_one_liner_many_fixits): Rename to...
26192 (selftest::test_one_liner_many_fixits_1): ...this, and update
26193 comment and expected output to reflect that the multiple fix-it
26194 hints are now consolidated into one insertion.
26195 (selftest::test_one_liner_many_fixits_2): New test.
26196 (selftest::test_diagnostic_show_locus_one_liner): Update for
26198 (selftest::test_fixit_consolidation): Update for fix-it API
26200 * diagnostic.c (print_parseable_fixits): Likewise.
26201 * edit-context.c (edited_line::m_line_events): Convert from
26202 auto_vec <line_event *> to auto_vec <line_event>.
26203 (class line_event): Convert from abstract base class to a concrete
26204 class, taking over the role of replace_event.
26205 (class insert_event): Delete.
26206 (class replace_event): Rename to class line_event. Convert to
26208 (edit_context::add_fixits): Reimplement.
26209 (edit_context::apply_insert): Delete.
26210 (edit_context::apply_replace): Rename to...
26211 (edit_context::apply_fixit): ...this. Convert to half-open range.
26212 (edited_file::apply_insert): Delete.
26213 (edited_file::apply_replace): Rename to...
26214 (edited_file::apply_fixit): ...this.
26215 (edited_line::~edited_line): Drop deletion of events.
26216 (edited_line::apply_insert): Delete.
26217 (edited_line::apply_replace): Rename to...
26218 (edited_line::apply_fixit): ...this. Convert to half-open range.
26219 Update for change to type of m_line_events.
26220 * edit-context.h (edit_context::apply_insert): Delete.
26221 (edit_context::apply_replace): Rename to...
26222 (edit_context::apply_fixit): ...this.
26224 2017-05-01 Martin Sebor <msebor@redhat.com>
26226 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26229 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26232 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26233 __get_cpuid_max returns 0.
26234 (__get_cpuid_count): Ditto.
26236 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26238 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26239 replacement expression is another instance of one of its arguments.
26241 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26244 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26245 check for stack push/pop autoinc.
26246 * config/i386/i386.c (ix86_agi_dependent): Return false
26247 if the only reason why modified_in_p returned true is that
26248 addr is SP based and set_insn is a push or pop.
26250 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26252 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26255 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26258 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26259 (account_size_time): Use two predicates - exec_pred and
26261 (evaluate_conditions_for_known_args): Compute both clause and
26263 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26264 (inline_summary_t::duplicate): Update.
26265 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26267 (compute_inline_parameters): Likewise.
26268 (estimate_edge_size_and_time): Update caluclation of time.
26269 (estimate_node_size_and_time): Compute both time and nonspecialized
26271 (estimate_ipcp_clone_size_and_time): Update.
26272 (inline_merge_summary): Update.
26273 (do_estimate_edge_time): Update.
26274 (do_estimate_edge_size): Update.
26275 (do_estimate_edge_hints): Update.
26276 (inline_read_section, inline_write_summary): Stream both new predicates.
26277 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26279 (compute_inlined_call_time): Cleanup.
26280 (big_speedup_p): Update.
26281 (edge_badness): Update.
26282 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26283 (size_time_entry): Replace predicate by exec_predicate and
26284 nonconst_predicate.
26285 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26286 (estimate_edge_time): Return also nonspec_time.
26287 (reset_edge_growth_cache): Update.
26289 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26291 PR rtl-optimization/80491
26292 * ifcvt.c (noce_process_if_block): When looking for x setter
26293 with missing else_bb, don't check only the insn right before
26294 cond_earliest, but look for the last insn that x is modified in
26295 within the same bb.
26297 PR rtl-optimization/80491
26298 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26300 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26302 PR tree-optimization/80487
26303 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26305 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26307 PR tree-optimization/79697
26308 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26309 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26310 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26312 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26313 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26315 2017-04-28 Martin Sebor <msebor@redhat.com>
26317 PR tree-optimization/80523
26318 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26319 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26321 (maybe_warn, format_directive, parse_directive): Use new functions.
26322 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26324 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26326 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26328 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26330 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26331 target_header_dir): Set correctly.
26332 * configure: Regenerated.
26333 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26334 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26335 instead of SYSTEM_HEADER_DIR.
26337 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26339 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26340 (estimate_local_effects): Likewise.
26341 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26342 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26343 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26344 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26345 do_estimate_edge_time, estimate_edge_time): Likewise.
26346 * ipa-inline-analysis.c (estimate_node_size_and_time,
26347 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26348 (estimate_time_after_inlining): Remove.
26350 2017-04-28 Martin Liska <mliska@suse.cz>
26352 * doc/gcov.texi: Enhance documentation of gcov.
26354 2017-04-28 Martin Liska <mliska@suse.cz>
26356 * doc/gcov.texi: Sort options in alphabetic order.
26357 * doc/gcov-dump.texi: Likewise.
26358 * doc/gcov-tool.texi: Likewise.
26359 * gcov.c (print_usage): Likewise.
26360 * gcov-dump.c (print_usage): Likewise.
26361 * gcov-tool.c (print_merge_usage_message): Likewise.
26362 (print_rewrite_usage_message): Likewise.
26363 (print_overlap_usage_message): Likewise.
26365 2017-04-28 Martin Liska <mliska@suse.cz>
26367 PR gcov-profile/53915
26368 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
26370 2017-04-28 Martin Liska <mliska@suse.cz>
26372 PR gcov-profile/79891
26373 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
26374 is marked by compiler as living on a line.
26375 (get_cycles_count): Remove usage of the union.
26376 (output_intermediate_file): Likewise.
26377 (find_source): Fix GNU coding style.
26378 (accumulate_line_counts): Remove old non-all block mode.
26379 (output_lines): Remove usage of the union.
26380 * profile.c (output_location): Include all BBs, even if
26381 belonging to a same line (and file) as a previous BB.
26383 2017-04-28 Martin Liska <mliska@suse.cz>
26385 * gcov.c (process_args): Handle new argument 'w'.
26386 (read_graph_file): Assign ID to BBs.
26387 (output_branch_count): Display BB # if verbose flag is set.
26388 (output_lines): Likewise for arcs.
26389 (print_usage): Add '--verbose' option help.
26390 * doc/gcov.texi: Document --verbose (-w) option.
26392 2017-04-28 Martin Liska <mliska@suse.cz>
26394 * gcov.c (struct block_location_info): New struct.
26395 (process_file): Fill up the new structure.
26396 (read_graph_file): Replace usage of encoding by the newly added
26398 (add_line_counts): Likewise.
26399 (accumulate_line_counts): Remove usage of the union.
26400 (function_info::function_info): New function.
26401 (function_info::~function_info): Likewise.
26402 (process_file): Call delete instead of release_function.
26403 (release_function): Release the function.
26404 (release_structures): Call delete instead of release_function.
26405 (solve_flow_graph): Replace usage of num_blocks.
26406 (find_exception_blocks): Likewise.
26407 (output_lines): Fix GNU coding style.
26409 2017-04-28 Martin Liska <mliska@suse.cz>
26412 * coverage.c (coverage_remove_note_file): New function.
26413 * coverage.h: Declare the function.
26414 * toplev.c (finalize): Clean if an error has been seen.
26416 2017-04-28 Martin Liska <mliska@suse.cz>
26418 PR gcov-profile/80031
26419 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
26420 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
26421 * gcov.c (read_graph_file): Read just number of blocks.
26422 * profile.c (branch_prob): Do not stream 0 flags per a basic
26425 2017-04-28 Martin Liska <mliska@suse.cz>
26427 * gcov-dump.c (tag_*): Add new argument to declarations.
26428 (dump_gcov_file): Likewise.
26429 (tag_blocks): Add and use new argument depth.
26430 (tag_arcs): Likewise.
26431 (tag_lines): Likewise.
26432 (tag_counters): Likewise.
26433 (tag_summary): Likewise.
26434 (dump_working_sets): Use depth to do a proper indentation.
26436 2017-04-28 Jakub Jelinek <jakub@redhat.com>
26439 * cgraph.h (symtab_node::debug_symtab): No longer inline.
26440 * symtab.c (symtab_node::debug_symtab): Move definition here.
26442 2017-04-28 Richard Biener <rguenther@suse.de>
26444 * lto-streamer.h (LTO_major_version): Bump to 7.
26446 2017-04-28 Richard Biener <rguenther@suse.de>
26448 * tree-vrp.c (assert_info): New struct.
26449 (add_assert_info): New helper.
26450 (register_edge_assert_for_2): Refactor to add asserts to a vector
26452 (register_edge_assert_for_1): Likewise.
26453 (register_edge_assert_for): Likewise.
26454 (finish_register_edge_assert_for): New helper actually registering
26455 asserts where live on edge.
26456 (find_conditional_asserts): Adjust.
26457 (find_switch_asserts): Likewise.
26458 (evrp_dom_walker::try_find_new_range): Generalize.
26459 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
26461 2017-04-27 Marek Polacek <polacek@redhat.com>
26464 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
26465 arg10 and arg11 to itype.
26467 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
26469 * doc/extend.texi (Object Size Checking): Improve grammar.
26471 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
26474 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
26475 that the logic for permitting reciprocal estimates matches that
26478 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26481 * tree.c (type_cache_hasher::equal): Only compare
26482 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
26483 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
26484 non-aggregate element types.
26485 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
26486 about the flag on ARRAY_TYPEs in the comment, formatting fix.
26488 2017-04-27 Richard Biener <rguenther@suse.de>
26490 PR middle-end/80533
26491 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
26492 stripping ARRAY_REFs from MEM_EXPR make sure we're not
26493 keeping a reference to a trailing array.
26495 2017-04-27 Richard Biener <rguenther@suse.de>
26497 PR middle-end/80539
26498 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
26499 being in loop-closed SSA form conservatively.
26500 (chrec_fold_multiply_poly_poly): Likewise.
26502 2017-04-27 Tamar Christina <tamar.christina@arm.com>
26504 PR middle-end/79665
26505 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
26506 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
26508 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26511 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
26512 (aarch64_function_arg_alignment): Return unsigned int again, but still
26513 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
26514 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26515 Don't emit -Wpsabi note.
26516 (aarch64_function_arg_boundary): Likewise.
26517 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26520 2017-04-26 Nathan Sidwell <nathan@acm.org>
26522 * tree.h (crc32_unsigned_n): Declare.
26523 (crc32_unsigned, crc32_unsigned): Make inline.
26524 * tree.c (crc32_unsigned_bits): Replace with ...
26525 (crc32_unsigned_n): ... this.
26526 (crc32_unsigned, crc32_byte): Remove.
26527 (crc32_string): Remove unnecessary braces.
26529 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26531 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
26532 * ipa-inline-analysis.c (MAX_TIME): Remove.
26533 (account_size_time): Use sreal for time.
26534 (dump_inline_summary): Update.
26535 (estimate_function_body_sizes): Update.
26536 (estimate_edge_size_and_time): Update.
26537 (estimate_calls_size_and_time): Update.
26538 (estimate_node_size_and_time): Update.
26539 (inline_merge_summary): Update.
26540 (inline_update_overall_summary): Update.
26541 (estimate_time_after_inlining): Update.
26542 (inline_read_section): Update.
26543 (inline_write_summary): Update.
26544 * ipa-inline.c (compute_uninlined_call_time): Update.
26545 (compute_inlined_call_time): Update.
26546 (recursive_inlining): Update.
26547 (inline_small_functions): Update.
26548 (dump_overall_stats): Update.
26549 * ipa-inline.h: Include sreal.h.
26550 (size_time_entry): Turn time to sreal.
26551 (inline_summary): Turn self_time nad time to sreal.
26553 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26555 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
26557 (sreal::stream_out, sreal::stream_in): New.
26558 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
26560 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26562 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
26565 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
26568 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
26570 (dimode_scalar_chain::compute_convert_gain): Ditto.
26571 (dimode_scalar_chain::make_vector_copies): Ditto.
26572 (dimode_scalar_chain::convert_reg): Ditto.
26573 (dimode_scalar_chain::convert_insn): Ditto.
26574 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
26575 (VI248_AVX512BW_1): New mode iterator.
26576 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
26577 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
26580 2017-04-25 Martin Sebor <msebor@redhat.com>
26582 PR tree-optimization/80497
26583 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
26584 constants are representable in HOST_WIDE_INT.
26585 (parse_directive): Ditto.
26587 2017-04-25 Martin Sebor <msebor@redhat.com>
26590 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
26591 (new_zero_array): Adjust signature.
26592 (dom_info::dom_init): Used unsigned rather that size_t.
26593 (dom_info::dom_info): Same.
26595 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26596 Jakub Jelinek <jakub@redhat.com>
26599 * config/arm/arm.c: Include gimple.h.
26600 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26601 returns negative, increment ncrn only if it returned positive.
26602 (arm_needs_doubleword_align): Return int instead of bool,
26603 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
26604 members, but if there is any such non-FIELD_DECL
26605 > PARM_BOUNDARY aligned decl, return -1 instead of false.
26606 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26607 returns negative, increment nregs only if it returned positive.
26608 (arm_setup_incoming_varargs): Likewise.
26609 (arm_function_arg_boundary): Emit -Wpsabi note if
26610 arm_needs_doubleword_align returns negative, return
26611 DOUBLEWORD_ALIGNMENT only if it returned positive.
26613 2017-04-25 Marek Polacek <polacek@redhat.com>
26616 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26617 first argument to type.
26619 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
26622 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
26623 type checks to test for compatibility instead of equality.
26625 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26626 Jakub Jelinek <jakub@redhat.com>
26629 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
26631 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
26632 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
26633 the alignment computation, but return their maximum in warn_alignment.
26634 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26635 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
26637 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
26638 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26641 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26643 * config/arc/simdext.md (dmpyh): Fix typo.
26645 2017-04-25 Richard Biener <rguenther@suse.de>
26647 PR tree-optimization/80492
26648 * alias.c (compare_base_decls): Handle registers with asm
26649 specification conservatively.
26650 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26651 compare_base_decls returning dont-know properly.
26653 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26655 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
26656 (legitimate_offset_address_p): New function.
26657 (arc_legitimate_address_p): Use above function.
26659 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26661 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
26663 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26665 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
26666 ACCH registers whenever they are available.
26668 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26670 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
26671 double regs fix when not used.
26673 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26675 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
26677 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
26678 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
26680 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26682 * config/arc/arc.c (arc_output_addsi): Check for h-register class
26683 when emitting short ADD instructions.
26685 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26687 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
26689 (cmpsi_cc_c_insn): Likewise.
26690 (cbranchsi4_scratch): Compute proper instruction length using
26691 compact_hreg_operand.
26692 * config/arc/predicates.md (compact_hreg_operand): New predicate.
26694 2017-04-25 Richard Biener <rguenther@suse.de>
26696 PR middle-end/80509
26697 * passes.c (pass_manager::pass_manager): Initialize
26698 m_name_to_pass_map.
26700 2017-04-25 Richard Biener <rguenther@suse.de>
26702 PR tree-optimization/79201
26703 * tree-ssa-sink.c (statement_sink_location): Handle calls.
26705 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26708 * config/s390/vector.md: Split MEM->GPR vector moves for
26709 non-s_operand addresses.
26711 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26714 * config/s390/predicates.md (reload_const_wide_int_operand): New
26716 * config/s390/s390.md ("movti"): Remove d/P alternative.
26717 ("movti_bigconst"): New pattern definition.
26719 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26722 * s390-protos.h (s390_expand_cs_hqi): Removed.
26723 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
26724 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
26725 modes as well as CCZ1mode and CCZmode.
26726 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
26727 signature of s390_emit_compare_and_swap.
26728 (s390_expand_cs_hqi): Likewise, make static.
26729 (s390_expand_cs_tdsi): Generate an explicit compare before trying
26730 compare-and-swap, in some cases.
26731 (s390_expand_cs): Wrapper function.
26732 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
26734 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
26735 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
26736 patterns for small and large integers. Forbid symref memory operands.
26737 Move expander to s390.c. Require cc register.
26738 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
26739 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
26740 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
26741 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
26742 symref memory operands. Remove CC mode and call s390_match_ccmode
26744 ("atomic_exchange<mode>"): Allow and implement all integer modes.
26746 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26748 * config/s390/s390.md (define_peephole2): New peephole to help
26749 combining the load-and-test pattern with volatile memory.
26751 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26753 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
26754 with CCZmode for TARGET_Z196.
26756 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26758 PR rtl-optimization/80501
26759 * combine.c (make_compound_operation_int): Set subreg_code to SET
26760 even for AND with mask of the sign bit of mode.
26762 PR rtl-optimization/80500
26763 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
26764 sum's initial value.
26766 2017-04-25 Julian Brown <julian@codesourcery.com>
26767 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26769 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
26771 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
26773 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
26775 2017-04-25 Julian Brown <julian@codesourcery.com>
26776 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26778 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
26779 (thunderx2t99_sha): New Reservation.
26781 2017-04-25 Julian Brown <julian@codesourcery.com>
26782 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26784 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
26785 type for 1-element load.
26787 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
26789 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
26791 2017-04-24 Martin Jambor <mjambor@suse.cz>
26793 PR tree-optimization/80293
26794 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
26795 char arrays not totally scalarizable if it is false.
26796 (analyze_all_variable_accesses): Pass correct value in the new
26797 parameter. Add a statistics counter.
26799 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
26801 PR middle-end/79931
26802 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
26804 2017-04-24 Richard Biener <rguenther@suse.de>
26806 PR tree-optimization/80494
26807 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
26808 out for complex types.
26810 2017-04-24 Richard Biener <rguenther@suse.de>
26812 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
26813 * tree-ssa-sccvn.c (print_scc): Print SCC size.
26814 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
26815 (DFS): Adjust and never fail.
26816 (sccvn_dom_walker::fail): Remove.
26817 (sccvn_dom_walker::before_dom_children): Adjust.
26818 (run_scc_vn): Likewise and never fail.
26819 * tree-ssa-pre.c (pass_pre::execute): Adjust.
26820 (pass_fre::execute): Likewise.
26822 2017-04-24 Richard Biener <rguenther@suse.de>
26824 PR tree-optimization/79725
26825 * tree-ssa-sink.c (statement_sink_location): Return whether
26826 failure reason was zero uses. Move that check later.
26827 (sink_code_in_bb): Deal with zero uses by removing the stmt
26830 2017-04-24 Richard Biener <rguenther@suse.de>
26833 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
26834 pointer-based references.
26836 2017-04-24 Richard Biener <rguenther@suse.de>
26839 * pass_manager.h (pass_manager::operator new): Remove.
26840 (pass_manager::operator delete): Likewise.
26841 * passes.c (pass_manager::operator new): Remove.
26842 (pass_manager::operator delete): Likewise.
26843 (pass_manager::pass_manager): Zero individual pass members.
26845 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
26848 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
26849 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
26850 Check "XEXP (src, 1)" operand here.
26851 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
26852 Check "XEXP (src, 1)" operand here.
26853 (dimode_scalar_chain::make_vector_copies): Detect count register
26854 of a shift instruction. Zero extend count register from QImode
26855 to DImode to satisfy vector shift pattern count operand predicate.
26856 Substitute vector shift count operand with a DImode copy.
26857 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
26860 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
26862 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
26863 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26864 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
26865 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26866 (UNSPEC_NOREX_MEM): Remove definition.
26868 2017-04-21 Richard Biener <rguenther@suse.de>
26870 PR tree-optimization/79547
26871 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26872 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
26873 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
26874 without any constraints.
26876 2017-04-21 Richard Biener <rguenther@suse.de>
26878 PR tree-optimization/78847
26879 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
26881 2017-04-21 Richard Biener <rguenther@suse.de>
26883 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
26884 (build_distinct_type_copy): Likewise.
26885 (build_variant_type_copy): Likewise.
26886 * tree.c (build_qualified_type): Pass down mem-stat info.
26887 (build_distinct_type_copy): Likewise.
26888 (build_variant_type_copy): Likewise.
26890 2017-04-21 Richard Biener <rguenther@suse.de>
26892 PR tree-optimization/80237
26893 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
26895 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
26896 for a simplified result.
26898 2016-04-21 Richard Biener <rguenther@suse.de>
26900 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
26901 sth as strict as a simple_iv but a chrec without symbols and an
26902 operand defined in the loop we are peeling (and not some subloop).
26903 (propagate_constants_for_unrolling): Propagate all constants.
26905 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26908 * config/i386/i386.c (print_reg): Remove assert for disalowed
26909 regno values, call output_operand_lossage instead.
26911 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26914 * config/i386/constraints.md (Yc): New register constraint.
26915 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
26916 Use Yc constraint for alternative 2 of operand 0. Remove
26917 preferred_for_speed attribute.
26919 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26921 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
26922 lastprivate clauses in SIMT case.
26924 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
26926 * doc/invoke.texi (-Wextra-semi): Document new warning option.
26928 2017-04-20 Richard Biener <rguenther@suse.de>
26930 PR tree-optimization/57796
26931 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
26932 as N scalar stores.
26933 (vect_model_load_cost): Cost gathers as N scalar loads.
26935 2017-04-20 Richard Biener <rguenther@suse.de>
26937 * ggc-page.c (ggc_allocated_p): Rename to ...
26938 (safe_lookup_page_table_entry): ... this and return the lookup
26940 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
26942 2017-04-20 Richard Biener <rguenther@suse.de>
26944 PR tree-optimization/80453
26945 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
26946 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
26947 from the conditions.
26948 (vn_phi_eq): Pass them down.
26949 (vn_phi_lookup): Record them.
26950 (vn_phi_insert): Likewise.
26952 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
26954 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
26955 uninitialized variable warning to avoid buffer overrun.
26957 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26960 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
26961 is suppressed for '{ 0 }' in C.
26963 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26965 * BASE-VER: Set to 8.0.0.
26967 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
26969 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
26970 priority .init_array and .fini_array section with SECTION_NOTYPE
26973 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26975 PR middle-end/80423
26976 * tree.h (build_array_type): Add typeless_storage default argument.
26977 * tree.c (type_cache_hasher::equal): Also compare
26978 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
26979 (build_array_type): Add typeless_storage argument, set
26980 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
26982 (build_nonshared_array_type): Adjust build_array_type_1 caller.
26983 (build_array_type): Likewise. Add typeless_storage argument.
26985 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26986 Jakub Jelinek <jakub@redhat.com>
26988 PR tree-optimization/80426
26989 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
26990 operation on symbolic operands, also compute the overflow for the
26991 invariant part when the operation degenerates into a negation.
26993 2017-04-19 Jakub Jelinek <jakub@redhat.com>
26996 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
26997 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
27000 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
27002 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
27005 * config/avr/avr.c (tree.h): Include it.
27006 (cgraph.h): Include it.
27007 (avr_encode_section_info): Don't warn for uninitialized progmem
27008 variable if it's just an alias.
27010 2017-04-19 Richard Biener <rguenther@suse.de>
27013 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
27014 when needed by AutoPGO.
27016 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27019 * doc/lto.texi: Remove an extra 'that'.
27021 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27023 PR rtl-optimization/80429
27024 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27025 are only used in debug insns.
27027 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27028 Vladimir Makarov <vmakarov@redhat.com>
27030 * config/sparc/predicates.md (input_operand): Add comment. Return
27031 true for any memory operand when LRA is in progress.
27032 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27034 2017-04-18 Jeff Law <law@redhat.com>
27037 * mips.md ({return,simple_return}_internal): Do not overwrite
27040 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27042 PR tree-optimization/80443
27043 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27044 instead of adding 1, subtract -1 and similarly instead of subtracting
27047 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27049 PR rtl-optimization/80357
27050 * haifa-sched.c (tmp_bitmap): New variable.
27051 (model_recompute): Handle duplicate use records.
27052 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27053 (free_global_sched_pressure_data): Free it.
27055 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27058 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27059 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27060 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27061 instead of SYSTEM_HEADER_DIR.
27063 2017-04-18 Jeff Law <law@redhat.com>
27065 PR middle-end/80422
27066 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27067 predecessors after walking up the insn chain.
27069 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27072 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27073 sizetype type into debug info.
27075 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27078 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27079 unneeded test for TARGET_UPPER_REGS_SF.
27080 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27082 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27085 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27086 instead of gsi_after_labels.
27088 2017-04-18 Jeff Law <law@redhat.com>
27090 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27094 2017-04-13 Jeff Law <law@redhat.com>
27095 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27096 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27098 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27101 * config/avr/avr.c (intl.h): Include it.
27102 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27104 2017-04-18 Martin Liska <mliska@suse.cz>
27106 PR gcov-profile/78783
27107 * gcov-tool.c (gcov_output_files): Validate that destination
27108 file is either removed by the tool or by a user.
27110 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27111 Guy Benyei <guybe@mellanox.com>
27113 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27114 block, and do not negate it, the stored id is already negative.
27116 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27118 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27120 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27123 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27124 masks of options that should be turned off if the VSX vector
27125 options are turned off.
27126 (OTHER_P8_VECTOR_MASKS): Likewise.
27127 (OTHER_VSX_VECTOR_MASKS): Likewise.
27128 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27129 rs6000_disable_incompatible_switches to validate no type switches
27131 (rs6000_incompatible_switch): New function to disallow turning on
27132 other vector options if -mno-vsx, -mno-power8-vector, or
27133 -mno-power9-vector are specified.
27135 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27137 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27139 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27141 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27142 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27143 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27144 (ARG_POINTER_CFA_OFFSET): Likewise.
27146 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27148 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27149 conditions to take advantage of various optimizations.
27151 2017-04-13 Jeff Law <law@redhat.com>
27153 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27154 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27155 (zero_extendsidi2_dext): Likewise.
27157 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27160 * fold-const.c (fold_ternary_loc): Revert
27161 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27164 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27166 PR rtl-optimization/80343
27167 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27168 new scratch pseudo.
27170 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27173 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27174 to ubsan_encode_value.
27176 2017-04-13 Jeff Law <law@redhat.com>
27178 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27179 appearing in DEBUG_INSNs.
27181 2017-04-13 Martin Liska <mliska@suse.cz>
27183 PR gcov-profile/80413
27184 * gcov-io.c (gcov_write_string): Copy to buffer just when
27185 allocated size is greater than zero.
27187 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27190 * dwarf2out.c (decls_for_scope): Ignore declarations of
27191 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27193 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27196 * ipa-visibility.c (non_local_p): Fix typos.
27197 (localize_node): When localizing symbol in same comdat group,
27198 dissolve the group only when we know external symbols are going
27200 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27202 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27204 PR tree-optimization/79390
27205 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27206 order does not result in usable sequence, retry with reversed operand
27212 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27213 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27214 op0 instead of fold_convert_loc (loc, type, arg0).
27216 2017-04-12 Jeff Law <law@redhat.com>
27218 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27219 has a delay slot in the generated code.
27221 * config/cris/cris.md (cris_preferred_reload_class): Return
27222 GENNONACR_REGS rather than GENERAL_REGS.
27224 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27227 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27228 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27229 signedness of the result type.
27231 2017-04-12 Richard Biener <rguenther@suse.de>
27232 Jeff Law <law@redhat.com>
27234 PR tree-optimization/80359
27235 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27236 trim stores to TARGET_MEM_REFs.
27238 2017-04-12 Richard Biener <rguenther@suse.de>
27240 PR tree-optimization/79390
27241 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27242 threading case even more.
27244 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27247 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27248 for quad_address_p for TImode, instead of just not indexed_address.
27250 2017-04-12 Richard Biener <rguenther@suse.de>
27251 Bernd Edlinger <bernd.edlinger@hotmail.de>
27253 PR middle-end/79671
27254 * alias.c (component_uses_parent_alias_set_from): Handle
27255 TYPE_TYPELESS_STORAGE.
27256 (get_alias_set): Likewise.
27257 * tree-core.h (tree_type_common): Add typeless_storage flag.
27258 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27259 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27260 for types containing members with TYPE_TYPELESS_STORAGE.
27261 (place_field): Likewise.
27262 (layout_type): Likewise for ARRAY_TYPE.
27263 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27264 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27265 TYPE_TYPELESS_STORAGE.
27266 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27268 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27271 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27272 first argument to type.
27274 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27278 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27279 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27280 (rs6000_expand_binop_builtin): Likewise.
27281 (rs6000_expand_ternop_builtin): Likewise; also add missing
27282 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27283 vshasigma built-ins.
27284 * doc/extend.texi: Document that vec_xxpermdi's third argument
27285 must be a constant.
27287 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27289 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27290 Use shift_const cost parameter when calculating gain of STV shifts.
27292 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27294 PR rtl-optimization/70478
27295 * lra-constraints.c (process_alt_operands): Check memory for
27296 disfavoring memory insn operand.
27298 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27300 PR middle-end/80100
27301 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27302 left shift in unsigned HOST_WIDE_INT type.
27304 PR rtl-optimization/80385
27305 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27306 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27309 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27310 if they have any depend clauses.
27312 2017-04-11 Martin Liska <mliska@suse.cz>
27315 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27316 * ipa-split.c (split_function): Create a local comdat symbol
27317 if caller is in a comdat group.
27319 2017-04-11 Martin Liska <mliska@suse.cz>
27322 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27325 2017-04-11 Martin Sebor <msebor@redhat.com>
27327 PR middle-end/80364
27328 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27329 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27331 (directive::set_width, directive::set_precision, format_character):
27333 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27336 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27339 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27340 conflict, set target->arch_name instead of target->cpu_name.
27342 2017-04-11 Richard Biener <rguenther@suse.de>
27344 PR tree-optimization/80374
27345 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27346 build_zero_cst, remove fold_convertible_p check again.
27348 2017-04-11 Martin Liska <mliska@suse.cz>
27351 * ubsan.c (instrument_object_size): Do not instrument register
27354 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27357 * config/i386/i386-builtin-types.def
27358 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
27359 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
27360 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
27361 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
27362 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
27363 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
27364 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
27365 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
27366 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
27367 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
27368 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
27369 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
27370 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
27371 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
27372 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
27373 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
27374 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
27375 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
27376 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
27377 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
27378 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
27379 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27380 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27381 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27382 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27383 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27384 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27385 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
27386 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
27387 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
27388 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
27389 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
27390 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
27391 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
27392 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
27393 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
27394 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
27395 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
27396 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
27397 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
27398 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
27399 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
27400 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
27401 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
27402 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
27403 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
27405 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
27406 flag to second_arg_count, handle 4 argument function type _COUNT
27407 aliases, handle second_arg_count on second argument rather than last.
27409 2017-04-10 Jeff Law <law@redhat.com>
27411 PR tree-optimization/80374
27412 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
27413 record anything if we can not convert integer_zero_node to the
27416 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27419 * config/rs6000/rs6000.c (rs6000_option_override_internal):
27420 Enhance special handling given to the TARGET_P9_MINMAX option in
27421 relation to certain other options.
27423 2017-04-10 Bin Cheng <bin.cheng@arm.com>
27425 PR tree-optimization/80153
27426 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
27427 remove POINTER_PLUS_EXPR's base part directly, rather than through
27430 2017-04-10 Richard Biener <rguenther@suse.de>
27431 Bin Cheng <bin.cheng@arm.com>
27433 PR tree-optimization/80153
27434 * tree-affine.c (aff_combination_to_tree): Get base pointer from
27435 the first element of pointer type aff_tree. Build result expr in
27437 (add_elt_to_tree): Convert to type unconditionally. Remove other
27438 fold_convert calls.
27439 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
27440 (rewrite_use_nonlinear_expr): Check invariant using iv information.
27442 2017-04-10 Richard Biener <rguenther@suse.de>
27444 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27447 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
27449 PR rtl-optimization/70478
27450 * lra-constraints.c (curr_small_class_check): New.
27451 (update_and_check_small_class_inputs): New.
27452 (process_alt_operands): Update curr_small_class_check. Disfavor
27453 alternative insn memory operands. Check available regs for small
27456 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27459 * config/mips/mips.opt (-mvirt): Update description.
27460 * doc/invoke.texi (-mvirt): Likewise.
27462 2017-04-10 Richard Biener <rguenther@suse.de>
27464 PR middle-end/80362
27465 * fold-const.c (fold_binary_loc): Look at unstripped ops when
27466 looking for NEGATE_EXPR in -A / -B to A / B folding.
27468 2017-04-10 Martin Liska <mliska@suse.cz>
27470 PR gcov-profile/80224
27471 * gcov.c (print_usage): Fix usage string.
27472 (get_gcov_intermediate_filename): Remove.
27473 (output_gcov_file): Use both for normal and intermediate format.
27474 (generate_results): Do not initialize special file for
27475 intermediate format.
27477 2017-04-10 Richard Biener <rguenther@suse.de>
27479 PR tree-optimization/80304
27480 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
27483 2017-04-10 Nathan Sidwell <nathan@acm.org>
27486 * config/rs6000/rs6000.c (rs6000_vector_type): New.
27487 (rs6000_init_builtins): Use it.
27489 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27491 * config/arm/arm.md (<mrc>): Add mode to SET source.
27492 (<mrrc>): Likewise.
27494 2017-04-10 Richard Biener <rguenther@suse.de>
27496 PR middle-end/80344
27497 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
27499 2017-04-10 Jakub Jelinek <jakub@redhat.com>
27502 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
27503 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
27504 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
27505 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
27506 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
27507 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
27508 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
27509 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
27510 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
27511 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
27512 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
27513 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
27514 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
27515 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
27516 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
27517 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
27518 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
27519 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
27520 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
27521 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
27522 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
27523 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
27524 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
27526 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
27528 PR rtl-optimization/70478
27529 * lra-constraints.c: Reverse the last patch.
27531 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
27533 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
27534 Add comment for WCHAR_T.
27536 2017-04-08 Martin Liska <mliska@suse.cz>
27539 2017-04-07 Martin Liska <mliska@suse.cz>
27542 * ipa-split.c (split_function): Add function part to a same comdat
27545 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27548 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
27550 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
27552 * rs6000/rs6000.c (vec_load_pendulum): Rename...
27553 (vec_pairing): ...to this.
27554 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
27555 (rs6000_sched_init): Adjust for name change.
27556 (struct rs6000_sched_context): Likewise.
27557 (rs6000_init_sched_context): Likewise.
27558 (rs6000_set_sched_context): Likewise.
27560 2017-04-07 Jakub Jelinek <jakub@redhat.com>
27566 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
27568 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
27569 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
27570 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
27572 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
27574 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
27576 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27578 PR rtl-optimization/70703
27579 * ira-color.c (update_conflict_hard_regno_costs): Use
27580 int64_t instead of HOST_WIDE_INT.
27582 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27584 PR rtl-optimization/70478
27585 * lra-constraints.c (process_alt_operands): Disfavor alternative
27586 insn memory operands.
27588 2017-04-07 Jeff Law <law@redhat.com>
27590 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
27591 CALL and NOTE_INSN_CALL_ARG_LOCATION.
27593 2017-04-07 Martin Liska <mliska@suse.cz>
27596 * config/aarch64/aarch64.c (aarch64_process_target_attr):
27597 Show error message instead of an ICE.
27599 2017-04-07 Martin Liska <mliska@suse.cz>
27602 * ipa-split.c (split_function): Add function part to a same comdat
27605 2017-04-07 Richard Biener <rguenther@suse.de>
27607 PR middle-end/80341
27608 * tree.c (get_unwidened): Also handle ! for_type case for
27610 * convert.c (do_narrow): Split out from ...
27611 (convert_to_integer_1): ... here. Do not pass final truncation
27612 type to get_unwidened for TRUNC_DIV_EXPR.
27614 2017-04-07 Richard Biener <rguenther@suse.de>
27616 * tree-affine.c (wide_int_ext_for_comb): Take type rather
27618 (aff_combination_const): Adjust.
27619 (aff_combination_scale): Likewise.
27620 (aff_combination_add_elt): Likewise.
27621 (aff_combination_add_cst): Likewise.
27622 (aff_combination_convert): Likewise.
27623 (add_elt_to_tree): Likewise. Remove unused argument.
27624 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
27626 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
27628 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
27630 * config/arm/arm.c (arm_default_short_enums): Use
27631 ARM_DEFAULT_SHORT_ENUMS.
27632 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
27634 2017-04-06 Jakub Jelinek <jakub@redhat.com>
27637 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
27638 members with redundant out-of-class redeclaration.
27640 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27643 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
27644 * config/i386/i386.md (*zero_extendsidi2):
27645 Add (?*x,*x) and (?*v,*v) alternatives.
27647 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27650 * config/i386/i386.c (ix86_expand_builtin)
27651 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
27652 mode from insn data. Convert operands to insn operand mode.
27653 Copy operands that don't satisfy insn predicate to a register.
27655 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
27657 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
27660 2017-04-06 Richard Biener <rguenther@suse.de>
27662 PR tree-optimization/80334
27663 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
27664 preserve alignment of accesses.
27666 2017-04-06 Richard Biener <rguenther@suse.de>
27668 PR tree-optimization/80262
27669 * tree-sra.c (build_ref_for_offset): Preserve address-space
27671 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
27672 Drop useless address-space information on MEM_REF offsets.
27674 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
27676 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
27678 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27680 PR rtl-optimization/70703
27681 * ira-color.c (update_conflict_hard_regno_costs): Use
27682 HOST_WIDE_INT instead of long.
27684 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
27687 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
27688 not defined for x86_64 target. Add -mmmx target option when __SSE2__
27690 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
27691 for x86_64 target. Handle -m3dnowa option.
27693 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27695 PR rtl-optimization/70703
27696 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
27697 (update_conflict_hard_regno_costs): Use long instead of unsigned
27698 arithmetic for cost calculation.
27700 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27701 Bernd Edlinger <bernd.edlinger@hotmail.de>
27704 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
27707 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
27710 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
27711 ptr_mode with Pmode throughout.
27712 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
27713 into probe_stack_range and use DImode.
27715 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
27718 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
27719 call_eh_return is true.
27721 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27723 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
27724 Initialize last_match_fntype_index.
27726 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27729 * tree-nvr.c: Include internal-fn.h.
27730 (pass_return_slot::execute): Ignore internal calls without
27733 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27734 Richard Biener <rguenther@suse.de>
27737 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
27738 captures used multiple times, except for the last use.
27739 * generic-match-head.c: Include gimplify.h.
27741 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27743 PR tree-optimization/79390
27744 * target.h (struct noce_if_info): Declare.
27745 * targhooks.h (default_noce_conversion_profitable_p): Declare.
27746 * target.def (noce_conversion_profitable_p): New target hook.
27747 * ifcvt.h (struct noce_if_info): New type, moved from ...
27748 * ifcvt.c (struct noce_if_info): ... here.
27749 (noce_conversion_profitable_p): Renamed to ...
27750 (default_noce_conversion_profitable_p): ... this. No longer
27752 (noce_try_store_flag_constants, noce_try_addcc,
27753 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
27754 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
27755 instead of noce_conversion_profitable_p.
27756 * config/i386/i386.c: Include ifcvt.h.
27757 (ix86_option_override_internal): Don't override
27758 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
27759 (ix86_noce_conversion_profitable_p): New function.
27760 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
27761 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
27762 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
27763 * doc/tm.texi: Regenerated.
27765 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27767 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
27770 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
27773 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
27774 instructions for small multiply cores.
27776 2017-04-04 Jeff Law <law@redhat.com>
27778 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
27780 (mips_expand_vec_perm_const): Initialize elements in orig_perm
27781 that are not set by the loop over the elements.
27783 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27786 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
27787 int mode, convert_modes it to mode as unsigned, otherwise use
27788 lowpart_subreg to mode rather than SImode.
27789 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
27790 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
27791 Use DImode instead of SImode for the shift count operand.
27792 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
27795 2017-04-04 Richard Biener <rguenther@suse.de>
27797 PR middle-end/80281
27798 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
27799 arithmetic done for the negate or the plus. Simplify.
27800 (A - (-B) -> A + B): Likewise.
27801 * fold-const.c (split_tree): Make sure to not negate pointers.
27803 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
27805 PR rtl-optimization/60818
27806 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
27807 a compare of comparisons with the thing compared if this results
27808 in a different machine mode.
27810 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
27812 * alias.c (base_alias_check): Fix typo in comment.
27813 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
27814 * cgraphunit.c (symbol_table::compile): Likewise.
27815 * collect2.c (maybe_run_lto_and_relink): Likewise.
27816 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
27817 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
27818 * config/avr/avr.c (avr_map_op_t): Likewise.
27819 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
27820 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
27821 * config/epiphany/epiphany.md (movcc): Likewise.
27822 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
27823 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
27825 * config/mips/mips.c (mips_save_restore_reg): Likewise.
27826 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
27827 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
27828 * config/sh/sh.c (sh_rtx_costs): Likewise.
27829 * fold-const.c (fold_truth_andor): Likewise.
27830 * genautomata.c (collapse_flag): Likewise.
27831 * gengtype.h (struct type::u::s): Likewise.
27832 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
27833 * input.c (FORMAT_AMOUNT): Likewise.
27834 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
27835 (known_aggs_to_agg_replacement_list): Likewise.
27836 * ipa-inline-analysis.c: Likewise.
27837 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
27838 * ipa-polymorphic-call.c
27839 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
27840 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
27841 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
27843 * modulo-sched.c (apply_reg_moves): Likewise.
27844 * omp-expand.c (build_omp_regions_1): Likewise.
27845 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
27846 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
27847 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
27848 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
27849 * value-prof.c: Likewise.
27850 * var-tracking.c (val_reset): Likewise.
27852 2017-04-03 Richard Biener <rguenther@suse.de>
27854 PR tree-optimization/80275
27855 * fold-const.c (split_address_to_core_and_offset): Handle
27858 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
27860 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
27861 descriptors is at least equal to that of functions.
27863 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27865 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
27867 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27870 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
27871 (mov<IMOD4:mode>): New expander.
27872 (*mov<IMOD4:mode>_internal): New insn and split pattern.
27874 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
27876 PR rtl-optimization/79405
27877 * fwprop.c (propagations_left): New variable.
27878 (forward_propagate_into): Decrement it.
27879 (fwprop_init): Initialize it.
27880 (fw_prop): If the variable has reached zero, stop propagating.
27881 (fwprop_addr): Ditto.
27883 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27886 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
27887 a FUNCTION_DECL, pass it as decl instead of origin to
27890 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
27892 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
27895 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
27898 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
27899 TARGET_VSX_SMALL_INTEGER.
27901 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27903 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27904 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
27906 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27908 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
27909 extraction from odd-numbered MSA register.
27911 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27913 PR middle-end/80173
27914 * expmed.c (store_bit_field_1): Don't attempt to create
27915 a word subreg out of hard registers wider than word if they
27916 have HARD_REGNO_NREGS of 1 for their mode.
27918 PR middle-end/80163
27919 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
27920 conversions to integer types wider than word and pointer.
27923 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
27924 (rtx_equal_for_cselib_p): Pass 0 to it.
27925 * cselib.c (cselib_hasher::equal): Likewise.
27926 (rtx_equal_for_cselib_1): Add depth argument. If depth
27927 is 128, don't look up VALUE locs and punt. Increment
27928 depth in recursive calls when walking VALUE locs.
27930 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
27932 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
27933 (make_gcov_file_name): Use the canonical path name for generating
27935 (read_line): Fix handling of files with ascii null bytes.
27937 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
27939 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
27940 to initialise a vector register instead
27941 of using a const_int.
27943 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27945 PR translation/80189
27946 * gimplify.c (omp_default_clause): Use %qs instead of %s in
27947 diagnostic messages.
27949 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
27952 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
27953 (dfp_diex_<mode>): Update mode of operand 1.
27954 * doc/extend.texi (dxex, dxexq): Document change to return type.
27955 (diex, diexq): Document change to argument type.
27957 2017-03-30 Martin Jambor <mjambor@suse.cz>
27960 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
27961 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
27962 it reflects the signature changes performed at the callee side.
27963 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
27964 to cgraph_build_function_type_skip_args.
27965 (build_function_decl_skip_args): Adjust call to the above function.
27967 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27970 * config/i386/sse.md
27971 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
27972 register as dest whenever it is a MEM not rtx_equal_p to the
27973 corresponding dup operand, and when forcing into reg move the
27974 reg into the memory afterwards.
27975 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
27976 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
27977 for the force_reg mode.
27978 (avx512vl_vextractf128<mode>): Use register as dest either
27979 always when a MEM, or when it is a MEM not rtx_equal_p to the
27980 corresponding dup operand, or even not when it is a CONST_VECTOR
27981 depending on the mode and lo vs. hi.
27982 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
27984 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
27985 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
27986 Likewise. Require that operands[2] is even.
27987 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
27988 Remove extraneous parens. Require that operands[2] is a multiple
27990 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
27991 operands[0] is a MEM if <mask_applied>, the predicates/constraints
27992 disallow memory then.
27994 2017-03-30 Richard Biener <rguenther@suse.de>
27996 PR tree-optimization/77498
27997 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
27998 to non-constants over backedges.
28000 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
28002 PR rtl-optimization/80233
28003 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
28004 as last_combined_insn. Do not test for BARRIER_P separately.
28006 2017-03-29 Andreas Schwab <schwab@suse.de>
28009 * calls.c (prepare_call_address): Convert funexp to Pmode before
28010 copying to temp reg.
28012 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28014 PR tree-optimization/80158
28015 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28016 Handle possible future case of more than one alternate
28018 (replace_rhs_if_not_dup): Likewise.
28019 (replace_one_candidate): Likewise.
28021 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28023 PR rtl-optimization/80193
28024 * ira.c (ira): Do not check allocation for LRA.
28026 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28028 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28029 (nvptx_output_simt_exit): Declare.
28030 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28031 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28032 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28033 (nvptx_declare_function_name): ...here. Emit declaration of local
28034 memory space buffer for omp_simt_enter insn.
28035 (nvptx_output_unisimt_switch): New.
28036 (nvptx_output_softstack_switch): New.
28037 (nvptx_output_simt_enter): New.
28038 (nvptx_output_simt_exit): New.
28039 * config/nvptx/nvptx.h (struct machine_function): New fields
28040 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28041 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28042 (UNSPECV_SIMT_EXIT): Ditto.
28043 (omp_simt_enter_insn): New insn.
28044 (omp_simt_enter): New expansion.
28045 (omp_simt_exit): New insn.
28046 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28048 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28049 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28050 (expand_GOMP_SIMT_EXIT): New.
28051 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28052 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28053 (GOMP_SIMT_EXIT): Ditto.
28054 * target-insns.def (omp_simt_enter): New insn.
28055 (omp_simt_exit): Ditto.
28056 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28058 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28059 (lower_rec_input_clauses): Likewise.
28060 (lower_lastprivate_clauses): Handle SIMT privatization.
28062 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28063 (ompdevlow_adjust_simt_enter): New.
28064 (find_simtpriv_var_op): New.
28065 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28066 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28068 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28069 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28070 (copy_decl_for_dup_finish): Ditto.
28072 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28074 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28077 * config/i386/i386.c (ix86_option_override_internal): Always
28078 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28080 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28082 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28084 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28086 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28087 mark new edge's irreducible flag accordign to it.
28088 (vect_do_peeling): Check loop preheader edge's irreducible flag
28089 and pass it to function slpeel_add_loop_guard.
28091 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28093 PR tree-optimization/80218
28094 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28095 Update block frequencies and counts.
28097 2017-03-28 Richard Biener <rguenther@suse.de>
28099 PR tree-optimization/78644
28100 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28101 of a simplification result we may not use it at all.
28103 2017-03-28 Richard Biener <rguenther@suse.de>
28106 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28107 without arguments, generate default definition of a SSA name.
28109 2017-03-28 Richard Biener <rguenther@suse.de>
28111 PR middle-end/80222
28112 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28113 TYPE_REF_CAN_ALIAS_ALL references.
28114 * fold-const.c (fold_indirect_ref_1): Likewise.
28116 2017-03-28 Martin Liska <mliska@suse.cz>
28119 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28120 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28122 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28123 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28125 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28126 (EXTRA_SPECS): Define.
28127 (SUBTARGET_EXTRA_SPECS): Likewise.
28128 (SUBTARGET_CPP_SPEC): Likewise.
28129 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28130 SUBTARGET_EXTRA_SPECS.
28131 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28133 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28135 * config/arc/simdext.md (vst64_insn): Update pattern.
28136 (vld32wh_insn): Likewise.
28137 (vld32wl_insn): Likewise.
28138 (vld64_insn): Likewise.
28139 (vld32_insn): Likewise.
28141 2017-03-28 Marek Polacek <polacek@redhat.com>
28144 * fold-const.c (fold_comparison): Use protected_set_expr_location
28145 instead of SET_EXPR_LOCATION.
28147 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28149 * tree.c (add_expr): Avoid name lookup warning.
28151 2017-03-27 Jeff Law <law@redhat.com>
28153 PR tree-optimization/80216
28154 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28155 function name. Limit recursion depth.
28156 (record_temporary_equivalences): Corresponding changes.
28158 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28160 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28163 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28166 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28168 * cfgcleanup.c (reg_note_cfa_p): New array.
28169 (insns_have_identical_cfa_notes): New function.
28170 (old_insns_match_p): Don't cross-jump in between /f
28171 and non-/f instructions. If both i1 and i2 are frame related,
28172 verify all CFA notes, their order and content.
28174 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28177 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28178 HImode and SImode with zero extend to DImode to one insn.
28179 (bswap<mode>2_extenddi): Likewise.
28180 (bswapsi2_extenddi): Likewise.
28181 (bswaphi2_extendsi): Likewise.
28182 (bswaphi2): Combine bswap HImode and SImode into one insn.
28183 Separate memory insns from swapping register.
28184 (bswapsi2): Likewise.
28185 (bswap<mode>2): Likewise.
28186 (bswaphi2_internal): Delete, no longer used.
28187 (bswapsi2_internal): Likewise.
28188 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28189 store, and gpr<-gpr swap insns.
28190 (bswap<mode>2_store): Likewise.
28191 (bswaphi2_reg): Register only splitter, combine with the splitter.
28192 (bswaphi2 splitter): Likewise.
28193 (bswapsi2_reg): Likewise.
28194 (bswapsi2 splitter): Likewise.
28195 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28196 the insns into load, store, and register/register insns.
28197 (bswapdi2_ldbrx): Likewise.
28198 (bswapdi2_load): Likewise.
28199 (bswapdi2_store): Likewise.
28200 (bswapdi2_reg): Likewise.
28202 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28204 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28205 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28207 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28210 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28212 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28213 special handling for target option conflicts between dform
28214 options (-mpower9-dform, -mpower9-dform-vector,
28215 -mpower9-dform-scalar) and -mno-direct-move.
28217 2017-03-27 Richard Biener <rguenther@suse.de>
28219 PR tree-optimization/80181
28220 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28222 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28224 * config/arc/predicates.md (move_double_src_operand): Replace the
28225 call to move_double_src_operand with a call to address_operand.
28227 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28229 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28230 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28231 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28233 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28235 * config/arc/predicates.md (long_immediate_loadstore_operand):
28236 Consider scaled addresses cases.
28238 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28240 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28241 restored when in interrupt.
28242 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28243 doesn't have delay slot.
28245 2017-03-27 Richard Biener <rguenther@suse.de>
28248 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28249 inlined thunk clones.
28251 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28254 * asan.c (instrument_derefs): Copy over last operand from
28255 original COMPONENT_REF to the new COMPONENT_REF with
28256 DECL_BIT_FIELD_REPRESENTATIVE.
28257 * ubsan.c (instrument_object_size): Likewise.
28259 2017-03-27 Richard Biener <rguenther@suse.de>
28261 PR tree-optimization/80170
28262 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28263 sure DR/SCEV didnt fold in constants we do not see when looking
28264 at the reference base alignment.
28266 2017-03-27 Richard Biener <rguenther@suse.de>
28268 PR middle-end/80171
28269 * gimple-fold.c (fold_ctor_reference): Properly guard against
28270 NULL return value from canonicalize_constructor_val.
28272 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28275 * config/i386/i386.c (ix86_expand_builtin)
28276 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28277 flags reg setting and flags reg using instructions.
28278 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28279 clobbering instructions to zero extend op2.
28281 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28283 * doc/install.texi (Configuration) <--with-aix-soname>:
28284 Update link to AIX ld.
28286 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28288 PR rtl-optimization/80160
28289 PR rtl-optimization/80159
28290 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28291 reg_alternate_class into account.
28293 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28296 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28297 to consider in curr_insn_transform.
28299 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28301 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28302 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28303 and emit_mode_inner.
28305 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28307 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28308 argument to the overloaded builtin variants. Use the new flag to
28309 deprecate certain builtin variants.
28310 * config/s390/s390-builtin-types.def: Add new builtin types.
28311 * config/s390/s390-builtins.h: Support new flags field for
28312 overloaded builtins.
28313 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28314 (s390_macro_to_expand): Enable vector float data type.
28315 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28316 builtins by incrementing the __VEC__ version number.
28317 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28319 (s390_resolve_overloaded_builtin): Emit error messages depending
28320 on the builtin flags.
28321 * config/s390/s390.c (s390_expand_builtin): Support additional
28322 flags argument. Change error message to match the messages
28323 emitted in s390-c.c.
28324 * config/s390/s390.md: New UNSPEC_* constants.
28325 (op_type): Add new instruction types.
28326 * config/s390/vecintrin.h: Add new builtins and test data class
28328 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28329 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28330 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28331 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28332 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28333 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28335 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28336 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28337 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28338 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28340 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28341 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28342 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28343 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28345 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28346 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28347 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28348 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28349 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28350 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28351 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28353 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28354 ("vec_scatter_element<V_HW_4:mode>_DI")
28355 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28356 ("vec_fpint<mode>", "vflls")
28357 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28358 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
28359 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
28360 ("*vec_cmphe<mode>_cc"): ... these.
28362 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
28363 mode constant instead of magic value.
28365 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28367 * config/s390/s390.c (s390_expand_vec_compare): Support other
28368 vector floating point modes than just V2DF.
28369 (s390_expand_vcond): Likewise.
28370 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
28371 (s390_cannot_change_mode_class): Prevent mode changes between TF
28372 and V1TF in vector registers.
28373 * config/s390/s390.md (DF, SF): New mode attributes.
28374 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
28375 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
28376 SFmode support for VRs.
28377 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
28379 (VFT, VF_HW): New mode iterators.
28380 (vw, sdx): New mode attributes.
28381 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28382 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28383 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28384 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28385 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
28386 also the new vector floating point modes. Renaming to ...
28388 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
28389 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
28390 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
28391 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
28392 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
28393 ("vec_unordered<mode>"): ... these.
28395 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
28396 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
28397 ("*vec_extendv2df"): New insn definitions.
28399 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28401 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
28402 ("mulditi3_2", "*muldi3_sign"): New patterns.
28403 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
28404 rename the pattern definition.
28406 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28408 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
28410 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
28412 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28414 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
28415 instruction if possible.
28416 * config/s390/vector.md (vec_halfnumelts): New mode
28418 ("*vec_vllezlf<mode>"): New pattern.
28420 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28422 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
28423 ("popcountv4si2", "popcountv2di2"): Rename to ...
28424 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
28425 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
28427 ("popcount<mode>2_vxe"): New pattern.
28429 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28431 * common/config/s390/s390-common.c (processor_flags_table): Add
28433 * config.gcc: Add arch12.
28434 * config/s390/driver-native.c (s390_host_detect_local_cpu):
28435 Default to arch12 for unknown CPU model numbers.
28436 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
28437 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
28438 PROCESSOR_max sanity check.
28439 * config/s390/s390-opts.h (enum processor_type): Add
28441 * config/s390/s390.c (processor_table): Add arch12.
28442 (s390_expand_builtin): Add check for B_VXE flag.
28443 (s390_issue_rate): Add PROCESSOR_ARCH12.
28444 (s390_get_sched_attrmask): Likewise.
28445 (s390_get_unit_mask): Likewise.
28446 (s390_sched_score): Enable z13 scheduling for arch12.
28447 (s390_sched_reorder): Likewise.
28448 (s390_sched_variable_issue): Likewise.
28449 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
28451 (s390_tune_attr): Use z13 scheduling also for arch12.
28452 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
28453 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
28454 (TARGET_VXE_P): New macros.
28455 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
28456 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
28457 * config/s390/s390.opt: Add arch12 as processor_type.
28459 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28461 * config/s390/s390.md
28462 ("fixuns_truncdddi2", "fixuns_trunctddi2")
28463 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
28464 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
28466 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
28467 Rename expanders to ...
28469 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
28470 ("fixuns_truncdddi2_emu"): ... these.
28472 ("fixuns_trunc<mode>si2_emu"): New expander.
28474 ("*fixuns_truncdfdi2_z13"): Rename to ...
28475 ("*fixuns_truncdfdi2_vx"): ... this.
28477 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28479 * config/s390/2964.md: Remove the single element vector compare
28480 instructions which are no longer used.
28481 * config/s390/s390.c (s390_select_ccmode): Remove handling of
28483 (s390_canonicalize_comparison): Remove handling of DFmode
28485 (s390_expand_vec_compare_scalar): Remove function.
28486 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
28487 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
28489 ("*cmp<mode>_ccs"): Add wfcdb instruction.
28491 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28493 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
28495 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
28496 will anyway by matched by mov<mode>_64dfp.
28498 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28500 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
28501 vlef/vstef. Add missing operand to vleif.
28503 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28505 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
28506 pair for all vector types with 64 bit elements.
28507 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
28508 * config/s390/vector.md (V_HW_64): ... here.
28509 (V_128_NOSINGLE): New mode iterator.
28510 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
28511 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
28512 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
28513 ("*vec_load_pairv2di"): Change to ...
28514 ("*vec_load_pair<mode>"): ... this one.
28516 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28518 * config/s390/constraints.md: Add comments.
28519 (jKK): Reject element sizes > 8 bytes.
28520 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
28522 * config/s390/s390.md: Add the s_operand checks formerly in
28523 s390_split_ok_p to various splitters where they are still
28525 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
28526 for 128 bit vectors. Plus two splitters.
28528 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28530 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
28533 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28536 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
28537 error if the boundary argument is not constant.
28539 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28541 PR rtl-optimization/80112
28542 * loop-doloop.c (doloop_condition_get): Don't check condition
28543 if cmp isn't SET with IF_THEN_ELSE src.
28545 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28547 PR tree-optimization/80158
28548 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
28549 replacing a candidate statement, also replace it for the
28550 candidate's alternate interpretation.
28551 (replace_rhs_if_not_dup): Likewise.
28552 (replace_one_candidate): Likewise.
28554 2017-03-24 Richard Biener <rguenther@suse.de>
28556 PR tree-optimization/80167
28557 * graphite-isl-ast-to-gimple.c
28558 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
28560 (translate_isl_ast_to_gimple::get_rename): Likewise.
28562 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28564 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
28565 handling of certain combinations of target options, including the
28566 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
28567 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
28569 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28572 * config/arm/arm.md (*load_multiple): Add reload_completed to
28573 matching condition.
28575 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28576 Richard Biener <rguenth@suse.de>
28578 PR tree-optimization/79908
28579 PR tree-optimization/80136
28580 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28581 been cast away, gimplify_and_add suffices.
28583 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
28585 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
28587 2017-03-23 Richard Biener <rguenther@suse.de>
28589 PR tree-optimization/80032
28590 * gimplify.c (gimple_push_cleanup): Forced unconditional
28591 cleanups still have to go to the conditional_cleanups
28594 2017-03-22 Jakub Jelinek <jakub@redhat.com>
28596 PR tree-optimization/80072
28597 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
28599 (next_operand_entry_id): Change type to unsigned int.
28600 (sort_by_operand_rank): Make sure to return the right return value
28601 even if unsigned fields are bigger than INT_MAX.
28602 (struct oecount): Change cnt and id type to unsigned int.
28603 (oecount_hasher::equal): Formatting fix.
28604 (oecount_cmp): Make sure to return the right return value
28605 even if unsigned fields are bigger than INT_MAX.
28606 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
28609 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
28610 TREE_READONLY on result if writing it more than once.
28613 * doc/invoke.texi (-fsanitize=thread): Document that with
28614 -fnon-call-exceptions atomics are not able to throw
28618 * tsan.c: Include tree-eh.h.
28619 (instrument_builtin_call): Call maybe_clean_eh_stmt or
28620 maybe_clean_or_replace_eh_stmt where needed.
28621 (instrument_memory_accesses): Add cfg_changed argument.
28622 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
28623 if it returned true.
28624 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
28626 PR rtl-optimization/63191
28627 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
28628 wrapper function, moved the whole old content into ...
28629 (ix86_delegitimize_address_1): ... this. New inline function.
28630 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
28631 true as last argument instead of ix86_delegitimize_address.
28633 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28635 * config/aarch64/aarch64.c (generic_branch_cost): Copy
28636 cortexa57_branch_cost.
28638 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28640 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
28642 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28645 * doc/md.texi (Constraints): Document wA constraint.
28646 * config/rs6000/constraints.md (wA): New.
28647 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
28648 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
28649 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
28650 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
28652 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
28655 * gimplify.c (is_oacc_declared): New function.
28656 (oacc_default_clause): Use it to set default flags for acc declared
28657 variables inside parallel regions.
28658 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
28659 declared variables.
28660 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
28661 declare attribute to any decl as necessary.
28663 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
28666 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
28667 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
28668 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
28669 (arm_arch_lpae): This.
28670 * config/arm/arm.c (arm_arch7ve): Rename into ...
28671 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
28672 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
28675 2017-03-22 Martin Liska <mliska@suse.cz>
28678 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
28679 error message instead of an ICE.
28681 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28683 * doc/extend.texi (6.11 Additional Floating Types): Revise.
28685 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
28687 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28689 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28692 2017-03-21 Martin Sebor <msebor@redhat.com>
28694 * doc/extend.texi: Use "cannot" instead of "can't."
28695 * doc/hostconfig.texi: Same.
28696 * doc/install.texi: Same.
28697 * doc/invoke.texi: Same.
28698 * doc/loop.texi: Same.
28699 * doc/md.texi: Same.
28700 * doc/objc.texi: Same.
28701 * doc/rtl.texi: Same.
28702 * doc/tm.texi: Same.
28703 * doc/tm.texi.in: Same.
28704 * doc/trouble.texi: Same.
28706 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
28709 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
28710 (collect_checksum_attributes): Set it.
28711 (die_checksum_ordered): Use it.
28713 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28715 PR tree-optimization/79908
28716 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
28717 change: For a VA_ARG whose LHS has been cast away, use
28718 force_gimple_operand to construct the side effects.
28720 2017-03-21 David Malcolm <dmalcolm@redhat.com>
28722 PR translation/80001
28723 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
28724 more amenable to translation.
28725 (oacc_loop_auto_partitions): Likewise.
28727 2017-03-21 Marek Polacek <polacek@redhat.com>
28728 Martin Sebor <msebor@redhat.com>
28730 PR tree-optimization/80109
28731 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
28732 on INTEGRAL_TYPE_P.
28734 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28735 Segher Boessenkool <segher@kernel.crashing.org>
28738 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
28739 check reg_used_between_p between insn and one of succ or succ2
28740 depending on if succ is artificial insn not inserted into insn
28743 2017-03-21 Martin Liska <mliska@suse.cz>
28745 PR gcov-profile/80081
28746 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
28747 * doc/gcc.texi: Include gcov-dump stuff.
28748 * doc/gcov-dump.texi: New file.
28750 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
28752 PR rtl-optimization/79150
28753 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
28754 conditional jump, if the jump is the last insn of the loop.
28756 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28757 Richard Biener <rguenth@suse.de>
28759 PR tree-optimization/79908
28760 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28761 been cast away, use force_gimple_operand to construct the side
28764 2017-03-21 Martin Liska <mliska@suse.cz>
28766 PR libfortran/79956
28767 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
28770 2017-03-21 Brad Spengler <spender@grsecurity.net>
28773 * plugin.c (htab_hash_plugin): New function.
28774 (add_new_plugin): Use it and adjust.
28775 (parse_plugin_arg_opt): Adjust.
28776 (init_one_plugin): Likewise.
28778 2017-03-21 Richard Biener <rguenther@suse.de>
28780 PR tree-optimization/80032
28781 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
28782 if set force the cleanup to happen unconditionally.
28783 (gimplify_target_expr): Push inserted clobbers with force_uncond
28784 to avoid them being removed by control-dependent DCE.
28786 2017-03-21 Richard Biener <rguenther@suse.de>
28788 PR tree-optimization/80122
28789 * tree-inline.c (copy_bb): Do not expans va-arg packs or
28790 va_arg_pack_len when the inlined call stmt requires pack
28792 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
28794 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28797 * tsan.c (instrument_builtin_call): If the memory model argument
28798 is not a constant, assume it is valid.
28801 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
28804 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28806 PR rtl-optimization/79910
28807 * combine.c (can_combine_p): Do not allow combining an I0 or I1
28808 if its dest is used by an insn before I2 (other than the combined
28809 insns themselves, which are properly handled already).
28811 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28814 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28816 * combine.c (record_used_regs): New static function.
28817 (try_combine): Handle situations where there is an additional
28818 instruction between I2 and I3 which needs to have a LOG_LINK
28822 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28824 * combine.c (try_combine): Delete redundant i1 test. Call
28825 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28827 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28830 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
28831 alternatives 13/14.
28833 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28835 PR tree-optimization/80054
28836 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
28837 the optimization if a PHI or any of its arguments is not dominated
28838 by the candidate's basis. Use gphi* rather than gimple* as
28840 (replace_profitable_candidates): Clean up a gimple* variable that
28841 should be a gphi* variable.
28843 2017-03-20 Martin Sebor <msebor@redhat.com>
28846 * doc/extend.texi (attribute constructor): Document present limitation.
28848 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
28851 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
28852 __POWER9_VECTOR__ #ifdef control, change template definition to
28853 use Power9-specific built-in function.
28854 (vec_any_eq): Likewise.
28855 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
28856 to control outcomes from this test.
28857 (vector_ae_<mode>p): For VEC_F modes, likewise.
28859 2017-03-20 Ian Lance Taylor <iant@google.com>
28861 * config/i386/i386.c (ix86_function_regparm): Save an extra
28862 register for -fsplit-stack with DECL_STATIC_CHAIN.
28864 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28867 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
28868 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
28870 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28872 * config/riscv/riscv.c (riscv_print_operand): Use "fence
28874 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
28877 2017-03-20 Marek Polacek <polacek@redhat.com>
28880 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
28882 2017-03-20 Richard Biener <rguenther@suse.de>
28884 PR tree-optimization/80113
28885 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
28886 allocate extra SSA name for PHI def.
28887 (add_close_phis_to_outer_loops): Likewise.
28888 (add_close_phis_to_merge_points): Likewise.
28889 (copy_loop_close_phi_args): Likewise.
28890 (copy_cond_phi_nodes): Likewise.
28892 2017-03-20 Martin Liska <mliska@suse.cz>
28894 PR middle-end/79753
28895 * tree-chkp.c (chkp_build_returned_bound): Do not build
28896 returned bounds for a LHS that's not a BOUNDED_P type.
28898 2017-03-20 Martin Liska <mliska@suse.cz>
28902 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
28903 COMPLEX_CST and VECTOR_CST.
28905 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28908 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
28909 target operand. A new splitter adds the clobber statement in case
28910 the target operand is dead anyway.
28912 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
28914 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
28915 to age-old versions of binutils and glibc.
28917 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
28919 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
28921 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28923 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
28925 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28927 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
28928 requirement for binutils 2.13.
28930 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28932 * combine.c (try_combine): Delete redundant i1 test. Call
28933 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28935 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
28937 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
28938 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
28940 <riscv64-*-elf>: Re-arrange section
28941 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
28942 <riscv32-*-linux>: Likewise.
28943 <riscv64-*-elf>: Likewise
28944 <riscv64-*-linux>: Likewise.
28946 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
28949 * aarch64.opt(verbose-cost-dump): Fix typo.
28951 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
28954 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
28955 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
28957 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28959 * reload.c (find_reloads): When reloading a nonoffsettable address,
28960 use RELOAD_OTHER for it and its address reloads.
28962 PR rtl-optimization/79910
28963 * combine.c (record_used_regs): New static function.
28964 (try_combine): Handle situations where there is an additional
28965 instruction between I2 and I3 which needs to have a LOG_LINK
28968 2017-03-17 Jeff Law <law@redhat.com>
28970 PR tree-optimization/71437
28971 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
28972 conditional in the hash table first.
28973 (vrp_dom_walker::before_dom_children): Extract condition from
28974 ASSERT_EXPR. Record condition, its inverion and any implied
28975 conditions as well.
28977 2017-03-17 Marek Polacek <polacek@redhat.com>
28978 Markus Trippelsdorf <markus@trippelsdorf.de>
28980 PR tree-optimization/80079
28981 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
28984 2017-03-17 Richard Biener <rguenther@suse.de>
28986 PR middle-end/80075
28987 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
28988 Properly verify the LHS before the RHS possibly claims to be
28990 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
28993 2017-03-17 Martin Jambor <mjambor@suse.cz>
28995 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
28996 (List of -O2 options): Likewise.
28997 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
29000 2017-03-17 Tom de Vries <tom@codesourcery.com>
29002 * gcov-dump.c (print_usage): Print bug_report_url.
29004 2017-03-17 Richard Biener <rguenther@suse.de>
29006 PR middle-end/80050
29007 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
29008 (parser::peek): Likewise.
29010 2017-03-17 Richard Biener <rguenther@suse.de>
29012 PR tree-optimization/80048
29013 * sese.c (free_sese_info): Properly release rename_map and
29014 copied_bb_map elements.
29016 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29018 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29019 Add linked-list forward and backlinks. Insert on
29020 construction, remove on destruction.
29021 (class pass_store_merging): Add m_stores_head field.
29022 (pass_store_merging::terminate_and_process_all_chains):
29023 Iterate over m_stores_head list.
29024 (pass_store_merging::terminate_all_aliasing_chains):
29026 (pass_store_merging::execute): Check for debug stmts first.
29027 Push new chains onto the m_stores_head stack.
29029 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29032 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29033 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29034 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29036 2017-03-16 Jeff Law <law@redhat.com>
29038 PR tree-optimization/71437
29039 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29040 member function. Implementation moved into after_dom_children
29041 member function and into the threader's thread_outgoing_edges
29043 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29044 some code into new thread_outgoing_edges.
29045 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29046 definition. Simplify marker handling (do it here). Assume we always
29047 have the available expression and the const/copies tables.
29048 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29050 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29051 * tree-vrp.c (equiv_stack): No longer file scoped.
29052 (vrp_dom_walker): New class.
29053 (vrp_dom_walker::before_dom_children): New member function.
29054 (vrp_dom_walker::after_dom_children): Likewise.
29055 (identify_jump_threads): Setup domwalker. Use it rather than
29056 walking edges in a random order by hand. Simplify setup/finalization.
29057 (finalize_jump_threads): Remove.
29058 (vrp_finalize): Do not call identify_jump_threads here.
29059 (execute_vrp): Do it here instead and call thread_through_all_blocks
29062 PR tree-optimization/71437
29063 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29065 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29067 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29068 (dom_opt_dom_walker::thread_across_edge): Remove
29069 handle_dominating_asserts argument. All callers changed.
29070 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29071 changes. Remove calls to lhs_of_dominating_assert. Other
29072 uses of handle_dominating_asserts turn into unconditional code
29073 (simplify_control_stmt_condition_1): Likewise.
29074 (simplify_control_stmt_condition): Likewise.
29075 (thread_through_normal_block, thread_across_edge): Likewise.
29076 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29077 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29078 object if it is not an SSA_NAME.
29079 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29080 before calling into the VRP specific simplifiers.
29081 (identify_jump_threads): Remove handle_dominating_asserts
29084 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29087 * tree-diagnostic.c (default_tree_printer): No longer static.
29088 * tree-diagnostic.h (default_tree_printer): New prototype.
29090 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29092 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29093 Change ins into fmov.
29095 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29097 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29098 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29099 Use h_con constraint for operand 1.
29100 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29101 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29103 2017-03-15 Jeff Law <law@redhat.com>
29105 PR tree-optimization/71437
29106 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29107 (record_temporary_equivalences): Use it.
29109 PR tree-optimization/71437
29110 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29111 tree-ssa-scopedtables.
29112 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29113 (record_conditions, record_cond, vuse_eq): Likewise.
29114 (record_edge_info): Adjust to API tweak of record_conditions.
29115 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29116 (record_temporary_equivalences, optimize_stmt): Likewise.
29117 (eliminate_redundant_computations): Likewise.
29118 (record_equivalences_from_stmt): Likewise.
29119 * tree-ssa-scopedtables.c: Include options.h and params.h.
29120 (vuse_eq): New function, moved from tree-ssa-dom.c
29121 (build_and_record_new_cond): Likewise.
29122 (record_conditions): Likewise. Accept vector of conditions rather
29123 than edge_equivalence structure for first argument.
29124 for the first argument.
29125 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29126 from tree-ssa-dom.c.
29127 (avail_exprs_stack::record_cond): Likewise.
29128 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29129 from tree-ssa-dom.c.
29130 (avail_exprs_stack): Add new member functions lookup_avail_expr
29132 (record_conditions): Declare.
29134 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29137 * lra-constraints.c (process_alt_operands): Increase reject for
29138 reloading an input/output operand.
29140 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29143 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29144 insns to convert from signed/unsigned char/short to IEEE 128-bit
29146 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29148 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29151 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29152 subreg of inner mode for values already in registers.
29154 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29156 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29157 iteration reg is used after the loop.
29159 2017-03-14 Martin Sebor <msebor@redhat.com>
29161 PR tree-optimization/79800
29162 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29163 precision in negative-positive range.
29164 (format_floating): Call non-const overload with adjusted precision.
29166 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29169 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29172 2017-03-14 Martin Sebor <msebor@redhat.com>
29174 PR middle-end/80020
29175 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29176 * builtins.def (aligned_alloc): Use it.
29179 * Makefile.in (GTFILES): Add calls.c.
29180 * calls.c: Include "gt-calls.h".
29182 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29184 PR rtl-optimization/79728
29185 * regs.h (struct target_regs): New field
29186 x_contains_allocatable_regs_of_mode.
29187 (contains_allocatable_regs_of_mode): New macro.
29188 * reginfo.c (init_reg_sets_1): Initialize it, and change
29189 contains_reg_of_mode so it includes global regs as well.
29190 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29191 rather than contains_regs_of_mode.
29193 2017-03-14 Martin Liska <mliska@suse.cz>
29195 * doc/invoke.texi: Document options that can't be combined with
29196 -fcheck-pointer-bounds.
29198 2017-03-14 Martin Liska <mliska@suse.cz>
29200 PR middle-end/79831
29201 * doc/invoke.texi (-Wchkp): Document the option.
29203 2017-03-14 Martin Liska <mliska@suse.cz>
29205 * Makefile.in: Install gcov-dump.
29207 2017-03-14 Martin Liska <mliska@suse.cz>
29209 * multiple_target.c (expand_target_clones): Bail out for
29210 an invalid attribute.
29212 2017-03-14 Richard Biener <rguenther@suse.de>
29214 * alias.c (struct alias_set_entry): Pack properly.
29215 * cfgloop.h (struct loop): Likewise.
29216 * cse.c (struct set): Likewise.
29217 * ipa-utils.c (struct searchc_env): Likewise.
29218 * loop-invariant.c (struct invariant): Likewise.
29219 * lra-remat.c (struct cand): Likewise.
29220 * recog.c (struct change_t): Likewise.
29221 * rtl.h (struct address_info): Likewise.
29222 * symbol-summary.h (function_summary): Likewise.
29223 * tree-loop-distribution.c (struct partition): Likewise.
29224 * tree-object-size.c (struct object_size_info): Likewise.
29225 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29226 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29227 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29228 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29229 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29230 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29231 (struct _stmt_vec_info): Likewise.
29233 2017-03-14 Martin Liska <mliska@suse.cz>
29236 * multiple_target.c (create_dispatcher_calls): Check that
29237 a target can create a function dispatcher.
29239 2017-03-14 Martin Liska <mliska@suse.cz>
29242 * multiple_target.c (expand_target_clones): Drop local.local
29243 flag for default implementation.
29245 2017-03-14 Richard Biener <rguenther@suse.de>
29247 PR tree-optimization/80030
29248 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29250 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29252 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29253 gcc_fallthrough() instead of __attribute__((fallthrough));
29255 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29257 * doc/gcc.texi: Remove "up" link to (DIR).
29258 * doc/gccint.texi: Ditto.
29260 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29262 * doc/install.texi (Specific) <avr>: Remove reference to
29265 2017-03-13 Jeff Law <law@redhat.com>
29267 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29268 attribute rather than comments.
29270 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29271 match_scratch operand is highest.
29273 2017-03-13 Martin Liska <mliska@suse.cz>
29275 PR middle-end/78339
29276 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29277 is a CHKP clone, use original declaration.
29279 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29281 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29282 (arc_conditional_register_usage): Use a different allocation order
29283 when optimizing for size.
29284 * common/config/arc/arc-common.c (arc_option_optimization_table):
29285 Section anchors default on when optimizing for size.
29287 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29289 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29291 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29293 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29294 * config/arc/arc.md (cpu_facility): Add cd variant.
29295 (*movqi_insn): Add code density variant.
29296 (*movhi_insn): Likewise.
29297 (*movqi_insn): Likewise.
29298 (*addsi3_mixed): Likewise.
29299 (subsi3_insn): Likewise.
29301 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29303 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29305 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29307 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29308 expressions with MINUS and UNARY ops.
29310 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29313 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29315 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29316 between vec_select and vector argument.
29317 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29318 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29319 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29320 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29321 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29322 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29324 2017-03-13 Richard Biener <rguenther@suse.de>
29327 * params.def (vect-max-peeling-for-alignment): Fix typo.
29329 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29331 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29332 issue that only occurred with binutils below 2.18.
29334 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29336 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29337 refer to binutils 2.11/2.12 minimum.
29339 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29341 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29342 ftp.kernel.org and simplify binutils requirement.
29344 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29346 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29348 (Optimize Options): Ditto. Also remove redundancy.
29350 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29352 PR translation/79848
29353 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29355 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29356 to G_ to avoid double translation.
29358 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29360 PR translation/79923
29361 * auto-profile.c (get_combined_location): Convert leading
29362 character of diagnostics to lower case and remove trailing period.
29363 (read_profile): Likewise for various diagnostics.
29364 * config/arm/arm.c (arm_option_override): Remove trailing period
29365 from various diagnostics.
29366 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
29367 (msp430_expand_delay_cycles): Likewise.
29369 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29372 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
29373 full command-line argument, rather than just "str".
29374 (aarch64_validate_march): Likewise.
29375 (aarch64_validate_mtune): Likewise.
29377 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
29379 PR rtl-optimization/78911
29380 * lra-assigns.c (must_not_spill_p): New function.
29381 (spill_for): Use it.
29383 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29385 PR tree-optimization/79981
29386 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
29387 ATOMIC_COMPARE_EXCHANGE ifn result.
29388 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
29389 IFN_ATOMIC_COMPARE_EXCHANGE.
29391 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29394 * opts.c (parse_sanitizer_options): Add missing question mark to
29395 "did you mean" message.
29397 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29399 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
29401 (VMULEUH_UNS): Likewise.
29402 (VMULOUB_UNS): Likewise.
29403 (VMULOUH_UNS): Likewise.
29404 * config/rs6000/rs6000.c (builtin_function_type): Remove
29405 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
29407 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29410 * read-rtl-function.c (function_reader::read_rtx_operand): Update
29411 x with result of extra_parsing_for_operand_code_0.
29412 (function_reader::extra_parsing_for_operand_code_0): Convert
29413 return type from void to rtx, returning x. When reading
29414 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
29415 larger size containing struct block_symbol.
29417 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
29419 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
29420 -mfloat128-hardware without -m64.
29422 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
29425 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
29426 entries to the case statement that marks unsigned arguments to
29427 overloaded functions.
29429 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29431 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
29432 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
29434 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
29437 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
29438 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
29440 2017-03-10 Martin Liska <mliska@suse.cz>
29444 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
29445 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
29448 2017-03-10 Olivier Hainque <hainque@adacore.com>
29450 * tree-switch-conversion (array_value_type): Start by resetting
29451 candidate type to it's main variant.
29453 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29455 PR rtl-optimization/79909
29456 * combine.c (try_combine): Use simplify_replace_rtx on individual
29457 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
29458 of the whole CALL_INSN_FUNCTION_USAGE.
29460 PR tree-optimization/79972
29461 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
29462 get_range_info on SSA_NAMEs. Formatting fixes.
29464 2017-03-10 Richard Biener <rguenther@suse.de>
29465 Jakub Jelinek <jakub@redhat.com>
29467 PR tree-optimization/77975
29468 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
29469 edge to be constant.
29470 (get_val_for): For constant x return it. Formatting fix.
29471 (loop_niter_by_eval): Avoid pointless looping if the next iteration
29472 would use the same bases as the current one.
29474 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29476 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
29477 instead of vec_select for V1TImode.
29478 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
29480 (VSX_LE_128): Add V1TI to this mode iterator.
29481 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
29482 (*vsx_le_perm_store_<mode>): Likewise.
29483 (pre-reload splitter for VSX stores): Likewise.
29484 (post-reload splitter for VSX stores): Likewise.
29485 (*vsx_xxpermdi2_le_<mode>): Likewise.
29486 (*vsx_lxvd2x2_le_<mode>): Likewise.
29487 (*vsx_stxvd2x2_le_<mode>): Likewise.
29489 2017-03-09 Michael Eager <eager@eagercon.com>
29491 Correct failures with --enable-checking=yes,rtl.
29493 * config/microblaze/microblaze.c (microblaze_expand_shift):
29494 Replace GET_CODE test with CONST_INT_P and INTVAL test with
29495 test for const0_rtx.
29496 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
29497 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
29499 2017-03-09 Richard Biener <rguenther@suse.de>
29501 PR tree-optimization/79977
29502 * graphite-scop-detection.c (scop_detection::merge_sese):
29503 Handle the case of extra exits to blocks dominating the entry.
29505 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
29507 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29510 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
29512 PR rtl-optimization/79949
29513 * lra-constraints.c (process_alt_operands): Check memory when
29514 trying to predict a cycle. Print about the overall increase.
29516 2017-03-09 Richard Biener <rguenther@suse.de>
29518 PR middle-end/79971
29519 * gimple-expr.c (useless_type_conversion_p): Preserve
29520 TYPE_SATURATING for fixed-point types.
29522 2017-03-09 Richard Biener <rguenther@suse.de>
29525 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
29526 alignment of BLKmode params.
29528 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29531 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
29532 (VALL_NO_V2Q): Likewise.
29534 * config/aarch64/aarch64-simd.md
29535 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
29537 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
29538 VALL_NO_V2Q mode iterator.
29539 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
29541 2017-03-09 Martin Liska <mliska@suse.cz>
29543 PR tree-optimization/79631
29544 * tree-chkp-opt.c (chkp_is_constant_addr): Call
29545 tree_int_cst_sign_bit just for INTEGER constants.
29547 2017-03-09 Martin Liska <mliska@suse.cz>
29551 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
29554 2017-03-09 Marek Polacek <polacek@redhat.com>
29557 * tree.c (inchash::add_expr): Handle TREE_VEC.
29559 2017-03-09 Martin Liska <mliska@suse.cz>
29562 (chkp_narrow_size_and_offset): New function.
29563 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
29564 (void chkp_parse_bit_field_ref): New function.
29565 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
29566 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
29568 2017-03-09 Martin Liska <mliska@suse.cz>
29571 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
29572 (chkp_find_bounds_1): Remove gcc_unreachable.
29574 2017-03-09 Jakub Jelinek <jakub@redhat.com>
29577 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
29578 BUILT_IN_SYNC*, determine the access type from the size suffix and
29579 always build a MEM_REF with that type. Handle forgotten
29580 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
29583 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
29584 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
29585 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
29586 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
29587 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
29588 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
29589 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
29590 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
29591 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
29592 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
29593 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
29594 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
29595 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
29596 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
29597 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
29598 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
29599 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
29600 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
29601 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
29602 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
29603 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
29604 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
29605 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
29606 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
29607 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
29608 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
29609 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
29610 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
29611 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
29612 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
29613 definitions outside of __OPTIMIZE__ guarded section.
29616 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
29617 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
29618 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
29619 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
29622 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29624 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
29625 ("vfenez<mode>"): Add missing constraints.
29627 2017-03-08 Martin Sebor <msebor@redhat.com>
29630 * config/nds32/nds32.c (nds32_option_override):
29631 Fix misspelled diagnostic.
29633 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29636 * gimplify.c (gimplify_omp_for): Replace index var in outer
29637 taskloop statement with an artificial variable and add
29638 OMP_CLAUSE_PRIVATE clause for it.
29640 2017-03-08 Richard Biener <rguenther@suse.de>
29642 PR tree-optimization/79955
29643 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
29644 for accesses that are completely outside of the variable.
29646 2017-03-08 Andrew Haley <aph@redhat.com>
29648 PR tree-optimization/79943
29649 * tree-ssa-loop-split.c (compute_new_first_bound): When
29650 calculating the new upper bound, (END-BEG) should be added, not
29653 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29655 * config/avr/avr.md (setmemhi): Make sure match_dup
29656 operand number comes before match_scratch.
29658 2017-03-08 Richard Biener <rguenther@suse.de>
29660 PR tree-optimization/79920
29661 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
29662 with ncopies == 1 to ...
29663 (vect_transform_slp_perm_load): ... here. Properly compute
29664 all element loads by iterating VF times over the group. Do
29665 not handle ncopies (computed in a broken way) in
29666 vect_create_mask_and_perm.
29668 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29671 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
29672 is a uniform vector, use uniform_vector_p return value instead of
29673 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
29675 2017-03-07 Marek Polacek <polacek@redhat.com>
29677 PR middle-end/79809
29678 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
29679 (alloca_call_type): Likewise.
29681 2017-03-07 Martin Liska <mliska@suse.cz>
29683 * gcov.c (process_args): Put comment to correct location.
29685 2017-03-07 Martin Liska <mliska@suse.cz>
29687 PR middle-end/68270
29688 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
29689 Use array_at_struct_end_p instead of DECL_CHAIN (field).
29690 (chkp_narrow_bounds_for_field): Likewise.
29691 (chkp_parse_array_and_component_ref): Pass one more argument to
29694 2017-03-07 Richard Biener <rguenther@suse.de>
29696 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
29699 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
29701 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
29702 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
29704 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29707 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
29708 to end of description.
29709 (PARAM_MAX_STORES_TO_MERGE): Likewise.
29711 2017-03-07 Jakub Jelinek <jakub@redhat.com>
29713 PR rtl-optimization/79901
29714 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
29716 (*avx512f_<code><mode>3<mask_name>): ... this.
29717 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
29718 iterator instead of VI8_AVX2_AVX512BW.
29720 PR rtl-optimization/79901
29721 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
29722 min/max expander, expand it using expand_vec_cond_expr.
29725 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
29728 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29731 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
29732 to void * for PCH reasons.
29733 * dwarf2out.c (output_loc_operands, output_die): Cast
29734 v.val_vec.array to unsigned char *.
29736 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
29739 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
29742 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
29744 PR rtl-optimization/79571
29745 * lra-constraints.c (process_alt_operands): Calculate static
29746 reject and subtract it from overall when only addresses will be
29749 2017-03-06 Julia Koval <julia.koval@intel.com>
29752 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
29753 incoming stack boundary to 128 for 64-bit targets.
29755 2017-03-06 Richard Biener <rguenther@suse.de>
29757 PR tree-optimization/79894
29758 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
29759 to NULL after folding it.
29761 2017-03-06 Richard Biener <rguenther@suse.de>
29763 PR tree-optimization/79824
29764 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
29765 check disabling peeling for gaps.
29767 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
29769 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
29770 attributes): Document gettimeofday.
29772 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29774 * config/s390/s390.c (s390_option_override_internal): Set
29775 PARAM_MIN_VECT_LOOP_BOUND
29777 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29779 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
29780 * config/s390/s390.md: Likewise.
29782 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29785 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
29786 (<avx2_avx512>_perm<mode>): Rename to ...
29787 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
29789 (<avx512>_perm<mode>_mask): Rename to ...
29790 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
29792 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
29793 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
29794 instead of VI8F_256_512.
29795 (avx512f_perm<mode>): New define_expand.
29796 (avx512f_perm<mode>_mask): Likewise.
29797 (avx512f_perm<mode>_1<mask_name>): New define_insn.
29798 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
29800 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29802 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
29803 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
29805 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
29807 2017-03-06 Martin Liska <mliska@suse.cz>
29810 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
29811 when having a SSA NAME w/o VAR_DECL assigned to it.
29813 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29815 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
29816 msa_dpsub_<su>_d): Fix MODE for vec_select.
29818 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29820 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
29822 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
29824 2017-03-06 Richard Biener <rguenther@suse.de>
29826 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
29827 * plugin.c (register_plugin_info): Likewise.
29828 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
29830 2017-03-05 Jakub Jelinek <jakub@redhat.com>
29832 * config/i386/sse.md (sse_storehps, sse_storelps,
29833 avx_<castmode><avxsizesuffix>_<castmode>,
29834 avx512f_<castmode><avxsizesuffix>_<castmode>,
29835 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
29836 in condition that at least one operand is not a MEM.
29838 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29840 PR middle-end/79805
29841 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
29842 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
29844 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
29845 gimple_call_nothrow_p flag based on whether original builtin can throw.
29846 If it can, emit following stmts on the fallthrough edge.
29847 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
29848 don't create new bb if inserting just debug stmts on the edge, try to
29849 insert them on the fallthru bb or just reset debug stmts.
29851 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
29854 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
29855 restore recog_data (including the operand rtxes inside it) around
29856 the call to get_insn_template.
29858 2017-03-03 Martin Sebor <msebor@redhat.com>
29860 PR tree-optimization/79699
29861 * context.c (context::~context): Free MPFR caches to avoid
29862 a memory leak on program exit.
29864 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29866 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
29867 Use wide_int::ulow () instead of .elt (0).
29869 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29871 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
29872 (*pushxf): Limit oF constraint to 32bit targets and add oC
29873 constraint for 64bit targets.
29874 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
29875 (*pushdf): Change rmF constraint to rmC.
29877 2017-03-03 Martin Liska <mliska@suse.cz>
29879 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
29880 Remove unused variable.
29882 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29885 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
29886 is a memory operand, increase num_memory.
29887 (ix86_expand_args_builtin): Likewise.
29889 2017-03-03 Jan Hubicka <jh@suse.cz>
29892 * ipa-devirt.c (maybe_record_node): Properly handle
29893 __cxa_pure_virtual visibility.
29895 2017-03-03 Martin Liska <mliska@suse.cz>
29897 PR tree-optimization/79803
29898 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
29900 (pass_loop_prefetch::execute): Disabled optimization if an
29901 assumption about L1 cache size is not met.
29903 2017-03-03 Martin Liska <mliska@suse.cz>
29905 PR rtl-optimization/79574
29906 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
29907 (hash_scan_set): Likewise.
29908 (dump_hash_table): Likewise.
29909 (hoist_code): Likewise.
29911 2017-03-03 Richard Biener <rguenther@suse.de>
29913 * fixed-value.c (fixed_from_string): Restore use of elt (1)
29914 in place of uhigh ().
29915 (fixed_convert_from_real): Likewise.
29917 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29920 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
29922 2017-03-03 Richard Biener <rguenther@suse.de>
29924 PR middle-end/79818
29925 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
29926 TYPE_OVERFLOW_UNDEFINED check.
29928 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29930 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
29932 (vector_ae_<mode>_p): Likewise.
29933 (vector_nez_<mode>_p): Likewise.
29934 (vector_ne_v2di_p): Likewise.
29935 (vector_ae_v2di_p): Likewise.
29936 (vector_ne_<mode>_p): Likewise.
29937 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
29939 (vsx_tsqrt<mode>2_fe): Likewise.
29941 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
29944 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
29946 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29948 PR rtl-optimization/79780
29949 * cprop.c (one_cprop_pass): When second and further conditional trap
29950 in a single basic block is turned into an unconditional trap, turn it
29951 into a deleted note to avoid RTL verification failures.
29953 2017-03-02 Richard Biener <rguenther@suse.de>
29955 * fold-const.c (const_binop): Use ulow () instead of elt (0).
29957 2017-03-02 Richard Biener <rguenther@suse.de>
29959 PR tree-optimization/79345
29961 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
29962 param and abort the walk, returning -1 if it is hit.
29963 (walk_aliased_vdefs): Take a limit param and pass it on.
29964 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
29965 defaulting to 0 and return a signed int.
29966 * tree-ssa-uninit.c (struct check_defs_data): New struct.
29967 (check_defs): New helper.
29968 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
29969 about uninitialized memory.
29970 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
29971 bogus uninitialized warning.
29972 (fixed_convert_from_real): Likewise.
29974 2017-03-02 Bin Cheng <bin.cheng@arm.com>
29976 PR tree-optimization/66768
29977 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
29978 iv_use if base object can't be determined.
29980 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29982 PR tree-optimization/79345
29983 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
29984 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
29985 (get_pattern_stats): Initialize it.
29986 * genemit.c (gen_expand): Verify match_scratch numbers come after
29987 match_operand/match_dup numbers.
29988 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
29989 match_scratch numbers.
29990 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
29992 * config/s390/s390.md (trunctdsd2): Likewise.
29994 2017-03-02 Richard Biener <rguenther@suse.de>
29996 * wide-int.h (wide_int_storage::operator=): Implement in terms
29999 2017-03-02 Richard Biener <rguenther@suse.de>
30001 PR tree-optimization/79777
30002 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
30003 the to insert expression to sth existing.
30005 2017-03-01 Martin Sebor <msebor@redhat.com>
30007 PR middle-end/79692
30008 * gimple-ssa-sprintf.c
30009 (directive::known_width_and_precision): New function.
30010 (format_integer): Use it.
30011 (get_mpfr_format_length): Consider the full range of precision
30012 when computing %g output with the # flag. Set the likely byte
30013 count to 3 rather than 1 when precision is indeterminate.
30014 (format_floating): Correct the lower bound of precision.
30016 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30018 * doc/invoke.texi: Document default code model for 64-bit Linux.
30020 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30023 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30024 udiv rather than div since input pattern is unsigned.
30026 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30028 * config/i386/i386.c (print_reg): Warn for values of
30029 unsupported size in integer register.
30031 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30034 * config/rs6000/predicates.md (current_file_function_operand): Do
30035 not allow self calls to be local if the function is replaceable.
30037 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30040 * config/rs6000/altivec.h (vec_ctz and others): Change the
30041 preprocessor macro that controls conditional compilation from
30042 _ARCH_PWR9 to __POWER9_VECTOR__.
30043 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30044 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30045 control (instead of _ARCH_PWR9 control) so that template
30046 definition uses power9-specific function.
30047 (vec_any_eq): Likewise.
30048 (vec_all_ne): Change macro definition to use a power9-specific
30049 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30050 _ARCH_PWR9 control).
30051 (vec_any_eq) Likewise.
30052 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30053 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30054 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30055 support for xvcmpnedp instruction.
30056 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30057 macro expansion so that Power9 implementation of vec_all_ne does
30058 not use the AltiVec predicate framework.
30059 (VCMPNEH_P): Likewise.
30060 (VCMPNEW_P): Likewise.
30061 (VCMPNED_P): Likewise.
30062 (VCMPNEFP_P): Likewise.
30063 (VCMPNEDP_P): Likewise.
30064 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30065 implementation of vec_any_eq to not use AltiVec predicate
30067 (VCMPAEH_P): Likewise.
30068 (VCMPAEW_P): Likewise.
30069 (VCMPAED_P): Likewise.
30070 (VCMPAEFP_P): Likewise.
30071 (VCMPAEDP_P): Likewise.
30072 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30073 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30074 not use the AltiVec predicate framework.
30075 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30076 of vec_any_eq to not use AltiVec predicate framework.
30077 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30078 support for predefined __POWER9_VECTOR__ macro to indicate that
30079 Power9 instruction selection is enabled.
30080 (altivec_overloaded_builtins): Remove extraneous
30081 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30082 function argument types RS6000_BTI_bool_V16QI and
30083 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30084 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30085 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30086 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30087 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30088 Power9 for implementations of vec_cmpne. Change the signature for
30089 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30090 (representing vec_all_ne) to remove the previously described first
30091 argument of type RS6000_BTI_INTSI, as this was an artifact of
30092 reliance on the AltiVec predicate framework, which is no longer
30093 used in the implementation of these functions. Add
30094 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30095 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30096 since, unlike the AltiVec predicate framework implementation, we
30097 do not share function descriptors between vec_alle and vec_anyeq.
30098 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30099 set of modes that receive special treatment even when
30100 TARGET_P9_VECTOR is true. The special treatment emits code that
30101 does not depend on Power9 instructions.
30102 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30103 define_expand to not rely on AltiVec predicate framework.
30104 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30106 (vector_ne_v2di_p): Change this define_expand to not rely on
30107 AltiVec predicate framework.
30108 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30110 (vector_ne_<mode>_p): Change this define_expand to not rely on
30111 AltiVec predicate framework.
30112 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30114 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30115 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30116 define_insn pattern.
30117 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30118 define_insn pattern because the xvcmpne<VSs>. instruction is not
30120 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30121 instruction is not supported.
30123 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30125 * config/nvptx/nvptx.c: Include intl.h.
30127 2017-03-01 Martin Jambor <mjambor@suse.cz>
30130 * ipa-prop.h (ipa_bits): Removed field known.
30131 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30132 to pointers. Adjusted their comments to warn about their sharing.
30133 (ipcp_transformation_summary): Change bits to a vector of pointers.
30134 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30135 (ipa_get_ipa_bits_for_value): Declare.
30136 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30137 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30138 (ipa_bits_hash_table): Likewise.
30139 (ipa_vr_ggc_hash_traits): Likewise.
30140 (ipa_vr_hash_table): Likewise.
30141 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30142 being pointers and vr_known being removed.
30143 (ipa_set_jf_unknown): Likewise.
30144 (ipa_get_ipa_bits_for_value): New function.
30145 (ipa_set_jfunc_bits): Likewise.
30146 (ipa_get_value_range): New overloaded functions.
30147 (ipa_set_jfunc_vr): Likewise.
30148 (ipa_compute_jump_functions_for_edge): Use the above functions to
30149 construct bits and vr parts of jump functions.
30150 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30151 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30153 (ipcp_grow_transformations_if_necessary): Also allocate
30154 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30156 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30157 them. Fix too long lines.
30158 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30159 vr_known being removed.
30160 (ipa_read_jump_function): Use new setter functions to construct bits
30161 and vr parts of jump functions or set them to NULL.
30162 (write_ipcp_transformation_info): Adjust for bits being pointers.
30163 (read_ipcp_transformation_info): Likewise.
30164 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30166 Include gt-ipa-prop.h.
30167 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30169 (ipcp_store_bits_results): Likewise.
30170 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30171 Do not write to existing jump functions but use a temporary instead.
30173 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30176 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30177 attempt to use its first operand as BIT_FIELD_REF base.
30179 2017-03-01 Richard Biener <rguenther@suse.de>
30181 PR middle-end/79721
30182 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30183 interpolating formula in wrapping arithmetic.
30184 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30186 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30188 PR tree-optimization/79734
30189 * tree-vect-generic.c (expand_vector_condition): Optimize
30190 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30191 Handle VEC_COND_EXPR where comparison has different inner width from
30192 type's inner width.
30194 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30196 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30197 markup, and similar issues. Remove @opindex entries for things
30198 that aren't options. Add missing -mmpy-option entries.
30200 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30202 PR tree-optimization/79737
30203 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30204 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30205 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30206 instead of byte_size. Formatting fix.
30207 (shift_bytes_in_array_right): Formatting fix.
30209 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30212 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30213 condition on optimize for the leaf function test.
30215 2017-02-28 Martin Liska <mliska@suse.cz>
30218 * read-rtl-function.c (function_reader::handle_unknown_directive):
30219 Bail out when one uses -flto.
30221 2017-02-28 Martin Liska <mliska@suse.cz>
30223 * common.opt: Replace space with tabular for options of <number>
30225 * config/i386/i386.opt: Show <number> value for
30226 -mlarge-data-threshold.
30227 * opts.c (print_filtered_help): Do not display number in hexadecimal
30230 2017-02-28 Martin Liska <mliska@suse.cz>
30232 * common.opt: Fix --help=option -Q for options which are of
30235 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30237 * config/i386/i386.c (print_reg): Error out for values
30238 of 8-bit size in invalid integer register.
30240 2017-02-28 Martin Sebor <msebor@redhat.com>
30242 PR tree-optimization/79691
30243 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30245 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30248 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30249 gcc_unreachable with output_operand_lossage.
30251 2017-02-28 Richard Biener <rguenther@suse.de>
30253 PR tree-optimization/79740
30254 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30256 (visit_nary_op): Insert the nary into the hashtable if we
30257 pattern-matched sth.
30258 * tree-ssa-pre.c (eliminate_insert): Robustify.
30260 2017-02-28 Richard Biener <rguenther@suse.de>
30262 PR middle-end/79731
30263 * fold-const.c (decode_field_reference): Reject out-of-bound
30266 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30268 * config/i386/i386.c: Include intl.h.
30269 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30270 instead of just cond ? "..." : "...".
30271 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30272 * coverage.c (read_counts_file): Likewise.
30273 * omp-offload.c: Include intl.h.
30274 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30275 of just cond ? "..." : "...".
30276 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30277 of just cond ? "..." : "...".
30279 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30282 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30284 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30285 'tune for' CPU name.
30286 * config/arm/arm-cpu-data.h: Regenerated.
30288 2017-02-28 Richard Biener <rguenther@suse.de>
30290 PR tree-optimization/79732
30291 * tree-inline.c (expand_call_inline): Do not shadow var.
30293 2017-02-28 Richard Biener <rguenther@suse.de>
30295 PR tree-optimization/79723
30296 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30297 address-space properly.
30299 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30301 * doc/optinfo.texi (Optimization groups): Fix option used for
30303 * doc/invoke.texi (-fopt-info): Document "omp".
30304 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30305 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30306 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30307 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30308 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30310 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30312 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30313 OPTGROUP_OMP with "omp".
30315 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30318 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30319 for arithmetic shift of unsigned V2DI.
30321 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30323 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30324 arc/linux.h headers.
30325 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30326 (LINK_SPEC): Likewise.
30327 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30328 (EXTRA_SPECS): Likewise.
30329 (STARTFILE_SPEC): Likewise.
30330 (ENDFILE_SPEC): Likewise.
30331 (LIB_SPEC): Likewise.
30332 (TARGET_SDATA_DEFAULT): Likewise.
30333 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30334 (MULTILIB_DEFAULTS): Likewise.
30335 (DWARF2_UNWIND_INFO): Likewise.
30336 * config/arc/big.h: New file.
30337 * config/arc/elf.h: Likewise.
30338 * config/arc/linux.h: Likewise.
30339 * config/arc/t-uClibc: Remove.
30341 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30343 PR tree-optimization/77536
30344 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30345 (tree_transform_and_unroll_loop): Use above function to compute the
30346 estimated niter of unrolled loop and use it when scaling profile.
30347 Also use count info rather than frequency if it's non-zero.
30348 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30349 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30350 (vect_transform_loop): Call above function.
30352 2017-02-27 Richard Biener <rguenther@suse.de>
30354 PR tree-optimization/45397
30355 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30356 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30357 (visit_nary_op): Add pattern matching for CSEing sign-changed
30358 or truncated operations with wider ones.
30360 2017-02-27 Richard Biener <rguenther@suse.de>
30362 PR tree-optimization/79690
30363 * tree-vect-stmts.c (vectorizable_store): Use vector type
30364 built from the DR with address-space.
30366 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
30368 * doc/invoke.texi (Optimize Options): Refine the description
30369 of asan-use-after-return.
30371 2017-02-25 Alan Modra <amodra@gmail.com>
30373 PR rtl-optimization/79584
30374 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
30375 base, not ad->base_term, the reg within base. Remove assertion
30376 that ad->base == ad->base_term. Replace gen_int_mode using
30377 bogus mode with const0_rtx.
30379 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30381 PR middle-end/79396
30382 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30383 FMA_EXPR like tcc_binary or tcc_unary.
30385 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
30388 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
30390 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
30391 (output_loc_operands): Handle DW_OP_call_ref and
30392 DW_OP_GNU_variable_value.
30393 (struct variable_value_struct): New type.
30394 (struct variable_value_hasher): Likewise.
30395 (variable_value_hash): New variable.
30396 (string_types): Remove.
30397 (copy_loc_descr): New function.
30398 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
30399 (prepend_loc_descr_to_each): New function.
30400 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
30401 instead of add_loc_descr_to_each if the first argument is single
30402 location list and the second has multiple.
30403 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
30404 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
30405 when looking for variable value which doesn't have other location info.
30406 (loc_list_from_tree): Formatting fix.
30407 (gen_array_type_die): Simplify DW_AT_string_length handling.
30408 (adjust_string_types): Remove.
30409 (gen_subprogram_die): Don't call adjust_string_types nor test/set
30410 string_types. Call resolve_variable_values.
30411 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
30412 (resolve_addr_in_expr): Likewise. Add A argument.
30413 (copy_deref_exprloc): Remove deref argument. Adjust for the
30414 original expression being DW_OP_GNU_variable_value with optionally
30415 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
30416 optionally after it.
30417 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
30418 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
30419 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
30420 (variable_value_hasher::hash, variable_value_hasher::equal): New
30422 (resolve_variable_value_in_expr, resolve_variable_value,
30423 resolve_variable_values, note_variable_value_in_expr,
30424 note_variable_value): New functions.
30425 (dwarf2out_early_finish): Call note_variable_value on all toplevel
30428 2017-02-24 Jakub Jelinek <jakub@redhat.com>
30431 * opts.h (handle_generated_option): Add GENERATED_P argument.
30432 * opts-common.c (handle_option): Adjust function comment.
30433 (handle_generated_option): Add GENERATED_P argument, pass it to
30435 (control_warning_option): Pass false to handle_generated_option
30437 * opts.c (maybe_default_option): Pass true to handle_generated_option
30439 * optc-gen.awk: Likewise.
30441 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30443 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
30444 a REG, look at the REG it is a SUBREG of.
30445 (splitter for cmpeqsi_t): Ditto.
30447 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30449 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
30450 the special USEs with the pattern of the insn, not the insn itself.
30452 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
30455 * doc/invoke.texi: Document -mload-store-pairs.
30457 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30458 Sandra Loosemore <sandra@codesourcery.com>
30460 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
30461 argument isn't a CONST_INT.
30462 (nios2_alternate_compare_const): Assert op is a CONST_INT.
30463 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
30464 (nios2_validate_compare): Bypass alternate compare logic if *op2
30465 is not a CONST_INT.
30466 (ldstwm_operation_p): Return false if first_base is not a REG or
30467 if first_offset is not a CONST_INT.
30469 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30471 * config/cris/cris.md: Use correct operand in a define_peephole2.
30473 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30475 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
30477 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30479 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
30480 this_insn if it is an INSN or JUMP_INSN.
30481 (force_offsettable): Look at base, not at addr.
30482 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
30483 on things that aren't necessarily CONST_INTs.
30485 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
30487 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
30488 -mfpmath=sse is the default also for x86-32 targets with SSE2
30489 instruction set when @option{-ffast-math} is enabled
30491 2017-02-24 Jeff Law <law@redhat.com>
30493 PR rtl-optimizatoin/79286
30494 * ira.c (update_equiv_regs): Drop may_trap_p exception to
30497 2017-02-24 Richard Biener <rguenther@suse.de>
30499 PR tree-optimization/79389
30500 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
30503 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
30505 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
30506 function comment to reflect reality.
30507 (loop_exits_before_overflow): Fix typo in function description.
30509 2017-02-24 Richard Biener <rguenther@suse.de>
30511 PR tree-optimization/79389
30512 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
30513 properly that a threading opportunity exists. Detect conditional
30514 copy/constant propagation opportunities.
30516 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
30518 * config/visium/visium.md (type): Add trap.
30519 (b): New mode attribute.
30520 (*btst): Rename into...
30521 (*btst<mode>): ...this and adjust.
30522 (*cbranchsi4_btst_insn): Rename into...
30523 (*cbranch<mode>4_btst_insn): ...this and adjust.
30524 (trap): New define_insn.
30526 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30528 PR tree-optimization/79389
30529 * ifcvt.c (struct noce_if_info): Add rev_cond field.
30530 (noce_reversed_cond_code): New function.
30531 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
30532 reversed_comparison_code. Formatting fix.
30533 (noce_try_store_flag): Test rev_cond != NULL in addition to
30534 reversed_comparison_code.
30535 (noce_try_store_flag_constants): Likewise.
30536 (noce_try_store_flag_mask): Likewise.
30537 (noce_try_addcc): Use rev_cond if non-NULL instead of
30538 reversed_comparison_code.
30539 (noce_try_cmove_arith): Likewise. Formatting fixes.
30540 (noce_try_minmax, noce_try_abs): Clear rev_cond.
30541 (noce_find_if_block): Initialize rev_cond.
30542 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
30543 instead of false as last argument never attempt to reverse it
30546 2017-02-23 Bin Cheng <bin.cheng@arm.com>
30548 PR tree-optimization/79663
30549 * tree-predcom.c (combine_chains): Process refs in reverse order
30550 only for ZERO length chains, and add explaining comment.
30552 2017-02-23 Jeff Law <law@redhat.com>
30554 PR tree-optimization/79578
30555 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
30556 in call to operand_equal_p.
30558 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
30561 * config/i386/cpuid.h: Fix another undefined behavior.
30563 2017-02-23 Richard Biener <rguenther@suse.de>
30565 PR tree-optimization/79683
30566 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
30567 vector types for data-refs.
30569 2017-02-23 Martin Liska <mliska@suse.cz>
30571 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
30573 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30575 PR middle-end/79665
30576 * internal-fn.c (get_range_pos_neg): Moved to ...
30577 * tree.c (get_range_pos_neg): ... here. No longer static.
30578 * tree.h (get_range_pos_neg): New prototype.
30579 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
30580 are known to be in between 0 and signed maximum inclusive, try to
30581 expand both unsigned and signed divmod and use the cheaper one from
30584 2017-02-22 Jeff Law <law@redhat.com>
30586 PR tree-optimization/79578
30587 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
30588 to compare base operands.
30590 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
30593 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
30594 gpc_reg_operand instead of fpr_reg_operand.
30596 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
30598 * config/mips/mips.c (mips_return_in_memory): Force FP
30599 vector types to be returned in memory for o32 ABI.
30601 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30603 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
30604 instead of DW_TAG_member for static data member declarations and don't
30605 set no_linkage_name for static inline data members.
30606 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
30609 2017-02-22 Martin Liska <mliska@suse.cz>
30611 * doc/invoke.texi: Replace inequality signs with square brackets
30614 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
30617 * lra-constraints.c (simplify_operand_subreg): Handle
30618 WORD_REGISTER_OPERATIONS targets.
30620 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30623 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
30624 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
30625 elimination by swapping fld*.
30627 2017-02-22 Richard Biener <rguenther@suse.de>
30629 PR tree-optimization/79673
30630 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
30631 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
30632 irrelevant address-space qualifiers and avoiding a
30633 ADDR_SPACE_CONVERT_EXPR from fold_convert.
30635 2017-02-22 Richard Biener <rguenther@suse.de>
30637 PR tree-optimization/79666
30638 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
30639 to not symbolically negate if that may introduce undefined
30642 2017-02-22 Martin Liska <mliska@suse.cz>
30645 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
30646 * data-streamer-out.c (streamer_write_gcov_count_stream):
30648 * value-prof.c (stream_out_histogram_value): Make assert more
30649 precise based on type of counter.
30651 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
30654 * config/i386/i386.md (standard_x87sse_constant_load splitter):
30655 Use nonimmediate_operand instead of memory_operand for operand 1.
30656 (float-extend standard_x87sse_constant_load splitter): Ditto.
30658 2017-02-21 Jeff Law <law@redhat.com>
30660 PR tree-optimization/79621
30661 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
30662 blocks with edges to themselves.
30664 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30667 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
30668 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
30669 Use gimple_call_builtin_p.
30672 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
30673 on temporarily removed DEBUG_INSNs.
30675 PR tree-optimization/79649
30676 * tree-loop-distribution.c (classify_partition): Give up on
30677 non-generic address space loads/stores.
30679 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
30681 * doc/loop.texi (Loop manipulation): Remove nonexistent
30682 tree_ssa_loop_version from the documentation.
30683 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
30685 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30688 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
30689 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
30690 * config/rs6000/rs6000.c: Include except.h.
30691 (rs6000_expand_split_stack_prologue): Call
30692 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
30694 2017-02-21 Martin Jambor <mjambor@suse.cz>
30697 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
30698 have been analyzed.
30700 2017-02-21 Martin Jambor <mjambor@suse.cz>
30702 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
30703 for backward compatibility only.
30704 * doc/invoke.texi (Option Summary): Remove all references to
30705 -fipa-cp-alignment.
30707 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
30711 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30713 * lra-constraints.c (curr_insn_transform): Handle
30714 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30716 2017-02-21 Martin Liska <mliska@suse.cz>
30718 * config/i386/i386.opt: Replace -masm-dialect with -masm.
30720 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
30722 PR translation/79638
30723 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
30725 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
30728 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
30729 (arm_function_ok_for_sibcall): Return false for an indirect call by
30730 descriptor if all the argument registers are used.
30731 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
30732 alignment of the function.
30734 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30736 PR tree-optimization/61441
30737 * simplify-rtx.c (simplify_const_unary_operation): For
30738 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
30739 the sNaN unmodified.
30741 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30743 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30744 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30745 instead of SYSTEM_HEADER_DIR.
30747 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
30748 Martin Liška <mliska@suse.cz>
30750 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
30751 Fix typos and grammar, use active voice, and clarify.
30753 2017-02-20 Marek Polacek <polacek@redhat.com>
30755 PR middle-end/79537
30756 * gimplify.c (gimplify_expr): Handle unused *&&L;.
30759 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
30761 2017-02-20 Jakub Jelinek <jakub@redhat.com>
30764 * config/i386/i386.c (ix86_expand_builtin): Handle
30765 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
30766 ix86_builtins_isa[fcode].isa as a requirement of those
30767 flags and any other flag in the bitmask.
30768 (ix86_init_mmx_sse_builtins): Use 0 instead of
30769 ~OPTION_MASK_ISA_64BIT as mask.
30770 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
30771 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
30772 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
30773 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
30775 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30778 * lra-constraints.c (split_reg): Check requested split mode
30779 is supported by the register.
30781 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30783 * lra-constraints.c (simplify_operand_subreg): Remove early
30786 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30789 * lra-constraints.c (curr_insn_transform): Tighten condition
30790 for converting SUBREG reloads from OP_OUT to OP_INOUT.
30792 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30795 * lra-constraints.c (curr_insn_transform): Handle
30796 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30798 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
30801 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
30803 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
30805 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
30808 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
30811 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30813 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
30816 2017-02-18 Jakub Jelinek <jakub@redhat.com>
30818 * final.c (last_columnnum, override_columnnum): New variables.
30819 (final_start_function): Set last_columnnum, pass it to begin_prologue
30820 hook and pass 0 to dwarf2out_begin_prologue.
30821 (final_scan_insn): Update override_columnnum. Pass last_columnnum
30822 to source_line debug hook.
30823 (notice_source_line): Compute last_columnnum and for debug_column_info
30824 return true on column changes.
30825 * debug.h (struct gcc_debug_hooks): Add column argument to
30826 source_line and begin_prologue hooks.
30827 (debug_nothing_int_charstar_int_bool): Remove prototype.
30828 (debug_nothing_int_int_charstar,
30829 debug_nothing_int_int_charstar_int_bool): New prototypes.
30830 (dwarf2out_begin_prologue): Add column argument.
30831 * debug.c (do_nothing_debug_hooks): Adjust source_line and
30832 begin_prologue hooks.
30833 (debug_nothing_int_charstar_int_bool): Remove.
30834 (debug_nothing_int_int_charstar,
30835 debug_nothing_int_int_charstar_int_bool): New functions.
30836 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
30837 through to dwarf2out_source_line.
30838 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
30839 (dwarf2out_source_line): Add column argument, emit it if requested.
30840 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
30842 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30843 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30844 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
30845 through to dwarf2out_begin_prologue.
30846 (vmsdbgout_source_line): Add column argument, pass it through to
30847 dwarf2out_source_line.
30848 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
30849 dbxout_source_line caller.
30850 (dbxout_source_line): Add column argument.
30852 * common.opt (gno-column-info, gcolumn-info): New options.
30853 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
30854 (check_die): Also test for multiple DW_AT_decl_column attributes.
30855 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
30856 DW_AT_decl_column if requested.
30857 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
30859 (gen_variable_die): Likewise.
30860 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
30861 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
30864 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
30865 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
30866 (ix86_handle_option): Handle OPT_m3dnowa.
30867 * doc/invoke.texi (-m3dnowa): Document.
30868 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
30869 -m3dnowa instead of -m3dnow -march=athlon.
30872 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
30873 instead of gcc_assert for K, r and R code checks. Formatting fixes.
30875 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30878 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
30879 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
30880 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
30881 generator for vsx_xxpermdi_<mode>_be.
30882 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
30883 force big-endian semantics.
30884 (vsx_xxpermdi_<mode>_be): New define_expand with same
30885 implementation as previous version of vsx_xxpermdi_<mode>.
30887 2017-02-17 Jakub Jelinek <jakub@redhat.com>
30889 PR tree-optimization/79327
30890 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
30891 variable, its initialization and use.
30893 2017-02-17 Julia Koval <julia.koval@intel.com>
30895 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
30896 (OPTION_MASK_ISA_PKU_UNSET): New.
30897 (ix86_handle_option): Handle -mrdpid.
30898 * config/i386/cpuid.h (bit_RDPID): New.
30899 * config/i386/driver-i386.c (host_detect_local_cpu):
30900 Detect RDPID feature.
30901 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
30902 * config/i386/i386-c.c (ix86_target_macros_internal):
30904 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
30905 (ix86_valid_target_attribute_inner_p): Add "rdpid".
30906 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
30907 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
30908 * config/i386/i386.md (define_insn "rdpid"): New.
30909 * config/i386/i386.opt Add -mrdpid.
30910 * config/i386/immintrin.h (_rdpid_u32): New.
30912 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
30914 PR rtl-optimization/79541
30915 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
30916 instead of transforming it into USE.
30918 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
30920 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
30921 If HONOR_SNANS (SFmode) force the input to a register.
30922 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
30923 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
30924 an frsp or similar insn.
30926 2017-02-17 Martin Liska <mliska@suse.cz>
30928 PR rtl-optimization/79577
30929 * params.def (selsched-max-sched-times): Increase minimum to 1.
30931 2017-02-17 Martin Liska <mliska@suse.cz>
30933 PR rtl-optimization/79574
30934 * gcse.c (want_to_gcse_p): Prevent integer overflow.
30936 2017-02-17 Martin Liska <mliska@suse.cz>
30938 PR tree-optimization/79529
30939 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
30940 ssa_defined_default_def_p to handle cases which are implicitly
30942 * tree-ssa.c (ssa_defined_default_def_p): New function.
30943 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
30944 which are implicitly defined.
30945 * tree-ssa.h (ssa_defined_default_def_p): Declare.
30947 2017-02-17 Richard Biener <rguenther@suse.de>
30949 PR middle-end/79576
30950 * params.def (max-ssa-name-query-depth): Limit to 10.
30952 2017-02-17 Richard Biener <rguenther@suse.de>
30954 PR tree-optimization/79552
30955 * tree-ssa-structalias.c (visit_loadstore): Properly verify
30958 2017-02-17 Richard Biener <rguenther@suse.de>
30961 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
30963 2017-02-17 Marek Polacek <polacek@redhat.com>
30965 PR middle-end/79536
30966 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
30967 (fold_negate_expr): New wrapper.
30969 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
30971 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
30972 Correct terminology and de-emphasize pre-standard behavior.
30974 2017-02-16 Alan Modra <amodra@gmail.com>
30976 PR rtl-optimization/79286
30977 * ira.c (def_dominates_uses): New function.
30978 (update_equiv_regs): Don't create an equivalence for insns that
30979 may trap where the register def does not dominate the use.
30981 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
30983 PR rtl-optimization/78127
30984 * lra.c (lra): Call lra_eliminate before finish the loop after
30987 2017-02-16 Richard Biener <rguenther@suse.de>
30989 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
30991 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
30992 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
30993 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
30994 (isl_val_int_from_wi): New function.
30995 (extract_affine_gmp): Rename to ...
30996 (extract_affine_wi): ... this, take a widest_int.
30997 (extract_affine_int): Just wrap extract_affine_wi.
30998 (add_param_constraints): Use isl_val_int_from_wi.
30999 (add_loop_constraints): Likewise, and extract_affine_wi.
31001 2017-02-15 Jeff Law <law@redhat.com>
31003 PR middle-end/79521
31004 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
31005 ira_init_register_move_cost_if_necessary.
31007 2017-02-15 Martin Sebor <msebor@redhat.com>
31009 PR middle-end/32003
31010 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
31011 removed in a prior commit.
31013 2017-02-15 Bin Cheng <bin.cheng@arm.com>
31015 PR tree-optimization/79347
31016 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31017 counters during peeling.
31019 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31021 * Makefile.in (site.exp): Remove "set ISLVER".
31023 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31026 * real.c (real_from_integer): Call real_convert even for decimal.
31028 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31031 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31033 2017-02-14 Andrew Pinski <apinski@cavium.com>
31035 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31036 cores and change the partno/implementer to be correct.
31037 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31038 the 'B" as the implementer.
31039 * config/aarch64/aarch64-tune.md: Regenerate.
31041 2017-02-14 Carl Love <cel@us.ibm.com>
31043 * config/rs6000/rs6000.c: Add case statement entry to make the
31044 xvcvuxdsp built-in argument unsigned.
31045 * config/rs6000/vsx.md: Fix the source and return operand types so they
31046 match the instruction definitions from the ISA document. Fix typo
31047 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31050 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31053 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31054 member early_clobber_alts.
31055 * lra-lives.c (reg_early_clobber_p): New.
31056 (process_bb_lives): Use it.
31057 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31058 (debug_operand_data): Initialize early_clobber_alts.
31059 (setup_operand_alternative): Set up early_clobber_alts.
31060 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31061 alternatives to new_insn_reg.
31062 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31064 (lra_update_insn_regno_info): Pass the new arg.
31066 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31068 PR middle-end/79505
31069 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31070 (new_oacc_loop_raw): Don't clear already cleared fields.
31073 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31074 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31075 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31077 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31080 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31082 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31085 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31086 the extra instruction to the right place to store 128-bit constant
31089 2017-02-14 Martin Sebor <msebor@redhat.com>
31091 PR middle-end/79448
31092 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31093 warning for strings of unknown length.
31095 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31097 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31099 2017-02-14 Jeff Law <law@redhat.com>
31102 * ira-costs.c (scan_one_insn): Initialize register move costs
31103 for pseudos seen in USE/CLOBBER insns.
31105 PR tree-optimization/79095
31106 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31107 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31108 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31109 if the operands are known to be not equal, then the resulting range
31111 (intersect_ranges): If the new range is ~[0,0] and the old range is
31112 wide, then prefer ~[0,0].
31113 * tree-vrp.c (overflow_comparison_p_1): New function.
31114 (overflow_comparison_p): New function.
31115 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31116 if NAME is used in an overflow test.
31117 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31118 overflow check that can be expressed as an equality test, then adjust
31119 ops to be that equality test.
31121 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31123 * config/s390/s390-builtin-types.def: Remove flags argument.
31124 * config/s390/s390.c (s390_init_builtins): Likewise.
31126 2017-02-14 Martin Liska <mliska@suse.cz>
31128 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31129 vector. Fix trailing white spaces.
31131 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31133 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31136 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31138 PR rtl-optimization/68664
31139 * config/arm/arm.c (arm_sched_can_speculate_insn):
31140 New function. Declare prototype.
31141 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31143 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31145 PR rtl-optimization/68664
31146 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31148 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31150 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31152 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31153 max skip bytes for function, loop and jump.
31155 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31157 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31158 ABS_EXPR for gimple dump.
31160 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31163 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31165 PR tree-optimization/79408
31166 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31167 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31168 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31169 also if rhs1 is INTEGER_CST.
31171 2017-02-14 Richard Biener <rguenther@suse.de>
31173 PR middle-end/79432
31174 * tree-into-ssa.c (insert_phi_nodes): When the function can
31175 have abnormal edges rewrite SSA names with broken use-def
31176 dominance out of SSA and register them for PHI insertion.
31178 2017-02-13 Martin Sebor <msebor@redhat.com>
31180 PR middle-end/79496
31181 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31182 clearing info.nowrite flag when snprintf size argument is a range.
31184 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31186 * cprop.c (cprop_jump): Add missing space in string literal.
31187 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31188 (get_constraint_for_component_ref): Likewise.
31189 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31190 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31191 * lra-constraints.c (process_alt_operands): Likewise.
31192 * ipa-inline.c (inline_small_functions): Likewise.
31193 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31194 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31195 * trans-mem.c (diagnose_tm_1_op): Likewise.
31196 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31197 (grid_parallel_clauses_gridifiable): Likewise.
31199 * config/nvptx/mkoffload.c (process): Add space in between
31202 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31203 "MOD4_SSE_REGS" and "ALL_REGS".
31205 * spellcheck.c (test_data): Add , in between "foo" and "food".
31207 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31210 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31211 boundary crossing check and subsequent code generation agree.
31213 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31215 * config/aarch64/aarch64.c (has_memory_op): Delete.
31216 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31219 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31221 PR rtl-optimization/79388
31222 PR rtl-optimization/79450
31223 * combine.c (distribute_notes): When removing TEM_INSN for which
31224 corresponding dest has last value recorded, invalidate that last
31227 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31229 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31230 of explicit '@'. Add missing assembly comment marker on branch costs
31233 2017-02-13 Nathan Sidwell <nathan@acm.org>
31235 * gengtype-lex.l (<in_struct>): Add '/'.
31237 2017-02-13 Martin Liska <mliska@suse.cz>
31240 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31242 2017-02-13 Richard Biener <rguenther@suse.de>
31244 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31246 * configure: Re-generate.
31247 * config.in: Likewise.
31248 * graphite-dependences.c: Simplify as if
31249 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31250 * graphite-isl-ast-to-gimple.c: Likewise.
31251 * graphite-optimize-isl.c: Likewise.
31252 * graphite-poly.c: Likewise.
31253 * graphite-sese-to-poly.c: Likewise.
31254 * graphite.h: Likewise.
31255 * toplev.c: Include isl/version.h and use isl_version () for
31256 printing the ISL version.
31257 * doc/install.texi: Update ISL requirement.
31259 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31261 * doc/standards.texi (Standards): Update reference to
31264 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31266 * doc/extend.texi (Named Address Spaces): sourceware.org now
31268 * doc/install.texi (Binaries): Ditto.
31271 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31273 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31274 standard terminology "stringize"/"stringizing" throughout.
31275 * doc/cppinternals.texi: Likewise.
31277 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31279 * doc/extend.texi: Fix some spelling mistakes and typos.
31280 * doc/invoke.texi: Likewise.
31282 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31285 * params.def (inline-min-speedup) Change from 10 to 8.
31287 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31289 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31292 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31295 * ipa-inline-analysis.c (get_minimal_bb): New function.
31296 (record_modified): Use it.
31297 (remap_edge_change_prob): Handle also ancestor functions.
31299 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31301 * doc/contrib.texi (Contributors): Remove broken link into
31302 the Mauve CVS repository.
31304 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31306 PR middle-end/79454
31307 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31308 result computation whenever lhs doesn't have vector mode, not
31309 just when it has BLKmode.
31311 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31313 * doc/makefile.texi (profiledbootstrap): Refer to the
31314 installation instructions only in textual form.
31316 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31319 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31321 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31323 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31324 (Specific): Update mingw-w64 reference.
31326 (Specific): Remove broken link to Renesas RX processor.
31328 2017-02-10 Richard Biener <rguenther@suse.de>
31330 * toplev.c (process_options): Do not mention obsolete graphite
31331 options when printing sorry message about missing graphite support.
31332 Mention -floop-nest-optimize.
31334 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31336 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31337 (vtst_p16): Likewise.
31338 (vtstq_p8): Likewise.
31339 (vtstq_p16): Likewise.
31341 (vtstq_p64): Likewise.
31342 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31343 (vset_lane_p64): New.
31344 (vsetq_lane_p64): New.
31346 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31348 PR tree-optimization/79411
31349 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31350 stmt operands are SSA_NAMEs used in abnormal phis.
31351 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31354 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31357 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
31360 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31362 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
31364 2017-02-09 Jakub Jelinek <jakub@redhat.com>
31366 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
31370 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
31371 not arbitrary TREE_CONSTANT.
31374 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
31375 "omp declare target link" attribute unless is_global_var.
31376 * omp-offload.c (find_link_var_op): Likewise.
31378 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
31379 Chung-Lin Tang <cltang@codesourcery.com>
31381 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31383 (gimplify_adjust_omp_clauses): Don't delete TILE.
31384 (gimplify_omp_for): Deal with TILE.
31385 * internal-fn.c (expand_GOACC_TILE): New function.
31386 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
31388 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
31389 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
31391 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
31392 avoid DIV for outermost collapse var.
31393 (expand_oacc_for): Insert tile element loop as needed. Adjust.
31394 Remove out of date comments, fix whitespace.
31395 * omp-general.c (omp_extract_for_data): Deal with tiling.
31396 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
31397 adjust OLF_DIM_BASE value.
31398 (struct omp_for_data): Add tiling field.
31399 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
31400 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
31401 for auto loops. Remove default auto determining, moved to
31402 oacc_loop_fixed_partitions.
31403 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
31404 stmts, add e_mask field.
31405 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
31406 (oacc_thread_numbers): Use oacc_dim_call.
31407 (oacc_xform_tile): New.
31408 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
31409 (finish_oacc_loop): Adjust for ifns vector.
31410 (oacc_loop_discover_walk): Append loop abstraction sites to list,
31411 add case for GOACC_TILE fns.
31412 (oacc_loop_xform_loop): Delete.
31413 (oacc_loop_process): Iterate over call list directly, and add
31414 handling for GOACC_TILE fns.
31415 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
31417 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
31418 vector partitioning to outer loops. Assign 2 partitions to loops
31419 when available. Add TILE handling.
31420 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
31421 (execite_oacc_device_lower): Process GOACC_TILE fns,
31422 ignore unknown specs.
31423 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
31424 * tree.c (omp_clause_num_ops): Adjust TILE ops.
31425 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
31427 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
31429 * configure.ac (ACX_BUGURL): Update.
31430 * configure: Regenerate.
31432 2017-02-09 Richard Biener <rguenther@suse.de>
31434 PR tree-optimization/69823
31435 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
31436 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
31438 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31440 * config/arc/arc-c.def: Add __NPS400__ definition.
31441 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
31442 (TARGET_NPS400): Define.
31444 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31446 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
31448 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
31449 pointer, arch_info.
31450 (arc_cpu_types): Fill the arch_info field with a pointer into the
31451 arc_arch_types table.
31452 (arc_selected_cpu): Declare.
31453 * config/arc/arc.c (arc_selected_cpu): Make global.
31454 (arc_selected_arch): Delete.
31455 (arc_base_cpu): Delete.
31456 (arc_override_options): Remove references to deleted variables,
31457 update access to arch information.
31458 (ARC_OPT): Update access to arch information.
31459 (ARC_OPTX): Likewise.
31460 * config/arc/arc.h (arc_base_cpu): Remove declaration.
31461 (TARGET_ARC600): Update access to arch information.
31462 (TARGET_ARC601): Likewise.
31463 (TARGET_ARC700): Likewise.
31464 (TARGET_EM): Likewise.
31465 (TARGET_HS): Likewise.
31466 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
31469 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
31472 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
31473 condition/operands for integer GE/LE/GEU/LEU operations.
31475 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
31477 PR translation/79397
31478 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
31481 2017-02-08 Martin Jambor <mjambor@suse.cz>
31484 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
31485 whether allocation happened.
31486 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
31487 nothing was allocated.
31489 2017-02-08 Jakub Jelinek <jakub@redhat.com>
31491 PR tree-optimization/79408
31492 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
31493 constant, but SSA_NAME with a known integer range, use the minimum
31494 of that range instead of op1 to determine if modulo can be replaced
31495 with its first operand.
31497 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31499 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
31501 2017-02-08 Richard Biener <rguenther@suse.de>
31503 PR tree-optimization/71824
31504 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31505 Check all loops contained in the merged region.
31507 2017-02-07 Andrew Pinski <apinski@cavium.com>
31509 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
31511 2017-02-07 Andrew Pinski <apinski@cavium.com>
31513 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
31514 (thunderxt88): Likewise.
31515 (thunderxt81): Disable LSE and change v8.1 to v8.
31516 (thunderxt83): Likewise.
31518 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31519 Richard Biener <rguenther@suse.de>
31521 PR middle-end/79399
31522 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
31523 type from int to size_t.
31524 * ira-costs.c (struct_costs_size): Change type from int to size_t.
31526 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31528 PR rtl-optimization/79386
31529 * cprop.c (bypass_conditional_jumps): Initialize
31530 bypass_last_basic_block already before splitting bbs after
31531 unconditional traps...
31532 (bypass_conditional_jumps): ... rather than here.
31535 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
31536 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
31537 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
31538 fix -masm=intel patterns.
31540 2017-02-07 Richard Biener <rguenther@suse.de>
31542 PR tree-optimization/79256
31543 PR middle-end/79278
31544 * builtins.c (get_object_alignment_2): Use min_align_of_type
31545 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
31546 and ADJUST_FIELD_ALIGN.
31548 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
31550 * doc/tm.texi: Regenerate.
31551 * stor-layout.c (layout_decl): Adjust.
31552 (update_alignment_for_field): Likewise.
31553 (place_field): Likewise.
31554 (min_align_of_type): Likewise.
31555 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
31556 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
31557 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
31558 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
31559 * config/frv/frv.c (frv_adjust_field_align): Likewise.
31560 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
31561 * config/i386/i386.c (x86_field_alignment): Likewise.
31562 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
31563 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
31564 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
31565 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
31566 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
31567 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
31571 2017-01-30 Richard Biener <rguenther@suse.de>
31573 PR tree-optimization/79256
31574 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31575 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31578 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
31580 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
31581 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
31582 builtins to SImode and emit a zero-extend, if necessary.
31584 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31586 * docs/invoke.texi (RISC-V Options): Alphabetize.
31588 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31590 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
31593 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31595 * config/riscv/riscv.c: New file.
31596 * common/config/riscv/riscv-common.c: Likewise.
31597 * config.gcc: Likewise.
31598 * config/riscv/constraints.md: Likewise.
31599 * config/riscv/elf.h: Likewise.
31600 * config/riscv/generic.md: Likewise.
31601 * config/riscv/linux.h: Likewise.
31602 * config/riscv/multilib-generator: Likewise.
31603 * config/riscv/peephole.md: Likewise.
31604 * config/riscv/pic.md: Likewise.
31605 * config/riscv/predicates.md: Likewise.
31606 * config/riscv/riscv-builtins.c: Likewise.
31607 * config/riscv/riscv-c.c: Likewise.
31608 * config/riscv/riscv-ftypes.def: Likewise.
31609 * config/riscv/riscv-modes.def: Likewise.
31610 * config/riscv/riscv-opts.h: Likewise.
31611 * config/riscv/riscv-protos.h: Likewise.
31612 * config/riscv/riscv.h: Likewise.
31613 * config/riscv/riscv.md: Likewise.
31614 * config/riscv/riscv.opt: Likewise.
31615 * config/riscv/sync.md: Likewise.
31616 * config/riscv/t-elf-multilib: Likewise.
31617 * config/riscv/t-linux: Likewise.
31618 * config/riscv/t-linux-multilib: Likewise.
31619 * config/riscv/t-riscv: Likewise.
31620 * configure.ac: Likewise.
31621 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
31622 Waterman as RISC-V maintainers.
31623 * doc/install.texi: Add RISC-V entries.
31624 * doc/invoke.texi: Add RISC-V options section.
31625 * doc/md.texi: Add RISC-V constraints section.
31626 * configure: Regenerated.
31628 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
31631 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
31632 false values to be constant vectors with all 0 or all 1 bits set.
31633 (vcondu<mode><mode>): Likewise.
31634 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
31636 (fpmask_comparison_operator): Update comment.
31637 (vecint_comparison_operator): New predicate.
31638 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
31639 vector conditionals when the true and false values are constant
31640 vectors with all 0 bits or all 1 bits set.
31642 2017-02-06 Martin Sebor <msebor@redhat.com>
31644 PR tree-optimization/79376
31645 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
31647 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
31649 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
31650 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
31651 to simplify split condition.
31653 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31655 * omp-expand.c (oxpand_omp_atomic_fetch_op,
31656 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
31659 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
31661 PR rtl-optimization/68664
31662 * target.def (can_speculate_insn): New hook.
31663 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
31664 * doc/tm.texi: Regenerate.
31665 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
31666 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
31667 (rs6000_sched_can_speculate_insn): New function.
31669 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31671 PR tree-optimization/79284
31672 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
31673 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
31674 vectorizable_mask_load_store, vectorizable_operation,
31675 vect_is_simple_cond, get_same_sized_vectype): Use it instead
31676 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
31677 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
31678 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
31679 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
31680 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
31681 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
31682 is_gimple_assign (stmt). Replace another such test with
31683 is_gimple_assign (stmt).
31685 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31688 * config/avr/avr.c (rtl-iter.h): Include it.
31689 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
31690 (avr_legitimate_combined_insn): ...and implementation.
31692 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
31694 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
31695 * config/s390/s390.c (s390_const_operand_ok)
31696 (s390_canonicalize_comparison, s390_extract_part)
31697 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
31698 (s390_contiguous_bitmask_p, s390_rtx_costs)
31699 (legitimize_pic_address): Likewise.
31700 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
31701 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
31702 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
31703 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
31704 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
31706 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31708 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
31709 REGNO($0) == REGNO($1).
31711 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31713 * config/s390/linux.h(SIZE_TYPE): Add comment.
31715 2017-02-06 Julian Brown <julian@codesourcery.com>
31716 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31717 Virendra Pathak <virendra.pathak@broadcom.com>
31719 * config/aarch64/aarch64-cores.def: Change the scheduler
31721 * config/aarch64/aarch64.md: Include thunderx2t99.md.
31722 * config/aarch64/thunderx2t99.md: New file.
31724 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31726 * doc/standards.texi (Go Language): Update link to language
31729 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31731 * tree-eh.c (lower_resx): Sanitize profile.
31732 (cleanup_empty_eh_move_lp): Likewise.
31734 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31737 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
31739 * cfgloopmanip.h (loop_version): Update prototype.
31740 * modulo-sched.c (sms_schedule): Update call of loop_version.
31741 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
31742 * tree-parloops.c (gen_parallel_loop): Likewise.
31743 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
31744 * tree-ssa-loop-split.c (split_loop): Likewise.
31745 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
31746 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
31748 2017-02-05 Martin Liska <mliska@suse.cz>
31751 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
31753 (print_operand_address): Initialize a struct to zero.
31755 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31757 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
31758 garbage collector only in textual form.
31760 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31762 * doc/extend.texi (x86 specific memory model extensions for
31763 transactional memory): Simplify a phrase.
31765 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
31768 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
31769 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
31770 (atomic_storedi_1): Likewise.
31772 2017-02-04 Jakub Jelinek <jakub@redhat.com>
31774 PR tree-optimization/79338
31775 * tree-parloops.c (gather_scalar_reductions): Don't call
31776 vect_analyze_loop_form for loop->inner before destroying loop's
31779 2017-02-03 Martin Sebor <msebor@redhat.com>
31781 PR tree-optimization/79327
31782 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
31783 when precision has resulted in leading zeros.
31784 (format_integer): Adjust the likely counter to assume an unknown
31785 argument that may be zero is non-zero.
31787 2017-02-03 Jason Merrill <jason@redhat.com>
31790 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
31791 avoid copying non-taken branch.
31793 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31795 PR tree-optimization/79340
31796 * tree-vect-loop.c (vectorizable_reduction): Release
31797 vec_defs elements after safe_splicing them into other vectors.
31800 PR tree-optimization/79327
31801 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
31802 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
31804 (format_integer): Use wide_int_to_tree instead of build_int_cst
31805 + to_?hwi. If argmin is NULL, just set argmin and argmax to
31806 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
31807 of shortest and longest sequence.
31809 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
31811 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
31812 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
31814 2017-02-03 Walter Lee <walt@tilera.com>
31817 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
31818 after initial stackframe link reg save.
31819 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
31821 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31824 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
31825 wu for stxssp alternative.
31827 2017-02-03 Martin Sebor <msebor@redhat.com>
31829 PR tree-optimization/79352
31830 * gimple-fold.c (get_range_strlen): Add argument.
31831 (get_range_strlen): Change return type to bool.
31832 (get_maxval_strlen): Pass in a dummy argument.
31833 * gimple-fold.h (get_range_strlen): Change return type to bool.
31834 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
31835 * tree.h (array_at_struct_end_p): Add argument.
31836 * tree.c (array_at_struct_end_p): Handle it.
31838 2017-02-03 Martin Liska <mliska@suse.cz>
31841 * multiple_target.c (create_dispatcher_calls): Redirect edge
31842 from a caller of a dispatcher.
31843 (expand_target_clones): Make the clones local.
31844 (ipa_target_clone): Do both target clones and resolvers.
31845 (ipa_dispatcher_calls): Remove the pass.
31846 (pass_dispatcher_calls::gate): Likewise.
31847 (make_pass_dispatcher_calls): Likewise.
31848 * passes.def (pass_target_clone): Put as very first IPA early
31851 2017-02-03 Martin Liska <mliska@suse.cz>
31853 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
31854 in case of a function with ifunc attribute.
31856 2017-02-03 Martin Liska <mliska@suse.cz>
31858 * cgraph.c (cgraph_node::dump): Dump function version info.
31859 * symtab.c (symtab_node::dump_base): Add missing new line.
31861 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31863 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
31864 (ifcombine_ifandif): Use it.
31866 2017-02-03 Martin Liska <mliska@suse.cz>
31868 * doc/invoke.texi: Document default value for
31869 use-after-scope-direct-emission-threshold.
31871 2017-02-03 Martin Liska <mliska@suse.cz>
31873 PR tree-optimization/79339
31874 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
31875 (format_floating): Likewise.
31877 2017-02-03 Martin Liska <mliska@suse.cz>
31880 * ipa-prop.c (ipa_node_params_t::insert): Remove current
31882 (ipa_node_params_t::remove): Likewise.
31883 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
31884 initialization from removed ipa_node_params_t::insert.
31885 (ipa_node_params::~ipa_node_params): Move from removed
31886 ipa_node_params_t::release.
31887 * symbol-summary.h (symbol_summary::m_released): New member.
31888 Do not release a summary twice. Do not allow to call finalizer
31889 for types of a summary that live in GGC memory.
31891 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31893 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
31896 2017-02-02 Martin Sebor <msebor@redhat.com>
31898 PR middle-end/79275
31899 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
31900 (format_string): Tighten up the range of output for non-constant
31901 strings and correct the expected range for wide non-constant strings.
31903 2017-02-02 Martin Sebor <msebor@redhat.com>
31905 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
31907 PR middle-end/32003
31908 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
31910 (-fdump-tree-@var): Add to index and document how to come up
31911 with pass-specific option and dump file names.
31912 (-fdump-passes): Clarify where to look for output.
31914 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31916 PR middle-end/77445
31917 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
31918 statistics of the analyzed path; allow threading for speed when
31919 any of BBs along the path are optimized for speed.
31921 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
31923 PR middle-end/78468
31924 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
31925 settings of the virtual registers.
31928 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
31930 * explow.c (get_dynamic_stack_size): Take known alignment of stack
31931 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
31934 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31936 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
31937 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
31939 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31941 * config/s390/s390.md: Add missing comments with the expanded
31943 * config/s390/vector.md: Likewise.
31944 * config/s390/vx-builtins.md: Likewise.
31946 2017-02-02 Jakub Jelinek <jakub@redhat.com>
31949 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
31950 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
31951 conditions on a single line.
31953 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31955 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31956 __S390_VX__ to __VX__.
31958 2017-02-01 Andrew Pinski <apinski@cavium.com>
31960 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
31961 stmt_info to record_stmt_cost.
31962 (vect_get_known_peeling_cost): Pass stmt_info if known to
31964 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
31965 cpu_vector_cost field into
31966 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
31967 field into vec_int_stmt_cost and vec_fp_stmt_cost.
31968 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
31969 splitting of scalar_stmt_cost and vec_stmt_cost.
31970 (thunderx_vector_cost): Likewise.
31971 (cortexa57_vector_cost): LIkewise.
31972 (exynosm1_vector_cost): Likewise.
31973 (xgene1_vector_cost): Likewise.
31974 (thunderx2t99_vector_cost): Improve after the splitting of the two
31976 (aarch64_builtin_vectorization_cost): Update for the splitting of
31977 scalar_stmt_cost and vec_stmt_cost.
31979 2017-02-01 Torvald Riegel <triegel@redhat.com>
31980 Richard Henderson <rth@redhat.com>
31982 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
31983 conditional on existance of a fast atomic load.
31984 * optabs-query.c (can_atomic_load_p): New function.
31985 * optabs-query.h (can_atomic_load_p): Declare it.
31986 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
31987 no fast atomic load is available for the particular size of access.
31988 (expand_atomic_compare_and_swap): Likewise.
31989 (expand_atomic_load): Likewise.
31990 (expand_atomic_store): Likewise.
31991 (expand_atomic_fetch_op): Likewise.
31992 * testsuite/lib/target-supports.exp
31993 (check_effective_target_sync_int_128): Remove x86 because it provides
31994 no fast atomic load.
31995 (check_effective_target_sync_int_128_runtime): Likewise.
31997 2017-02-01 Richard Biener <rguenther@suse.de>
31999 * graphite.c: Include tree-vectorizer.h for find_loop_location.
32000 (graphite_transform_loops): Provide opt-info for optimized nests.
32001 * tree-parloop.c (parallelize_loops): Provide opt-info for
32002 parallelized loops.
32004 2017-02-01 Richard Biener <rguenther@suse.de>
32006 PR middle-end/79315
32007 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
32008 was not set before.
32010 2017-02-01 Richard Biener <rguenther@suse.de>
32012 PR tree-optimization/71824
32013 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32014 Verify the loops are valid in the merged SESE region.
32015 (scop_detection::can_represent_loop_1): Check analyzing the
32016 evolution of the number of iterations in the region succeeds.
32018 2017-01-31 Ian Lance Taylor <iant@golang.org>
32020 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32021 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32023 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32025 PR preprocessor/79210
32026 * input.c (get_substring_ranges_for_loc): Replace line_width
32027 assertion with error-handling.
32029 2017-01-31 Richard Biener <rguenther@suse.de>
32031 PR tree-optimization/77318
32032 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32033 (create_pw_aff_from_tree): Take loop parameter.
32034 (add_condition_to_pbb): Pass loop of the condition to
32035 create_pw_aff_from_tree.
32037 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32039 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32040 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32042 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32046 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32048 (convert_int_to_float128): Likewise.
32049 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32050 (convert_int_to_float128): Likewise.
32051 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32052 (UNSPEC_IEEE128_CONVERT): Likewise.
32053 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32054 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32055 Use local variables for IBM extended format.
32056 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32057 (fix_trunc<mode>si2_fprs): Likewise.
32058 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32059 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32060 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32061 to know that we can now have integers of all sizes in vector
32063 (fix<uns>_<mode>di2_hw): Likewise.
32064 (float<uns>_<mode>si2_hw): Likewise.
32065 (fix_<mode>si2_hw): Likewise.
32066 (fixuns_<mode>si2_hw): Likewise.
32067 (float<uns>_<mode>di2_hw): Likewise.
32068 (float_<mode>di2_hw): Likewise.
32069 (float_<mode>si2_hw): Likewise.
32070 (floatuns_<mode>di2_hw): Likewise.
32071 (floatuns_<mode>si2_hw): Likewise.
32072 (xscvqp<su>wz_<mode>): Delete, no longer used.
32073 (xscvqp<su>dz_<mode>): Likewise.
32074 (xscv<su>dqp_<mode>): Likewise.
32075 (ieee128_mfvsrd_64bit): Likewise.
32076 (ieee128_mfvsrd_32bit): Likewise.
32077 (ieee128_mfvsrwz): Likewise.
32078 (ieee128_mtvsrw): Likewise.
32079 (ieee128_mtvsrd_64bit): Likewise.
32080 (ieee128_mtvsrd_32bit): Likewise.
32082 2017-01-31 Martin Liska <mliska@suse.cz>
32085 * ipa-prop.c (ipa_free_all_node_params): Call release method
32086 instead of ~sumbol_summary to not to trigger double times
32089 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32091 PR tree-optimization/71691
32092 * bitmap.h (class auto_bitmap): New.
32093 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32094 is_maybe_undefined instead of ssa_undefined_value_p.
32096 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32098 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32099 __S390_ARCH_LEVEL__ to __ARCH__.
32101 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32103 PR tree-optimization/79267
32104 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32105 if should_remove_lhs_p is true.
32107 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32110 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32111 (add_alignment_attribute): New.
32112 (base_type_die): Add alignment attribute.
32113 (subrange_type_die): Likewise.
32114 (modified_type_die): Likewise.
32115 (gen_array_type_die): Likewise.
32116 (gen_descr_array_type_die: Likewise.
32117 (gen_enumeration_type_die): Likewise.
32118 (gen_subprogram_die): Likewise.
32119 (gen_variable_die): Likewise.
32120 (gen_field_die): Likewise.
32121 (gen_ptr_to_mbr_type_die): Likewise.
32122 (gen_struct_or_union_type_die): Likewise.
32123 (gen_subroutine_type_die): Likewise.
32124 (gen_typedef_die): Likewise.
32125 (base_type_cmp): Compare alignment attribute.
32127 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32130 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32131 (setb_unsigned) New pattern for setb with CCUNS.
32132 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32133 subfc./subfe sequence to avoid overflow problems. Generate a
32134 shorter sequence with cmpld/setb for power9.
32135 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32136 for generating subfc. instruction.
32137 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32138 now uses this instruction.
32140 2017-01-30 Ian Lance Taylor <iant@google.com>
32143 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32144 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32146 2017-01-30 Martin Sebor <msebor@redhat.com>
32148 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32149 Move constant to the right of a relational operator.
32150 (get_mpfr_format_length, format_character, format_string): Ditto.
32151 (should_warn_p, maybe_warn): Same.
32153 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32155 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32158 * asan.c (get_translation_unit_decl): Remove function.
32159 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32161 2017-01-30 Martin Liska <mliska@suse.cz>
32163 PR gcov-profile/79259
32164 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32165 -fprofile-generate.
32167 2017-01-30 Martin Liska <mliska@suse.cz>
32170 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32171 Initialize variables with NULL value.
32173 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32176 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32178 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32180 2017-01-30 Richard Biener <rguenther@suse.de>
32182 PR tree-optimization/79276
32183 * tree-vrp.c (process_assert_insertions): Properly adjust common
32184 when removing a duplicate.
32186 2017-01-30 Richard Biener <rguenther@suse.de>
32188 PR tree-optimization/79256
32189 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32190 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32192 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32194 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32197 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32198 ("*r<noxa>sbg_<mode>_sll_bitmask")
32199 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32200 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32201 Use contiguous_bitmask_nowrap_operand.
32203 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32206 * config/rs6000/altivec.h (vec_xl): Revise #define.
32207 (vec_xst): Likewise.
32209 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32211 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32213 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32215 PR rtl-optimization/79194
32216 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32217 traps before call to bypass_conditional_jumps.
32219 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32221 PR tree-optimization/71374
32222 * lra-constraints.c (check_conflict_input_operands): New.
32223 (match_reload): Use it.
32225 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32228 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32229 account to calculate conflict_set.
32231 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32233 PR rtl-optimization/78559
32234 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32235 other_insn in combine.
32237 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32239 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32240 uint16_type_node for BT_UINT16.
32242 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32244 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32245 "RTL Tests" to menu.
32246 (GIMPLE Tests): New node.
32247 (RTL Tests): New node.
32249 2017-01-27 Richard Biener <rguenther@suse.de>
32251 PR tree-optimization/79245
32252 * tree-loop-distribution.c (distribute_loop): Apply cost
32253 modeling also to detected patterns.
32255 2017-01-27 Richard Biener <rguenther@suse.de>
32257 PR tree-optimization/71433
32258 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32259 (compare_assert_loc): New function.
32260 (process_assert_insertions): Sort and optimize assert locations
32261 to remove duplicates and push down identical assertions on
32262 edges to their destination block.
32264 2017-01-27 Richard Biener <rguenther@suse.de>
32266 PR tree-optimization/79244
32267 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32268 out SSA names even if abnormal.
32270 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32272 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32273 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32274 instead of MPFR_RNDN.
32276 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32279 * arm.c (arm_option_override): Don't call build_target_option_node
32280 until after doing all option overrides.
32281 (arm_valid_target_attribute_tree): Likewise.
32283 2017-01-27 Martin Liska <mliska@suse.cz>
32285 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32286 for {cd}tors and C++ {cd}tors.
32288 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32290 * config/s390/s390.md ("*setmem_long_and")
32291 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32293 2017-01-26 Martin Sebor <msebor@redhat.com>
32295 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32298 2017-01-26 Martin Sebor <msebor@redhat.com>
32300 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32301 HAVE_DFmode before using XFmode or DFmode.
32302 (parse_directive): Avoid using the z length modifier to avoid
32303 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32305 PR middle-end/78703
32306 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32307 to accept adjustment as an array.
32308 (get_int_range): New function.
32309 (struct directive): Make width and prec arrays.
32310 (directive::set_width, directive::set_precision): Call get_int_range.
32311 (format_integer, format_floating): Handle width and precision ranges.
32312 (format_string, parse_directive): Same.
32314 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32317 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32318 comdat_type_p set, just clone them, but keep the children in the
32322 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32323 which have direct callers with -fvar-tracking-assignments enabled
32325 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32326 inside of type units.
32328 2017-01-26 Martin Sebor <msebor@redhat.com>
32330 PR middle-end/78703
32331 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32333 (struct format_result): Replace number_chars, number_chars_min,
32334 and number_chars_max with a single member of struct result_range.
32336 (format_result::operator+=): Adjust.
32337 (struct fmtresult): Remove bounded. Handle likely and unlikely
32339 (fmtresult::adjust_for_width_or_precision): New function.
32340 (fmtresult:type_max_digits): New function.
32341 (bytes_remaining): Handle likely and unlikely counters.
32342 (min_bytes_remaining): Remove.
32343 (format_percent): Simplify.
32344 (format_integer, format_floating): Set likely and unlikely counters.
32345 (get_string_length, format_character, format_string): Same.
32346 (format_plain, should_warn_p): New function.
32347 (maybe_warn): Call should_warn_p. Update diagnostic messages
32348 and handle those for all directives, including plain strings.
32349 (format_directive): Handle likely and unlikely counters.
32350 Remove unnecessary quoting from diagnostics. Add an informational
32352 (add_bytes): Remove.
32353 (pass_sprintf_length::compute_format_length): Simplify.
32354 (try_substitute_return_value): Handle likely and unlikely counters.
32356 2017-01-26 Carl Love <cel@us.ibm.com>
32358 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
32359 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
32361 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
32364 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
32365 endianess for subregs into account.
32366 * lra-constraints.c (lra_constraints): Do risky transformations
32367 always on the first iteration.
32368 * lra-lives.c (check_pseudos_live_through_calls): Add arg
32369 last_call_used_reg_set.
32370 (process_bb_lives): Define and use last_call_used_reg_set.
32371 * lra.c (lra): Always continue after lra_constraints on the first
32374 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
32376 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
32378 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32380 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32382 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32383 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32384 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32385 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
32386 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
32387 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
32388 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
32389 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
32390 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
32392 2017-01-26 Marek Polacek <polacek@redhat.com>
32395 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
32396 for the third operand.
32398 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32400 PR middle-end/79236
32401 * omp-low.c (struct omp_context): Add simt_stmt field.
32402 (scan_omp_for): Return omp_context *.
32403 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
32404 context to the _simt_ SIMD stmt.
32405 (lower_omp_for): For combined SIMD with sibling _simt_
32406 SIMD, make sure to use the same decls in _looptemp_
32407 clauses as in the sibling.
32409 2017-01-26 David Sherwood <david.sherwood@arm.com>
32411 PR middle-end/79212
32412 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
32415 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32418 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
32419 emit fld b; fld a; if possible.
32421 * brig-builtins.def: Update copyright years.
32422 * config/arm/arm_acle_builtins.def: Update copyright years.
32424 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
32427 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
32428 constraint instead of o for the stxsd instruction.
32430 2017-01-25 Carl Love <cel@us.ibm.com>
32432 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
32433 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
32435 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
32437 * doc/invoke.texi (C++ Dialect Options): Fix typo.
32439 2017-01-25 Richard Biener <rguenther@suse.de>
32441 PR tree-optimization/69264
32442 * target.def (vector_alignment_reachable): Improve documentation.
32443 * doc/tm.texi: Regenerate.
32444 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
32446 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
32447 earlier changes with respect to TYPE_USER_ALIGN.
32448 (vector_alignment_reachable_p): Likewise. Improve dumping.
32450 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32453 * config/arm/arm.md (xordi3): Force constant operand into a register
32456 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32458 * doc/invoke.texi (-fstore-merging): Correct default optimization
32459 levels at which it is enabled.
32460 (-O): Move -fstore-merging from list to...
32463 2017-01-25 Richard Biener <rguenther@suse.de>
32466 * omp-expand.c: Include debug.h.
32467 (expand_omp_taskreg): Make sure to generate early debug before
32468 outlining anything from a function.
32469 (expand_omp_target): Likewise.
32470 (grid_expand_target_grid_body): Likewise.
32472 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
32475 * asan.c (get_translation_unit_decl): New function.
32476 (asan_add_global): Extract modules file name from globals
32477 TRANSLATION_UNIT_DECL name.
32479 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
32482 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
32483 for long calls with APCS frame and VFP.
32485 2017-01-24 David Malcolm <dmalcolm@redhat.com>
32487 * cfg.c (original_copy_tables_initialized_p): New function.
32488 * cfg.h (original_copy_tables_initialized_p): New decl.
32489 * cfgrtl.c (relink_block_chain): Guard the call to
32490 free_original_copy_tables with a call to
32491 original_copy_tables_initialized_p.
32492 * cgraph.h (symtab_node::native_rtl_p): New decl.
32493 * cgraphunit.c (symtab_node::native_rtl_p): New function.
32494 (symtab_node::needed_p): Don't assert for early assembly output
32495 for __RTL functions.
32496 (cgraph_node::finalize_function): Set "force_output" for __RTL
32498 (cgraph_node::analyze): Bail out early for __RTL functions.
32499 (analyze_functions): Update assertion to support __RTL functions.
32500 (cgraph_node::expand): Bail out early for __RTL functions.
32501 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
32503 * function.h (struct function): Update comment for field
32505 * gimple-expr.c: Include "tree-pass.h".
32506 (gimple_has_body_p): Return false for __RTL functions.
32507 * Makefile.in (OBJS): Add run-rtl-passes.o.
32508 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
32510 (gcc::pass_manager::get_clean_slate): New accessor.
32511 * passes.c: Include "insn-addr.h".
32512 (should_skip_pass_p): Add logging. Update logic for running
32513 "expand" to be compatible with both __GIMPLE and __RTL. Guard
32514 property-provider override so it is only done for gimple passes.
32516 (skip_pass): New function.
32517 (execute_one_pass): Call skip_pass when skipping passes.
32518 * read-md.c (md_reader::read_char): Support filtering
32519 the input to a subset of line numbers.
32520 (md_reader::md_reader): Initialize fields
32521 m_first_line and m_last_line.
32522 (md_reader::read_file_fragment): New function.
32523 * read-md.h (md_reader::read_file_fragment): New decl.
32524 (md_reader::m_first_line): New field.
32525 (md_reader::m_last_line): New field.
32526 * read-rtl-function.c (function_reader::create_function): Only
32527 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
32528 curr_properties. Set DECL_INITIAL to a dummy block.
32529 (read_rtl_function_body_from_file_range): New function.
32530 * read-rtl-function.h (read_rtl_function_body_from_file_range):
32532 * run-rtl-passes.c: New file.
32533 * run-rtl-passes.h: New file.
32535 2017-01-24 Jeff Law <law@redhat.com>
32537 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
32540 2017-01-24 Bin Cheng <bin.cheng@arm.com>
32542 PR tree-optimization/79159
32543 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
32544 (record_nonwrapping_iv): Improve boundary using above function if no
32545 value range information.
32547 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
32548 Martin Jambor <mjambor@suse.cz>
32550 * brig-builtins.def: New file.
32551 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
32552 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
32553 (DEF_HSAIL_SAT_BUILTIN): Likewise.
32554 (DEF_HSAIL_INTR_BUILTIN): Likewise.
32555 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
32556 * builtin-types.def (BT_INT8): New.
32557 (BT_INT16): Likewise.
32558 (BT_UINT8): Likewise.
32559 (BT_UINT16): Likewise.
32560 (BT_FN_ULONG): Likewise.
32561 (BT_FN_UINT_INT): Likewise.
32562 (BT_FN_UINT_ULONG): Likewise.
32563 (BT_FN_UINT_LONG): Likewise.
32564 (BT_FN_UINT_PTR): Likewise.
32565 (BT_FN_ULONG_PTR): Likewise.
32566 (BT_FN_INT8_FLOAT): Likewise.
32567 (BT_FN_INT16_FLOAT): Likewise.
32568 (BT_FN_UINT32_FLOAT): Likewise.
32569 (BT_FN_UINT16_FLOAT): Likewise.
32570 (BT_FN_UINT8_FLOAT): Likewise.
32571 (BT_FN_UINT64_FLOAT): Likewise.
32572 (BT_FN_UINT16_UINT32): Likewise.
32573 (BT_FN_UINT32_UINT16): Likewise.
32574 (BT_FN_UINT16_UINT16_UINT16): Likewise.
32575 (BT_FN_INT_PTR_INT): Likewise.
32576 (BT_FN_UINT_PTR_UINT): Likewise.
32577 (BT_FN_LONG_PTR_LONG): Likewise.
32578 (BT_FN_ULONG_PTR_ULONG): Likewise.
32579 (BT_FN_VOID_UINT64_UINT64): Likewise.
32580 (BT_FN_UINT8_UINT8_UINT8): Likewise.
32581 (BT_FN_INT8_INT8_INT8): Likewise.
32582 (BT_FN_INT16_INT16_INT16): Likewise.
32583 (BT_FN_INT_INT_INT): Likewise.
32584 (BT_FN_UINT_FLOAT_UINT): Likewise.
32585 (BT_FN_FLOAT_UINT_UINT): Likewise.
32586 (BT_FN_ULONG_UINT_UINT): Likewise.
32587 (BT_FN_ULONG_UINT_PTR): Likewise.
32588 (BT_FN_ULONG_ULONG_ULONG): Likewise.
32589 (BT_FN_UINT_UINT_UINT): Likewise.
32590 (BT_FN_VOID_UINT_PTR): Likewise.
32591 (BT_FN_UINT_UINT_PTR: Likewise.
32592 (BT_FN_UINT32_UINT64_PTR): Likewise.
32593 (BT_FN_INT_INT_UINT_UINT): Likewise.
32594 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
32595 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
32596 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
32597 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
32598 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
32599 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
32600 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
32601 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
32602 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
32603 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
32604 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
32605 * doc/frontends.texi: List BRIG FE.
32606 * doc/install.texi (Testing): Add BRIG tesring requirements.
32607 * doc/invoke.texi (Overall Options): Mention BRIG.
32608 * doc/standards.texi (Standards): Doucment BRIG HSA version.
32610 2017-01-24 Richard Biener <rguenther@suse.de>
32612 PR translation/79208
32613 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
32615 2017-01-24 Martin Jambor <mjambor@suse.cz>
32618 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
32619 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
32620 and known_contexts.
32622 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
32624 PR middle-end/79123
32625 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
32626 casts from signed to unsigned really don't have a range.
32628 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
32630 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
32631 GMP_RNDx for compatiblity.
32633 2017-01-24 Martin Liska <mliska@suse.cz>
32636 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
32637 that would prevent us to call alloca with -1 as argument.
32639 2017-01-24 Jakub Jelinek <jakub@redhat.com>
32641 * dwarf2out.c (output_compilation_unit_header, output_file_names):
32642 Avoid -Wformat-security warning.
32644 2017-01-23 Andrew Pinski <apinski@cavium.com>
32646 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
32649 2017-01-23 Martin Sebor <msebor@redhat.com>
32651 PR middle-end/78703
32652 * gimple-ssa-sprintf.c (warn_level): New global.
32653 (format_integer): Use it here and throughout the rest of the file.
32654 Use the same switch to compute sign as base.
32655 (maybe_warn): New function.
32656 (format_directive): Factor out warnings into maybe_warn.
32657 Add debugging output. Use warn_level.
32658 (add_bytes): Use warn_level.
32659 (pass_sprintf_length::compute_format_length): Add debugging output.
32660 (try_substitute_return_value): Same.
32661 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
32663 PR middle-end/78703
32664 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
32665 (struct fmtresult, format_integer, format_floating): Adjust.
32666 (fmtresult::fmtresult): Set max correctly in two argument ctor.
32667 (get_string_length, format_string,format_directive): Same.
32668 (pass_sprintf_length::compute_format_length): Same.
32669 (try_substitute_return_value): Simplify slightly.
32671 PR middle-end/78703
32672 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
32673 (fmtresult::operator+=): Outlined.
32674 (struct fmtresult): Add ctors.
32675 (struct conversion_spec): Rename...
32676 (struct directive): ...to this. Add and remove data members.
32677 (directive::set_width, directive::set_precision): New functions.
32678 (format_percent): Use fmtresult ctor.
32679 (get_width_and_precision): Remove.
32680 (format_integer): Make naming changes. Avoid computing width and
32682 (format_floating): Same. Adjust indentation.
32683 (format_character, format_none): New functions.
32684 (format_string): Moved character handling to format_character.
32685 (format_directive): Remove arguments, change return type.
32686 (parse_directive): New function.
32687 (pass_sprintf_length::compute_format_length): Move directive
32688 parsing to parse_directive.
32690 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32692 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
32693 (assign_assembler_name_if_needed): ... this.
32694 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
32695 (assign_assembler_name_if_needed): ... this.
32696 (free_lang_data_in_cgraph): Adjust callers.
32697 * cgraphunit.c (cgraph_node::analyze): Likewise.
32698 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
32701 2017-01-23 Richard Biener <rguenther@suse.de>
32703 PR tree-optimization/79088
32704 PR tree-optimization/79188
32705 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
32706 resetting loop bounds after last path deletion. Reset loop
32707 bounds of the target loop, make code match the comments.
32708 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
32709 Make sure loops need no fixups.
32711 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32713 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
32714 exponent support with double type for first argument.
32715 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
32716 type returned by __builtin_vec_extract_sig,
32717 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
32718 functions from "vector int" to "vector unsigned int" or from
32719 "vector long long int" to "vector unsigned long long int".
32720 Changed type returned by __builtin_vec_extract_exp,
32721 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
32722 functions from "vector int" to "vector unsigned int" or from
32723 "vector long long int" to "vector unsigned long long int".
32724 Changed return type of __builtin_vec_test_data_class,
32725 __builtin_vec_test_data_class_sp, and
32726 __builtin_vec_test_data_class_dp from "vector int" to
32727 "vector bool int" or from "vector long long int" to "vector bool
32728 long long int" and changed second argument type from "unsigned
32729 int" to "int". Added new overloaded function forms "vector float
32730 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
32731 "vector float __builtin_vec_insert_exp_sp (vector float, vector
32732 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
32733 double, vector unsigned long long int)" and "vector double
32734 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
32735 long int)". Changed return type of
32736 __builtin_scalar_test_data_class and
32737 __builtin_scalar_test_data_class_sp and
32738 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
32739 int" and changed second argument from "unsigned int" to "int".
32740 Changed type returned by __builtin_scalar_test_neg,
32741 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
32742 from "int" to "bool int". Added new overloaded function form
32743 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
32744 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
32745 exponent double-precision with floating point first argument.
32746 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
32747 documentation of scalar_test_data_class, scalar_test_neg,
32748 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
32749 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
32750 vec_test_data_class built-in functions to reflect refinements in
32751 their type signatures.
32753 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
32755 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
32757 (aarch64_elf_asm_destructor): Likewise.
32759 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
32761 PR rtl-optimization/78634
32762 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
32763 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
32764 * ifcvt.c (noce_try_cmove): Add missing cost check.
32766 PR rtl-optimization/71724
32767 * combine.c (if_then_else_cond): Look for situations where it is
32768 beneficial to undo the work of one of the recursive calls.
32770 2017-01-23 Bin Cheng <bin.cheng@arm.com>
32772 PR tree-optimization/70754
32773 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
32774 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
32775 combined stmt before it if not NULL.
32776 (combine_chains): Process refs reversely and compute dominance point
32779 2017-01-23 Martin Liska <mliska@suse.cz>
32781 PR tree-optimization/79196
32782 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
32783 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
32785 (strlen_optimize_stmt): Call the renamed function.
32787 2017-01-23 Michael Matz <matz@suse.de>
32789 PR tree-optimization/78384
32790 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
32792 2017-01-23 Richard Biener <rguenther@suse.de>
32794 PR tree-optimization/79186
32795 * tree-vrp.c (register_new_assert_for): Make sure we've seen
32796 both incoming edges before moving an assert.
32798 2017-01-23 Martin Jambor <mjambor@suse.cz>
32800 * ipa-prop.c (load_from_param_1): Removed.
32801 (load_from_unmodified_param): Bits from load_from_param_1 put back
32803 (load_from_param): Removed.
32804 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
32805 with stmt. Reverted back to use of load_from_unmodified_param.
32807 2017-01-23 Martin Jambor <mjambor@suse.cz>
32810 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
32811 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
32812 field a pointer to garbage collected vector, mark lattices and
32813 ipcp_orig_node with GTY((skip)).
32814 (ipa_get_param_count): Adjust to descriptors being a pointer.
32815 (ipa_get_param): Likewise.
32816 (ipa_get_type): Likewise.
32817 (ipa_get_param_move_cost): Likewise.
32818 (ipa_set_param_used): Likewise.
32819 (ipa_get_controlled_uses): Likewise.
32820 (ipa_set_controlled_uses): Likewise.
32821 (ipa_is_param_used): Likewise.
32822 (ipa_node_params_t): Move into garbage collector. New methods insert
32824 (ipa_node_params_sum): Annotate wth GTY(()).
32825 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
32827 (ipa_load_from_parm_agg): Adjust declaration.
32828 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
32829 * ipa-profile.c (ipa_profile): Likewise.
32830 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
32831 (ipa_populate_param_decls): Make descriptors parameter garbage
32833 (ipa_dump_param): Adjust to descriptors being a pointer.
32834 (ipa_alloc_node_params): Likewise.
32835 (ipa_initialize_node_params): Likewise.
32836 (load_from_param_1): Make descriptors parameter garbage collected.
32837 (load_from_unmodified_param): Likewise.
32838 (load_from_param): Likewise.
32839 (ipa_load_from_parm_agg): Likewise.
32840 (ipa_node_params::~ipa_node_params): Removed.
32841 (ipa_free_all_node_params): Remove call to delete operator.
32842 (ipa_node_params_t::insert): New.
32843 (ipa_node_params_t::remove): Likewise.
32844 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
32845 copy known_csts and known_contexts vectors.
32846 (ipa_read_node_info): Adjust to descriptors being a pointer.
32847 (ipcp_modif_dom_walker): Make m_descriptors field garbage
32849 (ipcp_transform_function): Make descriptors variable garbage
32852 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
32854 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
32855 * config/i386/avx512dqintrin.h: Ditto.
32856 * config/i386/avx512fintrin.h: Ditto.
32857 * config/i386/i386.c: Handle new builtins.
32858 * config/i386/i386-builtin.def: Add new builtins.
32859 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
32860 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
32862 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32863 Martin Liska <mliska@suse.cz>
32865 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
32866 * asan.c (asan_expand_poison_ifn): Support stores and use
32867 appropriate ASAN report function.
32868 * internal-fn.c (expand_ASAN_POISON_USE): New function.
32869 * internal-fn.def (ASAN_POISON_USE): Declare.
32870 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
32871 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
32872 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
32873 ASAN_POISON calls w/o LHS.
32874 * tree-ssa.c (execute_update_addresses_taken): Create clobber
32875 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
32876 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
32877 * gimplify.c (asan_poison_variables): Add attribute
32878 use_after_scope_memory to variables that really needs to live
32880 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
32881 having the attribute.
32883 2017-01-23 Martin Liska <mliska@suse.cz>
32885 * asan.c (create_asan_shadow_var): New function.
32886 (asan_expand_poison_ifn): Likewise.
32887 * asan.h (asan_expand_poison_ifn): New declaration.
32888 * internal-fn.c (expand_ASAN_POISON): Likewise.
32889 * internal-fn.def (ASAN_POISON): New builtin.
32890 * sanopt.c (pass_sanopt::execute): Expand
32891 asan_expand_poison_ifn.
32892 * tree-inline.c (copy_decl_for_dup_finish): Make function
32894 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
32895 * tree-ssa.c (is_asan_mark_p): New function.
32896 (execute_update_addresses_taken): Rewrite local variables
32897 (identified just by use-after-scope as addressable) into SSA.
32899 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32901 * doc/install.texi (Specific): opensource.apple.com uses https
32902 now. Remove trailing slash.
32904 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32906 * README.Portability: Remove note on an Irix compatibility issue.
32908 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
32910 * gcov.c (INCLUDE_ALGORITHM): Define.
32911 (INCLUDE_VECTOR): Define.
32912 No longer include <vector> and <algorithm> directly.
32914 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
32916 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
32918 * doc/invoke.texi (Code Gen Options): Ditto.
32920 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
32923 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
32925 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
32927 rtl-optimization/79125
32928 * cprop.c (local_cprop_pass): Handle cases where we make an
32929 unconditional trap.
32931 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
32935 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
32936 read from, for big endian.
32938 2017-01-20 Jiong Wang <jiong.wang@arm.com>
32940 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
32941 register pauth builtins for LP64 only.
32943 2017-01-20 Marek Polacek <polacek@redhat.com>
32946 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
32949 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32951 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
32953 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
32954 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
32955 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
32957 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32960 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
32961 in big-endian mode when they are not a single duplicated value.
32963 2017-01-20 Richard Biener <rguenther@suse.de>
32965 * BASE-VER: Bump to 7.0.1.
32967 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32969 * omp-low.c (omplow_simd_context): New struct. Use it...
32970 (lower_rec_simd_input_clauses): ...here and...
32971 (lower_rec_input_clauses): ...here to hold common data. Adjust all
32972 references to idx, lane, max_vf, is_simt.
32974 2017-01-20 Graham Markall <graham.markall@embecosm.com>
32976 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
32979 2017-01-20 Martin Jambor <mjambor@suse.cz>
32981 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
32982 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
32984 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
32985 (GTFILES): Rename hsa.c to hsa-common.c.
32986 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
32987 * hsa-dump.c: Likewise.
32988 * hsa-gen.c: Likewise.
32989 * hsa-regalloc.c: Likewise.
32990 * ipa-hsa.c: Likewise.
32991 * omp-expand.c: Likewise.
32992 * omp-low.c: Likewise.
32993 * toplev.c: Likewise.
32995 2017-01-20 Marek Polacek <polacek@redhat.com>
32998 * doc/invoke.texi: Document -Wduplicated-branches.
32999 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
33000 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
33001 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
33002 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
33003 return 0 only when not OEP_LEXICOGRAPHIC.
33004 (fold_build_cleanup_point_expr): Use the expression
33005 location when building CLEANUP_POINT_EXPR.
33006 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
33007 * tree.c (add_expr): Handle error_mark_node.
33009 2017-01-20 Martin Liska <mliska@suse.cz>
33012 * tree-profile.c (init_ic_make_global_vars): Do not call
33014 (gimple_init_gcov_profiler): Likewise.
33016 2017-01-20 Martin Liska <mliska@suse.cz>
33019 * cgraph.h (maybe_create_reference): Remove argument and
33021 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33023 * ipa-cp.c (create_specialized_node): Likewise.
33024 * symtab.c (symtab_node::maybe_create_reference): Handle
33025 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33027 2017-01-20 Martin Liska <mliska@suse.cz>
33029 * read-rtl-function.c (function_reader::create_function): Use
33030 build_decl instread of build_decl_stat.
33032 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33034 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33035 * config/i386/avx512dqintrin.h: Ditto.
33036 * config/i386/avx512fintrin.h: Ditto.
33037 * config/i386/i386-builtin-types.def: Add new types.
33038 * config/i386/i386.c: Handle new types.
33039 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33040 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33041 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33042 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33043 (__builtin_ia32_kshiftridi): New.
33044 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33046 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33050 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33051 define to rs6000_init_stack_protect_guard.
33052 (rs6000_init_stack_protect_guard): New function.
33054 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33055 Yunqiang Su <yunqiang.su@imgtec.com>
33057 * config.gcc (supported_defaults): Add madd4.
33058 (with_madd4): Add validation.
33059 (all_defaults): Add madd4.
33060 * config/mips/mips.opt (mmadd4): New option.
33061 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33063 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33065 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33066 (ISA_HAS_FUSED_MADD4): Likewise.
33067 * doc/invoke.texi (-mmadd4): Document the new option.
33068 * doc/install.texi (--with-madd4): Document the new option.
33070 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33072 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33073 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33074 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33075 (aarch64_init_pauth_hint_builtins): New.
33076 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33077 (aarch64_expand_builtin): Expand new builtins.
33079 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33081 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33082 * combine-stack-adj.c (no_unhandled_cfa): Handle
33083 REG_CFA_TOGGLE_RA_MANGLE.
33084 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33085 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33086 info for return address signing.
33087 (aarch64_expand_epilogue): Likewise.
33089 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33091 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33092 * config/aarch64/aarch64-protos.h
33093 (aarch64_return_address_signing_enabled): New declaration.
33094 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33096 (aarch64_expand_prologue): Sign return address before it's pushed onto
33098 (aarch64_expand_epilogue): Authenticate return address fetched from
33100 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33101 (aarch64_attributes): New function attributes for "sign-return-address".
33102 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33103 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33104 ("*do_return"): Generate combined instructions according to key index.
33105 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33106 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33108 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33109 * config/aarch64/aarch64.opt (msign-return-address=): New.
33110 * doc/extend.texi (AArch64 Function Attributes): Documents
33111 "sign-return-address=".
33112 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33114 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33116 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33117 overall option summary.
33119 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33121 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33122 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33123 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33124 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33126 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33128 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33129 -mpower9-minmax by default for -mcpu=power9.
33130 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33131 128-bit floating point.
33133 2017-01-20 Alan Modra <amodra@gmail.com>
33135 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33136 optimizing for size.
33138 2017-01-20 Alan Modra <amodra@gmail.com>
33141 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33142 for strcmp and strncmp from corresponding builtin decl.
33144 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33146 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33147 instead of i386/rtems-64.h.
33148 * config/i386/rtems-64.h: Remove.
33150 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33154 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33156 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33158 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33160 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33161 Change int to HOST_WIDE_INT.
33162 * config/aarch64/aarch64-protos.h
33163 (aarch64_simd_gen_const_vector_dup): Likewise.
33164 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33166 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33168 * langhooks-def.h (lhd_type_for_size): New decl.
33169 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33170 * langhooks.c (lhd_type_for_size): New function, taken from
33173 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33175 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33176 define_bypass for CR latency.
33177 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33178 (power9-alu2): Add define_bypass for CR latency.
33180 (power9-mul): Update insn latency.
33181 (power9-mul-compare): Update insn latency, bypass latency and remove
33184 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33186 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33188 * config/aarch64/aarch64.md
33189 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33190 aarch64_nopcrelative_literal_loads.
33191 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33193 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33195 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33196 TARGET_LOONGSON_3A.
33197 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33199 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33202 * config.gcc (supported_defaults): Add lxc1-sxc1.
33203 (with_lxc1_sxc1): Add validation.
33204 (all_defaults): Add lxc1-sxc1.
33205 * config/mips/mips.opt (mlxc1-sxc1): New option.
33206 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33208 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33209 __mips_no_lxc1_sxc1.
33210 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33211 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33212 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33214 2017-01-19 Richard Biener <rguenther@suse.de>
33216 PR tree-optimization/72488
33217 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33218 sure to restore SSA info.
33219 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33221 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33223 PR rtl-optimization/79121
33224 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33225 of the inner type when shifting an extended value.
33227 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33230 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33231 interposable aliases.
33233 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33236 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33237 Use the evmergelohi instruction.
33238 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33239 (mov_sitf_e500_subreg8_2_be): Likewise.
33240 (mov_sitf_e500_subreg12_2_le): Likewise.
33241 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33242 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33243 (mov_sitf_e500_subreg8_2_le): Likewise.
33244 (mov_sitf_e500_subreg12_2_be): Likewise.
33246 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33248 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33249 attribute from vecsimple to vecperm.
33250 (altivec_vbpermq2): Likewise.
33252 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33255 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33257 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33258 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33259 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33260 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33261 case where N arg is SIZE_MAX.
33262 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33263 (cmpstrsi): Add pattern.
33265 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33267 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33268 __builtin_vec_revb builtins.
33269 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33270 built-in functions to support generation of the ISA 3.0 XXBR<x>
33271 vector byte reverse instructions.
33272 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33273 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33274 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33275 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33276 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33277 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33278 (P9V_BUILTIN_VEC_REVB): Likewise.
33279 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33280 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33281 (p9_xxbrq_v16qi): Likewise.
33282 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33283 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33284 (p9_xxbrh_v8hi): Likewise.
33285 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33286 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33287 vec_revb built-in functions.
33289 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33291 PR rtl-optimization/78952
33292 * config/i386/i386.md (any_extract): New code iterator.
33293 (*insvqi_2): Use any_extract for source operand.
33294 (*insvqi_3): Use any_shiftrt for source operand.
33296 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33298 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33300 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33302 2017-01-18 Matthias Klose <doko@ubuntu.com>
33304 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33306 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33308 * config/rs6000/altivec.h (vec_bperm): Change #define.
33309 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33310 (altivec_vbpermq2): New define_insn.
33311 (altivec_vbpermd): Likewise.
33312 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33313 function interface.
33314 (VBPERMD): Likewise.
33315 (VBPERM): New polymorphic function interface.
33316 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33317 Add entries for P9V_BUILTIN_VEC_VBPERM.
33318 * doc/extend.texi: Add interfaces for vec_bperm.
33320 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33322 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33323 first letter of error messages.
33324 (s390_resolve_overloaded_builtin): Likewise.
33325 * config/s390/s390.c (s390_expand_builtin): Likewise.
33326 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33327 (s390_valid_target_attribute_inner_p): Likewise.
33328 * config/s390/s390.md ("tabort"): Likewise.
33330 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33332 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33333 (ISA_AVOID_DIV_HILO): New macro.
33334 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33335 (ISA_HAS_DDIV): Likewise.
33337 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33339 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33341 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33343 * doc/invoke.texi (fabi-version): Spelling fix.
33345 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33348 * doc/invoke.texi (fabi-version): Mention mangling fix for
33351 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33354 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33356 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
33359 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
33360 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
33362 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
33364 (stack_protect_test): Ditto.
33365 * config/rs6000/rs6000.opt (mstack-protector-guard=,
33366 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
33368 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
33369 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
33370 -mstack-protector-guard-offset=.
33371 (RS/6000 and PowerPC Options): Ditto.
33373 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33375 * config/i386/i386.h (MASK_CLASS_P): New define.
33376 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
33377 there are no registers from different register sets also when
33378 mask registers are used. Update function comment.
33379 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33380 to (*k/*r) and (*k/*km) alternatives.
33382 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33384 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33385 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
33386 (EH_RETURN_HANDLER_RTX): New define.
33387 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
33388 Force frame pointer in EH return functions.
33389 (aarch64_expand_epilogue): Add barrier for eh_return.
33390 (aarch64_final_eh_return_addr): Remove.
33391 (aarch64_eh_return_handler_rtx): New function.
33392 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
33394 (aarch64_eh_return_handler_rtx): New prototype.
33396 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33398 * config/rs6000/altivec.h (vec_rlmi): New #define.
33399 (vec_vrlnm): Likewise.
33400 (vec_rlnm): Likewise.
33401 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
33402 (UNSPEC_VRLNM): Likewise.
33403 (VIlong): New mode iterator.
33404 (altivec_vrl<VI_char>mi): New define_insn.
33405 (altivec_vrl<VI_char>nm): Likewise.
33406 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
33408 (VRLDNM): Likewise.
33409 (RLNM): New polymorphic function entry.
33410 (VRLWMI): New monomorphic function entry.
33411 (VRLDMI): Likewise.
33412 (RLMI): New polymorphic function entry.
33413 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
33414 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
33415 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
33418 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33421 * dwarf2out.c (field_byte_offset): Restore the
33422 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
33423 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
33424 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
33427 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33430 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
33432 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33435 * dwarf2out.c (add_data_member_location_attribute): For constant
33436 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
33437 instead of DW_AT_data_member_location, DW_AT_bit_offset and
33438 DW_AT_byte_size attributes.
33440 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33442 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
33443 after forcing to constant memory when the code model is medium.
33445 2017-01-17 Julia Koval <julia.koval@intel.com>
33448 * config/i386/avx512fintrin.h
33449 (_mm512_i32gather_ps): Change __addr type to void const*.
33450 (_mm512_mask_i32gather_ps): Ditto.
33451 (_mm512_i32gather_pd): Ditto.
33452 (_mm512_mask_i32gather_pd): Ditto.
33453 (_mm512_i64gather_ps): Ditto.
33454 (_mm512_mask_i64gather_ps): Ditto.
33455 (_mm512_i64gather_pd): Ditto.
33456 (_mm512_mask_i64gather_pd): Ditto.
33457 (_mm512_i32gather_epi32): Ditto.
33458 (_mm512_mask_i32gather_epi32): Ditto.
33459 (_mm512_i32gather_epi64): Ditto.
33460 (_mm512_mask_i32gather_epi64): Ditto.
33461 (_mm512_i64gather_epi32): Ditto.
33462 (_mm512_mask_i64gather_epi32): Ditto.
33463 (_mm512_i64gather_epi64): Ditto.
33464 (_mm512_mask_i64gather_epi64): Ditto.
33465 (_mm512_i32scatter_ps): Change __addr type to void*.
33466 (_mm512_mask_i32scatter_ps): Ditto.
33467 (_mm512_i32scatter_pd): Ditto.
33468 (_mm512_mask_i32scatter_pd): Ditto.
33469 (_mm512_i64scatter_ps): Ditto.
33470 (_mm512_mask_i64scatter_ps): Ditto.
33471 (_mm512_i64scatter_pd): Ditto.
33472 (_mm512_mask_i64scatter_pd): Ditto.
33473 (_mm512_i32scatter_epi32): Ditto.
33474 (_mm512_mask_i32scatter_epi32): Ditto.
33475 (_mm512_i32scatter_epi64): Ditto.
33476 (_mm512_mask_i32scatter_epi64): Ditto.
33477 (_mm512_i64scatter_epi32): Ditto.
33478 (_mm512_mask_i64scatter_epi32): Ditto.
33479 (_mm512_i64scatter_epi64): Ditto.
33480 (_mm512_mask_i64scatter_epi64): Ditto.
33481 * config/i386/avx512pfintrin.h
33482 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
33483 (_mm512_mask_prefetch_i32gather_ps): Ditto.
33484 (_mm512_mask_prefetch_i64gather_pd): Ditto.
33485 (_mm512_mask_prefetch_i64gather_ps): Ditto.
33486 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
33487 (_mm512_prefetch_i32scatter_ps): Ditto.
33488 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
33489 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
33490 (_mm512_prefetch_i64scatter_pd): Ditto.
33491 (_mm512_prefetch_i64scatter_ps): Ditto.
33492 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
33493 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
33494 * config/i386/avx512vlintrin.h
33495 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
33496 (_mm_mmask_i32gather_ps): Ditto.
33497 (_mm256_mmask_i32gather_pd): Ditto.
33498 (_mm_mmask_i32gather_pd): Ditto.
33499 (_mm256_mmask_i64gather_ps): Ditto.
33500 (_mm_mmask_i64gather_ps): Ditto.
33501 (_mm256_mmask_i64gather_pd): Ditto.
33502 (_mm_mmask_i64gather_pd): Ditto.
33503 (_mm256_mmask_i32gather_epi32): Ditto.
33504 (_mm_mmask_i32gather_epi32): Ditto.
33505 (_mm256_mmask_i32gather_epi64): Ditto.
33506 (_mm_mmask_i32gather_epi64): Ditto.
33507 (_mm256_mmask_i64gather_epi32): Ditto.
33508 (_mm_mmask_i64gather_epi32): Ditto.
33509 (_mm256_mmask_i64gather_epi64): Ditto.
33510 (_mm_mmask_i64gather_epi64): Ditto.
33511 (_mm256_i32scatter_ps): Change __addr type to void*.
33512 (_mm256_mask_i32scatter_ps): Ditto.
33513 (_mm_i32scatter_ps): Ditto.
33514 (_mm_mask_i32scatter_ps): Ditto.
33515 (_mm256_i32scatter_pd): Ditto.
33516 (_mm256_mask_i32scatter_pd): Ditto.
33517 (_mm_i32scatter_pd): Ditto.
33518 (_mm_mask_i32scatter_pd): Ditto.
33519 (_mm256_i64scatter_ps): Ditto.
33520 (_mm256_mask_i64scatter_ps): Ditto.
33521 (_mm_i64scatter_ps): Ditto.
33522 (_mm_mask_i64scatter_ps): Ditto.
33523 (_mm256_i64scatter_pd): Ditto.
33524 (_mm256_mask_i64scatter_pd): Ditto.
33525 (_mm_i64scatter_pd): Ditto.
33526 (_mm_mask_i64scatter_pd): Ditto.
33527 (_mm256_i32scatter_epi32): Ditto.
33528 (_mm256_mask_i32scatter_epi32): Ditto.
33529 (_mm_i32scatter_epi32): Ditto.
33530 (_mm_mask_i32scatter_epi32): Ditto.
33531 (_mm256_i32scatter_epi64): Ditto.
33532 (_mm256_mask_i32scatter_epi64): Ditto.
33533 (_mm_i32scatter_epi64): Ditto.
33534 (_mm_mask_i32scatter_epi64): Ditto.
33535 (_mm256_i64scatter_epi32): Ditto.
33536 (_mm256_mask_i64scatter_epi32): Ditto.
33537 (_mm_i64scatter_epi32): Ditto.
33538 (_mm_mask_i64scatter_epi32): Ditto.
33539 (_mm256_i64scatter_epi64): Ditto.
33540 (_mm256_mask_i64scatter_epi64): Ditto.
33541 (_mm_i64scatter_epi64): Ditto.
33542 (_mm_mask_i64scatter_epi64): Ditto.
33543 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
33544 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
33545 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
33546 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
33547 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
33548 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
33549 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
33550 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
33551 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
33552 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
33553 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
33554 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
33555 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
33556 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
33557 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
33558 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
33559 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
33560 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
33561 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
33562 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
33563 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
33564 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
33565 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
33566 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
33567 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
33568 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
33569 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
33570 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
33571 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
33572 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
33573 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
33574 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
33575 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
33576 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
33577 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
33578 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
33579 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
33580 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
33581 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
33582 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
33583 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
33584 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
33585 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
33586 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
33587 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
33588 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
33589 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
33590 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
33591 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
33592 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
33593 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
33594 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
33595 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
33596 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
33597 definitions accordingly.
33599 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
33600 Kuan-Lin Chen <kuanlinchentw@gmail.com>
33603 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
33606 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
33609 * ira-conflicts.c (ira_build_conflicts): Update total conflict
33610 hard regs for inner regno.
33612 2017-01-17 Martin Liska <mliska@suse.cz>
33615 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
33616 assumption and add comment.
33618 2017-01-17 Nathan Sidwell <nathan@acm.org>
33620 * ipa-visibility.c (localize_node): New function, broken out of ...
33621 (function_and_variable_visibility): ... here. Call it.
33623 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33625 PR middle-end/77445
33626 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
33627 correctly set frequency of oudgoing edge.
33628 (duplicate_thread_path): Fix profile updating.
33630 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33633 * configure.ac: Add GCC_BASE_VER.
33634 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
33635 version from BASE-VER file.
33636 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
33637 (gcc.o): Depend on $(BASEVER).
33638 * common.opt (dumpfullversion): New option.
33639 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
33640 * doc/invoke.texi: Document -dumpfullversion.
33641 * doc/install.texi: Document --with-gcc-major-version-only.
33642 * configure: Regenerated.
33644 2017-01-17 Richard Biener <rguenther@suse.de>
33646 PR tree-optimization/71433
33647 * tree-vrp.c (register_new_assert_for): Merge same asserts
33648 on all incoming edges.
33649 (process_assert_insertions_for): Handle insertions at the
33652 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
33654 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
33655 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
33657 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
33660 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
33663 2017-01-17 Alan Modra <amodra@gmail.com>
33666 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
33667 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
33668 symbolic stack limit when pic.
33670 2017-01-16 Martin Sebor <msebor@redhat.com>
33672 PR tree-optimization/78608
33673 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
33675 2017-01-16 Jeff Law <law@redhat.com>
33678 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
33679 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
33680 for several include directories that may be relative to sysroot.
33681 * config/i386/x-mingw32 (gplus_includedir): Define.
33682 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
33683 (native_system_includedir): Likewise.
33684 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
33685 override if TARGET_SYSTEM_ROOT is defined.
33686 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
33688 PR tree-optimization/79090
33689 PR tree-optimization/33562
33690 PR tree-optimization/61912
33691 PR tree-optimization/77485
33692 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
33693 and computed trims into the dump file.
33695 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33697 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
33699 2017-01-16 Jakub Jelinek <jakub@redhat.com>
33702 * gimplify.c (gimplify_init_constructor): If want_value and
33703 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
33707 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
33708 sequence. Formatting fixes.
33709 (doloop_optimize): Formatting fixes.
33712 * gcc.c (debug_level_greater_than_spec_func): New function.
33713 (static_spec_functions): Add debug-level-gt spec function.
33714 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
33716 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33717 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33718 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
33719 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
33720 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
33721 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
33723 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
33725 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
33726 QImode fixups to general and mask registers only.
33728 2017-01-16 Carl Love <cel@us.ibm.com>
33730 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
33731 for built-in functions
33732 vector signed char vec_nabs (vector signed char)
33733 vector signed short vec_nabs (vector signed short)
33734 vector signed int vec_nabs (vector signed int)
33735 vector signed long long vec_nabs (vector signed long long)
33736 vector float vec_nabs (vector float)
33737 vector double vec_nabs (vector double)
33738 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
33740 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
33741 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
33742 * doc/extend.texi: Update the documentation file for the new built-in
33745 2017-01-16 Martin Sebor <msebor@redhat.com>
33747 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
33750 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33752 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
33753 UNSPEC_VSX__XXSPLTD to require special splat handling.
33755 2017-01-16 David Malcolm <dmalcolm@redhat.com>
33758 * system.h: Poison strndup.
33760 2017-01-16 Alan Modra <amodra@gmail.com>
33763 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
33766 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
33768 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
33770 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
33772 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
33773 call recog here. Assert that INSN_CODE (insn) is non-negative.
33775 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33778 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
33780 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
33781 in the currently scheduled RTL fragment.
33783 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33785 PR rtl-optimization/78751
33786 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
33789 2017-01-14 Jeff Law <law@redhat.com>
33791 PR tree-optimization/79090
33792 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
33793 variable length stores.
33794 (compute_trims): Delete dead assignment to *trim_tail.
33795 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
33798 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
33800 PR rtl-optimization/78626
33801 PR rtl-optimization/78727
33802 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
33803 of a block, and split such blocks after everything else is finished.
33805 2017-01-14 Alan Modra <amodra@gmail.com>
33808 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
33809 target legitimate_combined_insn.
33810 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
33811 (rs6000_legitimate_combined_insn): New function.
33812 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
33814 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
33815 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
33816 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
33818 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
33820 * doc/frontends.texi (G++ and GCC): Remove references to Java.
33822 2017-01-13 Jeff Law <law@redhat.com>
33824 PR tree-optimization/33562
33825 PR tree-optimization/61912
33826 PR tree-optimization/77485
33827 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
33829 (delete_dead_assignment): Likewise.
33830 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
33831 statement to delete_dead_call and delete_dead_assignment.
33833 2017-01-13 David Malcolm <dmalcolm@redhat.com>
33836 * substring-locations.c (format_warning_va): Strengthen case 1 so
33837 that both endpoints of the substring must be within the format
33838 range for just the substring to be printed.
33840 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
33842 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
33843 * config/i386/i386.c (ix86_target_string): Add missing options
33844 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
33845 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
33846 flags_other and ix86_target_other to flags2_other. Display unknown
33848 (ix86_valid_target_attribute_inner_p): Add missing options and
33849 reorder options by implied ISAs, as in ix86_target_string.
33851 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33853 * hash-table.h (hash_table::too_empty_p): New function.
33854 (hash_table::expand): Use it.
33855 (hash_table::traverse): Likewise.
33856 (hash_table::empty_slot): Use sizeof (value_type) instead of
33857 sizeof (PTR) to convert bytes to elements. Shrink the table
33858 if the current size is excessive for the current number of
33861 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33863 * ira-costs.c (record_reg_classes): Break from the inner loop
33864 early once alt_fail is known to be true. Update outer loop
33865 handling accordingly.
33867 2017-01-13 Jeff Law <law@redhat.com>
33869 * tree-ssa-dse.c (decrement_count): New function.
33870 (increment_start_addr, maybe_trim_memstar_call): Likewise.
33871 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
33872 when we know the partially dead statement is a mem* function.
33874 PR tree-optimization/61912
33875 PR tree-optimization/77485
33876 * tree-ssa-dse.c: Include expr.h.
33877 (maybe_trim_constructor_store): New function.
33878 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
33880 PR tree-optimization/33562
33881 PR tree-optimization/61912
33882 PR tree-optimization/77485
33883 * doc/invoke.texi: Document new dse-max-object-size param.
33884 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
33885 * tree-ssa-dse.c: Include params.h.
33886 (dse_store_status): New enum.
33887 (initialize_ao_ref_for_dse): New, partially extracted from
33889 (valid_ao_ref_for_dse, normalize_ref): New.
33890 (setup_live_bytes_from_ref, compute_trims): Likewise.
33891 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
33892 (maybe_trim_partially_dead_store): Likewise.
33893 (maybe_trim_complex_store): Likewise.
33894 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
33895 Track what bytes live from the original store. Return tri-state
33896 for dead, partially dead or live.
33897 (dse_dom_walker): Add constructor, destructor and new private members.
33898 (delete_dead_call, delete_dead_assignment): New extracted from
33900 (dse_optimize_stmt): Make a member of dse_dom_walker.
33901 Use initialize_ao_ref_for_dse.
33903 PR tree-optimization/33562
33904 PR tree-optimization/61912
33905 PR tree-optimization/77485
33906 * sbitmap.h (bitmap_count_bits): Prototype.
33907 (bitmap_clear_range, bitmap_set_range): Likewise.
33908 * sbitmap.c (bitmap_clear_range): New function.
33909 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
33911 2017-01-13 Martin Liska <mliska@suse.cz>
33914 * function.c (set_cfun): Add new argument force.
33915 * function.h (set_cfun): Likewise.
33916 * ipa-inline-transform.c (inline_call): Use the function when
33917 strict alising from is dropped for function we inline to.
33919 2017-01-13 Richard Biener <rguenther@suse.de>
33921 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
33922 for dumping GIMPLE INTEGER_CSTs.
33924 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33926 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
33927 to 201112L since C++17.
33929 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
33932 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
33933 if -fsanitize=kernel-address is present.
33935 2017-01-13 Richard Biener <rguenther@suse.de>
33937 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
33938 as _Literal ( type ) number in case usual suffixes do not
33939 preserve all information.
33941 2017-01-13 Richard Biener <rguenther@suse.de>
33943 PR tree-optimization/77283
33944 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
33945 and ssa-iterators.h.
33946 (is_feasible_trace): Implement a cost model based on joiner
33949 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
33952 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
33953 char or short to __float128/_Float128 directly.
33955 2017-01-12 Martin Sebor <msebor@redhat.com>
33957 to -Wformat-overflow.
33958 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
33959 (min_bytes_remaining): Same.
33960 (get_string_length): Same.
33961 (format_string): Same.
33962 (format_directive): Same.
33964 (pass_sprintf_length::handle_gimple_call): Same.
33966 2017-01-12 Jakub Jelinek <jakub@redhat.com>
33968 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
33969 info.nowrite calls with no lhs that can't throw. Return bool
33970 whether gsi_remove has been called or not.
33971 (pass_sprintf_length::handle_gimple_call): Return bool whether
33972 try_substitute_return_value called gsi_remove. Formatting fix.
33973 (pass_sprintf_length::execute): Don't use gsi_remove if
33974 handle_gimple_call returned true.
33977 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
33978 be removed due to side-effects, don't remove following barrier nor
33979 turn the successor edge into fallthru edge.
33981 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33984 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
33985 element-reversing loads and stores as not swappable.
33987 2017-01-12 Nathan Sidwell <nathan@acm.org>
33988 Nicolai Stange <nicstange@gmail.com>
33990 * combine.c (try_combine): Don't ignore result of overlap checking
33991 loop. Combine overlap & asm check into single loop.
33993 2017-01-12 Richard Biener <rguenther@suse.de>
33995 * tree-pretty-print.c (dump_generic_node): Provide -gimple
33996 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
33998 2017-01-12 Richard Biener <rguenther@suse.de>
34000 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
34001 and TS_TARGET_OPTION directly derive from TS_BASE.
34002 * tree-core.h (tree_optimization_option): Derive from tree_base.
34003 (tree_target_option): Likewise.
34005 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
34007 * config/i386/i386.c (memory_address_length): Increase len
34008 only when rip_relative_addr_p returns false.
34010 2017-01-11 Julia Koval <julia.koval@intel.com>
34012 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
34013 (OPTION_MASK_ISA_SGX_SET): New.
34014 (ix86_handle_option): Handle OPT_msgx.
34015 * config.gcc: Added sgxintrin.h.
34016 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34017 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34018 * config/i386/i386.c (ix86_target_string): Add -msgx.
34020 (ix86_option_override_internal): Handle new options.
34021 (ix86_valid_target_attribute_inner_p): Add sgx.
34022 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34023 * config/i386/i386.opt: Add msgx.
34024 * config/i386/sgxintrin.h: New file.
34025 * config/i386/x86intrin.h: Add sgxintrin.h.
34027 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34030 * fold-const.c (maybe_nonzero_address): Return 1 for function
34032 (tree_single_nonzero_warnv_p): Don't handle function local objects
34036 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34039 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34042 * opts.c: Include "spellcheck.h"
34043 (struct string_fragment): New struct.
34044 (struct edit_distance_traits<const string_fragment &>): New
34046 (get_closest_sanitizer_option): New function.
34047 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34049 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34051 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34053 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34054 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34055 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34056 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34057 for initial die_offset if dwarf_split_debug_info.
34058 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34059 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34061 (output_skeleton_debug_sections): Formatting fix. Use
34062 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34063 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34065 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34067 * config/arm/cortex-a53.md: Add bypasses for
34068 cortex_a53_r2f_cvt.
34069 (cortex_a53_r2f): Only use for transfers.
34070 (cortex_a53_f2r): Likewise.
34071 (cortex_a53_r2f_cvt): Add reservation for conversions.
34072 (cortex_a53_f2r_cvt): Likewise.
34074 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34076 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34077 to all inlined functions, change static to extern.
34079 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34082 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34084 (arm_assemble_integer): Likewise.
34086 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34088 * config.gcc: Use new awk script to check CPU, FPU and architecture
34089 parameters for --with-... options.
34090 * config/arm/parsecpu.awk: New file
34091 * config/arm/arm-cpus.in: New file.
34092 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34094 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34096 * config/arm/t-arm: Update dependency rules.
34097 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34098 of processing .def files.
34099 * config/arm/genopt.sh: Deleted.
34100 * config/arm/gentune.sh: Deleted.
34101 * config/arm/arm-cores.def: Deleted.
34102 * config/arm/arm-arches.def: Deleted.
34103 * config/arm/arm-fpus.def: Deleted.
34104 * config/arm/arm-tune.md: Regenerated.
34105 * config/arm/arm-tables.opt: Regenerated.
34106 * config/arm/arm-cpu.h: New generated file.
34107 * config/arm/arm-cpu-data.h: New generated file.
34108 * config/arm/arm-cpu-cdata.h: New generated file.
34110 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34113 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34115 (input_varpool_node): Unpack dynamically_initialized bit.
34117 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34119 PR rtl-optimization/79032
34120 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34121 the alignment of the adjusted memory reference against that of MODE,
34122 instead of the alignment of the original memory reference.
34124 2017-01-11 Martin Jambor <mjambor@suse.cz>
34126 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34128 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34129 decorated functions.
34131 2017-01-11 Richard Biener <rguenther@suse.de>
34133 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34134 set range/nonnull info for PHI results. Do not set it on
34135 stmts marked for removal.
34137 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34139 * expr.c (store_field): In the bitfield case, fetch the return value
34140 from the registers before applying a single big-endian adjustment.
34141 Always do a final load for a BLKmode value not larger than a word.
34143 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34146 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34147 that we correctly handle column numbers greater than
34148 LINE_MAP_MAX_COLUMN_NUMBER.
34150 2017-01-10 Martin Sebor <msebor@redhat.com>
34152 PR middle-end/78245
34153 * gimple-ssa-sprintf.c (get_destination_size): Call
34154 {init,fini}object_sizes.
34155 * tree-object-size.c (addr_object_size): Adjust.
34156 (pass_through_call): Adjust.
34157 (pass_object_sizes::execute): Adjust.
34158 * tree-object-size.h (fini_object_sizes): Declare.
34160 2017-01-10 Martin Sebor <msebor@redhat.com>
34162 PR tree-optimization/78775
34163 * builtins.c (get_size_range): Move...
34164 * calls.c: ...to here.
34165 (alloc_max_size): Accept zero argument.
34166 (operand_signed_p): Remove.
34167 (maybe_warn_alloc_args_overflow): Call get_size_range.
34168 * calls.h (get_size_range): Declare.
34170 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34172 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34173 from TI's devices.csv file as of September 2016.
34174 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34176 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34178 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34179 * doc/invoke.texi: Likewise.
34180 * doc/md.texi: Likewise.
34181 * doc/objc.texi: Likewise.
34183 2017-01-10 Joshua Conner <joshconner@google.com>
34185 * config/arm/fuchsia-elf.h: New file.
34186 * config/fuchsia.h: New file.
34187 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34188 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34190 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34192 2016-01-10 Richard Biener <rguenther@suse.de>
34194 PR tree-optimization/79034
34195 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34196 Propagate out degenerate PHIs in the joiner.
34198 2017-01-10 Martin Liska <mliska@suse.cz>
34200 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34201 (sort_congruence_classes_by_decl_uid): Likewise.
34202 (sort_congruence_class_groups_by_decl_uid): Likewise.
34203 (sem_item_optimizer::merge_classes): Sort class, groups in these
34204 classes and members in the groups by DECL_UID of declarations.
34205 This would make merge operations stable.
34207 2017-01-10 Martin Liska <mliska@suse.cz>
34209 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34210 usage of m_classes_vec.
34211 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34212 (sem_item_optimizer::get_group_by_hash): Likewise.
34213 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34214 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34215 (sem_item_optimizer::verify_classes): Likewise.
34216 (sem_item_optimizer::process_cong_reduction): Likewise.
34217 (sem_item_optimizer::dump_cong_classes): Likewise.
34218 (sem_item_optimizer::merge_classes): Likewise.
34219 * ipa-icf.h (congruence_class_hash): Rename from
34220 congruence_class_group_hash. Remove declaration of m_classes_vec.
34222 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34224 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34225 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34226 * config.gcc: Add avx512vpopcntdqintrin.h.
34227 * config/i386/avx512vpopcntdqintrin.h: New.
34228 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34229 * config/i386/i386-builtin-types.def: Add new types.
34230 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34231 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34232 __builtin_ia32_vpopcountq_v8di_mask): New.
34233 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34234 __AVX512VPOPCNTDQ__.
34235 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34236 (PTA_AVX512VPOPCNTDQ): Define.
34237 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34238 TARGET_AVX512VPOPCNTDQ_P): Define.
34239 * config/i386/i386.opt: Add mavx512vpopcntdq.
34240 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34241 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34243 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34245 PR middle-end/77484
34246 * predict.def (PRED_CALL): Set to 67.
34248 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34250 * expr.c (store_field): In the bitfield case, if the value comes from
34251 a function call and is of an aggregate type returned in registers, do
34252 not modify the field mode; extract the value in all cases if the mode
34253 is BLKmode and the size is not larger than a word.
34255 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34258 * config/i386/cpuid.h: Fix undefined behavior.
34260 2017-01-04 Jeff Law <law@redhat.com>
34262 PR tree-optimization/79007
34263 PR tree-optimization/67955
34264 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34265 conservative for pt.null when flag_non_call_exceptions is on.
34267 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34269 PR translation/79019
34270 PR translation/79020
34271 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34272 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34273 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34275 * config/avr/avr.opt (maccumulate-args): Likewise.
34276 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34277 * common.opt (freport-bug): Likewise.
34278 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34279 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34280 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34281 translatable string.
34282 * config/i386/i386.c (function_value_32): Likewise.
34283 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34284 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34286 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34287 * common/config/msp430/msp430-common.c (msp430_handle_option):
34289 * symtab.c (symtab_node::verify_base): Likewise.
34290 * opts.c (set_debug_level): Likewise.
34291 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34292 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34293 missing whitespace to translatable strings.
34294 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34295 * config/sh/superh.h: Likewise.
34296 * config/i386/xopintrin.h: Likewise.
34297 * config/i386/znver1.md: Likewise.
34298 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34299 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34300 * double-int.h (struct double_int): Likewise.
34301 * double-int.c (div_and_round_double): Likewise.
34302 * wide-int.cc: Likewise.
34303 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34304 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34305 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34306 (crossjumps_occurred): ... this.
34307 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34310 PR tree-optimization/78899
34311 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34312 returning bool return struct loop *, NULL for failure and the new
34314 (versionable_outer_loop_p): Don't version outer loop if it has
34315 dont_vectorized bit set.
34316 (tree_if_conversion): When versioning outer loop, ensure
34317 tree_if_conversion is performed also on the inner loop of the
34318 non-vectorizable outer loop copy.
34319 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34320 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34321 prevent vectorization of it.
34322 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34323 the outer loop vectorization of the non-scalar version is attempted
34324 before vectorization of the inner loop in scalar version. If
34325 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34326 vectorization of its inner loop.
34327 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34328 has 2 inner loops, rename also on edges from bb whose single pred
34329 is outer_loop->header. Fix typo in function comment.
34331 2017-01-09 Martin Sebor <msebor@redhat.com>
34334 * asan.c (asan_emit_stack_protection): Increase local buffer size
34335 to avoid snprintf truncation warning.
34337 2017-01-09 Andrew Pinski <apinski@cavium.com>
34339 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34340 to reference thunderx2t99 for the tuning structure
34341 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34343 (thunderx2t99_extra_costs): This.
34344 * config/aarch64/aarch64-tune.md: Regenerate.
34345 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34346 (vulcan_addrcost_table): This.
34347 (vulcan_regmove_cost): Rename to ...
34348 (thunderx2t99_regmove_cost): This.
34349 (vulcan_vector_cost): Rename to ...
34350 (thunderx2t99_vector_cost): this.
34351 (vulcan_branch_cost): Rename to ...
34352 (thunderx2t99_branch_cost): This.
34353 (vulcan_tunings): Rename to ...
34354 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34355 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34357 2017-01-09 Martin Jambor <mjambor@suse.cz>
34361 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
34362 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
34363 (propagate_vr_accross_jump_function): Use the above function for all
34364 value range computations for pass-through jump functions and type
34365 converasion from explicit value range values.
34366 (ipcp_propagate_stage): Do not attempt to deduce types of formal
34367 parameters from TYPE_ARG_TYPES.
34368 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
34369 (ipa_write_node_info): Stream type of the actual argument.
34370 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
34372 2017-01-09 Martin Liska <mliska@suse.cz>
34375 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
34376 (lookup_compiler): Do not show error message with have_E.
34378 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34380 PR tree-optimization/78938
34381 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34382 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34383 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34384 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
34387 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34389 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
34392 2017-01-09 Richard Biener <rguenther@suse.de>
34394 PR tree-optimization/78997
34395 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
34396 name condition properly.
34398 2017-01-09 Richard Biener <rguenther@suse.de>
34401 * dwarf2out.c (is_cxx): New overload with context.
34402 (is_naming_typedef_decl): Use it.
34404 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34406 * invoke.texi (Option Summary): Correct spacing in option lists
34407 and add line breaks to fix over-long lines.
34409 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34411 PR middle-end/17660
34413 * extend.texi (Common Variable Attributes): Add xref to GCC
34414 Internals manual to explain mode attribute keywords.
34416 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34419 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
34420 and Preprocessor Options.
34421 (Options for Linking): Document -pthread here....
34422 (RS/6000 and PowerPC Options): ...not here.
34423 (Solaris 2 Options): ...or here.
34424 * doc/cppopts.texi: Document -pthread.
34426 2017-01-08 Martin Sebor <msebor@redhat.com>
34428 PR middle-end/77708
34429 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
34430 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
34431 New member functions.
34432 (format_directive): Used them.
34434 (pass_sprintf_length::handle_gimple_call): Same.
34435 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
34436 to avoid truncation for any argument.
34437 (extract_affine_mul): Same.
34438 * tree.c (get_file_function_name): Same.
34440 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34442 PR middle-end/77484
34443 * predict.def (PRED_INDIR_CALL): Set to 86.
34445 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34447 PR preprocessor/54124
34448 * doc/cppopts.texi: Reformat -d subtable to list the full name
34449 of the options. Add cross-reference to the docs for the general
34450 compiler -d options.
34451 * doc/invoke.texi (Developer Options): Add cross-reference to the
34452 preprocessor-specific -d option documentation.
34454 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34456 PR preprocessor/13498
34457 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
34458 redudant material, and reflect new command-line options.
34459 (System Headers): Likewise.
34461 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34463 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
34464 -isystem, and -idirafter. Copy-edit.
34465 * doc/cppopts.texi: Copy-edit. Remove contradiction about
34466 default for -ftrack-macro-expansion. Delete obsolete and
34467 badly-formatted implementation details about -fdebug-cpp output.
34468 * doc/cppwarnopts.texi: Copy-edit.
34470 2017-01-07 David Malcolm <dmalcolm@redhat.com>
34473 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34474 that the transition from a max line width >= 1<<10 to narrower
34475 lines works correctly.
34477 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
34479 * doc/options.texi (PerFunction): New.
34480 * opt-functions.awk (switch_flags): Map both Optimization and
34481 PerFunction to CL_OPTIMIZATION.
34482 * opth-gen.awk: Test for PerFunction flag along with
34484 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
34485 it only when the latter is present. Skip those that don't in
34486 the hash function generator.
34487 * common.opt (fvar-tracking): Mark as PerFunction instead of
34489 (fvar-tracking-assignments): Likewise.
34490 (fvar-tracking-assignments-toggle): Likewise.
34491 (fvar-tracking-uninit): Likewise.
34493 2017-01-07 Jakub Jelinek <jakub@redhat.com>
34495 PR translation/79018
34496 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
34499 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
34502 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
34503 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
34504 TARGET_LONG_JUMP_TABLE_OFFSETS.
34505 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
34506 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
34507 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
34508 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
34509 * config/m68k/m68k.md (tablejump expander): Likewise.
34510 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
34511 TARGET_LONG_JUMP_TABLE_OFFSETS.
34512 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
34513 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
34515 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
34516 David Holsgrove <david.holsgrove@xilinx.com>
34518 * common/config/microblaze/microblaze-common.c
34519 (TARGET_EXCEPT_UNWIND_INFO): Remove.
34520 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
34522 * config/microblaze/microblaze.c (microblaze_must_save_register)
34523 (microblaze_expand_epilogue, microblaze_return_addr): Handle
34525 (microblaze_eh_return): New function.
34526 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
34527 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
34528 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
34529 * config/microblaze/microblaze.md (eh_return): New pattern.
34531 2017-01-06 Jakub Jelinek <jakub@redhat.com>
34533 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
34534 GCC_DIAGNOSTIC_STRINGIFY): Define.
34536 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
34538 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34540 * config/arm/arm.md (<mcrr>): New.
34542 * config/arm/arm.c (arm_arch5te): New.
34543 (arm_option_override): Set arm_arch5te.
34544 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
34546 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
34547 (arm_mcrr_qualifiers): ... this. New.
34548 (MRRC_QUALIFIERS): Define to...
34549 (arm_mrrc_qualifiers): ... this. New.
34550 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
34552 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
34553 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
34554 (MRRCI, mrrc, MRRC): New.
34555 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
34556 VUNSPEC_MRRC2): New.
34558 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34560 * config/arm/arm.md (<mcr>): New.
34562 * config/arm/arm.c (arm_coproc_builtin_available): Add
34563 support for mcr, mrc, mcr2 and mrc2.
34564 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
34565 (arm_mcr_qualifiers): ... this. New.
34566 (MRC_QUALIFIERS): Define to ...
34567 (arm_mrc_qualifiers): ... this. New.
34568 (MCR_QUALIFIERS): Define to ...
34569 (arm_mcr_qualifiers): ... this. New.
34570 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
34572 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
34573 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
34574 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
34575 VUNSPEC_MRC2): New.
34577 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34579 * config/arm/arm.md (*ldc): New.
34583 * config/arm/arm.c (arm_coproc_builtin_available): Add
34584 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
34585 (arm_coproc_ldc_stc_legitimate_address): New.
34586 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
34587 'qualifier_const_pointer'.
34588 (LDC_QUALIFIERS): Define to...
34589 (arm_ldc_qualifiers): ... this. New.
34590 (STC_QUALIFIERS): Define to...
34591 (arm_stc_qualifiers): ... this. New.
34592 * config/arm/arm-protos.h
34593 (arm_coproc_ldc_stc_legitimate_address): New.
34594 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
34595 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
34596 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
34597 stc2, stcl, stc2l): New.
34598 * config/arm/constraints.md (Uz): New.
34599 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
34600 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
34601 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
34602 VUNSPEC_STC2L): New.
34604 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34606 * config/arm/arm.md (<cdp>): New.
34607 * config/arm/arm.c (neon_const_bounds): Rename this ...
34608 (arm_const_bounds): ... this.
34609 (arm_coproc_builtin_available): New.
34610 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
34611 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
34612 (CDP_QUALIFIERS): Define to...
34613 (arm_cdp_qualifiers): ... this. New.
34615 (arm_expand_builtin_args): Add case for 6 arguments.
34616 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
34617 (arm_const_bounds): ... this.
34618 (arm_coproc_builtin_available): New.
34619 * config/arm/arm_acle.h (__arm_cdp): New.
34621 * config/arm/arm_acle_builtins.def (cdp): New.
34623 * config/arm/iterators.md (CDPI,CDP,cdp): New.
34624 * config/arm/neon.md: Rename all 'neon_const_bounds' to
34625 'arm_const_bounds'.
34626 * config/arm/types.md (coproc): New.
34627 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
34628 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
34629 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
34630 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
34632 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34634 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
34635 (UBINOP_QUALIFIERS): New.
34637 (acle_builtin_data): New. Change comment.
34638 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
34639 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
34640 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
34641 arm_acle_builtins.def.
34642 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
34643 (arm_init_acle_builtins): New.
34644 (CRC32_BUILTIN): Remove.
34645 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
34646 crc32cb, crc32ch and crc32cw.
34647 (arm_init_crc32_builtins): Remove.
34648 (arm_init_builtins): Use arm_init_acle_builtins rather
34649 than arm_init_crc32_builtins.
34650 (arm_expand_acle_builtin): New.
34651 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
34652 * config/arm/arm_acle_builtins.def: New.
34654 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34656 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
34657 (arm_builtin_datum): ... this.
34658 (arm_init_neon_builtin): Rename to ...
34659 (arm_init_builtin): ... this. Add a new parameters PREFIX
34660 and USE_SIG_IN_NAME.
34661 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
34662 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
34663 'arm_builtin_datum'.
34664 (arm_init_vfp_builtins): Likewise.
34665 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
34666 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
34667 (arm_expand_neon_args): Rename to ...
34668 (arm_expand_builtin_args): ... this. Rename builtin_arg
34669 enum values and differentiate between ARG_BUILTIN_MEMORY
34670 and ARG_BUILTIN_NEON_MEMORY.
34671 (arm_expand_neon_builtin_1): Rename to ...
34672 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
34673 values, arm_expand_builtin_args and add bool parameter NEON.
34674 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
34675 (arm_expand_vfp_builtin): Likewise.
34676 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
34678 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34680 PR middle-end/77484
34681 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
34682 * predict.c (tree_estimate_probability_bb): Reverse direction of
34683 polymorphic call predictor.
34685 2017-01-06 David Malcolm <dmalcolm@redhat.com>
34687 * passes.c (execute_one_pass): Split out pass-skipping logic into...
34688 (determine_pass_name_match): ...this new function and...
34689 (should_skip_pass_p): ...this new function.
34691 2017-01-06 Nathan Sidwell <nathan@acm.org>
34693 * ipa-visibility.c (function_and_variable_visibility): Reformat
34694 comments and long lines. Remove extrneous if.
34695 * symtab.c (symtab_node::make_decl_local): Fix code format.
34696 (symtab_node::set_section_for_node): Fix comment typo.
34698 2017-01-06 Martin Liska <mliska@suse.cz>
34701 * lra-constraints.c: Rename invariant to lra_invariant.
34702 * predict.c (set_even_probabilities): Initialize e to NULL.
34704 2017-01-05 Martin Sebor <msebor@redhat.com>
34706 PR tree-optimization/78910
34707 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
34708 (format_integer): Correct off-by-one error in the handling
34709 of precision with negative numbers in signed conversions..
34711 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
34713 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
34715 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34717 PR tree-optimization/71016
34718 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
34719 factor_out_conditional_conversion. Formatting fix.
34720 (factor_out_conditional_conversion): Add cond_stmt argument.
34721 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
34722 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
34725 2017-01-05 David Malcolm <dmalcolm@redhat.com>
34727 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
34728 read-rtl-function.o, and selftest-rtl.o.
34729 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
34730 (selftest::aarch64_test_loading_full_dump): New function.
34731 (selftest::aarch64_run_selftests): New function.
34732 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
34733 selftest::aarch64_run_selftests.
34734 * config/i386/i386.c
34735 (selftest::ix86_test_loading_dump_fragment_1): New function.
34736 (selftest::ix86_test_loading_call_insn): New function.
34737 (selftest::ix86_test_loading_full_dump): New function.
34738 (selftest::ix86_test_loading_unspec): New function.
34739 (selftest::ix86_run_selftests): Call the new functions.
34740 * emit-rtl.c (maybe_set_max_label_num): New function.
34741 * emit-rtl.h (maybe_set_max_label_num): New decl.
34742 * function.c (instantiate_decls): Guard call to
34743 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
34744 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
34746 * gensupport.c (gen_reader::gen_reader): Pass "false"
34747 for new "compact" param of rtx_reader.
34748 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
34749 rather than an empty string for NULL strings.
34750 * read-md.c: Potentially include config.h rather than bconfig.h.
34751 Wrap include of errors.h with #ifdef GENERATOR_FILE.
34752 (have_error): New global, copied from errors.c.
34753 (md_reader::read_name): Rename to...
34754 (md_reader::read_name_1): ...this, adding "out_loc" param,
34755 and converting "missing name or number" to returning false, rather
34757 (md_reader::read_name): Reimplement in terms of read_name_1.
34758 (md_reader::read_name_or_nil): New function.
34759 (md_reader::read_string): Handle "(nil)" by returning NULL.
34760 (md_reader::md_reader): Add new param "compact".
34761 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
34762 (md_reader::read_file): New method.
34763 * read-md.h (md_reader::md_reader): Add new param "compact".
34764 (md_reader::read_file): New method.
34765 (md_reader::is_compact): New accessor.
34766 (md_reader::read_name): Convert return type from void to file_location.
34767 (md_reader::read_name_or_nil): New decl.
34768 (md_reader::read_name_1): New decl.
34769 (md_reader::m_compact): New field.
34770 (noop_reader::noop_reader): Pass "false" for new "compact" param
34772 (rtx_reader::rtx_reader): Add new "compact" param.
34773 (rtx_reader::read_rtx_operand): Make virtual and convert return
34774 type from void to rtx.
34775 (rtx_reader::read_until): New decl.
34776 (rtx_reader::handle_any_trailing_information): New virtual function.
34777 (rtx_reader::postprocess): New virtual function.
34778 (rtx_reader::finalize_string): New virtual function.
34779 (rtx_reader::m_in_call_function_usage): New field.
34780 (rtx_reader::m_reuse_rtx_by_id): New field.
34781 * read-rtl-function.c: New file.
34782 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
34783 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
34784 (selftest::verify_three_block_rtl_cfg): New decl.
34785 * read-rtl-function.h: New file.
34786 * read-rtl.c: Potentially include config.h rather than bconfig.h.
34787 For host, include function.h, memmodel.h, and emit-rtl.h.
34788 (one_time_initialization): New function.
34789 (struct compact_insn_name): New struct.
34790 (compact_insn_names): New array.
34791 (find_code): Handle insn codes in compact dumps.
34792 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
34793 (bind_subst_iter_and_attr): Likewise.
34794 (add_condition_to_string): Likewise.
34795 (add_condition_to_rtx): Likewise.
34796 (apply_attribute_uses): Likewise.
34797 (add_current_iterators): Likewise.
34798 (apply_iterators): Likewise.
34799 (initialize_iterators): Guard usage of apply_subst_iterator with
34800 #ifdef GENERATOR_FILE.
34801 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
34802 (md_reader::read_mapping): Likewise.
34803 (add_define_attr_for_define_subst): Likewise.
34804 (add_define_subst_attr): Likewise.
34805 (read_subst_mapping): Likewise.
34806 (check_code_iterator): Likewise.
34807 (rtx_reader::read_rtx): Likewise. Move one-time initialization
34809 (one_time_initialization): New function.
34810 (rtx_reader::read_until): New method.
34811 (read_flags): New function.
34812 (parse_reg_note_name): New function.
34813 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
34814 Handle reuse_rtx ids.
34815 Wrap iterator lookup within #ifdef GENERATOR_FILE.
34816 Add parsing support for RTL dumps, mirroring the special-cases in
34817 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
34818 values, and calling handle_any_trailing_information.
34819 (rtx_reader::read_rtx_operand): Convert return type from void
34820 to rtx, returning return_rtx. Handle case 'e'. Call
34821 finalize_string on XSTR and XTMPL fields.
34822 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
34823 "(nil)" values were omitted. Call the postprocess vfunc on the
34825 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
34826 class ctor. Initialize m_in_call_function_usage. Call
34827 one_time_initialization.
34828 * rtl-tests.c (selftest::test_uncond_jump): Call
34829 set_new_first_and_last_insn.
34830 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
34831 * selftest-rtl.c: New file.
34832 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
34833 (selftest::get_insn_by_uid): New decl.
34834 * selftest-run-tests.c (selftest::run_tests): Call
34835 read_rtl_function_c_tests.
34836 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
34837 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
34840 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
34842 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
34843 operands in a special way. Assert that pos+len <= mode precision.
34845 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34847 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
34848 3 argument Alias with unlimited for the negative form.
34849 (fno-vect-cost-model): Removed.
34851 2017-01-05 Martin Liska <mliska@suse.cz>
34853 * hsa-gen.c (gen_hsa_divmod): New function.
34854 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
34856 2017-01-05 Martin Liska <mliska@suse.cz>
34859 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
34862 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34864 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
34865 small constant length operands.
34867 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34869 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
34870 between loop iterations.
34872 2017-01-05 Martin Liska <mliska@suse.cz>
34875 * gimplify.c (gimplify_decl_expr): Compare to
34876 asan_poisoned_variables instread of checking flags.
34877 (gimplify_target_expr): Likewise.
34878 (gimplify_expr): Likewise.
34879 (gimplify_function_tree): Conditionally initialize
34880 asan_poisoned_variables.
34882 2017-01-04 Jeff Law <law@redhat.com>
34884 PR tree-optimizatin/78812
34885 * rtl.h (contains_mem_rtx_p): Prototype.
34886 * ifcvt.c (containts_mem_rtx_p): Move from here to...
34887 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
34888 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
34889 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
34890 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
34892 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34894 * input.c (assert_char_at_range): Default-initialize actual_range.
34896 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34898 * df-scan.c (df_ref_create_structure): Make regno unsigned,
34899 to match the caller.
34901 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34903 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
34904 insns after final jump in test to emit dummy move.
34906 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34908 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
34909 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
34911 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34913 * multiple_target.c (create_dispatcher_calls): Init e_next.
34914 * tree-ssa-loop-split.c (split_loop): Init border.
34915 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
34918 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
34923 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
34924 (altivec_register_operand): Do not return true if the operand
34925 contains a SUBREG mixing SImode and SFmode.
34926 (vsx_register_operand): Likewise.
34927 (vsx_reg_sfsubreg_ok): New predicate.
34928 (vfloat_operand): Do not return true if the operand contains a
34929 SUBREG mixing SImode and SFmode.
34930 (vint_operand): Likewise.
34931 (vlogical_operand): Likewise.
34932 (gpc_reg_operand): Likewise.
34933 (int_reg_operand): Likewise.
34934 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
34935 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
34936 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
34938 (rs6000_emit_move_si_sf_subreg): New helper function.
34939 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
34940 fixup SUBREGs involving SImode and SFmode.
34941 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
34942 numbers for the new peephole2 optimization.
34943 (peephole2 for SFmode unions): New peephole2 to optimize cases in
34944 the GLIBC math library that do AND/IOR/XOR operations on single
34945 precision floating point.
34946 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
34947 target macros to say whether we need to avoid SUBREGs mixing
34949 (TARGET_ALLOW_SF_SUBREG): Likewise.
34950 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
34951 (UNSPEC_SI_FROM_SF): Likewise.
34952 (iorxor): Change spacing.
34953 (and_ior_xor): New iterator for AND, IOR, and XOR.
34954 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
34955 (movdi_from_sf_zero_ext): Likewise.
34956 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
34957 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
34958 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
34959 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
34960 (fms<mode>4): Likewise.
34961 (fnma<mode>4): Likewise.
34962 (fnms<mode>4): Likewise.
34963 (nfma<mode>4): Likewise.
34964 (nfms<mode>4): Likewise.
34966 2017-01-04 Marek Polacek <polacek@redhat.com>
34969 * doc/invoke.texi: Document -Wpointer-compare.
34971 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34973 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
34976 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
34977 descriptions for -gdwarf-5 and emit them as uleb128 instead of
34980 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
34983 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
34984 documentation of the powerpc_popcntb_ok attribute.
34985 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34986 code to issue warning messages if a requested CPU configuration is
34987 not supported by the binary (assembler and loader) toolchain.
34988 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
34989 made to define a built-in function that has been disabled.
34990 (paired_init_builtins): Add assertion to prevent ICE if attempt is
34991 made to define a built-in function that has been disabled.
34992 (altivec_init_builtins): Add comment explaining why definition
34993 of the DST built-in functions is not preceded by an assertion
34994 check. Add assertions to prevent ICE if attempts are made to
34995 define an altivec predicate or an abs* built-in function that has
34997 (htm_init_builtins): Add comment explaining why definition of the
34998 htm built-in functions is not preceded by an assertion check.
35000 2017-01-04 Jeff Law <law@redhat.com>
35002 PR tree-optimizatin/67955
35003 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
35004 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
35005 the points-to solution does not include pt_null. Use DECL_PT_UID
35008 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
35010 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
35011 Use gen_int_mode instead of gen_lopwart for const_int operands.
35013 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35015 PR tree-optimization/71563
35016 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35017 out of range value - has low bits known to be zero.
35019 2017-01-04 Alan Modra <amodra@gmail.com>
35021 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35022 * configure: Regenerate.
35023 * config.in: Regenerate.
35025 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35028 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35029 a substring of the message, but strcmp with the whole message. Ifdef
35030 ENABLE_NLS, translate the message first using dgettext.
35032 2017-01-03 Jeff Law <law@redhat.com>
35034 PR tree-optimizatin/78856
35035 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35036 (mark_threaded_blocks): Remove code to truncate thread paths that
35037 cross multiple loop headers. Instead invalidate the cached loop
35038 iteration information and handle case of a thread path walking
35039 into an irreducible region.
35041 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35044 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35045 assertions. Add support for doing the signbit if the IEEE 128-bit
35046 floating point value is in a GPR.
35047 * config/rs6000/rs6000.md (Fsignbit): Delete.
35048 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35049 Update the length attribute if the value is in a GPR.
35050 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35051 the sign or zero extension instruction, since the value is always 0/1.
35052 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35055 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35056 extracting SImode to a GPR register so that we can generate a
35057 store, limit the vector to be in a traditional Altivec register
35058 for the vextuwrx instruction.
35060 2017-01-03 Ian Lance Taylor <iant@google.com>
35062 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35064 2017-01-03 Martin Sebor <msebor@redhat.com>
35066 PR tree-optimization/78696
35067 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35068 precision. Use MPFR for %f for greater fidelity. Correct handling
35070 (pass_sprintf_length::compute_format_length): Set width and precision
35071 specified by asrerisk to void_node for vararg functions.
35072 (try_substitute_return_value): Adjust dump output.
35074 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35076 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35078 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35080 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35081 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35082 -mlra/-mno-lra was passed to the compiler.
35084 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35086 PR rtl-optimization/65618
35087 * emit-rtl.c (try_split): Move initialization of "before" and
35088 "after" to just before the call to emit_insn_after_setloc.
35090 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35092 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35094 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35096 * dwarf2out.c (gen_enumeration_type_die): When
35097 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35099 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35101 PR tree-optimization/78965
35102 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35103 Change first argument from const call_info & to call_info &. For %n
35104 set info.nowrite to false.
35106 PR middle-end/78901
35107 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35108 possibly throwing calls.
35110 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35111 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35112 and fns handling, rather than in a separate case SSA_NAME.
35114 2017-01-02 Jeff Law <law@redhat.com>
35116 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35117 fixes for first_period and second_period variables.
35119 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35122 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35123 (*insvqi_1): New insn pattern.
35124 (*insvqi_1_mem_rex64): Ditto.
35125 (*insvqi_2): Ditto.
35126 (*insvqi_3): Rename from *insvqi.
35128 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35130 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35132 * doc/cfg.texi (Edges): Remove reference to Java.
35133 (Maintaining the CFG): Ditto.
35135 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35137 PR middle-end/77674
35138 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35139 transparent aliases.
35141 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35143 PR middle-end/77484
35144 * predict.def (PRED_CALL): Update hitrate.
35145 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35146 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35147 into direct/indirect/polymorphic variants.
35149 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35151 Update copyright years.
35153 * gcc.c (process_command): Update copyright notice dates.
35154 * gcov-dump.c (print_version): Ditto.
35155 * gcov.c (print_version): Ditto.
35156 * gcov-tool.c (print_version): Ditto.
35157 * gengtype.c (create_file): Ditto.
35158 * doc/cpp.texi: Bump @copying's copyright year.
35159 * doc/cppinternals.texi: Ditto.
35160 * doc/gcc.texi: Ditto.
35161 * doc/gccint.texi: Ditto.
35162 * doc/gcov.texi: Ditto.
35163 * doc/install.texi: Ditto.
35164 * doc/invoke.texi: Ditto.
35166 Copyright (C) 2017 Free Software Foundation, Inc.
35168 Copying and distribution of this file, with or without modification,
35169 are permitted in any medium without royalty provided the copyright
35170 notice and this notice are preserved.