1 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3 PR tree-optimization/82726
4 PR tree-optimization/70754
5 * tree-predcom.c (order_drefs_by_pos): New function.
6 (combine_chains): Move code setting has_max_use_after to...
7 (try_combine_chains): ...here. New parameter. Sort combined chains
8 according to position information.
9 (tree_predictive_commoning_loop): Update call to above function.
10 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
13 2017-11-15 Bin Cheng <bin.cheng@arm.com>
15 PR tree-optimization/82726
16 * gcc.dg/tree-ssa/pr82726.c: New test.
18 2017-11-15 Bin Cheng <bin.cheng@arm.com>
20 PR tree-optimization/82726
22 2017-01-23 Bin Cheng <bin.cheng@arm.com>
24 PR tree-optimization/70754
25 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
26 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
27 combined stmt before it if not NULL.
28 (combine_chains): Process refs reversely and compute dominance point
32 2017-02-23 Bin Cheng <bin.cheng@arm.com>
34 PR tree-optimization/79663
35 * tree-predcom.c (combine_chains): Process refs in reverse order
36 only for ZERO length chains, and add explaining comment.
38 2017-11-15 Tamar Christina <tamar.christina@arm.com>
40 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
41 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
42 * doc/invoke.texi (ARM Options): Add armv8.3-a.
44 2017-11-15 Tamar Christina <tamar.christina@arm.com>
46 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
48 2017-11-15 Martin Liska <mliska@suse.cz>
50 * tree-cfg.c (pass_warn_function_return::execute):
51 Compare warn_return_type for greater than zero.
53 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
57 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
58 to return true on Xeon and not on Xeon Phi.
59 (ix86_check_avx256_register): Changed to ...
60 (ix86_check_avx_upper_register): ... this. Add extra check for
61 VALID_AVX512F_REG_OR_XI_MODE.
62 (ix86_avx_u128_mode_needed): Changed
63 ix86_check_avx256_register to ix86_check_avx_upper_register.
64 (ix86_check_avx256_stores): Changed to ...
65 (ix86_check_avx_upper_stores): ... this. Changed
66 ix86_check_avx256_register to ix86_check_avx_upper_register.
67 (ix86_avx_u128_mode_after): Changed
68 avx_reg256_found to avx_upper_reg_found. Changed
69 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
70 (ix86_avx_u128_mode_entry): Changed
71 ix86_check_avx256_register to ix86_check_avx_upper_register.
72 (ix86_avx_u128_mode_exit): Ditto.
73 * config/i386/i386.h: (host_detect_local_cpu): New define.
75 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
77 * config/arm/xgene1.md (xgene1): Split into automatons
78 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
79 (xgene1_f_load): Adjust reservations and/or types.
80 (xgene1_f_store): Likewise.
81 (xgene1_load_pair): Likewise.
82 (xgene1_store_pair): Likewise.
83 (xgene1_fp_load1): Likewise.
84 (xgene1_load1): Likewise.
85 (xgene1_store1): Likewise.
86 (xgene1_move): Likewise.
87 (xgene1_alu): Likewise.
88 (xgene1_simd): Likewise.
89 (xgene1_bfm): Likewise.
90 (xgene1_neon_load1): Likewise.
91 (xgene1_neon_store1): Likewise.
92 (xgene1_neon_logic): Likewise.
93 (xgene1_neon_st1): Likewise.
94 (xgene1_neon_ld1r): Likewise.
95 (xgene1_alu_cond): Added.
96 (xgene1_shift_reg): Likwise.
97 (xgene1_bfx): Likewise.
98 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
100 2017-11-15 Jakub Jelinek <jakub@redhat.com>
103 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
105 (can_widen_mult_without_libcall): New function.
106 (expand_mul_overflow): If only checking unsigned mul overflow,
107 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
108 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
109 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
110 (expand_DIVMOD): Formatting fix.
111 * expmed.h (expand_mult): Add NO_LIBCALL argument.
112 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
113 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
115 PR tree-optimization/82977
116 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
117 constructed temporary to strlen_to_stridx.put.
119 2017-11-15 Martin Liska <mliska@suse.cz>
121 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
122 * configure: Regenerate.
124 2017-11-15 Martin Liska <mliska@suse.cz>
127 * config/sh/sh-mem.cc: Use proper probability for
130 2017-11-14 Jeff Law <law@redhat.com>
132 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
133 the red zone for stack_clash_protection_final_dynamic_probe targets
134 when the total dynamic stack size is zero bytes.
136 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
137 blocks is post order.
139 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
141 * dumpfile.h (TDF_COMPARE_DEBUG): New.
142 * final.c (rest_of_clean_state): Set it for the
143 -fcompare-debug dump.
144 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
145 class when TDF_COMPARE_DEBUG is set.
147 * dwarf2out.c (gen_producer_string): Discard
150 2017-11-15 Joseph Myers <joseph@codesourcery.com>
153 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
154 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
155 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
156 (__TGMATH_REAL_2_3): Remove macros.
157 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
158 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
160 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
162 (remquo): Define using __TGMATH_REAL_3.
164 2017-11-14 Jeff Law <law@redhat.com>
166 * vr-values.c: New file with contents extracted from tree-vrp.c.
167 * Makefile.in (OBJS): Add vr-values.o
168 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
169 (set_value_range, set_and_canonicalize_value_range): Likewise.
170 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
171 (value_range_constant_singleton, symbolic_range_p): Likewise.
172 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
173 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
174 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
175 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
176 (find_case_label_range, find_case_label_index): Likewise.
177 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
178 (range_int_cst_singleton_p, value_inside_range): Likewise.
179 (get_single_symbol): Likewise.
180 (switch_update): Move structure definition here.
181 (to_remove_edges, to_update_switch_stmts): Provide externs.
182 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
183 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
184 (vrp_val_is_min, set_value_range): Likewise.
185 (set_and_canonicalize_value_range, copy_value_range): Likewise.
186 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
187 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
188 (range_is_nonnull, range_int_cst_p): Likewwise.
189 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
190 (get_single_symbol, operand_less_p): Likewise
191 (compare_values_warnv, compare_values): Likewise.
192 (value_inside_range, value_range_constant_singleton): Likewise.
193 (zero_nonzero_bitgs_from_vr): Likewise.
194 (extract_range_from_binary_expr_1): Likewise.
195 (overflow_comparison_p): Likewise.
196 (to_remove_edges, to_update_switch_stmts): Likewise.
197 (find_case_label-index, find_case_label_range): Likewise.
198 (switch_update, set_value_range_to_nonnegative): Remove.
199 (set_value_range_to_truthvalue): Likewise.
200 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
201 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
202 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
203 (find_case_label_ranges, test_for_singularity): Likewise.
204 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
205 (x_vr_values): Move to its remaining use site.
207 2017-11-10 Jeff Law <law@redhat.com>
209 * vr-values.h (VR_INITIALIZER): Move #define here.
210 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
211 * Makefile.in (OBJS): Add tree-evrp.o
212 * tree-vrp.h (assert_info): Move structure definition here.
213 (set_value_range_to_varying): Prototype.
214 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
215 (infer_value_range, register_edge_assert_for): Likewise.
216 (stmt_interesting_for_vrp): Likewise.
217 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
218 (set_value_range_to_varying): No longer static.
219 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
220 (infer_value_range, register_edge_assert_for): Likewise.
222 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
224 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
225 generate the XXBRD instruction.
227 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
228 (rs6000_builtin_type_compatible): Treat _Float128 and long double
229 as being compatible if -mabi=ieeelongdouble.
230 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
231 to setup float128 built-ins with hardware support.
232 (BU_FLOAT128_HW_2): Likewise.
233 (BU_FLOAT128_HW_3): Likewise.
234 (BU_FLOAT128_HW_VSX_1): Likewise.
235 (BU_FLOAT128_HW_VSX_2): Likewise.
236 (scalar_extract_expq): Change float128 built-in functions to
237 accommodate having both KFmode and TFmode functions. Use the
238 KFmode variant as the default.
239 (scalar_extract_sigq): Likewise.
240 (scalar_test_neg_qp): Likewise.
241 (scalar_insert_exp_q): Likewise.
242 (scalar_insert_exp_qp): Likewise.
243 (scalar_test_data_class_qp): Likewise.
244 (sqrtf128_round_to_odd): Delete processing the round to odd
245 built-in functions as special built-in functions, and define them
246 as float128 built-ins. Use the KFmode variant as the default.
247 (truncf128_round_to_odd): Likewise.
248 (addf128_round_to_odd): Likewise.
249 (subf128_round_to_odd): Likewise.
250 (mulf128_round_to_odd): Likewise.
251 (divf128_round_to_odd): Likewise.
252 (fmaf128_round_to_odd): Likewise.
253 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
254 support for KFmode and TFmode xststdcqp calls.
255 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
256 point, switch the built-in handlers for the get/set float128
257 exponent, get float128 mantissa, float128 test built-ins, and the
258 float128 round to odd built-in functions. Eliminate creating the
259 float128 round to odd built-in functions as special built-ins.
260 (rs6000_init_builtins): Eliminate special creation of the float128
261 round to odd built-in functions.
262 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
263 function insns to support both TFmode and KFmode variants.
264 (xsxsigqp_<mode>): Likewise.
265 (xsiexpqpf_<mode>): Likewise.
266 (xsiexpqp_<mode>): Likewise.
267 (xststdcqp_<mode>): Likewise.
268 (xststdcnegqp_<mode>): Likewise.
269 (xststdcqp_<mode>): Likewise.
271 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
273 * tree-ssa-threadupdate.c (compute_path_counts): Remove
274 unused path_in_freq_ptr parameter.
275 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
277 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
279 * ipa-inline.c (edge_badness): Dump sreal frequency.
280 (compute_inlined_call_time): Match natural implementaiton ...
281 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
282 forgotten division by CGRAPH_FREQ_BASE.
284 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
286 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
287 Solaris 11. Update comment.
288 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
290 * config/sol2.h (STARTFILE_SPEC): Likewise.
291 * configure: Regenerate.
293 2017-11-14 Carl Love <cel@us.ibm.com>
295 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
296 le_ and be_ prefixes to swap* variables. Remove
297 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
300 2017-11-14 Jason Merrill <jason@redhat.com>
302 Support GTY((cache)) on hash_map.
303 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
304 (ggc_cache_remove): Override it instead of ggc_mx.
305 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
306 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
307 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
308 (simple_cache_map_traits): Override maybe_mx.
309 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
310 (hash_map): Friend gt_cleare_cache.
311 (gt_cleare_cache): New.
312 * tree.h (tree_cache_traits): New hash_map traits class.
313 (tree_cache_map): New typedef.
315 2017-11-14 Richard Biener <rguenther@suse.de>
317 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
318 paramter and handling.
319 (cleanup_control_flow_bb): Likewise.
320 (cleanup_control_flow_pre): New helper performing a DFS walk
321 to call cleanup_control_flow_bb in PRE order.
322 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
323 via cleanup_control_flow_pre.
325 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
327 * config/aarch64/aarch64-simd.md
328 (aarch64_simd_bsl<mode>_internal): Remove DImode.
329 (*aarch64_simd_bsl<mode>_alt): Likewise.
330 (aarch64_simd_bsldi_internal): New.
331 (aarch64_simd_bsldi_alt): Likewise.
333 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
335 * tracer.c (better_p): Do not compare frequencies.
336 * reg-stack.c (better_edge): Likewise.
337 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
340 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
342 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
343 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
344 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
345 * lto-streamer-in.c (input_function): Use update_max_bb_count.
346 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
347 * predict.c (maybe_hot_frequency_p): Inline to ...
348 (maybe_hot_count_p): ... here; rewrite to counts.
349 (counts_to_freqs): Rename to ...
350 (update_max_bb_count): ... this one.
351 (expensive_function_p): Use counts.
352 (estimate_bb_frequencies): Update.
353 (rebuild_frequencies): Update.
354 * predict.h (counts_to_freqs): Rename to ...
355 (update_max_bb_count): ... this one.
356 * profile.c (compute_branch_probabilities): Add debug info
357 * tree-inline.c (expand_call_inline): Update debug info.
358 (optimize_inline_calls): Use update_max_bb_count..
359 (tree_function_versioning): Use update_max_bb_count..
360 * value-prof.c (gimple_value_profile_transformations):
361 Do not use update_max_bb_count.
363 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
365 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
366 always use frequencies.
368 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
370 * bb-reorder.c: Remove frequencies from comments.
371 (better_edge_p): Use profile counts.
372 (find_traces): Dump profile counts.
373 (rotate_loop): Use profile counts.
374 (find_traces_1_round): Likewise.
375 (connect_better_edge_p): Use counts instead of probabilities for
377 (copy_bb_p): Drop early check for non-0 frequency.
378 (sanitize_hot_paths): Update comments.
380 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
382 * ipa-split.c (struct split_point): Add count.
383 (consider_split): Do not compute incoming frequency; compute incoming
384 count and store it to split_point.
385 (split_function): Set count of the call to split part correctly.
387 2017-11-13 Carl Love <cel@us.ibm.com>
389 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
391 2017-11-13 Tom Tromey <tom@tromey.com>
393 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
395 2017-11-13 Carl Love <cel@us.ibm.com>
397 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
398 Add support for builtins:
399 unsigned int vec_first_{,miss}_match_{,or_eos}index,
400 vector {un,}signed {char,int,short},
401 vector {un,}signed {char,int,short}) arguments.
402 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
403 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
404 Add BU_P9V_AV_2 expansions for the builtins.
405 * config/rs6000/altivec.h (vec_first_match_index,
406 vec_first_mismatch_index, vec_first_match_or_eos_index,
407 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
408 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
409 new extern declaration.
410 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
411 * config/rs6000/vsx.md (first_match_index_<mode>,
412 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
413 first_mismatch_or_eos_index_<mode>): Add define expand.
414 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
415 * doc/extend.texi: Update the built-in documenation file for the new
418 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
420 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
421 into the min/max operations for _Float<N> and _Float<N>X types.
423 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
426 * dwarf2out.c (do_eh_frame): New static variable.
427 (dwarf2out_begin_prologue): Set it.
428 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
430 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
432 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
434 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
436 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
439 * debug.h (dwarf2out_do_eh_frame): Declare.
440 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
441 (dwarf2out_do_frame): Use it.
442 (dwarf2out_do_cfi_asm): Likewise.
443 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
444 (dwarf2out_assembly_start): Likewise.
445 (dwarf2out_begin_prologue): Fix comment.
446 * toplev.c (compile_file): Always call dwarf2out_frame_finish
447 if the target needs either debug or unwind DWARF2 info.
448 * lto-opts.c (lto_write_options): Do not save -fexceptions,
449 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
450 -ftrapping-math, -ftrapv and -fwrapv.
452 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
454 * cgraph.c (cgraph_edge::sreal_frequency): New function.
455 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
456 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
457 (estimate_edge_size_and_time): Likewise.
458 (ipa_merge_fn_summary_after_inlining): Likewise.
459 * ipa-inline.c (cgraph_freq_base_rec): Remove.
460 (compute_uninlined_call_time): Use sreal_frequency.
461 (compute_inlined_call_time): Likewise.
462 (ipa_inline): Do not initialize cgraph_freq_base_rec.
463 * profile-count.c: Include sreal.h.
464 (profile_count::to_sreal_scale): New.
465 * profile-count.h: Forward declare sreal.
466 (profile_count::to_sreal_scale): Declare.
468 2017-11-13 Nathan Sidwell <nathan@acm.org>
470 * diagnostic.c (maybe_line_and_column): New.
471 (diagnostic_get_location_text): Use it.
472 (diagnostic_report_current_module): Likewise.
473 (test_diagnostic_get_location_text): Add tests.
475 2017-11-13 Luis Machado <luis.machado@linaro.org>
477 * doc/md.texi (Specifying processor pipeline description): Fix
478 incorrect latency for the div instruction example.
480 2017-11-13 Jakub Jelinek <jakub@redhat.com>
482 PR tree-optimization/78821
483 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
484 that bit_not_p is the same.
485 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
486 (split_group): Count precisely bit_not_p bits in each statement.
487 (invert_op): New function.
488 (imm_store_chain_info::output_merged_store): Use invert_op to
489 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
490 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
492 2017-11-13 Martin Liska <mliska@suse.cz>
494 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
495 (struct source_info): Likewise.
496 (add_branch_counts): Likewise.
497 (add_line_counts): Likewise.
498 (function_summary): Likewise.
499 (output_intermediate_line): Likewise.
500 (generate_results): Likewise.
502 2017-11-13 Martin Liska <mliska@suse.cz>
504 * gcov.c (struct block_info): Remove typedef for block_t.
505 (struct line_info): Likewise.
506 (line_info::has_block): Likewise.
507 (EXIT_BLOCK): Likewise.
510 (get_cycles_count): Likewise.
511 (process_file): Likewise.
512 (read_graph_file): Likewise.
513 (solve_flow_graph): Likewise.
514 (find_exception_blocks): Likewise.
515 (add_line_counts): Likewise.
516 (accumulate_line_info): Likewise.
517 (output_line_details): Likewise.
519 2017-11-13 Martin Liska <mliska@suse.cz>
521 * gcov.c (struct arc_info): Remove typedef for arc_t.
522 (struct line_info): Likewise.
523 (add_branch_counts): Likewise.
524 (output_branch_count): Likewise.
525 (function_info::~function_info): Likewise.
527 (output_intermediate_line): 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.
534 (output_function_details): Likewise.
536 2017-11-13 Martin Liska <mliska@suse.cz>
538 * gcov.c (struct function_info): Remove typedef for function_t.
539 (struct source_info): Likewise.
540 (source_info::get_functions_at_location): Likewise.
541 (solve_flow_graph): Likewise.
542 (find_exception_blocks): Likewise.
543 (add_line_counts): Likewise.
544 (output_intermediate_file): Likewise.
545 (process_file): Likewise.
546 (generate_results): Likewise.
547 (release_structures): Likewise.
548 (read_graph_file): Likewise.
549 (read_count_file): Likewise.
550 (accumulate_line_counts): Likewise.
551 (output_lines): Likewise.
553 2017-11-13 Martin Liska <mliska@suse.cz>
555 * gcov.c (function_info::function_info): Remove num_counts
556 and add vector<gcov_type>.
557 (function_info::~function_info): Use the vector.
558 (process_file): Likewise.
559 (read_graph_file): Likewise.
560 (read_count_file): Likewise.
561 (solve_flow_graph): Likewise.
563 2017-11-13 Martin Liska <mliska@suse.cz>
565 * gcov.c (function_info::is_artificial): New function.
566 (process_file): Erase all artificial early.
567 (generate_results): Skip as all artificial are already
570 2017-11-13 Martin Liska <mliska@suse.cz>
572 * gcov.c (read_graph_file): Store to global vector of functions.
573 (read_count_file): Iterate the vector.
574 (process_file): Likewise.
575 (generate_results): Likewise.
576 (release_structures): Likewise.
578 2017-11-13 Jakub Jelinek <jakub@redhat.com>
580 PR tree-optimization/82954
581 * gimple-ssa-store-merging.c
582 (imm_store_chain_info::coalesce_immediate_stores): If
583 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
585 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
587 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
588 Upddate call to ENDIAN_LANE_N.
589 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
590 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
591 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
592 and use aarch64_endian_lane_rtx.
593 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
595 2017-11-12 Tom de Vries <tom@codesourcery.com>
597 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
599 2017-11-12 Tom de Vries <tom@codesourcery.com>
601 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
603 2017-11-12 Tom de Vries <tom@codesourcery.com>
605 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
607 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
608 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
609 * defaults.h (ASM_OUTPUT_LABELREF): Same.
611 2017-11-11 Martin Sebor <msebor@redhat.com>
614 * doc/extend.texi (attribute nonstring): Remove spurious argument.
617 * prefic.c (translate_name): Replace strncpy with memcpy to
618 avoid -Wstringop-truncation.
620 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
622 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
624 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
626 * predict.c (maybe_hot_frequency_p): Do not use cfun.
628 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
630 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
633 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
635 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
636 enable generating XXBRH if the value is in a vector register.
637 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
638 value is in a vector register.
639 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
640 register to register bswap64's instead of doing the GPR sequence
641 used on previous machines.
642 (bswapdi2_xxbrd): New insn.
643 (bswapdi2_reg): Disallow on ISA 3.0.
644 (register to register bswap64 splitter): Do not split the insn on
645 ISA 3.0 systems that use XXBRD.
647 2017-11-10 Martin Sebor <msebor@redhat.com>
650 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
652 (find_subframework_file): Same.
654 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
656 * auto-profile.c (afdo_indirect_call): Drop frequency.
657 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
658 (cgraph_node::create_edge): Drop frequency argument.
659 (cgraph_node::create_indirect_edge): Drop frequency argument.
660 (cgraph_edge::make_speculative): Drop frequency arguments.
661 (cgraph_edge::resolve_speculation): Do not update frequencies
662 (cgraph_edge::dump_edge_flags): Do not dump frequency.
663 (cgraph_node::dump): Check consistency in IPA mode.
664 (cgraph_edge::maybe_hot_p): Use IPA counter.
665 (cgraph_edge::verify_count_and_frequency): Rename to ...
666 (cgraph_edge::verify_count): ... this one; drop frequency checking.
667 (cgraph_node::verify_node): Update.
668 * cgraph.h (struct cgraph_edge): Drop frequency.
669 (cgraph_edge::frequency): New function.
670 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
672 (cgraph_edge::rebuild_edges): Likewise.
673 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
674 (duplicate_thunk_for_node): Do not pass frequency.
675 (cgraph_node::create_clone): Scale only counts.
676 (cgraph_node::create_virtual_clone): Do not pass frequency.
677 (cgraph_node::create_edge_including_clones): Do not pass frequency.
678 (cgraph_node::create_version_clone): Do not pass frequency.
679 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
680 (cgraph_node::expand_thunk): Do not pass frequency.
681 (cgraph_node::create_wrapper): Do not pass frequency.
682 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
684 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
685 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
686 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
687 (gather_caller_stats): Use frequency function.
688 (ipcp_cloning_candidate_p): Use frequency function.
689 (ipcp_propagate_stage): Use frequency function.
690 (get_info_about_necessary_edges): Use frequency function.
691 (update_profiling_info): Update only IPA profile.
692 (update_specialized_profile): Use frequency functoin.
693 (perhaps_add_new_callers): Update only IPA profile.
694 * ipa-devirt.c (ipa_devirt): Use IPA profile.
695 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
696 (dump_ipa_call_summary): Use frequency function.
697 (estimate_edge_size_and_time): Use frequency function.
698 (ipa_merge_fn_summary_after_inlining): Use frequency function.
699 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
700 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
701 (update_noncloned_counts): ... ths one; scale counts only.
702 (clone_inlined_nodes): Do not scale frequency.
703 (inline_call): Do not pass frequency.
704 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
705 (compute_inlined_call_time): Use IPA profile.
706 (want_inline_small_function_p): Use IPA profile.
707 (want_inline_self_recursive_call_p): Use IPA profile.
708 (edge_badness): Use IPA profile.
709 (lookup_recursive_calls): Use IPA profile.
710 (recursive_inlining): Do not pass frequency.
711 (resolve_noninline_speculation): Do not update frequency.
712 (inline_small_functions): Collect max of IPA profile.
713 (dump_overall_stats): Dump IPA porfile.
714 (dump_inline_stats): Dump IPA porfile.
715 (ipa_inline): Collect IPA stats.
716 * ipa-inline.h (clone_inlined_nodes): Update prototype.
717 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
718 (ipa_propagate_frequency): Use frequency function.
719 (ipa_profile): Cleanup.
720 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
721 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
722 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
723 (input_node): Do not stream frequency.
724 (input_edge): Do not stream frequency.
725 (merge_profile_summaries): Scale only IPA profiles.
726 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
727 * predict.c (drop_profile): Do not recompute frequency.
728 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
729 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
730 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
731 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
732 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
733 * tree-inline.c (copy_bb): Do not scale frequency.
734 (expand_call_inline): Do not scale frequency.
735 (tree_function_versioning): Do not scale frequency.
736 * ubsan.c (ubsan_create_edge): Do not pass frequency.
738 2017-11-10 Julia Koval <julia.koval@intel.com>
740 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8,
741 _mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8,
742 _mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8,
743 _mm256_maskz_gf2p8affine_epi64_epi8,
744 _mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8
745 _mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
746 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi,
747 __builtin_ia32_vgf2p8affineqb_v32qi,
748 __builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
749 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
751 2017-11-10 Tamar Christina <tamar.christina@arm.com>
755 (arm_option_override): Refactor.
756 (arm_option_reconfigure_globals): New.
757 (arm_options_perform_arch_sanity_checks): New.
758 * config/arm/arm-protos.h (arm_option_reconfigure_globals): New prototype.
759 (arm_options_perform_arch_sanity_checks): Likewise
761 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
763 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
765 * rs6000/rs6000.md ("type" attr): Add qmul.
766 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
767 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
768 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
770 2017-11-10 Martin Sebor <msebor@redhat.com>
773 * builtins.c (compute_objsize): Handle arrays that
774 compute_builtin_object_size likes to fail for. Make extern.
775 * builtins.h (compute_objsize): Declare.
776 (check_strncpy_sizes): New function.
777 (expand_builtin_strncpy): Call check_strncpy_sizes.
778 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
779 -Wstringop-truncation.
780 (gimple_fold_builtin_strncat): Same.
781 * gimple.c (gimple_build_call_from_tree): Set call location.
782 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
783 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
784 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
785 (handle_builtin_strlen): Use strlen_to_stridx.
786 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
788 Use strlen_to_stridx.
789 (pass_strlen::execute): Release strlen_to_stridx.
790 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
791 (-Wstringop-truncation): Document new option.
793 2017-11-10 Martin Liska <mliska@suse.cz>
795 PR gcov-profile/82702
796 * gcov.c (main): Handle intermediate files in a different
798 (get_gcov_intermediate_filename): New function.
799 (output_gcov_file): Remove support of intermediate files.
800 (generate_results): Allocate intermediate file.
801 (release_structures): Clean-up properly fn_end.
802 (output_intermediate_file): Start iterating with line 1.
804 2017-11-10 Jakub Jelinek <jakub@redhat.com>
806 PR tree-optimization/82929
807 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
808 ops_swapped_p non-static data member.
809 (store_immediate_info::store_immediate_info): Clear it.
810 (imm_store_chain_info::coalesce_immediate_stores): If swapping
811 ops set ops_swapped_p.
812 (count_multiple_uses): Handle ops_swapped_p.
814 2017-11-10 Martin Liska <mliska@suse.cz>
816 * coverage.c (coverage_init): Stream information about
817 support of has_unexecuted_blocks.
818 * doc/gcov.texi: Document that.
819 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
820 * gcov.c (read_graph_file): Likewise.
821 (output_line_beginning): Fix a small issue with
824 2017-11-10 Bin Cheng <bin.cheng@arm.com>
826 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
827 reference of trivial component.
829 2017-11-10 Jakub Jelinek <jakub@redhat.com>
832 * gimple-ssa-store-merging.c
833 (pass_store_merging::terminate_all_aliasing_chains): For
834 gimple_store_p stmts also call refs_output_dependent_p.
836 PR rtl-optimization/82913
837 * compare-elim.c (try_merge_compare): Punt if def_insn is not
840 2017-11-09 Jeff Law <law@redhat.com>
842 * vr-values.h: New file with vr_values class.
843 * tree-vrp.c: Include vr-values.h
844 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
845 data objects into the vr_values class.
846 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
847 (get_value_range): Make it a member function within vr_values class.
848 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
849 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
850 (op_with_constant_singleton_value_range): Likewise.
851 (extract_range_for_var_from_comparison_expr): Likewise.
852 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
853 (extract_range_from_binary_expr): Likewise.
854 (extract_range_from_unary_expr): Likewise.
855 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
856 (check_for_binary_op_overflow, extract_range_basic): Likewise.
857 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
858 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
859 (compare_name_with_value, compare_names): Likewise.
860 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
861 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
862 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
863 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
864 (extract_range_from_phi_node): Likewise.
865 (simplify_truth_ops_using_ranges): Likewise.
866 (simplify_div_or_mod_using_ranges): Likewise.
867 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
868 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
869 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
870 (simplify_float_conversion_using_ranges): Likewise.
871 (simplify_internal_call_using_ranges): Likewise.
872 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
873 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
874 poitner via x_vr_values for calls into gimple folder.
875 (vrp_initialize_lattice): Make this the vr_values ctor.
876 (vrp_free_lattice): Make this the vr_values dtor.
877 (set_vr_value): New function.
878 (class vrp_prop): Add vr_values data member. Add various member
879 functions as well as member functions that delegate to vr_values.
880 (check_array_ref): Make a member function within vrp_prop class.
881 (search_for_addr_array, vrp_initialize): Likewise.
882 (vrp_finalize): Likewise. Revamp to avoid direct access to
883 vr_value, values_propagated, etc.
884 (check_array_bounds): Extract vrp_prop class instance pointer from
885 walk info structure. Use it to call member functions.
886 (check_all_array_refs): Make a member function within vrp_prop class.
887 Smuggle class instance pointer via walk info structure.
888 (x_vr_values): New local static.
889 (vrp_valueize): Use x_vr_values to get class instance.
890 (vr_valueize_1): Likewise.
891 (class vrp_folder): Add vr_values data member. Add various member
892 functions as well as member functions that delegate to vr_values.
893 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
894 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
895 class instance from vr_values. Use it to call member functions.
896 (vrp_dom_walker): Add vr_values data member.
897 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
898 instance via x_vr_values.
899 (identify_jump_threads): Accept vr_values as argument. Store
900 it into the walker structure.
901 (evrp_dom_walker): Add vr_values class data member. Add various
903 (evrp_dom_walker::try_find_new_range): Use vr_values data
904 member to access the memory allocator.
905 (evrp_dom_walker::before_dom_children): Store vr_values class
906 instance into the vrp_folder class.
907 (evrp_dom_walker::push_value_range): Rework to avoid direct
908 access to num_vr_values and vr_value.
909 (evrp_dom_walker::pop_value_range): Likewise.
910 (execute_early_vrp): Remove call to vrp_initialize_lattice.
911 Use vr_values to get to dump_all_value_ranges member function.
912 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
913 and simplify_cond_using_ranges_2 via vrp_prop class instance.
914 Pass vr_values class instance down to identify_jump_threads.
915 Remove call to vrp_free_lattice.
916 (debug_all_value_ranges): Remove.
918 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
919 (vrp_folder): Likewise.
921 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
922 Get it from the existing bitmap instead.
923 (vrp_intersect_ranges_1): Likewise.
925 2017-11-09 Jakub Jelinek <jakub@redhat.com>
927 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
929 (store_immediate_info::store_immediate_info): Add bitnotp argument,
931 (imm_store_chain_info::coalesce_immediate_stores): Break group
932 if bit_not_p is different.
933 (count_multiple_uses, split_group,
934 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
935 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
936 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
937 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
938 to store_immediate_info ctor.
940 2017-11-09 Jim Wilson <jimw@sifive.com>
942 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
943 (scan_prog_file): Likewise.
945 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
947 * bb-reorder.c (max_entry_frequency): Remove.
948 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
949 connect_traces, push_to_next_round_p): Remove prototypes.
950 (find_traces_1_round): Use counts only.
951 (push_to_next_round_p): Likewise.
952 (find_traces): Likewise.
953 (rotate_loop): Likewise.
954 (find_traces_1_round): Likewise.
955 (connect_traces): Likewise.
956 (edge_order): Likewise.
958 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
960 * config/arm/arm.c (output_return_instruction): Add comments to
961 indicate requirement for cmse_nonsecure_entry return to account
962 for the size of clearing instruction output here.
963 (thumb_exit): Likewise.
964 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
965 return in hardfloat mode.
967 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
969 * config/rs6000/rs6000.c (machine_function): Add a bool,
970 "toc_is_wrapped_separately".
971 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
972 if it wasn't explicitly set or unset, we are optimizing for speed, and
973 doing separate shrink-wrapping.
974 (rs6000_get_separate_components): Enable the TOC component if
975 saving the TOC register in the prologue.
976 (rs6000_components_for_bb): Handle the TOC component.
977 (rs6000_emit_prologue_components): Store the TOC register where needed.
978 (rs6000_set_handled_components): Mark TOC as handled, if handled.
979 (rs6000_emit_prologue): Don't save the TOC if that is already done.
981 2017-11-09 Martin Jambor <mjambor@suse.cz>
983 * ipa-param-manipulation.c: New file.
984 * ipa-param-manipulation.h: Likewise.
985 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
986 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
987 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
988 (ipa_parm_adjustment): Likewise.
989 (ipa_parm_adjustment_vec): Likewise.
990 (ipa_get_vector_of_formal_parms): Moved declaration to
991 ipa-param-manipulation.h.
992 (ipa_get_vector_of_formal_parm_types): Likewise.
993 (ipa_modify_formal_parameters): Likewise.
994 (ipa_modify_call_arguments): Likewise.
995 (ipa_combine_adjustments): Likewise.
996 (ipa_dump_param_adjustments): Likewise.
997 (ipa_modify_expr): Likewise.
998 (ipa_get_adjustment_candidate): Likewise.
999 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1000 ipa-param-manipulation.c.
1001 (ipa_get_vector_of_formal_parm_types): Likewise.
1002 (ipa_modify_formal_parameters): Likewise.
1003 (ipa_modify_call_arguments): Likewise.
1004 (ipa_modify_expr): Likewise.
1005 (get_ssa_base_param): Likewise.
1006 (ipa_get_adjustment_candidate): Likewise.
1007 (index_in_adjustments_multiple_times_p): Likewise.
1008 (ipa_combine_adjustments): Likewise.
1009 (ipa_dump_param_adjustments): Likewise.
1010 * tree-sra.c: Also include ipa-param-manipulation.h
1011 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
1014 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1015 Alan Hayward <alan.hayward@arm.com>
1016 David Sherwood <david.sherwood@arm.com>
1018 * doc/sourcebuild.texi (vect_masked_store): Document.
1020 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1021 Alan Hayward <alan.hayward@arm.com>
1022 David Sherwood <david.sherwood@arm.com>
1024 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
1026 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1027 Alan Hayward <alan.hayward@arm.com>
1028 David Sherwood <david.sherwood@arm.com>
1030 * doc/sourcebuild.texi (vect_variable_length): Document.
1032 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1033 Alan Hayward <alan.hayward@arm.com>
1034 David Sherwood <david.sherwood@arm.com>
1036 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
1038 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1039 Alan Hayward <alan.hayward@arm.com>
1040 David Sherwood <david.sherwood@arm.com>
1042 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1044 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1045 Alan Hayward <alan.hayward@arm.com>
1046 David Sherwood <david.sherwood@arm.com>
1048 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1049 previously undocumented selectors.
1050 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1052 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1054 * doc/rtl.texi (const_vector): Say that elements can be
1055 const_wide_ints too.
1056 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1057 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1058 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1059 * optabs.c (expand_vector_broadcast): Likewise.
1061 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1062 Alan Hayward <alan.hayward@arm.com>
1063 David Sherwood <david.sherwood@arm.com>
1065 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1066 scaled index even if the unscaled address was invalid.
1067 Don't increase the complexity of using a scale in that case.
1069 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1070 Alan Hayward <alan.hayward@arm.com>
1071 David Sherwood <david.sherwood@arm.com>
1073 * doc/rtl.texi: Rewrite the subreg rules so that they partition
1074 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1075 UNITS_PER_WORD bytes.
1076 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1077 based on REGMODE_NATURAL_SIZE of the inner mode.
1078 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1079 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1080 * expmed.c (lowpart_bit_field_p): Divide the value up into
1081 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1082 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1083 whether something is likely to occupy more than one register.
1085 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1088 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1089 profile_count::adjust_for_ipa_scaling.
1090 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1091 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1093 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1095 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1097 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1098 (split_group): Add total_orig and total_new arguments, estimate the
1099 number of statements related to the store group without store merging
1100 and with store merging.
1101 (imm_store_chain_info::output_merged_store): Adjust split_group
1102 callers, punt if estimated number of statements with store merging
1103 is not smaller than estimated number of statements without it.
1105 (handled_load): Remove has_single_use checks.
1106 (pass_store_merging::process_store): Likewise.
1108 2017-11-09 Richard Biener <rguenther@suse.de>
1110 PR tree-optimization/82902
1111 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1113 2017-11-09 Martin Liska <mliska@suse.cz>
1116 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1119 2017-11-09 Martin Liska <mliska@suse.cz>
1121 PR tree-optimization/82669
1122 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1124 2017-11-09 Martin Liska <mliska@suse.cz>
1126 PR gcov-profile/48463
1127 * coverage.c (coverage_begin_function): Output also end locus
1128 of a function and information whether the function is
1130 * gcov-dump.c (tag_function): Parse and print the information.
1131 * gcov.c (INCLUDE_MAP): Add include.
1132 (INCLUDE_SET): Likewise.
1133 (struct line_info): Move earlier in the source file because
1134 of vector<line_info> in function_info structure.
1135 (line_info::line_info): Likewise.
1136 (line_info::has_block): Likewise.
1137 (struct source_info): Add new member index.
1138 (source_info::get_functions_at_location): New function.
1139 (function_info::group_line_p): New function.
1140 (output_intermediate_line): New function.
1141 (output_intermediate_file): Use the mentioned function.
1142 (struct function_start): New.
1143 (struct function_start_pair_hash): Likewise.
1144 (process_file): Add code that identifies group functions.
1145 Assign lines either to global or function scope.
1146 (generate_results): Skip artificial functions.
1147 (find_source): Assign index for each source file.
1148 (read_graph_file): Read new flag artificial and end_line.
1149 (add_line_counts): Assign it either to global of function scope.
1150 (accumulate_line_counts): Isolate core of the function to
1151 accumulate_line_info and call it for both function and global
1153 (accumulate_line_info): New function.
1154 (output_line_beginning): Fix GNU coding style.
1155 (print_source_line): New function.
1156 (output_line_details): Likewise.
1157 (output_function_details): Likewise.
1158 (output_lines): Iterate both source (global) scope and function
1160 (struct function_line_start_cmp): New class.
1161 * doc/gcov.texi: Reflect changes in documentation.
1163 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1166 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1167 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1168 and similarly for not instead of neg.
1170 2017-11-08 Andi Kleen <ak@linux.intel.com>
1172 * config/i386/i386.opt: Add -mforce-indirect-call.
1173 * config/i386/predicates.md: Check for flag_force_indirect_call.
1174 * doc/invoke.texi: Document -mforce-indirect-call
1176 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1178 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1180 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1181 config/riscv/riscv.c (predict.h): New include.
1182 (riscv_slow_unaligned_access_p): No longer static.
1183 (riscv_block_move_straight): Add require.
1184 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1186 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1189 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1190 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1191 nonimmediate_operand predicate for operand 1 instead of
1194 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1196 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1198 * config/aarch64/constraints.md (Uml): New constraint.
1199 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1202 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1204 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1205 of two vec_duplicates into a vec_concat.
1207 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1209 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1210 Simplify vec_merge of vec_duplicate and vec_concat.
1211 * config/aarch64/constraints.md (Utq): New constraint.
1212 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1215 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1217 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1218 Simplify vec_merge of vec_duplicate and const_vector.
1219 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1221 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1222 mode iterator. Update predicate on operand 1 to
1223 handle non-const_vec constants. Delete constraints.
1224 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1226 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1228 PR tree-optimization/78821
1229 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1231 (store_operand_info::store_operand_info): Initialize it to false.
1232 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1233 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1234 store in the group, and if chain_info is non-NULL, to ignore altogether
1236 (compatible_load_p): Fail if bit_not_p does not match.
1237 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1238 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1239 (pass_store_merging::process_store): Adjust
1240 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1241 call terminate_all_aliasing_chains newly when adding a store into
1242 a chain with non-NULL chain_info.
1244 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1246 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1248 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1250 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1252 (aarch64_layout_frame): Initialise emit_frame_chain.
1253 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1254 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1256 2017-11-08 Martin Liska <mliska@suse.cz>
1258 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1259 of gimple_call_internal_p.
1261 2017-11-07 Tom de Vries <tom@codesourcery.com>
1263 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1265 2017-11-07 Tom de Vries <tom@codesourcery.com>
1267 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1270 2017-11-08 Martin Liska <mliska@suse.cz>
1273 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1275 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1277 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1279 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1282 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1283 (*cmp<mode>_ccz_1): New insn with $k alternative.
1286 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1287 RTX_COMM_COMPARE as commutative as well.
1288 (ix86_binary_operator_ok): Formatting fix.
1289 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1290 *<code><mode>3<mask_name><round_saeonly_name>,
1291 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1292 *<s>mul<mode>3_highpart<mask_name>,
1293 *vec_widen_umult_even_v16si<mask_name>,
1294 *vec_widen_umult_even_v8si<mask_name>,
1295 *vec_widen_umult_even_v4si<mask_name>,
1296 *vec_widen_smult_even_v16si<mask_name>,
1297 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1298 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1299 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1300 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1301 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1302 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1303 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1304 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1305 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1306 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1307 ix86_binary_operator_ok. Formatting fixes.
1308 (*<plusminus_insn><mode>3<mask_name><round_name>,
1309 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1312 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1314 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1315 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1316 eq and ne if TARGET_ISEL.
1317 (cmp): New code_iterator.
1318 (UNS, UNSU_, UNSIK): New code_attrs.
1319 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1320 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1322 ("eq<mode>3"): ... this.
1323 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1325 ("ne<mode>3"): ... this.
1327 2017-11-07 Julia Koval <julia.koval@intel.com>
1330 * common/config/i386/i386-common.c
1331 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1332 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1333 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1334 * config/i386/i386.opt: Ditto.
1335 * config/i386/i386.c (ix86_target_string): Ditto.
1336 (ix86_option_override_internal): Ditto.
1337 (ix86_init_mpx_builtins): Move MPX to args2.
1338 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1339 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1340 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1341 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1342 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1343 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1344 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
1346 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
1349 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
1350 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
1351 (zero-extendsidi peephole2): Remove peephole.
1353 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1356 * stor-layout.c: Include attribs.h.
1357 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
1358 explicit lookup of "aligned" attribute.
1360 2017-11-07 Andrew Waterman <andrew@sifive.com>
1362 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
1363 (riscv_expand_block_move): Likewise.
1364 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1366 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1367 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1368 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1369 (riscv_adjust_block_mem): Likewise.
1370 (riscv_block_move_loop): Likewise.
1371 (riscv_expand_block_move): Likewise.
1372 gcc/config/riscv/riscv.md (movmemsi): New pattern.
1374 2017-11-07 Michael Clark <michaeljclark@mac.com>
1376 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
1377 (MUSL_DYNAMIC_LINKER): Likewise.
1379 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
1381 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
1382 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
1385 2017-11-07 Richard Biener <rguenther@suse.de>
1387 * match.pd: Fix build.
1389 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
1390 Jackson Woodruff <jackson.woodruff@arm.com>
1392 PR tree-optimization/71026
1393 * match.pd: Canonicalize negate in division.
1395 2017-11-07 Sudakshina Das <sudi.das@arm.com>
1398 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
1400 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1402 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
1403 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
1405 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1407 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
1408 non-scalar integral types.
1409 * match.pd (negate_expr_p): Handle MINUS_EXPR.
1410 (-(A-B), -(~A)): New transformations.
1412 2017-11-07 Tom de Vries <tom@codesourcery.com>
1414 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
1415 semicolon after "do {} while (0)".
1416 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1417 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1418 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1419 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1420 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1422 2017-11-07 Tom de Vries <tom@codesourcery.com>
1424 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
1425 after "do {} while (0)".
1426 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1427 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1428 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1429 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1430 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1432 2017-11-07 Tom de Vries <tom@codesourcery.com>
1435 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
1437 (arm_rtx_costs_internal): Add missing semicolon after
1438 HANDLE_NARROW_SHIFT_ARITH call.
1440 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1442 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1443 disable isel if it was not set explicitly.
1445 2017-11-06 James Bowman <james.bowman@ftdichip.com>
1447 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
1449 (add_type_attribute) likewise.
1451 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
1453 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
1454 of struct ix86_frame.
1455 (ix86_initial_elimination_offset): Likewise.
1456 (ix86_expand_split_stack_prologue): Likewise.
1458 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
1460 * tree-vrp.h (enum value_range_type): Update stale comment.
1462 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1463 Alan Hayward <alan.hayward@arm.com>
1464 David Sherwood <david.sherwood@arm.com>
1466 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
1467 (aarch64_expand_vec_perm_const): Take the number of units too.
1468 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
1469 (aarch64_expand_vec_perm_const): Likewise.
1470 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
1471 (vec_perm<mode>): Update accordingly.
1473 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1474 Alan Hayward <alan.hayward@arm.com>
1475 David Sherwood <david.sherwood@arm.com>
1477 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
1478 Take the number of units too.
1479 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
1480 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
1481 but check for a vector mode before rather than after the call.
1482 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
1483 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
1484 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
1485 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
1486 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
1487 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
1488 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
1489 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
1490 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
1491 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
1492 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
1493 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
1494 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
1495 (aarch64_sqdmull2_n<mode>): Update accordingly.
1497 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1498 Alan Hayward <alan.hayward@arm.com>
1499 David Sherwood <david.sherwood@arm.com>
1501 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
1502 the number of units too.
1503 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
1504 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
1505 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
1506 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
1507 (vec_store_lanesxi<mode>): Update accordingly.
1509 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1510 Alan Hayward <alan.hayward@arm.com>
1511 David Sherwood <david.sherwood@arm.com>
1513 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
1514 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
1515 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
1516 of units rather than the mode.
1517 * config/aarch64/iterators.md (nunits): New mode attribute.
1518 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
1519 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
1520 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
1521 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
1522 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1523 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
1524 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
1525 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
1526 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
1527 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
1528 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1529 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
1530 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
1531 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
1532 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1533 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1534 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
1535 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
1536 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
1537 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
1538 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
1539 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
1540 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
1541 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1542 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
1543 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
1544 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1545 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
1546 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
1547 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
1548 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
1549 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
1550 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
1551 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
1552 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
1553 (aarch64_simd_vec_setv2di): Likewise.
1555 2017-11-06 Carl Love <cel@us.ibm.com>
1557 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
1559 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
1560 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
1562 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
1564 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
1565 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
1566 Add power 8 macro expansions.
1567 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
1568 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
1569 power 8 instructions. (VSX_XXBR): Add iterator.
1571 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
1573 * config/arm/arm.md (predicable_short_it): Change default to "no",
1574 improve documentation, remove uses that are identical to the default.
1575 (enabled_for_depr_it): Rename to enabled_for_short_it.
1576 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
1577 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
1578 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
1579 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
1580 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
1582 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
1585 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
1586 float128 helper macros, which are no longer used after deleting
1587 the old 'q' built-in functions, and moving the round to odd
1588 built-in functions to being special built-in functions.
1589 (BU_FLOAT128_2): Likewise.
1590 (BU_FLOAT128_1_HW): Likewise.
1591 (BU_FLOAT128_2_HW): Likewise.
1592 (BU_FLOAT128_3_HW): Likewise.
1593 (FABSQ): Delete old 'q' built-in functions.
1594 (COPYSIGNQ): Likewise.
1595 (SQRTF128_ODD): Move round to odd built-in functions to be
1596 special built-in functions, so that we can handle
1597 -mabi=ieeelongdouble.
1598 (TRUNCF128_ODD): Likewise.
1599 (ADDF128_ODD): Likewise.
1600 (SUBF128_ODD): Likewise.
1601 (MULF128_ODD): Likewise.
1602 (DIVF128_ODD): Likewise.
1603 (FMAF128_ODD): Likewise.
1604 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
1605 built-in names to 'f128'.
1606 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
1607 old 'q' built-in functions, as the machine independent code for
1608 'f128' built-in functions handles this.
1609 (rs6000_expand_builtin): Add expansion for float128 round to odd
1610 functions, keying off on -mabi=ieeelongdouble of whether to use
1611 the KFmode or TFmode variant.
1612 (rs6000_init_builtins): Initialize the _Float128 round to odd
1614 * doc/extend.texi (PowerPC Built-in Functions): Document the old
1615 _Float128 'q' built-in functions are now mapped into the new
1616 'f128' built-in functions.
1618 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
1620 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
1621 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
1623 2017-11-06 David Malcolm <dmalcolm@redhat.com>
1626 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
1627 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
1628 * toplev.c: Include "ipa-fnsummary.h".
1629 (toplev::finalize): Call ipa_fnsummary_c_finalize.
1631 2017-11-06 Jakub Jelinek <jakub@redhat.com>
1633 PR tree-optimization/82838
1634 * gimple-ssa-store-merging.c
1635 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
1636 on a separate gimple_seq which is then appended to seq.
1638 2017-11-06 Jeff Law <law@redhat.com>
1641 * config/i386/i386.c (PROBE_INTERVAL): Remove.
1642 (get_probe_interval): New functions.
1643 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
1644 (ix86_adjust_stack_and_probe): Likewise.
1645 (output_adjust_stack_and_probe): Likewise.
1646 (ix86_emit_probe_stack_range): Likewise.
1647 (ix86_expand_prologue): Likewise.
1649 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1651 PR tree-optimization/82816
1652 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
1653 if the modes of the two types are the same.
1654 (convert_plusminus_to_widen): Likewise.
1656 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1658 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
1660 (usadv16qi): New define_expand.
1661 (usadv8hi): New define_expand.
1663 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
1666 * ipa-inline-transform.c (update_noncloned_frequencies): Always
1668 (inline_transform): Likewise.
1669 * predict.c (counts_to_freqs): Remove useless conditional.
1670 * profile-count.h (profile_count::apply_scale): Move sanity check.
1671 * tree-inline.c (copy_bb): Always scale.
1672 (copy_cfg_body): Likewise.
1674 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
1677 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
1679 (*cmp_ite0): Add enabled_for_depr_it attribute.
1680 (*cmp_ite1): Likewise.
1682 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1684 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
1687 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1689 * tree-vrp.c (vrp_int_const_binop): Return true on success and
1690 return the value by pointer.
1691 (extract_range_from_multiplicative_op_1): Update accordingly.
1692 Return as soon as an operation fails.
1694 2017-11-05 Tom de Vries <tom@codesourcery.com>
1697 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
1698 (DEF_SANITIZER_BUILTIN): ... here.
1699 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
1700 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
1702 2017-11-05 Tom de Vries <tom@codesourcery.com>
1705 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
1707 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
1708 ASM_OUTPUT_BEFORE_CASE_LABEL call.
1709 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
1711 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1712 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1713 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1715 2017-11-05 Tom de Vries <tom@codesourcery.com>
1718 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
1721 2017-11-04 Michael Clark <michaeljclark@mac.com>
1723 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
1724 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
1726 (*addsi3_extended): Likewise.
1727 (*addsi3_extended2): Likewise.
1728 (<optab>si3): Likewise.
1729 (<optab>di3): Likewise.
1730 (<optab><mode>3): Likewise.
1731 (<*optabe>si3_internal): Likewise.
1732 (zero_extendqi<SUPERQI:mode>2): Likewise.
1733 (*add<mode>hi3): Likewise.
1734 (*xor<mode>hi3): Likewise.
1735 (<optab>di3): Likewise.
1736 (*<optab>si3_extend): Likewise.
1737 (*sge<u>_<X:mode><GPR:mode>): Likewise.
1738 (*slt<u>_<X:mode><GPR:mode>): Likewise.
1739 (*sle<u>_<X:mode><GPR:mode>): Likewise.
1741 2017-11-04 Andrew Waterman <andrew@sifive.com>
1743 * config/riscv/riscv.c (riscv_option_override): Conditionally set
1744 TARGET_STRICT_ALIGN based upon -mtune argument.
1746 2017-11-04 Andrew Waterman <andrew@sifive.com>
1748 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
1750 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
1752 config/i386/i386.c (choose_basereg): Use optional scratch
1753 register and add assertion.
1754 (x86_emit_outlined_ms2sysv_save): Use scratch register when
1755 needed, and don't allocate stack.
1756 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
1757 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
1758 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
1760 2017-11-03 Jeff Law <law@redhat.com>
1762 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
1763 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
1764 to probe at the start of a noreturn function.
1766 2017-11-03 Jakub Jelinek <jakub@redhat.com>
1768 PR tree-optimization/78821
1769 * gimple-ssa-store-merging.c: Update the file comment.
1770 (MAX_STORE_ALIAS_CHECKS): Define.
1771 (struct store_operand_info): New type.
1772 (store_operand_info::store_operand_info): New constructor.
1773 (struct store_immediate_info): Add rhs_code and ops data members.
1774 (store_immediate_info::store_immediate_info): Add rhscode, op0r
1775 and op1r arguments to the ctor, initialize corresponding data members.
1776 (struct merged_store_group): Add load_align_base and load_align
1778 (merged_store_group::merged_store_group): Initialize them.
1779 (merged_store_group::do_merge): Update them.
1780 (merged_store_group::apply_stores): Pick the constant for
1781 encode_tree_to_bitpos from one of the two operands, or skip
1782 encode_tree_to_bitpos if neither operand is a constant.
1783 (class pass_store_merging): Add process_store method decl. Remove
1784 bool argument from terminate_all_aliasing_chains method decl.
1785 (pass_store_merging::terminate_all_aliasing_chains): Remove
1786 var_offset_p argument and corresponding handling.
1787 (stmts_may_clobber_ref_p): New function.
1788 (compatible_load_p): New function.
1789 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
1790 if there is overlap and rhs_code is not INTEGER_CST. For
1791 non-overlapping stores terminate group if rhs is not mergeable.
1792 (get_alias_type_for_stmts): Change first argument from
1793 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
1794 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
1795 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
1797 (get_location_for_stmts): Change first argument from
1798 auto_vec<gimple *> & to vec<gimple *> &.
1799 (struct split_store): Remove orig_stmts data member, add orig_stores.
1800 (split_store::split_store): Create orig_stores rather than orig_stmts.
1801 (find_constituent_stmts): Renamed to ...
1802 (find_constituent_stores): ... this. Change second argument from
1803 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
1804 to info structures rather than the statements.
1805 (split_group): Rename ALLOW_UNALIGNED argument to
1806 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
1807 it. Adjust find_constituent_stores caller.
1808 (imm_store_chain_info::output_merged_store): Handle rhs_code other
1809 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
1810 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
1811 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
1812 (mem_valid_for_store_merging): New function.
1813 (handled_load): New function.
1814 (pass_store_merging::process_store): New method.
1815 (pass_store_merging::execute): Use process_store method. Adjust
1816 terminate_all_aliasing_chains caller.
1818 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1820 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
1821 Return true for more constants, symbols and label references.
1822 (aarch64_valid_floating_const): Remove unused function.
1824 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1827 * config/aarch64/aarch64.c (aarch64_layout_frame):
1828 Undo forcing of LR at bottom of frame.
1830 2017-11-03 Jeff Law <law@redhat.com>
1833 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
1834 for int_registers_saved.
1836 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
1837 extracted from tree-ssa-dom.c.
1838 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
1839 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
1840 (record_equivalences_from_incoming_edge): Add additional argument
1841 to single_pred_edge_ignoring_loop_edges call.
1842 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
1843 (uncprop_dom_walker::before_dom_children): Add additional argument
1844 to single_pred_edge_ignoring_loop_edges call.
1845 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
1846 single_pred_edge_ignoring_loop_edges rather than open coding.
1847 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
1849 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
1851 * match.pd (-(-A)): Rewrite.
1853 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1855 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
1856 (rs6000_emit_int_cmove): New declaration.
1857 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
1858 (rs6000_emit_sISEL): Delete.
1859 (rs6000_emit_int_cmove): Make non-static.
1860 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
1861 instead of rs6000_emit_sISEL.
1863 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
1865 * asan.c (create_cond_insert_point): Maintain profile.
1866 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
1868 * basic-block.h (struct basic_block_def): Remove frequency.
1869 (EDGE_FREQUENCY): Use to_frequency
1870 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
1872 (find_traces): Update to use to_frequency.
1873 (find_traces_1_round): Likewise; use only IPA counts.
1874 (bb_to_key): Likewise.
1875 (connect_traces): Use IPA counts only.
1876 (copy_bb_p): Update to use to_frequency.
1877 (fix_up_crossing_landing_pad): Likewise.
1878 (sanitize_hot_paths): Likewise.
1879 * bt-load.c (basic_block_freq): Likewise.
1880 * cfg.c (init_flow): Set count_max to uninitialized.
1881 (check_bb_profile): Remove frequencies; check counts.
1882 (dump_bb_info): Do not dump frequencies.
1883 (update_bb_profile_for_threading): Update counts only.
1884 (scale_bbs_frequencies_int): Likewise.
1885 (MAX_SAFE_MULTIPLIER): Remove.
1886 (scale_bbs_frequencies_gcov_type): Update counts only.
1887 (scale_bbs_frequencies_profile_count): Update counts only.
1888 (scale_bbs_frequencies): Update counts only.
1889 * cfg.h (struct control_flow_graph): Add count-max.
1890 (update_bb_profile_for_threading): Update prototype.
1891 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
1892 (find_many_sub_basic_blocks): Likewise.
1893 * cfgcleanup.c (try_forward_edges): Likewise.
1894 (try_crossjump_to_edge): Likewise.
1895 * cfgexpand.c (expand_gimple_cond): Likewise.
1896 (expand_gimple_tailcall): Likewise.
1897 (construct_init_block): Likewise.
1898 (construct_exit_block): Likewise.
1899 * cfghooks.c (verify_flow_info): Check consistency of counts.
1900 (dump_bb_for_graph): Do not dump frequencies.
1901 (split_block_1): Do not update frequencies.
1902 (split_edge): Do not update frequencies.
1903 (make_forwarder_block): Do not update frequencies.
1904 (duplicate_block): Do not update frequencies.
1905 (account_profile_record): Do not update frequencies.
1906 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
1907 for global heuristics.
1908 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
1909 (expected_loop_iterations_unbounded): Use counts only.
1910 * cfgloopmanip.c (scale_loop_profile): Simplify.
1911 (create_empty_loop_on_edge): Simplify
1913 (duplicate_loop_to_header_edge): Simplify
1914 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
1915 (update_br_prob_note): Take care of removing note when profile
1917 (relink_block_chain): Do not dump frequency.
1918 (rtl_account_profile_record): Use to_frequency.
1919 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
1920 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
1921 (cgraph_update_edges_for_call_stmt_node): Likewise.
1922 (cgraph_edge::verify_count_and_frequency): Update.
1923 (cgraph_node::verify_node): Temporarily disable frequency verification.
1924 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
1925 to_cgraph_frequency.
1926 (cgraph_edge::rebuild_edges): Convert to ipa counts.
1927 * cgraphunit.c (init_lowered_empty_function): Do not initialize
1929 (cgraph_node::expand_thunk): Update profile.
1930 * except.c (dw2_build_landing_pads): Do not update frequency.
1931 * final.c (compute_alignments): Use to_frequency.
1932 (dump_basic_block_info): Do not dump frequency.
1933 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
1934 (dump_gimple_bb_header): Do not dump frequency.
1935 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
1937 * gimple-streamer-in.c (input_bb): Do not stream frequency.
1938 * gimple-streamer-out.c (output_bb): Do not stream frequency.
1939 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
1940 (init_before_recovery): Do not update frequency.
1941 (sched_create_recovery_edges): Do not update frequency.
1942 * hsa-gen.c (convert_switch_statements): Do not update frequency.
1943 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
1944 (ipa_cp_c_finalize): Set max_count to uninitialized.
1945 * ipa-fnsummary.c (get_minimal_bb): Use counts.
1946 (param_change_prob): Use counts.
1947 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
1949 * ipa-split.c (consider_split): Use to_frequency.
1950 (split_function): Use to_frequency.
1951 * ira-build.c (loop_compare_func): Likewise.
1952 (mark_loops_for_removal): Likewise.
1953 (mark_all_loops_for_removal): Likewise.
1954 * loop-doloop.c (doloop_modify): Do not update frequency.
1955 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
1957 * lto-streamer-in.c (input_function): Update count_max.
1958 * omp-expand.c (expand_omp_taskreg): Update count_max.
1959 * omp-simd-clone.c (simd_clone_adjust): Update profile.
1960 * predict.c (maybe_hot_frequency_p): Use to_frequency.
1961 (maybe_hot_count_p): Use ipa counts only.
1962 (maybe_hot_bb_p): Simplify.
1963 (maybe_hot_edge_p): Simplify.
1964 (probably_never_executed): Do not take frequency argument.
1965 (probably_never_executed_bb_p): Do not pass frequency.
1966 (probably_never_executed_edge_p): Likewise.
1967 (combine_predictions_for_bb): Check that profile is nonzero.
1968 (propagate_freq): Do not set frequency.
1969 (drop_profile): Simplify.
1970 (counts_to_freqs): Simplify.
1971 (expensive_function_p): Use to_frequency.
1972 (propagate_unlikely_bbs_forward): Simplify.
1973 (determine_unlikely_bbs): Simplify.
1974 (estimate_bb_frequencies): Add hack to silence graphite issues.
1975 (compute_function_frequency): Use ipa counts.
1976 (pass_profile::execute): Update.
1977 (rebuild_frequencies): Use counts only.
1978 (force_edge_cold): Use counts only.
1979 * profile-count.c (profile_count::dump): Dump new count types.
1980 (profile_count::differs_from_p): Check compatiblity.
1981 (profile_count::to_frequency): New function.
1982 (profile_count::to_cgraph_frequency): New function.
1983 * profile-count.h (struct function): Declare.
1984 (enum profile_quality): Add profile_guessed_local and
1985 profile_guessed_global0.
1986 (class profile_proability): Decrease number of bits to 29;
1987 update from_reg_br_prob_note and to_reg_br_prob_note.
1988 (class profile_count: Update comment; decrease number of bits
1989 to 61. Check compatibility.
1990 (profile_count::compatible_p): New private member function.
1991 (profile_count::ipa_p): New member function.
1992 (profile_count::operator<): Handle global zero correctly.
1993 (profile_count::operator>): Handle global zero correctly.
1994 (profile_count::operator<=): Handle global zero correctly.
1995 (profile_count::operator>=): Handle global zero correctly.
1996 (profile_count::nonzero_p): New member function.
1997 (profile_count::force_nonzero): New member function.
1998 (profile_count::max): New member function.
1999 (profile_count::apply_scale): Handle IPA scalling.
2000 (profile_count::guessed_local): New member function.
2001 (profile_count::global0): New member function.
2002 (profile_count::ipa): New member function.
2003 (profile_count::to_frequency): Declare.
2004 (profile_count::to_cgraph_frequency): Declare.
2005 * profile.c (OVERLAP_BASE): Delete.
2006 (compute_frequency_overlap): Delete.
2007 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2008 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2009 * sched-ebb.c (rank): Use counts only.
2010 * shrink-wrap.c (handle_simple_exit): Use counts only.
2011 (try_shrink_wrapping): Use counts only.
2012 (place_prologue_for_one_component): Use counts only.
2013 * tracer.c (find_best_predecessor): Use to_frequency.
2014 (find_trace): Use to_frequency.
2015 (tail_duplicate): Use to_frequency.
2016 * trans-mem.c (expand_transaction): Do not update frequency.
2017 * tree-call-cdce.c: Do not update frequency.
2018 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2019 (gimple_merge_blocks): Likewise.
2020 (gimple_split_edge): Likewise.
2021 (gimple_duplicate_sese_region): Likewise.
2022 (gimple_duplicate_sese_tail): Likewise.
2023 (move_sese_region_to_fn): Likewise.
2024 (gimple_account_profile_record): Likewise.
2025 (insert_cond_bb): Likewise.
2026 * tree-complex.c (expand_complex_div_wide): Likewise.
2027 * tree-eh.c (lower_resx): Update profile.
2028 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2030 (initialize_cfun): Do not initialize frequencies
2031 (freqs_to_counts): Delete.
2032 (copy_cfg_body): Ignore count parameter.
2033 (copy_body): Update.
2034 (expand_call_inline): Update count_max.
2035 (optimize_inline_calls): Update count_max.
2036 (tree_function_versioning): Update count_max.
2037 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2038 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2040 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2041 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2042 (try_peel_loop): Likewise.
2043 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2045 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2046 (tree_transform_and_unroll_loop): Do not use frequencies
2047 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2048 Use reliable prediction only.
2049 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2050 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2051 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2052 probability scaling.
2053 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2054 not update frequency
2055 (any_remaining_duplicated_blocks): Likewise.
2056 (update_profile): Likewise.
2057 (estimated_freqs_path): Delete.
2058 (freqs_to_counts_path): Delete.
2059 (clear_counts_path): Delete.
2060 (ssa_fix_duplicate_block_edges): Likewise.
2061 (duplicate_thread_path): Likewise.
2062 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2063 * tree-tailcall.c (decrease_profile): Do not update frequency.
2064 (eliminate_tail_call): Likewise.
2065 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2066 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2067 (optimize_mask_stores): Likewise.
2068 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2069 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2070 (ubsan_expand_ptr_ifn): Update profile.
2071 * value-prof.c (gimple_ic): Simplify.
2072 * value-prof.h (gimple_ic): Update prototype.
2073 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2074 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2076 (want_inline_self_recursive_call_p): Likewise.
2077 (resolve_noninline_speculation): Only cummulate defined counts.
2078 (inline_small_functions): Use nonzero_p.
2079 (ipa_inline): Do not access freed node.
2081 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2083 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2084 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2086 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2088 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2089 non-legitimate address.
2091 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2093 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2094 (*lt0_<mode>di, *lt0_<mode>si): New.
2096 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2098 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2099 TARGET_PAIRED_FLOAT.
2101 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2102 Jim Wilson <jim.wilson@linaro.org>
2104 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2105 * config/aarch64/aarch64-tune.md: Regenerated.
2106 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2107 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2109 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2111 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2112 (arc_expand_prologue): Restore blink for millicode.
2113 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2115 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2118 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2119 gen_vec_duplicate after forcing the scalar into a register.
2121 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2123 * combine (try_combine): Print the insns input to try_combine to the
2126 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2129 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2130 Remove second argument from aarch64_process_target_attr call.
2131 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2133 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2135 (aarch64_handle_attr_arch): Remove second argument.
2136 (aarch64_handle_attr_cpu): Ditto.
2137 (aarch64_handle_attr_tune): Ditto.
2138 (aarch64_handle_attr_isa_flags): Ditto.
2139 (aarch64_process_one_target_attr): Ditto.
2140 (aarch64_process_target_attr): Ditto.
2141 (aarch64_option_valid_attribute_p): Remove second argument.
2142 on aarch64_process_target_attr call.
2144 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2146 * diagnostic.c: Include "selftest-diagnostic.h".
2147 (selftest::assert_location_text): New function.
2148 (selftest::test_diagnostic_get_location_text): New function.
2149 (selftest::diagnostic_c_tests): Call it.
2151 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2153 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2154 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2155 (class selftest::test_diagnostic_context): Move to...
2156 * selftest-diagnostic.c: New file.
2157 * selftest-diagnostic.h: New file.
2159 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2161 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2162 offset range for FT32B.
2163 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2164 * config/ft32/ft32.md: Add TARGET_NOPM.
2165 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2166 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2168 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2170 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2172 2017-11-02 Jeff Law <law@redhat.com>
2174 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2175 virtual keyword on FINAL OVERRIDE members.
2177 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2178 virtuals together. Add virtual destructor.
2179 (substitute_and_fold_engine): Similarly.
2181 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2183 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2185 2017-11-02 Richard Biener <rguenther@suse.de>
2187 PR tree-optimization/82795
2188 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2190 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2192 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2194 (gcc_GAS_CHECK_FEATURE): Remove.
2195 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2196 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2197 * configure: Regenerate.
2199 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2201 * config/arc/arc.c (hwloop_optimize): Account for empty
2204 2017-11-02 Richard Biener <rguenther@suse.de>
2207 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2208 Truncate ARRAY_REF index and element size.
2210 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2212 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2214 2017-11-01 Jeff Law <law@redhat.com>
2216 * tree-ssa-ccp.c (ccp_folder): New class derived from
2217 substitute_and_fold_engine.
2218 (ccp_folder::get_value): New member function.
2219 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2220 (ccp_fold_stmt): Remove prototype.
2221 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2222 * tree-ssa-copy.c (copy_folder): New class derived from
2223 substitute_and_fold_engine.
2224 (copy_folder::get_value): Renamed from get_value.
2225 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2226 * tree-vrp.c (vrp_folder): New class derived from
2227 substitute_and_fold_engine.
2228 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2229 (vrp_folder::get_value): New member function.
2230 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2231 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2232 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2233 provide a class interface to folder/substitute routines.
2234 (ssa_prop_fold_stmt_fn): Remove typedef.
2235 (ssa_prop_get_value_fn): Likewise.
2236 (subsitute_and_fold): Remove prototype.
2237 (replace_uses_in): Likewise.
2238 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2239 Renamed from replace_uses_in. Call the virtual member function
2240 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2241 (substitute_and_fold_dom_walker): Remove initialization of
2242 data member entries for calbacks. Add substitute_and_fold_engine
2243 member and initialize it.
2244 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2245 member functions for get_value, replace_phi_args_in c
2246 replace_uses_in, and fold_stmt calls.
2247 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2248 substitute_and_fold. Remove assert. Update ctor call.
2250 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2251 (ssa_prop_visit_phi_fn): Likewise.
2252 (class ssa_propagation_engine): New class to provide an interface
2254 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2256 (ssa_prop_visit_phi): Likewise.
2257 (ssa_propagation_engine::simulate_stmt): Moved into class.
2258 Call visit_phi/visit_stmt from the class rather than via
2259 file scoped static variables.
2260 (ssa_propagation_engine::simulate_block): Moved into class.
2261 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2262 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2263 set file scoped statics for the visit_stmt/visit_phi callbacks.
2264 * tree-complex.c (complex_propagate): New class derived from
2265 ssa_propagation_engine.
2266 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2267 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2268 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2270 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2271 ssa_propagation_engine.
2272 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2273 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2274 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2275 * tree-ssa-copy.c (copy_prop): New class derived from
2276 ssa_propagation_engine.
2277 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2278 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2279 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2280 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2281 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2282 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2283 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2285 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2287 PR rtl-optimization/82778
2288 PR rtl-optimization/82597
2289 * compare-elim.c (struct comparison): Add in_a_setter field.
2290 (find_comparison_dom_walker::before_dom_children): Remove killed
2291 bitmap and df_simulate_find_defs call, instead walk the defs.
2292 Compute last_setter and initialize in_a_setter. Merge definitions
2293 with first initialization for a few variables.
2294 (try_validate_parallel): Use insn_invalid_p instead of
2295 recog_memoized. Return insn rather than just the pattern.
2296 (try_merge_compare): Fix up comment. Don't uselessly test if
2297 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2299 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2302 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2303 Alan Hayward <alan.hayward@arm.com>
2304 David Sherwood <david.sherwood@arm.com>
2306 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2307 aarch64_hard_regno_nregs to get the number of registers
2310 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2311 Alan Hayward <alan.hayward@arm.com>
2312 David Sherwood <david.sherwood@arm.com>
2314 * config/aarch64/constraints.md (Upl): Rename to...
2316 * config/aarch64/aarch64.md
2317 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
2320 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2321 Alan Hayward <alan.hayward@arm.com>
2322 David Sherwood <david.sherwood@arm.com>
2324 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2325 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2328 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2329 Alan Hayward <alan.hayward@arm.com>
2330 David Sherwood <david.sherwood@arm.com>
2332 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2333 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2334 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2336 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2337 of a single element.
2338 * config/aarch64/iterators.md: Add a comment above the permute
2339 unspecs to say that they are generated directly by
2340 aarch64_expand_vec_perm_const.
2341 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
2343 2017-11-01 Nathan Sidwell <nathan@acm.org>
2345 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
2347 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2349 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
2350 and medany code models, and describe what they do.
2352 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2354 Revert accidental duplicate:
2356 * combine.c (can_change_dest_mode): Reject changes in
2357 REGMODE_NATURAL_SIZE.
2359 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
2361 PR rtl-optimization/64682
2362 PR rtl-optimization/69567
2363 PR rtl-optimization/69737
2364 PR rtl-optimization/82683
2365 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2366 register mentioned in the note, drop the note, unless it came from I3,
2367 in which case it should go to I3 again.
2369 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2371 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
2372 and return false if not.
2373 (clear_bytes_written_by, live_bytes_read): Update accordingly.
2375 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2377 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
2378 range is known to be empty.
2380 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2381 Alan Hayward <alan.hayward@arm.com>
2382 David Sherwood <david.sherwood@arm.com>
2384 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
2385 and CONST_VECTOR_NUNITS instead of computing the number of units from
2386 the byte sizes of the vector and element.
2387 (simplify_binary_operation_1): Likewise.
2388 (simplify_const_binary_operation): Likewise.
2389 (simplify_ternary_operation): Likewise.
2391 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2392 Alan Hayward <alan.hayward@arm.com>
2393 David Sherwood <david.sherwood@arm.com>
2395 * var-tracking.c (INT_MEM_OFFSET): Replace with...
2396 (int_mem_offset): ...this new function.
2397 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
2398 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
2399 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
2402 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2403 Alan Hayward <alan.hayward@arm.com>
2404 David Sherwood <david.sherwood@arm.com>
2406 * lower-subreg.c (interesting_mode_p): New function.
2407 (compute_costs, find_decomposable_subregs, decompose_register)
2408 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
2409 (resolve_clobber, dump_choices): Use it.
2411 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2412 Alan Hayward <alan.hayward@arm.com>
2413 David Sherwood <david.sherwood@arm.com>
2415 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
2417 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2418 Alan Hayward <alan.hayward@arm.com>
2419 David Sherwood <david.sherwood@arm.com>
2421 * alias.c (find_base_value, find_base_term): Only process integer
2422 truncations. Check the precision rather than the size.
2424 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2425 Alan Hayward <alan.hayward@arm.com>
2426 David Sherwood <david.sherwood@arm.com>
2428 * machmode.h (is_narrower_int_mode): New function
2429 * optabs.c (expand_float, expand_fix): Use it.
2430 * dwarf2out.c (rotate_loc_descriptor): Likewise.
2432 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2433 Alan Hayward <alan.hayward@arm.com>
2434 David Sherwood <david.sherwood@arm.com>
2436 * rtl.h (narrower_subreg_mode): New function.
2437 * ira-color.c (update_costs_from_allocno): Use it.
2439 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2440 Alan Hayward <alan.hayward@arm.com>
2441 David Sherwood <david.sherwood@arm.com>
2443 * optabs-query.h (convert_optab_p): New function, split out from...
2444 (convert_optab_handler): ...here.
2445 (widening_optab_handler): Delete.
2446 (find_widening_optab_handler): Remove permit_non_widening parameter.
2447 (find_widening_optab_handler_and_mode): Likewise. Provide an
2448 override that operates on mode class wrappers.
2449 * optabs-query.c (widening_optab_handler): Delete.
2450 (find_widening_optab_handler_and_mode): Remove permit_non_widening
2451 parameter. Assert that the two modes are the same class and that
2452 the "from" mode is narrower than the "to" mode. Use
2453 convert_optab_handler instead of widening_optab_handler.
2454 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
2455 instead of widening_optab_handler.
2456 * expr.c (expand_expr_real_2): Update calls to
2457 find_widening_optab_handler.
2458 * optabs.c (expand_widen_pattern_expr): Likewise.
2459 (expand_binop_directly): Take the insn_code as a parameter.
2460 (expand_binop): Only call find_widening_optab_handler for
2461 conversion optabs; use optab_handler otherwise. Update calls
2462 to find_widening_optab_handler and expand_binop_directly.
2463 Use convert_optab_handler instead of widening_optab_handler.
2464 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
2465 find_widening_optab_handler and use scalar_mode rather than
2467 (convert_plusminus_to_widen): Likewise.
2469 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2470 Alan Hayward <alan.hayward@arm.com>
2471 David Sherwood <david.sherwood@arm.com>
2473 * machmode.h (fixed_size_mode): New class.
2474 * rtl.h (get_pool_mode): Return fixed_size_mode.
2475 * gengtype.c (main): Add fixed_size_mode.
2476 * target.def (get_raw_result_mode): Return a fixed_size_mode.
2477 (get_raw_arg_mode): Likewise.
2478 * doc/tm.texi: Regenerate.
2479 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
2480 * targhooks.c (default_get_reg_raw_mode): Likewise.
2481 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
2482 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
2483 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
2484 (msp430_get_raw_result_mode): Likewise.
2485 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
2486 * dbxout.c (dbxout_parms): Require fixed-size modes.
2487 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
2488 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2489 * omp-low.c (lower_oacc_reductions): Likewise.
2490 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
2491 (simplify_subreg): Update accordingly.
2492 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
2493 (force_const_mem): Update accordingly. Return NULL_RTX for modes
2494 that aren't fixed-size.
2495 (get_pool_mode): Return a fixed_size_mode.
2496 (output_constant_pool_2): Take a fixed_size_mode.
2498 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2499 Alan Hayward <alan.hayward@arm.com>
2500 David Sherwood <david.sherwood@arm.com>
2502 * doc/rtl.texi (vec_series): Document.
2503 (const): Say that the operand can be a vec_series.
2504 * rtl.def (VEC_SERIES): New rtx code.
2505 * rtl.h (const_vec_series_p_1): Declare.
2506 (const_vec_series_p): New function.
2507 * emit-rtl.h (gen_const_vec_series): Declare.
2508 (gen_vec_series): Likewise.
2509 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
2510 (gen_vec_series): Likewise.
2511 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
2512 * simplify-rtx.c (simplify_unary_operation): Handle negations
2514 (simplify_binary_operation_series): New function.
2515 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
2516 (test_vector_ops_series): New function.
2517 (test_vector_ops): Call it.
2518 * config/powerpcspe/altivec.md (altivec_lvsl): Use
2519 gen_const_vec_series.
2520 (altivec_lvsr): Likewise.
2521 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
2523 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2524 Alan Hayward <alan.hayward@arm.com>
2525 David Sherwood <david.sherwood@arm.com>
2527 * doc/rtl.texi (const): Update description of address constants.
2528 Say that vector constants are allowed too.
2529 * common.md (E, F): Use CONSTANT_P instead of checking for
2531 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
2532 checking for CONST_VECTOR.
2533 * expmed.c (make_tree): Use build_vector_from_val for a CONST
2535 * expr.c (expand_expr_real_2): Check for vector modes instead
2536 of checking for CONST_VECTOR.
2537 * rtl.h (const_vec_p): New function.
2538 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
2539 (unwrap_const_vec_duplicate): Handle them here too.
2541 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2542 David Malcolm <dmalcolm@redhat.com>
2543 Alan Hayward <alan.hayward@arm.com>
2544 David Sherwood <david.sherwood@arm.com>
2546 * rtl.h (vec_duplicate_p): New function.
2547 * selftest-rtl.c (assert_rtx_eq_at): New function.
2548 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
2549 (assert_rtx_eq_at): Declare.
2550 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
2551 * selftest-run-tests.c (selftest::run_tests): Call it.
2552 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
2553 (simplify_unary_operation_1): Recursively handle vector duplicates.
2554 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
2556 (simplify_subreg): Handle subregs of vector duplicates.
2557 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
2558 (selftest::simplify_rtx_c_tests): New functions.
2560 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2561 Alan Hayward <alan.hayward@arm.com>
2562 David Sherwood <david.sherwood@arm.com>
2564 * emit-rtl.h (gen_const_vec_duplicate): Declare.
2565 (gen_vec_duplicate): Likewise.
2566 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
2568 (gen_const_vector): ...here.
2569 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
2570 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
2571 whose elements are all equal.
2572 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
2573 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
2574 (simplify_relational_operation): Likewise.
2575 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
2577 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
2578 (aarch64_expand_vector_init): Likewise.
2579 * config/arm/arm.c (neon_vdup_constant): Likewise.
2580 (neon_expand_vector_init): Likewise.
2581 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
2582 (arm_block_set_unaligned_vect): Likewise.
2583 (arm_block_set_aligned_vect): Likewise.
2584 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
2585 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
2586 (expand_vec_perm_even_odd_pack): Likewise.
2587 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
2588 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
2589 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
2590 gen_const_vec_duplicate.
2591 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
2592 * config/mips/mips.c (mips_gen_const_int_vector): Use
2593 gen_const_vec_duplicate.
2594 (mips_expand_vector_init): Use CONST0_RTX.
2595 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
2596 (define_split): Use gen_const_vec_duplicate.
2597 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
2598 (define_split): Use gen_const_vec_duplicate.
2599 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
2600 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
2601 * config/spu/spu.c (spu_const): Likewise.
2603 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2604 Alan Hayward <alan.hayward@arm.com>
2605 David Sherwood <david.sherwood@arm.com>
2607 * combine.c (can_change_dest_mode): Reject changes in
2608 REGMODE_NATURAL_SIZE.
2610 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
2612 * configure.ac (--enable-libssp): New.
2613 (gcc_cv_libc_provides_ssp): Check for explicit setting before
2614 trying to determine target-specific default. Adjust indentation.
2615 * configure: Regenerated.
2616 * doc/install.texi (Configuration): Expand --disable-libssp
2619 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
2621 config/i386/i386.c (ix86_expand_epilogue): Correct stack
2624 2017-10-31 Martin Jambor <mjambor@suse.cz>
2627 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
2629 2017-10-31 David Malcolm <dmalcolm@redhat.com>
2631 * auto-profile.c (autofdo_source_profile::read): Use
2632 UNKNOWN_LOCATION rather than 0.
2633 * diagnostic-core.h (warning_at_rich_loc): Rename to...
2634 (warning_at): ...this overload.
2635 (warning_at_rich_loc_n): Rename to...
2636 (warning_n): ...this overload.
2637 (error_at_rich_loc): Rename to...
2638 (error_at): ...this overload.
2639 (pedwarn_at_rich_loc): Rename to...
2640 (pedwarn): ...this overload.
2641 (permerror_at_rich_loc): Rename to...
2642 (permerror): ...this overload.
2643 (inform_at_rich_loc): Rename to...
2644 (inform): ...this overload.
2645 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
2646 (diagnostic_n_impl_richloc): Rename to...
2647 (diagnostic_n_impl): ...this rich_location *-based decl.
2648 (inform_at_rich_loc): Rename to...
2649 (inform): ...this, and add an assertion.
2650 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
2651 (warning_at_rich_loc): Rename to...
2652 (warning_at): ...this, and add an assertion.
2653 (warning_at_rich_loc_n): Rename to...
2654 (warning_n): ...this, and add an assertion.
2655 (warning_n): Update location_t-based implementation for removal of
2656 location_t-based diagnostic_n_impl.
2657 (pedwarn_at_rich_loc): Rename to...
2658 (pedwarn): ...this, and add an assertion.
2659 (permerror_at_rich_loc): Rename to...
2660 (permerror): ...this, and add an assertion.
2661 (error_n): Update for removal of location_t-based diagnostic_n_impl.
2662 (error_at_rich_loc): Rename to...
2663 (error_at): ...this, and add an assertion.
2664 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
2665 (driver::do_spec_on_infiles): Likewise.
2666 * substring-locations.c (format_warning_va): Update for renaming
2667 of inform_at_rich_loc.
2669 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2671 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
2672 _Float<N>X built-in functions so that the variant without the
2673 "__builtin_" prefix is only enabled for the GNU C and Objective C
2674 languages when they are in non-strict ANSI/ISO mode.
2675 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
2676 * target.def (floatn_builtin_p): Add a target hook to control
2677 whether _Float<N> and _Float<N>X built-in functions without the
2678 "__builtin_" prefix are enabled, and return true for C and
2679 Objective C in the default hook. Include langhooks.h in
2681 * targhooks.h (default_floatn_builtin_p): Likewise.
2682 * targhooks.c (default_floatn_builtin_p): Likewise.
2683 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
2684 floatn_builtin_p target hook.
2685 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
2687 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
2688 Eric Botcazou <ebotcazou@adacore.com>
2690 PR rtl-optimization/81803
2691 * lra-constraints.c (curr_insn_transform): Also reload the whole
2692 register for a strict subreg no wider than a word if this is for
2693 a WORD_REGISTER_OPERATIONS target.
2695 2017-10-31 Jason Merrill <jason@redhat.com>
2697 * gdbinit.in: Skip over inlines from timevar.h.
2699 2017-10-31 Martin Liska <mliska@suse.cz>
2701 * doc/gcov.texi: Document new option.
2702 * gcov.c (print_usage): Likewise print it.
2703 (process_args): Support the argument.
2704 (format_count): New function.
2705 (format_gcov): Use the function.
2707 2017-10-31 Martin Liska <mliska@suse.cz>
2709 * gcov.c (struct name_map): do not use typedef.
2710 Define operator== and operator<.
2711 (name_search): Remove.
2712 (name_sort): Remove.
2713 (main): Do not allocate names.
2714 (process_file): Add vertical space.
2715 (generate_results): Use std::find.
2716 (release_structures): Do not release memory.
2717 (find_source): Use std::find.
2719 2017-10-31 Martin Liska <mliska@suse.cz>
2721 * gcov.c (struct line_info): Remove it's typedef.
2722 (line_info::line_info): Add proper ctor.
2723 (line_info::has_block): Do not use a typedef.
2724 (struct source_info): Do not use typedef.
2725 (circuit): Likewise.
2726 (get_cycles_count): Likewise.
2727 (output_intermediate_file): Iterate via vector iterator.
2728 (add_line_counts): Use std::vector methods.
2729 (accumulate_line_counts): Likewise.
2730 (output_lines): Likewise.
2732 2017-10-31 Martin Liska <mliska@suse.cz>
2734 * gcov.c (struct source_info): Remove typedef.
2735 (source_info::source_info): Add proper ctor.
2736 (accumulate_line_counts): Use struct, not it's typedef.
2737 (output_gcov_file): Likewise.
2738 (output_lines): Likewise.
2739 (main): Do not allocate an array.
2740 (output_intermediate_file): Use size of vector container.
2741 (process_file): Resize the vector.
2742 (generate_results): Do not preallocate, use newly added vector
2744 (release_structures): Do not release sources.
2745 (find_source): Use vector methods.
2746 (add_line_counts): Do not use typedef.
2748 2017-10-31 Martin Liska <mliska@suse.cz>
2750 * doc/gcov.texi: Document that.
2751 * gcov.c (add_line_counts): Mark lines with a non-executed
2753 (output_line_beginning): Handle such lines.
2754 (output_lines): Pass new argument.
2755 (output_intermediate_file): Print it in intermediate format.
2757 2017-10-31 Martin Liska <mliska@suse.cz>
2759 * color-macros.h: New file.
2760 * diagnostic-color.c: Factor out color related to macros to
2762 * doc/gcov.texi: Document -k option.
2763 * gcov.c (INCLUDE_STRING): Include string.h.
2764 (print_usage): Add -k option.
2765 (process_args): Parse it.
2766 (pad_count_string): New function.
2767 (output_line_beginning): Likewise.
2768 (DEFAULT_LINE_START): New macro.
2769 (output_lines): Support color output.
2771 2017-10-31 Martin Liska <mliska@suse.cz>
2773 PR gcov-profile/82633
2774 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
2775 their interaction with GCOV infrastructure.
2776 * configure.ac: Add -fkeep-{inline,static}-functions to
2778 * configure: Regenerate.
2780 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
2783 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
2785 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
2788 * config/rs6000/rs6000.md (allocate_stack): Force update interval
2789 into a register if it does not fit into an immediate offset field.
2791 2017-10-31 Olivier Hainque <hainque@adacore.com>
2793 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
2795 2017-10-31 Julia Koval <julia.koval@intel.com>
2797 * config.gcc: Add gfniintrin.h.
2798 * config/i386/gfniintrin.h: New.
2799 * config/i386/i386-builtin-types.def
2800 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2801 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2802 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2803 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2804 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2805 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
2806 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
2807 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
2808 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
2809 V64QI_FTYPE_V64QI_V64QI_INT): New types.
2810 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2811 * config/i386/immintrin.h: Include gfniintrin.h.
2812 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
2814 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
2816 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
2818 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2820 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
2821 (arm_ashldi3_1bit): Remove pattern.
2822 (ashrdi3): Remove shift by 1 expansion.
2823 (arm_ashrdi3_1bit): Remove pattern.
2824 (lshrdi3): Remove shift by 1 expansion.
2825 (arm_lshrdi3_1bit): Remove pattern.
2826 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
2827 cost of ashldi3 by 1.
2828 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
2829 (<shift>di3_neon): Likewise.
2831 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2833 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
2834 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
2835 and (*aarch64_simd_mov<VQ:mode>).
2836 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
2837 pattern alternative.
2838 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
2839 attributes to match pattern alternative.
2841 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
2843 * config.gcc (powerpc*-*-*): Add emmintrin.h.
2844 * config/rs6000/emmintrin.h: New file.
2845 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
2847 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2849 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
2850 * (movdi_vfp_cortexa8): Remove pattern.
2852 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2854 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
2856 (Specific, alpha*-dec-osf5.1): Remove.
2857 (Specific, mips-sgi-irix5): Remove.
2858 (Specific, mips-sgi-irix6): Remove.
2860 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2863 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
2864 arguments to clear_bit_region_be.
2866 2017-10-30 Jim Wilson <wilson@tuliptree.org>
2868 * gimplify.c: Include memmodel.h.
2870 2017-10-30 Martin Jambor <mjambor@suse.cz>
2872 * omp-grid.c (grid_attempt_target_gridification): Also insert a
2873 condition whether loop should be executed at all.
2875 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
2877 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2878 gimple folding of vec_madd() intrinsics.
2879 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
2880 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
2881 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
2883 2017-10-30 Richard Biener <rguenther@suse.de>
2885 PR tree-optimization/82762
2887 2017-10-23 Richard Biener <rguenther@suse.de>
2889 PR tree-optimization/82129
2891 2017-08-01 Richard Biener <rguenther@suse.de>
2893 PR tree-optimization/81181
2894 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2895 (compute_antic): ... end of iteration here.
2897 2017-10-30 Joseph Myers <joseph@codesourcery.com>
2899 * doc/invoke.texi (C Dialect Options): Document -std=c17,
2900 -std=iso9899:2017 and -std=gnu17.
2901 * doc/standards.texi (C Language): Document C17 support.
2902 * doc/cpp.texi (Overview): Mention -std=c17.
2903 (Standard Predefined Macros): Document C11 and C17 values of
2904 __STDC_VERSION__. Do not refer to C99 support as incomplete.
2905 * doc/extend.texi (Inline): Do not list individual options for
2906 standards newer than C99.
2907 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
2909 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
2912 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
2914 * asan.c (asan_finish_file): Align asan globals array by shadow
2917 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2920 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
2921 (struct store_immediate_info): Add bitregion_start and bitregion_end
2923 (store_immediate_info::store_immediate_info): Add brs and bre
2924 arguments and initialize bitregion_{start,end} from those.
2925 (struct merged_store_group): Add bitregion_start, bitregion_end,
2926 align_base and mask fields. Drop unnecessary struct keyword from
2927 struct store_immediate_info. Add do_merge method.
2928 (clear_bit_region_be): Use memset instead of loop storing zeros.
2929 (merged_store_group::do_merge): New method.
2930 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
2931 stores as long as the surrounding bitregions have no gaps.
2932 (merged_store_group::merge_overlapping): Use do_merge.
2933 (merged_store_group::apply_stores): Test that bitregion_{start,end}
2934 is byte aligned, rather than requiring that start and width are
2935 byte aligned. Drop unnecessary struct keyword from
2936 struct store_immediate_info. Allocate and populate also mask array.
2937 Make start of the arrays relative to bitregion_start rather than
2938 start and size them according to bitregion_{end,start} difference.
2939 (struct imm_store_chain_info): Drop unnecessary struct keyword from
2940 struct store_immediate_info.
2941 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
2942 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
2943 struct keyword from struct store_immediate_info.
2944 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
2945 between stores as long as the surrounding bitregions have no gaps.
2947 (struct split_store): Add orig non-static data member.
2948 (split_store::split_store): Initialize orig to false.
2949 (find_constituent_stmts): Return store_immediate_info *, non-NULL
2950 if there is exactly a single original stmt. Change stmts argument
2951 to pointer from reference, if NULL, don't push anything to it. Add
2952 first argument, use it to optimize skipping over orig stmts that
2953 are known to be before bitpos already. Simplify.
2954 (split_group): Return unsigned int count how many stores are or
2955 would be needed rather than a bool. Add allow_unaligned argument.
2956 Change split_stores argument from reference to pointer, if NULL,
2957 only do a dry run computing how many stores would be produced.
2958 Rewritten algorithm to use both alignment and misalign if
2959 !allow_unaligned and handle bitfield stores with gaps.
2960 (imm_store_chain_info::output_merged_store): Set start_byte_pos
2961 from bitregion_start instead of start. Compute allow_unaligned
2962 here, if true, do 2 split_group dry runs to compute which one
2963 produces fewer stores and prefer aligned if equal. Punt if
2964 new count is bigger or equal than original before emitting any
2965 statements, rather than during that. Remove no longer needed
2966 new_ssa_names tracking. Replace num_stmts with
2967 split_stores.length (). Use 32-bit stack allocated entries
2968 in split_stores auto_vec. Try to reuse original store lhs/rhs1
2969 if possible. Handle bitfields with gaps.
2970 (pass_store_merging::execute): Ignore bitsize == 0 stores.
2971 Compute bitregion_{start,end} for the stores and construct
2972 store_immediate_info with that. Formatting fixes.
2974 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
2977 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
2978 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
2980 2017-10-29 Jim Wilson <wilson@tuliptree.org>
2982 * gimplify.c: Include tm_p.h.
2984 * common.opt (gcoff): Re-add as ignored option.
2985 (gcoff1, gcoff2, gcoff3): Likewise.
2987 * Makefile.in (OBJS): Delete sdbout.o.
2988 (GTFILES): Delete $(srcdir)/sdbout.c.
2989 * debug.h: Delete sdb_debug_hooks.
2990 * final.c: Delete sdbout.h include.
2991 (final_scan_insn): Delete SDB_DEBUG check.
2992 (rest_of_clean_state): Likewise.
2993 * output.h: Delete sdb_begin_function_line.
2996 * toplev.c: Delete sdbout.h include.
2997 (process_options): Delete SDB_DEBUG check.
2998 * tree-core.h (tree_type_common): Delete pointer field of
3000 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3001 TYPE_SYMTAB_POINTER.
3002 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3003 (TYPE_SYMTAB_IS_POINTER): Delete.
3004 (TYPE_SYMTAB_IS_DIE): Renumber.
3005 * xcoffout.c: Refer to former sdbout.c file.
3006 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3008 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3009 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3010 (Debugging Options): Delete -gcoff.
3011 (-gstabs): Delete SDB reference.
3013 (-gcoff@var{level}): Delete.
3014 * doc/passes.texi (Debugging information output): Delete SDB and
3015 sdbout.c references.
3016 * doc/tm.texi: Regenerate.
3017 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3018 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3020 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3021 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3022 (SDB_DEBUGGING_INFO): Delete.
3023 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3024 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3025 * target.def (output_source_filename): Delete COFF reference.
3027 * common.opt (gcoff): Delete.
3028 (gxcoff+): Update Negative chain.
3029 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3031 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3032 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3033 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3035 (expand_function_start): Change sdb reference to past tense.
3036 (expand_function_end): Change sdb reference to past tense.
3037 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3038 * opts.c (debug_type_names): Delete coff entry.
3039 (common_handle_option): Delete OPT_gcoff case.
3040 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3042 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3043 * config/cris/cris.h: Delete SDB reference in comment.
3044 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3045 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3046 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3047 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3049 (ix86_expand_prologue): Likewise.
3050 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3051 * config/ia64/ia64.h: Likewise.
3052 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3053 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3055 * config/mmix/mmix.h: Likewise.
3056 * config/nds32/nds32.c: Likewise.
3057 * config/stormy/storym16.h: Likewise.
3058 * config/visium/visium.h: Likewise.
3059 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3061 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3063 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3064 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3065 FRAME_POINTER_REGNUM point at high end of local var area.
3067 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3069 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3070 Move comment around. Do not reset best_edge for a copiable
3071 destination if the copy would cause a partition change.
3072 (better_edge_p): Remove redundant check.
3074 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3076 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3078 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3080 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3081 for math functions that have _Float<N> and _Float<N>X variants.
3082 (mathfn_built_in_2): Add support for math functions that have
3083 _Float<N> and _Float<N>X variants.
3084 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3085 (expand_builtin_mathfn_ternary): Add support for fma with
3086 _Float<N> and _Float<N>X variants.
3087 (expand_builtin): Likewise.
3088 (fold_builtin_3): Likewise.
3089 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3090 create math function _Float<N> and _Float<N>X variants as external
3092 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3093 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3094 the __builtin_ prefix and if not strict ansi, without the prefix.
3095 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3096 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3097 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3098 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3099 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3100 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3101 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3102 function signatures for fma _Float<N> and _Float<N>X variants.
3103 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3104 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3105 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3106 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3107 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3108 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3109 * gencfn-macros.c (print_case_cfn): Add support for math functions
3110 that have _Float<N> and _Float<N>X variants.
3111 (print_define_operator_list): Likewise.
3112 (fltfn_suffixes): Likewise.
3114 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3115 for math functions that have _Float<N> and _Float<N>X variants.
3116 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3117 and _Float<N>X variants.
3118 (COPYSIGN): Likewise.
3121 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3122 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3124 (integer_valued_read_call_p): Likewise.
3125 * fold-const-call.c (fold_const_call_ss): Likewise.
3126 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3127 _Float<N> and _Float<N>X variants.
3128 (fold_const_call_ssss): Add support for fma _Float<N> and
3129 _Float<N>X variants.
3130 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3131 support for copysign and fma _Float<N> and _Float<N>X variants.
3132 (backprop::process_builtin_call_use): Likewise.
3133 * tree-call-cdce.c (can_test_argument_range); Add support for
3134 sqrt _Float<N> and _Float<N>X variants.
3135 (edom_only_function): Likewise.
3136 (get_no_error_domain): Likewise.
3137 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3138 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3139 copysign _Float<N> and _Float<N>X variants.
3140 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3141 handled by machine independent code.
3142 (FMAF128): Likewise.
3143 * doc/cpp.texi (Common Predefined Macros): Document defining
3144 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3145 fma _Float<N> and _Float<N>X variants.
3147 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3150 * config/i386/i386-modes.def (CCFPU): Remove definition.
3151 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3152 (ix86_cc_modes_compatible): Ditto.
3153 (ix86_expand_carry_flag_compare): Ditto.
3154 (ix86_expand_int_movcc): Ditto.
3155 (ix86_expand_int_addcc): Ditto.
3156 (ix86_reverse_condition): Ditto.
3157 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3158 Return true/false for unordered/ordered fp comparisons.
3159 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3160 (ix86_prepare_fp_compare_args): Update for rename.
3161 (ix86_expand_fp_compare): Update for rename. Generate unordered
3162 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3163 (ix86_expand_sse_compare_and_jump): Ditto.
3164 * config/i386/predicates.md (fcmov_comparison_operator):
3165 Remove CCFPU mode handling.
3166 (ix86_comparison_operator): Ditto.
3167 (ix86_carry_flag_operator): Ditto.
3168 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3169 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3170 (*cmpu<mode>_cc_i387): Ditto.
3171 (FPCMP): Remove mode iterator.
3172 (unord): Remove mode attribute.
3173 (unord_subst): New define_subst transformation
3174 (unord): New define_subst attribute.
3176 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3177 (*cmpi<unord>xf_i387): Ditto.
3178 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3179 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3180 using unord_subst transformation.
3181 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3182 (round_saeonly): Also handle CCFP mode.
3183 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3184 Remove UNSPEC_SAHF unspec handling.
3186 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3188 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3190 2017-10-27 Jeff Law <law@redhat.com>
3192 * gimple-ssa-sprintf.c: Include domwalk.h.
3193 (class sprintf_dom_walker): New class, derived from dom_walker.
3194 (sprintf_dom_walker::before_dom_children): New function.
3195 (struct call_info): Moved into sprintf_dom_walker class
3196 (compute_formath_length, handle_gimple_call): Likewise.
3197 (sprintf_length::execute): Call the dominator walker rather
3198 than walking the statements.
3200 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3201 gimple statement locations.
3202 (check_array_bounds): Corresponding changes. Get the statement's
3203 location directly from wi->stmt.
3205 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3208 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3210 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3212 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3213 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3215 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3218 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3219 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3220 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3221 maybe_get_pool_constant.
3222 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3225 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3227 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3228 2.26 caveat. Update gas and gld versions.
3229 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3232 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3234 * cgraph.h (set_malloc_flag): Declare.
3235 * cgraph.c (set_malloc_flag_1): New function.
3236 (set_malloc_flag): Likewise.
3237 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3238 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3240 (read_ipa_call_summary): Add support for reading is_return_callee.
3241 (write_ipa_call_summary): Stream is_return_callee.
3242 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3243 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3244 ipa-prop.h, ipa-fnsummary.h.
3245 (pure_const_names): Change to static.
3246 (malloc_state_e): Define.
3247 (malloc_state_names): Define.
3248 (funct_state_d): Add field malloc_state.
3249 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3250 (check_retval_uses): New function.
3251 (malloc_candidate_p): Likewise.
3252 (analyze_function): Add support for malloc attribute.
3253 (pure_const_write_summary): Stream malloc_state.
3254 (pure_const_read_summary): Add support for reading malloc_state.
3255 (dump_malloc_lattice): New function.
3256 (propagate_malloc): New function.
3257 (warn_function_malloc): New function.
3258 (ipa_pure_const::execute): Call propagate_malloc and
3259 ipa_free_fn_summary.
3260 (pass_local_pure_const::execute): Add support for malloc attribute.
3261 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3262 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3264 2017-10-27 Martin Liska <mliska@suse.cz>
3266 PR gcov-profile/82457
3267 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3268 for fork-like functions to be properly instrumented.
3270 2017-10-27 Richard Biener <rguenther@suse.de>
3273 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3274 info when we redirected EH.
3276 2017-10-26 Michael Collison <michael.collison@arm.com>
3278 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3280 (<optab>_trunchf<GPI:mode>2: New pattern.
3281 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3282 * config/aarch64/iterators.md (wv): New mode attribute.
3283 (vf, VF): New mode attributes.
3284 (vgp, VGP): New mode attributes.
3285 (s): Update attribute with SImode and DImode prefixes.
3287 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3289 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3290 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3291 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3292 (nios2_option_overide): Initialize it. Don't allow R0-relative
3293 addressing with PIC.
3294 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3295 (nios2_symbolic_constant_p): Likewise.
3296 (nios2_legitimate_address_p): Likewise.
3297 (nios2_r0rel_section_name_p): New.
3298 (nios2_symbol_ref_in_r0rel_data_p): New.
3299 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3300 (r0rel_constant_p): New.
3301 (nios2_print_operand_address): Handle r0rel_constant_p.
3302 (nios2_cdx_narrow_form_p): Likewise.
3303 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3304 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3305 (Nios II Options): Document -mr0rel-sec.
3307 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3309 * config/nios2/nios2.c: Include xregex.h.
3310 (nios2_gprel_sec_regex): New.
3311 (nios2_option_overide): Initialize it. Don't allow GP-relative
3312 addressing with PIC.
3313 (nios2_small_section_name_p): Check for regex match.
3314 * config/nios2/nios2.opt (mgprel-sec=): New option.
3315 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3316 (Nios II Options): Document -mgprel-sec.
3318 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3320 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3322 2017-10-26 Tom de Vries <tom@codesourcery.com>
3324 PR tree-optimization/82707
3325 * gimple.c (gimple_copy): Fix unsharing of
3326 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3328 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3330 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3331 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3332 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3333 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3334 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3335 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3336 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3337 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3338 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3339 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3340 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3341 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3342 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
3343 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
3344 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
3345 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
3346 _mm512_mask_cmpunord_ps_mask): New intrinsics.
3348 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3350 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
3352 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
3353 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
3354 warning to rs6000.c. Remove the Undocumented flag, since it has
3356 (-mabi=ibmlongdouble): Likewise.
3357 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
3358 already set, set the default format for long double.
3359 (rs6000_debug_reg_global): Print whether long double is IBM or
3361 (rs6000_option_override_internal): Rework setting long double
3362 format. Only warn if the user is changing the long double default
3363 and they did not use -Wno-psabi.
3364 * doc/invoke.texi (PowerPC options): Update the documentation for
3365 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3367 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3368 Alan Hayward <alan.hayward@arm.com>
3369 David Sherwood <david.sherwood@arm.com>
3371 * rtl.h (wider_subreg_mode): New function.
3372 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
3373 rather than an unsigned int *.
3374 * ira-color.c (regno_max_ref_width): Replace with...
3375 (regno_max_ref_mode): ...this new variable.
3376 (coalesced_pseudo_reg_slot_compare): Update accordingly.
3377 Use wider_subreg_mode.
3378 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
3379 rather than an unsigned int *.
3380 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
3381 (process_alt_operands): Likewise.
3382 (invariant_p): Likewise.
3383 * lra-spills.c (assign_mem_slot): Likewise.
3384 (add_pseudo_to_slot): Likewise.
3385 * lra.c (collect_non_operand_hard_regs): Likewise.
3386 (add_regs_to_insn_regno_info): Likewise.
3387 * reload1.c (regno_max_ref_width): Replace with...
3388 (regno_max_ref_mode): ...this new variable.
3389 (reload): Update accordingly. Update call to
3390 ira_sort_regnos_for_alter_reg.
3391 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
3392 (init_eliminable_invariants): Update to use regno_max_ref_mode.
3393 (scan_paradoxical_subregs): Likewise.
3395 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3397 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
3398 (aarch64_frame): Add emit_frame_chain boolean.
3399 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3400 Move eh_return case to aarch64_layout_frame.
3401 (aarch64_layout_frame): Initialize emit_frame_chain.
3402 (aarch64_expand_prologue): Use emit_frame_chain.
3404 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3406 * config/aarch64/aarch64.c (aarch64_layout_frame):
3407 Ensure LR is always stored at the bottom of the callee-saves.
3408 Remove rarely used frame layout which saves callee-saves at top of
3409 frame, so the store of LR can be used as a valid probe in all cases.
3411 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3413 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
3414 Improve unaligned TImode/TFmode base/offset split.
3416 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3417 Alan Hayward <alan.hayward@arm.com>
3418 David Sherwood <david.sherwood@arm.com>
3420 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
3421 * combine.c (find_single_use_1): Likewise.
3422 (expand_field_assignment): Likewise.
3423 (move_deaths): Likewise.
3424 * lra-constraints.c (simplify_operand_subreg): Likewise.
3425 (curr_insn_transform): Likewise.
3426 * lra.c (collect_non_operand_hard_regs): Likewise.
3427 (add_regs_to_insn_regno_info): Likewise.
3428 * rtlanal.c (reg_referenced_p): Likewise.
3429 (covers_regno_no_parallel_p): Likewise.
3431 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3433 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
3434 Don't print any bits outside the precision of the value.
3435 * wide-int.cc (test_printing): Add some new tests.
3437 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3439 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
3440 supports -xbrace_comment option.
3441 * configure: Regenerate.
3442 * config.in: Regenerate.
3443 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
3444 (ASM_CPU_SPEC): Use it.
3446 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3448 * target.def (static_rtx_alignment): New hook.
3449 * targhooks.h (default_static_rtx_alignment): Declare.
3450 * targhooks.c (default_static_rtx_alignment): New function.
3451 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
3452 * doc/tm.texi: Regenerate.
3453 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
3454 instead of targetm.constant_alignment. Remove call to
3456 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3457 (cris_preferred_mininum_alignment): New function, split out from...
3458 (cris_constant_alignment): ...here.
3459 (cris_static_rtx_alignment): New function.
3460 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
3462 (ix86_constant_alignment): ...here.
3463 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3464 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3465 (mmix_static_rtx_alignment): New function.
3466 * config/spu/spu.c (spu_static_rtx_alignment): New function.
3467 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3469 2017-10-26 Tamar Christina <tamar.christina@arm.com>
3472 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
3473 Add flag_trapping_math and flag_fp_int_builtin_inexact.
3475 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
3477 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
3478 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
3479 mark as a sign-extending load.
3480 (local_pic_load_u): Define.
3482 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
3485 * fold-const.c (operand_equal_for_comparison_p): Also return true
3486 if ARG0 is a simple variant of ARG1 with narrower precision.
3487 (fold_ternary_loc): Always pass unstripped operands to the predicate.
3489 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
3491 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
3493 * i386.h (processor_costs): Add gather_static, gather_per_elt,
3494 scatter_static, scatter_per_elt.
3495 * x86-tune-costs.h: Add new cost entries.
3497 2017-10-25 Richard Biener <rguenther@suse.de>
3499 * tree-ssa-sccvn.h (vn_eliminate): Declare.
3500 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
3501 class pass_fre): Move to ...
3502 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
3503 class pass_fre): ... here and adjust for statistics.
3505 2017-10-25 Jakub Jelinek <jakub@redhat.com>
3508 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
3509 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
3510 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
3511 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
3514 2017-10-25 Richard Biener <rguenther@suse.de>
3516 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
3517 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
3518 eliminate_push_avail, eliminate_insert): Move inside...
3519 (class eliminate_dom_walker): ... this class in preparation
3521 (fini_eliminate): Remove by merging with ...
3522 (eliminate): ... this function. Adjust for class changes.
3523 (pass_pre::execute): Remove fini_eliminate call.
3524 (pass_fre::execute): Likewise.
3526 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3529 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
3530 (VPERMI2, VPERMI2I): New mode iterators.
3531 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
3532 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
3534 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
3535 mode iterator. Remove 3 old define_insn patterns.
3536 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
3537 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
3538 VPERMI2 mode iterator, remove the other two expanders.
3539 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
3540 to use VPERMI2 mode iterator, add another alternative for vpermi2*
3541 instructions, remove the other two patterns.
3542 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
3543 mode iterator, remove the other two patterns.
3544 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
3545 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
3546 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
3547 and adjust argument order accordingly.
3548 (ix86_expand_vec_perm): Adjust caller.
3549 (expand_vec_perm_1): Likewise.
3550 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
3551 (expand_vec_perm_vpermt2_vpshub2): ... this.
3552 (ix86_expand_vec_perm_const_1): Adjust caller.
3553 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
3556 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
3557 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
3558 (vec_shl_<mode>): Remove unused expander.
3559 (avx512bw_<shift_insn><mode>3): New define_insn.
3560 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
3561 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
3563 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
3566 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
3569 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3571 PR rtl-optimization/82396
3572 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
3573 (autopref_multipass_init): Simplify initialization.
3574 (autopref_rank_data): Simplify sort order.
3575 * gcc/sched-int.h (autopref_multipass_data_): Remove
3576 multi_mem_insn_p, min_offset and max_offset.
3578 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3581 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3582 Check special value of flag_omit_frame_pointer.
3583 (aarch64_can_eliminate): Likewise.
3584 (aarch64_override_options_after_change_1): Simplify handling of
3585 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
3587 2017-10-24 Richard Biener <rguenther@suse.de>
3589 PR tree-optimization/82697
3590 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
3591 zero for conditional load and unconditional store.
3593 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3595 * doc/install.texi: Document bootstrap-cet.
3597 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3600 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
3601 ENDBR instruction at function entrance if function is only
3604 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3607 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
3608 patterns to better describe from which operation the CF is computed.
3609 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
3610 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
3611 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
3612 is 0, use _0 suffixed expanders instead of emitting a comparison
3615 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
3617 * config/i386/i386.md(*movsf_internal, *movdf_internal):
3618 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
3620 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
3623 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
3624 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
3625 * loop-iv.c (iv_get_reaching_def): Likewise.
3626 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
3627 variable is promoted and the partition contains undefined values.
3629 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3631 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
3633 (nios2_address_cost): Define.
3634 (nios2_legitimize_address): Recognize (exp + constant) directly.
3635 (TARGET_ADDRESS_COST): Define.
3637 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3639 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
3640 (nios2_symbolic_memory_operand_p): Declare.
3641 (nios2_split_large_constant): Declare.
3642 (nios2_split_symbolic_memory_operand): Declare.
3643 * config/nios2/nios2.c: Adjust includes.
3644 (nios2_symbolic_constant_allowed): New.
3645 (nios2_symbolic_constant_p): New.
3646 (nios2_plus_symbolic_constant_p): New.
3647 (nios2_valid_addr_expr_p): Recognize addresses involving
3649 (nios2_legitimate_address_p): Likewise, also LO_SUM.
3650 (nios2_symbolic_memory_operand_p): New.
3651 (nios2_large_constant_p): New.
3652 (nios2_split_large_constant): New.
3653 (nios2_split_plus_large_constant): New.
3654 (nios2_split_symbolic_memory_operand): New.
3655 (nios2_legitimize_address): Code refactoring. Handle addresses
3656 involving symbolic constants.
3657 (nios2_emit_move_sequence): Likewise.
3658 (nios2_print_operand): Improve error output.
3659 (nios2_print_operand_address): Handle LO_SUM.
3660 (nios2_cdx_narrow_form_p): Likewise.
3661 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
3662 operands involving symbolic constants.
3663 (movhi_internal, movsi_internal): Likewise.
3664 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
3665 (extendhisi2, extendqi<mode>2): Likewise.
3667 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3669 * tree-pass.h (PROP_rtl_split_insns): Define.
3670 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
3672 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3674 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
3676 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3679 * target.def (const_not_ok_for_debug_p): Default to
3680 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
3681 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
3682 * targhooks.c (default_const_not_ok_for_debug_p): New function.
3683 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
3684 which targetm.const_not_ok_for_debug_p returned true.
3685 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
3687 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
3689 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
3690 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
3691 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
3693 (ix86_const_not_ok_for_debug_p): New function.
3694 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
3695 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
3697 2017-10-23 David Malcolm <dmalcolm@redhat.com>
3700 * system.h: Conditionally include "unique-ptr.h" if
3701 INCLUDE_UNIQUE_PTR is defined.
3702 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
3703 of defining INCLUDE_UNIQUE_PTR before including "system.h".
3705 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
3707 * config/rl78/rl78.md: New define_expand "subdi3".
3709 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
3712 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
3713 DF_REF_INSN if DF_REF_INSN_INFO is false.
3715 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
3717 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
3718 xmm_move instead of sse_move.
3719 (sse_store_index): New function.
3720 (ix86_register_move_cost): Be more sensible about mismatch stall;
3721 model AVX moves correctly; make difference between sse->integer and
3723 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
3724 moves; make difference between SSE and AVX.
3725 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
3726 and zmm_move. Increase size of sse load and store tables;
3727 add unaligned load and store tables; add ssemmx_to_integer.
3728 * x86-tune-costs.h: Update all entries according to real
3729 move latencies from Agner Fog's manual and chip documentation.
3731 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3734 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
3735 * config/i386/constraints.md (Wf): New constraint.
3736 * config/i386/i386.md (UNSPEC_SBB): New unspec.
3737 (cmp<dwi>_doubleword): Removed.
3738 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
3739 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
3740 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
3741 expand with cmp<dwi>_doubleword. For LTU and GEU use
3742 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
3744 * common.opt (gcolumn-info): Enable by default.
3745 * doc/invoke.texi (gcolumn-info): Document new default.
3747 2017-10-23 Richard Biener <rguenther@suse.de>
3749 PR tree-optimization/82672
3750 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
3751 Fold the stmt if we propagated into it.
3753 2017-10-23 Richard Biener <rguenther@suse.de>
3755 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
3756 (bitmap_remove_expr_from_set): ... this. All callers call this
3757 for non-constant values.
3758 (bitmap_set_subtract): Rename to...
3759 (bitmap_set_subtract_expressions): ... this. Adjust and
3761 (bitmap_set_contains_value): Remove superfluous check.
3762 (bitmap_set_replace_value): Inline into single caller ...
3763 (bitmap_value_replace_in_set): ... here and simplify.
3764 (dependent_clean): Merge into ...
3765 (clean): ... this using an overload. Adjust.
3766 (prune_clobbered_mems): Adjust.
3767 (compute_antic_aux): Likewise.
3768 (compute_partial_antic_aux): Likewise.
3770 2017-10-23 Richard Biener <rguenther@suse.de>
3772 PR tree-optimization/82129
3774 2017-08-01 Richard Biener <rguenther@suse.de>
3776 PR tree-optimization/81181
3777 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3778 (compute_antic): ... end of iteration here.
3780 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3782 * target.def (starting_frame_offset): New hook.
3783 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
3784 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
3785 * doc/tm.texi.in: Regenerate.
3786 * hooks.h (hook_hwi_void_0): Declare.
3787 * hooks.c (hook_hwi_void_0): New function.
3788 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
3789 STARTING_FRAME_OFFSET.
3790 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
3791 * reload1.c (reload): Likewise.
3792 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
3793 instead of STARTING_FRAME_OFFSET.
3794 * function.c (try_fit_stack_local): Likewise.
3795 (assign_stack_local_1): Likewise
3796 (instantiate_virtual_regs): Likewise.
3797 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
3798 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
3799 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
3800 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
3801 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
3802 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
3803 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
3804 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
3805 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
3806 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
3807 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
3808 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
3809 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
3810 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
3811 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
3812 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
3813 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
3814 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
3815 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
3816 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
3817 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
3818 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
3819 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
3820 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
3821 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
3822 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
3823 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
3824 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
3825 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
3826 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
3827 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
3828 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
3829 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
3830 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
3831 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
3832 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
3833 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
3834 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
3835 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
3836 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
3837 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
3838 * config/avr/avr.c (avr_starting_frame_offset): Make static and
3839 return a HOST_WIDE_INT.
3840 (avr_builtin_setjmp_frame_value): Use it instead of
3841 STARTING_FRAME_OFFSET.
3842 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3843 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
3844 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
3846 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3847 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
3848 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
3849 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3850 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
3851 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
3852 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3853 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
3854 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
3855 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3856 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
3857 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
3859 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3860 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
3861 * config/mips/mips.c (mips_compute_frame_info): Refer to
3862 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3863 (mips_starting_frame_offset): New function.
3864 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3865 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
3866 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
3867 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
3868 and return a HOST_WIDE_INT.
3869 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3870 (mmix_initial_elimination_offset): Refer to
3871 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3872 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
3873 * config/pa/pa.c (pa_starting_frame_offset): New function.
3874 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
3875 (pa_expand_prologue): Likewise.
3876 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3877 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
3878 !FRAME_GROWS_DOWNWARD handling to...
3879 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3880 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
3881 !FRAME_GROWS_DOWNWARD handling to...
3882 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3883 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
3884 !FRAME_GROWS_DOWNWARD handling to...
3885 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3886 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
3888 (rs6000_starting_frame_offset): New function.
3889 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
3890 !FRAME_GROWS_DOWNWARD handling to...
3891 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3892 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
3893 !FRAME_GROWS_DOWNWARD handling to...
3894 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3895 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
3896 !FRAME_GROWS_DOWNWARD handling to...
3897 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3898 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
3899 (rs6000_starting_frame_offset): New function.
3900 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
3901 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
3902 * config/vax/vax.c (vax_starting_frame_offset): New function.
3903 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
3904 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3905 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
3906 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
3907 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3908 * system.h (STARTING_FRAME_OFFSET): Poison.
3910 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3912 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3913 SCALAR_TYPE_MODE instead of TYPE_MODE.
3915 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3916 Alan Hayward <alan.hayward@arm.com>
3917 David Sherwood <david.sherwood@arm.com>
3919 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
3921 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3922 Alan Hayward <alan.hayward@arm.com>
3923 David Sherwood <david.sherwood@arm.com>
3925 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
3927 2017-10-23 Richard Biener <rguenther@suse.de>
3929 PR tree-optimization/82129
3930 * tree-ssa-pre.c (bitmap_set_and): Remove.
3931 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
3932 canonicalizing expressions in the set to those with lowest
3933 ID rather than taking that from the first edge.
3935 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3937 * combine.c (rtx_equal_for_field_assignment_p): Use
3938 byte_lowpart_offset.
3940 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3941 Alan Hayward <alan.hayward@arm.com>
3942 David Sherwood <david.sherwood@arm.com>
3944 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
3945 to a SUBREG_PROMOTED_VAR.
3947 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3948 Alan Hayward <alan.hayward@arm.com>
3949 David Sherwood <david.sherwood@arm.com>
3951 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
3952 (expand_debug_source_expr): Likewise.
3953 * combine.c (combine_simplify_rtx): Likewise.
3954 * cse.c (fold_rtx): Likewise.
3955 * optabs.c (expand_float): Likewise.
3956 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3957 (simplify_binary_operation_1): Likewise.
3959 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3960 Alan Hayward <alan.hayward@arm.com>
3961 David Sherwood <david.sherwood@arm.com>
3963 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
3964 (record_promoted_value): Likewise.
3965 * expr.c (expand_expr_real_2): Likewise.
3966 * ree.c (update_reg_equal_equiv_notes): Likewise.
3967 (combine_set_extension): Likewise.
3968 * rtlanal.c (low_bitmask_len): Likewise.
3969 * simplify-rtx.c (neg_const_int): Likewise.
3970 (simplify_binary_operation_1): Likewise.
3972 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3973 Alan Hayward <alan.hayward@arm.com>
3974 David Sherwood <david.sherwood@arm.com>
3976 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
3977 * regcprop.c (maybe_mode_change): Likewise.
3978 * reload1.c (alter_reg): Likewise.
3980 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3982 * inchash.h (inchash::hash::add_wide_int): New function.
3983 * lto-streamer-out.c (hash_tree): Use it.
3985 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3987 * inchash.h (inchash::hash::add_wide_int): Rename to...
3988 (inchash::hash::add_hwi): ...this.
3989 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
3990 (polymorphic_call_target_hasher::hash): Likewise.
3991 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
3992 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
3993 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
3994 * lto-streamer-out.c (hash_tree): Likewise.
3995 * optc-save-gen.awk: Likewise.
3996 * tree.c (add_expr): Likewise.
3998 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4001 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4002 for ordered inequality comparisons even with TARGET_IEEE_FP.
4004 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4007 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4008 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4009 Expand with cmp<dwi>_doubleword.
4011 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4013 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4014 List CET intrinsics.
4015 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4016 specific to -fcf-protection option.
4018 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4020 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4021 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4022 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4023 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4024 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4025 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4026 (extra_objs): Add cet.o for Linux/x86 targets.
4027 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4028 * config/i386/cet.c: New file.
4029 * config/i386/cetintrin.h: Likewise.
4030 * config/i386/t-cet: Likewise.
4031 * config/i386/cpuid.h (bit_SHSTK): New.
4032 (bit_IBT): Likewise.
4033 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4034 pass IBT and SHSTK bits.
4035 * config/i386/i386-builtin-types.def
4036 (VOID_FTYPE_UNSIGNED_PVOID): New.
4037 (VOID_FTYPE_UINT64_PVOID): Likewise.
4038 * config/i386/i386-builtin.def: Add CET intrinsics.
4039 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4040 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4041 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4042 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4044 * config/i386/i386.c (rest_of_insert_endbranch): New.
4045 (pass_data_insert_endbranch): Likewise.
4046 (pass_insert_endbranch): Likewise.
4047 (make_pass_insert_endbranch): Likewise.
4048 (ix86_notrack_prefixed_insn_p): Likewise.
4049 (ix86_target_string): Add -mibt, -mshstk flags.
4050 (ix86_option_override_internal): Add flag_cf_protection
4052 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4053 (ix86_print_operand): Add 'notrack' prefix output.
4054 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4055 (ix86_expand_builtin): Expand CET intrinsics.
4056 (x86_output_mi_thunk): Add 'endbranch' instruction.
4057 * config/i386/i386.h (TARGET_IBT): New.
4058 (TARGET_IBT_P): Likewise.
4059 (TARGET_SHSTK): Likewise.
4060 (TARGET_SHSTK_P): Likewise.
4061 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4062 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4063 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4064 (builtin_setjmp_setup): New pattern.
4065 (builtin_longjmp): Likewise.
4066 (rdssp<mode>): Likewise.
4067 (incssp<mode>): Likewise.
4068 (saveprevssp): Likewise.
4069 (rstorssp): Likewise.
4070 (wrss<mode>): Likewise.
4071 (wruss<mode>): Likewise.
4072 (setssbsy): Likewise.
4073 (clrssbsy): Likewise.
4074 (nop_endbr): Likewise.
4075 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4077 * config/i386/immintrin.h: Include <cetintrin.h>.
4078 * config/i386/linux-common.h
4079 (file_end_indicate_exec_stack_and_cet): New prototype.
4080 (TARGET_ASM_FILE_END): New.
4082 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4084 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4085 latencies instead of having separate table; make difference between
4086 integer and float costs.
4087 * i386.h (processor_costs): Remove scalar_stmt_cost,
4088 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4089 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4091 * x86-tune-costs.h: Remove entries which has been removed in
4092 procesor_costs from all tables; make cond_taken_branch_cost
4093 and cond_not_taken_branch_cost COST_N_INSNS based.
4095 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4097 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4099 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4101 * config/i386/i386.md (isa): Remove fma_avx512f.
4102 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4103 <avx512>_fmadd_<mode>_mask3<round_name>,
4104 <avx512>_fmsub_<mode>_mask<round_name>,
4105 <avx512>_fmsub_<mode>_mask3<round_name>,
4106 <avx512>_fnmadd_<mode>_mask<round_name>,
4107 <avx512>_fnmadd_<mode>_mask3<round_name>,
4108 <avx512>_fnmsub_<mode>_mask<round_name>,
4109 <avx512>_fnmsub_<mode>_mask3<round_name>,
4110 <avx512>_fmaddsub_<mode>_mask<round_name>,
4111 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4112 <avx512>_fmsubadd_<mode>_mask<round_name>,
4113 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4114 (*vec_widen_umult_even_v16si<mask_name>,
4115 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4116 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4118 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4120 * extend.texi: Add 'nocf_check' documentation.
4121 * gimple.texi: Add second parameter to
4122 gimple_build_call_from_tree.
4123 * invoke.texi: Add -fcf-protection documentation.
4124 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4126 2017-10-20 Richard Biener <rguenther@suse.de>
4128 PR tree-optimization/82473
4129 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4130 the largest input type.
4132 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4134 * c-attribs.c (handle_nocf_check_attribute): New function.
4135 (c_common_attribute_table): Add 'nocf_check' handling.
4136 * gimple-parser.c: Add second argument NULL to
4137 gimple_build_call_from_tree.
4138 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4139 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4141 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4142 * common.opt: Add fcf-protection flag.
4143 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4144 * flag-types.h: Add enum cf_protection_level.
4145 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4146 Add 'nocf_check' attribute propagation to gimple call.
4147 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4148 (gimple_build_call_from_tree): Update prototype.
4149 (gimple_call_nocf_check_p): New function.
4150 (gimple_call_set_nocf_check): Likewise.
4151 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4152 * ipa-icf.c: Add nocf_check attribute in statement hash.
4153 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4154 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4155 * toplev.c (process_options): Add flag_cf_protection handling.
4157 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4159 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4161 2017-10-20 Richard Biener <rguenther@suse.de>
4163 PR tree-optimization/82603
4164 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4165 remove false predicated stores.
4167 2017-10-20 Richard Biener <rguenther@suse.de>
4169 * graphite-isl-ast-to-gimple.c
4170 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4171 Remove return value and simplify, dump copied stmt after lhs
4173 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4174 Reduce dump verbosity.
4175 (gsi_insert_earliest): Likewise.
4176 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4177 * graphite.c (print_global_statistics): Adjust dumping.
4178 (print_graphite_scop_statistics): Likewise.
4179 (print_graphite_statistics): Do not dump loops here.
4180 (graphite_transform_loops): But here.
4182 2017-10-20 Nicolas Roche <roche@adacore.com>
4184 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4185 * configure: Regenerate.
4187 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4190 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4191 functions when optimizing replace GIMPLE_RETURN stmts with
4192 calls to __builtin_unreachable ().
4195 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4196 for -fsanitize=thread link of executables.
4197 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4198 link of executables.
4201 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4203 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4204 define_insns for logical vector shifts to use VI248_AVX512BW
4205 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4206 condition, useless isa and prefix attributes. Change the first
4207 2 of these define_insns to ...
4208 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4209 define_insn for avx512vl.
4210 (<shift_insn><mode>3): ... and this, new define_insn without
4211 masking for non-avx512vl.
4214 * config/i386/sse.md (*andnot<mode>3,
4215 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4216 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4217 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4218 not applied use empty suffix even for TARGET_AVX512VL.
4219 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4220 is applied, supply evex,evex or evex,evex,evex instead of just
4223 2017-10-20 Julia Koval <julia.koval@intel.com>
4225 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4226 (OPTION_MASK_ISA_GFNI_UNSET): New.
4227 (ix86_handle_option): Handle OPT_mgfni.
4228 * config/i386/cpuid.h (bit_GFNI): New.
4229 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4230 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4231 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4232 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4233 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4234 * config/i386/i386.opt: Add mgfni.
4236 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4238 * config/msp430/msp430.c (msp430_option_override): Disable
4239 -fdelete-null-pointer-checks.
4240 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4242 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4244 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4245 of x87 and SSE instructions.
4247 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4249 * asan.c (create_cond_insert_point): Do not update edge count.
4250 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4251 (afdo_propagate_circuit): Likewise.
4252 (afdo_calculate_branch_prob): Likewise.
4253 (afdo_annotate_cfg): Likewise.
4254 * basic-block.h (struct edge_def): Remove count.
4255 (edge_def::count): New accessor.
4256 * bb-reorder.c (rotate_loop): Update.
4257 (find_traces_1_round): Update.
4258 (connect_traces): Update.
4259 (sanitize_hot_paths): Update.
4260 * cfg.c (unchecked_make_edge): Update.
4261 (make_single_succ_edge): Update.
4262 (check_bb_profile): Update.
4263 (dump_edge_info): Update.
4264 (update_bb_profile_for_threading): Update.
4265 (scale_bbs_frequencies_int): Update.
4266 (scale_bbs_frequencies_gcov_type): Update.
4267 (scale_bbs_frequencies_profile_count): Update.
4268 (scale_bbs_frequencies): Update.
4269 * cfganal.c (connect_infinite_loops_to_exit): Update.
4270 * cfgbuild.c (compute_outgoing_frequencies): Update.
4271 (find_many_sub_basic_blocks): Update.
4272 * cfgcleanup.c (try_forward_edges): Update.
4273 (try_crossjump_to_edge): Update
4274 * cfgexpand.c (expand_gimple_cond): Update
4275 (expand_gimple_tailcall): Update
4276 (construct_exit_block): Update
4277 * cfghooks.c (verify_flow_info): Update
4278 (redirect_edge_succ_nodup): Update
4279 (split_edge): Update
4280 (make_forwarder_block): Update
4281 (duplicate_block): Update
4282 (account_profile_record): Update
4283 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4284 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4285 * cfgloopmanip.c (scale_loop_profile): Update.
4287 (lv_adjust_loop_entry_edge): Update.
4288 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4289 (force_nonfallthru_and_redirect): Update.
4290 (purge_dead_edges): Update.
4291 (rtl_flow_call_edges_add): Update.
4292 * cgraphunit.c (init_lowered_empty_function): Update.
4293 (cgraph_node::expand_thunk): Update.
4294 * gimple-pretty-print.c (dump_probability): Update.
4295 (dump_edge_probability): Update.
4296 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4297 * haifa-sched.c (sched_create_recovery_edges): Update.
4298 * hsa-gen.c (convert_switch_statements): Update.
4299 * ifcvt.c (dead_or_predicable): Update.
4300 * ipa-inline-transform.c (inline_transform): Update.
4301 * ipa-split.c (split_function): Update.
4302 * ipa-utils.c (ipa_merge_profiles): Update.
4303 * loop-doloop.c (add_test): Update.
4304 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4305 * lto-streamer-in.c (input_cfg): Update.
4306 (input_function): Update.
4307 * lto-streamer-out.c (output_cfg): Update.
4308 * modulo-sched.c (sms_schedule): Update.
4309 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4310 * predict.c (maybe_hot_edge_p): Update.
4311 (unlikely_executed_edge_p): Update.
4312 (probably_never_executed_edge_p): Update.
4313 (dump_prediction): Update.
4314 (drop_profile): Update.
4315 (propagate_unlikely_bbs_forward): Update.
4316 (determine_unlikely_bbs): Update.
4317 (force_edge_cold): Update.
4318 * profile.c (compute_branch_probabilities): Update.
4319 * reg-stack.c (better_edge): Update.
4320 * shrink-wrap.c (handle_simple_exit): Update.
4321 * tracer.c (better_p): Update.
4322 * trans-mem.c (expand_transaction): Update.
4323 (split_bb_make_tm_edge): Update.
4324 * tree-call-cdce.c: Update.
4325 * tree-cfg.c (gimple_find_sub_bbs): Update.
4326 (gimple_split_edge): Update.
4327 (gimple_duplicate_sese_region): Update.
4328 (gimple_duplicate_sese_tail): Update.
4329 (gimple_flow_call_edges_add): Update.
4330 (insert_cond_bb): Update.
4331 (execute_fixup_cfg): Update.
4332 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4333 * tree-complex.c (expand_complex_div_wide): Update.
4334 * tree-eh.c (lower_resx): Update.
4335 (unsplit_eh): Update.
4336 (cleanup_empty_eh_move_lp): Update.
4337 * tree-inline.c (copy_edges_for_bb): Update.
4338 (freqs_to_counts): Update.
4339 (copy_cfg_body): Update.
4340 * tree-ssa-dce.c (remove_dead_stmt): Update.
4341 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4342 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4343 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4344 (unloop_loops): Update.
4345 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4346 * tree-ssa-loop-split.c (connect_loops): Update.
4347 (split_loop): Update.
4348 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4349 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4350 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4351 * tree-ssa-reassoc.c (branch_fixup): Update.
4352 * tree-ssa-tail-merge.c (replace_block_by): Update.
4353 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4354 (compute_path_counts): Update.
4355 (update_profile): Update.
4356 (recompute_probabilities): Update.
4357 (update_joiner_offpath_counts): Update.
4358 (estimated_freqs_path): Update.
4359 (freqs_to_counts_path): Update.
4360 (clear_counts_path): Update.
4361 (ssa_fix_duplicate_block_edges): Update.
4362 (duplicate_thread_path): Update.
4363 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4364 (case_bit_test_cmp): Update.
4365 (collect_switch_conv_info): Update.
4366 (gen_inbound_check): Update.
4367 (do_jump_if_equal): Update.
4368 (emit_cmp_and_jump_insns): Update.
4369 * tree-tailcall.c (decrease_profile): Update.
4370 (eliminate_tail_call): Update.
4371 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4372 (vect_do_peeling): Update.
4373 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4374 * ubsan.c (ubsan_expand_null_ifn): Update.
4375 (ubsan_expand_ptr_ifn): Update.
4376 * value-prof.c (gimple_divmod_fixed_value): Update.
4377 (gimple_mod_pow2): Update.
4378 (gimple_mod_subtract): Update.
4379 (gimple_ic): Update.
4380 (gimple_stringop_fixed_value): Update.
4382 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4385 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
4387 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
4389 PR rtl-optimization/82395
4390 * ira-color.c (allocno_priority_compare_func): Fix comparison step
4391 based on non_spilled_static_chain_regno_p.
4393 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4395 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
4396 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
4397 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
4399 2017-10-19 Martin Sebor <msebor@redhat.com>
4401 PR tree-optimization/82596
4402 * tree.c (array_at_struct_end_p): Handle STRING_CST.
4404 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4406 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
4407 (get_mem_refs_of_builtin_call): Likewise.
4408 * builtins.c (expand_builtin_apply): Adjust call to
4409 allocate_dynamic_stack_space.
4410 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
4411 the third argument to allocate_dynamic_stack_space, otherwise -1.
4412 (expand_builtin): Deal with all alloca variants.
4413 (is_inexpensive_builtin): Likewise.
4414 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
4415 * calls.c (special_function_p): Deal with all alloca variants.
4416 (initialize_argument_information): Adjust call to
4417 allocate_dynamic_stack_space.
4418 (expand_call): Likewise.
4419 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
4420 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
4421 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
4422 use it for the stack usage computation.
4423 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
4424 * function.c (gimplify_parameters): Call build_alloca_call_expr.
4425 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
4426 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
4427 (in_loop_p): Remove first argument and useless check.
4428 (pass_walloca::execute): Remove useless test and adjust call to above.
4429 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
4430 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
4431 (gimplify_call_expr): Deal with all alloca variants.
4432 * hsa-gen.c (gen_hsa_alloca): Likewise.
4433 (gen_hsa_insns_for_call): Likewise.
4434 * ipa-pure-const.c (special_builtin_state): Likewise.
4435 * tree-chkp.c (chkp_build_returned_bound): Likewise.
4436 * tree-object-size.c (alloc_object_size): Likewise.
4437 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4438 (call_may_clobber_ref_p_1): Likewise.
4439 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
4440 (ccp_fold_stmt): Likewise.
4441 (optimize_stack_restore): Likewise.
4442 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
4443 (mark_all_reaching_defs_necessary_1): Likewise.
4444 (propagate_necessity): Likewise.
4445 (eliminate_unnecessary_stmts): Likewise.
4446 * tree.c (build_common_builtin_nodes): Build
4447 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
4448 (build_alloca_call_expr): New function.
4449 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
4450 (CASE_BUILT_IN_ALLOCA): Likewise.
4451 (build_alloca_call_expr): Declare.
4452 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
4454 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4457 * dwarf2out.c (new_die_raw): New static inline function.
4458 (new_die): Use it to create the DIE.
4459 (add_AT_external_die_ref): Likewise.
4460 (clone_die): Likewise.
4461 (clone_as_declaration): Likewise.
4462 (dwarf2out_vms_debug_main_pointer): Likewise.
4463 (base_type_die): Likewise. Remove early return for corner cases.
4464 Do not call add_pubtype on the DIE here.
4465 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
4466 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
4467 typedefs for base types with DW_AT_endianity. Make sure a DIE with
4468 native order exists for base types, attach the DIE manually and call
4469 add_pubtype on it. Do not equate a reverse order DIE to the type.
4471 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
4473 * config/arm/arm.c (align_ok_ldrd_strd): New function.
4474 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
4476 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
4478 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4480 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
4481 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
4482 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
4483 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
4484 * opts.c (sanitizer_opts): Add builtin.
4485 * ubsan.c (instrument_builtin): New function.
4486 (pass_ubsan::execute): Call it.
4487 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
4488 * doc/invoke.texi: Document -fsanitize=builtin.
4490 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
4491 builtins, store max (log2 (align), 0) into uchar field instead of
4492 align into uptr field.
4493 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
4494 store uchar 0 field instead of uptr 0 field.
4495 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
4496 instead of passing one address of struct with 2 locations pass
4497 two addresses of structs with 1 location each.
4498 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
4499 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
4500 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
4501 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
4502 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
4503 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
4504 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
4505 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
4507 2017-10-19 Martin Liska <mliska@suse.cz>
4510 * file-find.c (remove_prefix): Remove.
4511 * file-find.h (remove_prefix): Likewise.
4512 * gcc-ar.c: Remove smartness of lookup.
4514 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
4516 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
4517 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
4518 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
4520 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4523 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
4524 (setcc + and to xor + setcc): New peephole2.
4526 2017-10-19 Tom de Vries <tom@codesourcery.com>
4528 * doc/sourcebuild.texi (Test Directives, Variants of
4529 dg-require-support): Add dg-require-stack-size.
4531 2017-10-19 Martin Liska <mliska@suse.cz>
4534 * gimplify.c (gimplify_decl_expr): Do not instrument variables
4535 that have a large alignment.
4536 (gimplify_target_expr): Likewise.
4538 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
4540 PR rtl-optimization/82602
4541 * ira.c (rtx_moveable_p): Return false for volatile asm.
4543 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
4546 * config/i386/i386-modes.def (CCGZ): New CC mode.
4547 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
4548 * config/i386/predicates.md (ix86_comparison_operator):
4550 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4551 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
4552 with double-word subtraction.
4553 (put_condition_code): Handle CCGZmode.
4555 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
4557 * wide-int.cc (debug (const wide_int &)): New.
4558 (debug (const wide_int *)): New.
4559 (debug (const widest_int &)): New.
4560 (debug (const widest_int *)): New.
4562 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
4565 * lra-constraints.c (curr_insn_transform): Use non-input operand
4566 instead of output one for matched reload.
4568 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4570 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
4571 (tree-ssa-loop-ivopts.h): New header file.
4572 (struct builtin_info): New fields.
4573 (classify_builtin_1): Compute and record base and offset parts for
4574 memset builtin partition by calling strip_offset.
4575 (offset_cmp, fuse_memset_builtins): New functions.
4576 (finalize_partitions): Fuse adjacent memset partitions by calling
4578 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
4579 Expose the interface.
4580 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
4582 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4584 PR tree-optimization/82574
4585 * tree-loop-distribution.c (find_single_drs): New parameter. Check
4586 that data reference must be executed exactly once per iteration
4587 against the outermost loop in nest.
4588 (classify_partition): Update call to above function.
4590 2017-10-18 Richard Biener <rguenther@suse.de>
4592 PR tree-optimization/82591
4593 * graphite.c (graphite_transform_loops): Move code gen message
4595 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4596 Here. Handle scop_to_isl_ast failing.
4597 (scop_to_isl_ast): Limit the number of ISL operations.
4599 2017-10-18 Richard Biener <rguenther@suse.de>
4601 * graphite-isl-ast-to-gimple.c
4602 (translate_isl_ast_to_gimple::set_rename): Simplify.
4603 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
4604 (graphite_copy_stmts_from_block): ... here.
4605 (copy_bb_and_scalar_dependences): Simplify.
4606 (add_parameters_to_ivs_params): Canonicalize.
4607 (generate_entry_out_of_ssa_copies): Simplify.
4608 * graphite-sese-to-poly.c (extract_affine_name): Simplify
4609 by passing in ISL dimension.
4610 (parameter_index_in_region_1): Rename to ...
4611 (parameter_index_in_region): ... this.
4612 (extract_affine): Adjust assert, pass down parameter index.
4613 (add_param_constraints): Use range-info when available.
4614 (build_scop_context): Adjust.
4615 * sese.c (new_sese_info): Adjust.
4616 (free_sese_info): Likewise.
4617 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
4618 Remove unused typedefs.
4619 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
4621 2017-10-18 Martin Liska <mliska@suse.cz>
4623 * combine.c (simplify_compare_const): Add gcc_fallthrough.
4625 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4627 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
4628 (s390_sched_init): Do not reset s390_sched_state if we entered the
4629 current basic block via a fallthru edge and all others are unlikely.
4631 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4633 * config/s390/s390.c (NUM_SIDES): New variable.
4634 (LONGRUNNING_THRESHOLD): New variable.
4635 (LATENCY_FACTOR): New variable.
4636 (s390_sched_score): Decrease score for long-running instructions on
4638 (s390_sched_variable_issue): Perform bookkeeping for long-running
4641 2017-10-18 Richard Biener <rguenther@suse.de>
4643 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
4644 Simplify with removal of the parameter rename map.
4645 (set_rename): Likewise.
4646 (should_copy_to_new_region): Likewise.
4647 (graphite_copy_stmts_from_block): Likewise.
4648 (copy_bb_and_scalar_dependences): Remove initialization of
4649 unused copied_bb_map.
4651 (copy_internal_parameters): Likewise.
4652 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
4653 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
4654 Use INTEGRAL_TYPE_P.
4655 (parameter_index_in_region_1): Rename to ...
4656 (assign_parameter_index_in_region): ... this. Assert we have
4657 a parameter we handle.
4658 (scan_tree_for_params): Adjust.
4659 * sese.h (parameter_rename_map_t): Remove.
4660 (struct sese_info_t): Remove unused parameter_rename_map and
4661 copied_bb_map members.
4662 * sese.c (new_sese_info): Adjust.
4663 (free_sese_info): Likewise.
4665 2017-10-18 Martin Liska <mliska@suse.cz>
4668 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
4669 on an abnormal edge.
4671 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
4673 * doc/invoke.texi (ffunction-sections and fdata-sections):
4676 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4678 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
4679 the use statement can throw internally.
4681 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4683 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
4684 any RTX present on the RHS of a SET.
4685 * compare-elim.c (try_eliminate_compare): Restore comment.
4687 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4689 * langhooks.h (struct lang_hooks): Document that tree_size langhook
4690 may be also called on tcc_type nodes.
4691 * langhooks.c (lhd_tree_size): Likewise.
4693 2017-10-17 David Malcolm <dmalcolm@redhat.com>
4695 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
4696 format_warning_at_substring.
4697 (maybe_warn): Convert source_range * param to a location_t. Pass
4698 UNKNOWN_LOCATION rather than NULL to fmtwarn.
4699 (format_directive): Remove code to extract source_ranges and
4700 source_range * in favor of just a location_t.
4701 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
4703 * substring-locations.c (format_warning_va): Convert
4704 source_range * param to a location_t.
4705 (format_warning_at_substring): Likewise.
4706 * substring-locations.h (format_warning_va): Likewise.
4707 (format_warning_at_substring): Likewise.
4709 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
4711 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
4713 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
4714 and scatter/gather ops.
4716 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
4717 vec_gather_load and vec_scatter_store.
4718 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
4719 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
4720 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
4721 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
4722 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
4723 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
4725 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
4727 * reg-stack.c (compare_for_stack_reg): Add bool argument.
4728 Detect FTST instruction and handle its register pops. Only pop
4729 second operand if can_pop_second_op is true.
4730 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
4731 set can_pop_second_op to false in the compare_for_stack_reg call.
4733 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
4734 output_fp_compare for stack register operands.
4735 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
4736 instructions here. Do not emit stack register pops here. Assert
4737 that FCOMPP pops next to top stack register. Rewrite function.
4739 2017-10-17 Nathan Sidwell <nathan@acm.org>
4742 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
4743 use DECL_ASSEMBLER_NAME_RAW.
4746 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
4749 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
4750 Wilco Dijkstra <wilco.dijkstra@arm.com>
4752 * builtins.c (expand_builtin_update_setjmp_buf): Add a
4753 converstion to Pmode from the buf_addr.
4755 2017-10-17 Richard Biener <rguenther@suse.de>
4757 * graphite-dependences.c (scop_get_reads_and_writes): Change
4758 output parameters to references.
4760 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
4762 PR 71026/tree-optimization
4763 * fold-const.c (distribute_real_division): Removed.
4764 (fold_binary_loc): Remove calls to distribute_real_divison.
4766 2017-10-17 Richard Biener <rguenther@suse.de>
4768 * graphite-scop-detection.c
4769 (scop_detection::stmt_has_simple_data_refs_p): Always use
4770 the full nest as region.
4771 (try_generate_gimple_bb): Likewise.
4772 * sese.c (scalar_evolution_in_region): Simplify now that
4773 SCEV can handle instantiation in regions.
4774 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
4775 in the non-loop part of a function if requested.
4777 2017-10-17 Richard Biener <rguenther@suse.de>
4779 PR tree-optimization/82563
4780 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
4782 (graphite_regenerate_ast_isl): Call it.
4783 * graphite-scop-detection.c (build_scops): Remove entry edge split.
4785 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4787 PR tree-optimization/82549
4788 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
4789 Formatting fixes. Instead of calling make_bit_field_ref with negative
4792 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
4794 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
4795 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
4796 _mm_maskz_reduce_ss): New.
4797 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
4798 __builtin_ia32_reducess_mask): Ditto..
4799 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
4800 * config/i386/sse.md (reduces<mode>): Renamed to ...
4801 (reduces<mode><mask_scalar_name>): ... this.
4802 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
4804 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
4805 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
4807 2017-10-16 David Malcolm <dmalcolm@redhat.com>
4809 * Makefile.in (OBJS): Add unique-ptr-tests.o.
4810 * selftest-run-tests.c (selftest::run_tests): Call
4811 selftest::unique_ptr_tests_cc_tests.
4812 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
4813 * unique-ptr-tests.cc: New file.
4815 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
4818 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4820 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4821 (make_hard_regno_born, make_hard_regno_dead): Update
4822 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
4824 2017-10-16 Jeff Law <law@redhat.com>
4826 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4828 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
4830 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
4832 2017-10-16 Olivier Hainque <hainque@adacore.com>
4834 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
4835 with_cpu if we were configured for an e500v2 target cpu name.
4837 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
4839 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
4840 * doc/invoke.texi: Document +nodsp as a valid extension for
4843 2017-10-16 Martin Liska <mliska@suse.cz>
4845 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
4846 (test_set_range): Likewise.
4847 (test_range_functions): Rename to ...
4848 (test_bit_in_range): ... this.
4849 (sbitmap_c_tests): Add new test.
4851 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4853 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
4855 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
4856 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
4858 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4860 * config/aarch64/aarch64-builtins.c
4861 (aarch64_types_quadopu_lane_qualifiers): New.
4862 (TYPES_QUADOPU_LANE): New.
4863 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
4864 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
4865 (aarch64_<sur>dot_laneq<vsi2qi>): New.
4866 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
4867 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
4868 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
4869 (Vdottype, DOTPROD): New.
4870 (sur): Add SDOT and UDOT.
4872 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4874 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
4875 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
4876 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
4878 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
4879 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
4880 Enable TARGET_DOTPROD.
4881 (cortex-a75.cortex-a55): Likewise.
4882 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
4884 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4886 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
4887 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
4889 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
4891 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
4892 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
4893 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
4894 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
4895 * config/arm/types.md (neon_dot, neon_dot_q): New.
4896 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
4898 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4900 * config/arm/arm.h (TARGET_DOTPROD): New.
4901 * config/arm/arm.c (arm_arch_dotprod): New.
4902 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
4903 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
4904 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
4905 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
4906 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
4907 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
4908 * doc/invoke.texi (armv8.2-a): Document dotprod
4910 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4912 * i386.c (ix86_vec_cost): New function.
4913 (ix86_rtx_costs): Handle vector operations better.
4914 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
4915 * x86-tune-costs.h: Add new costs to all tables.
4917 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4919 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
4921 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
4922 divsd, sqrtss and sqrtsd
4923 * x86-tune-costs.h: Add new entries to all costs.
4924 (znver1_cost): Fix to match real instruction latencies.
4926 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4927 Michael Collison <michael.collison@arm.com>
4929 * compare-elim.c: Include emit-rtl.h.
4930 (can_merge_compare_into_arith): New function.
4931 (try_validate_parallel): Likewise.
4932 (try_merge_compare): Likewise.
4933 (try_eliminate_compare): Call the above when no previous clobber
4935 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
4938 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4942 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
4943 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
4947 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
4948 to be any operand_equal_p operands. For & (B - 1) require
4949 B to be power of 2. Recognize
4950 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
4952 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4955 * optabs.c (expand_memory_blockage): Fix call of
4956 targetm.have_memory_blockage.
4958 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4961 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
4962 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
4963 objects to extra_objs instead of overwriting it.
4965 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4967 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
4968 Use any_fp_register_operand as operand[3] predicate. Simplify
4969 equality test for operands[2] and operands[4] memory location.
4970 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
4971 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
4972 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
4973 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
4974 any_fp_register_operand as operand[1] predicate. Simplify
4975 equality test for operands[0] and operands[3] memory location.
4976 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
4977 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
4978 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
4980 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4982 * target-insns.def: Add memory_blockage.
4983 * optabs.c (expand_memory_blockage): New function.
4984 (expand_asm_memory_barrier): Rename ...
4985 (expand_asm_memory_blockage): ... to this.
4986 (expand_mem_thread_fence): Call expand_memory_blockage
4987 instead of expand_asm_memory_barrier.
4988 (expand_mem_singnal_fence): Ditto.
4989 (expand_atomic_load): Ditto.
4990 (expand_atomic_store): Ditto.
4991 * doc/md.texi (Standard Pattern Names For Generation):
4992 Document memory_blockage instruction pattern.
4994 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
4996 * config/rl78/rl78.c (rl78_emit_libcall): New function.
4997 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
4998 * config/rl78/rl78.md: New define_expand "adddi3".
5000 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5002 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5005 2017-10-13 Jeff Law <law@redhat.com>
5007 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5009 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5012 * internal-fn.c (expand_mul_overflow): If both operands have
5013 the same highpart of -1 or 0 and the topmost bit of lowpart
5014 is different, overflow is if res <= 0 rather than res < 0.
5016 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5018 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5019 TARGET_P9_VECTOR code for unaligned_load case.
5021 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5023 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5025 2017-10-13 Nathan Sidwell <nathan@acm.org>
5027 * tree-core.h (tree_contains_struct): Make bool.
5028 * tree.c (tree_contains_struct): Likewise.
5029 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5030 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5031 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5032 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5033 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5035 2017-10-13 Richard Biener <rguenther@suse.de>
5037 * graphite-isl-ast-to-gimple.c
5038 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5039 parameters and dominance check.
5040 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5041 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5042 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5043 Do not update SSA form here or do intermediate IL verification.
5044 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5045 (graphite_initialize): Remove check on the number of loops in
5046 the function and inline into graphite_transform_loops.
5047 (graphite_finalize): Inline into graphite_transform_loops.
5048 (graphite_transform_loops): Perform SSA update and IL verification
5050 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5052 2017-10-13 Richard Biener <rguenther@suse.de>
5054 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5055 graphite_expression_type_precision): Avoid global constructor
5057 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5058 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5059 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5060 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5061 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5062 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5064 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5067 * config/i386/i386.h (ix86_red_zone_size): New.
5068 * config/i386/i386.md (push peephole2s): Replace
5069 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5071 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5072 Alan Hayward <alan.hayward@arm.com>
5073 David Sherwood <david.sherwood@arm.com>
5075 * combine.c (can_change_dest_mode): Reject changes in
5076 REGMODE_NATURAL_SIZE.
5078 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5079 Alan Hayward <alan.hayward@arm.com>
5080 David Sherwood <david.sherwood@arm.com>
5082 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5083 (expand_debug_source_expr): Likewise.
5084 * combine.c (combine_simplify_rtx): Likewise.
5085 * cse.c (fold_rtx): Likewise.
5086 * fwprop.c (canonicalize_address): Likewise.
5087 * targhooks.c (default_shift_truncation_mask): Likewise.
5089 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5090 Alan Hayward <alan.hayward@arm.com>
5091 David Sherwood <david.sherwood@arm.com>
5093 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5094 (widened_mode): Likewise.
5095 (expand_unop): Likewise.
5096 * ree.c (transform_ifelse): Likewise.
5097 (merge_def_and_ext): Likewise.
5098 (combine_reaching_defs): Likewise.
5099 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5101 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5102 Alan Hayward <alan.hayward@arm.com>
5103 David Sherwood <david.sherwood@arm.com>
5105 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5106 * combine.c (gen_lowpart_for_combine): Likewise.
5107 * dwarf2out.c (rtl_for_decl_location): Likewise.
5108 * final.c (alter_subreg): Likewise.
5109 * rtlhooks.c (gen_lowpart_general): Likewise.
5110 (gen_lowpart_if_possible): Likewise.
5112 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5113 Alan Hayward <alan.hayward@arm.com>
5114 David Sherwood <david.sherwood@arm.com>
5116 * calls.c (expand_call): Use subreg_lowpart_offset.
5117 * cse.c (cse_insn): Likewise.
5118 * regcprop.c (copy_value): Likewise.
5119 (copyprop_hardreg_forward_1): Likewise.
5121 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5124 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5125 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5126 =Q constraints to +Q and into insn condition add check
5127 that operands[0] and operands[1] are equal.
5128 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5129 =Q constraints to +Q and into insn condition add check
5130 that operands[0] is equal to either operands[1] or operands[2].
5133 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5134 instead of handling MINUS_EXPR twice (once for each argument),
5135 canonicalize operand order and handle just once, use rtype where
5136 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5139 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5140 any values of __C while still being pattern recognizable as a simple
5143 2017-10-13 Richard Biener <rguenther@suse.de>
5145 PR tree-optimization/82451
5147 2017-10-02 Richard Biener <rguenther@suse.de>
5149 PR tree-optimization/82355
5150 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5151 a mapping for the enclosing loop but avoid generating one for
5153 (copy_bb_and_scalar_dependences): Remove premature codegen
5154 error on PHIs in blocks duplicated into multiple places.
5155 * graphite-scop-detection.c
5156 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5157 in the region use it as loop and nest to analyze the DR in.
5158 (try_generate_gimple_bb): Likewise.
5159 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5160 (add_loop_constraints): For blocks in a loop not in the region
5161 create a dimension with a single iteration.
5162 * sese.h (gbb_loop_at_index): Remove assert.
5164 * cfgloop.c (loop_preheader_edge): For the loop tree root
5165 return the single successor of the entry block.
5166 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5167 Reset the SCEV hashtable and niters.
5168 * graphite-scop-detection.c
5169 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5170 assert that we only have POLYNOMIAL_CHREC that vary in loops
5171 contained in the region.
5172 (scop_detection::graphite_can_represent_expr): Adjust.
5173 (scop_detection::stmt_has_simple_data_refs_p): For loops
5174 not in the region set loop to NULL. The nest is now the
5175 entry edge to the region.
5176 (try_generate_gimple_bb): Likewise.
5177 * sese.c (scalar_evolution_in_region): Adjust for
5178 instantiate_scev change.
5179 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5180 Make nest parameter the edge into the region.
5181 (create_data_ref): Likewise.
5182 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5183 entry edge into a region and adjust instantiate_scev calls.
5184 (create_data_ref): Likewise.
5185 (graphite_find_data_references_in_stmt): Likewise.
5186 (find_data_references_in_stmt): Pass the loop preheader edge
5187 from the nest argument.
5188 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5189 parameter the edge into the region.
5190 (instantiate_parameters): Use the loop preheader edge as entry.
5191 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5193 (get_instantiated_value_entry): Make instantiate_below parameter
5194 the edge into the region.
5195 (instantiate_scev_name): Likewise. Adjust dominance checks,
5196 when we cannot use loop-based instantiation instantiate by
5197 walking use-def chains.
5198 (instantiate_scev_poly): Adjust.
5199 (instantiate_scev_binary): Likewise.
5200 (instantiate_scev_convert): Likewise.
5201 (instantiate_scev_not): Likewise.
5202 (instantiate_array_ref): Remove.
5203 (instantiate_scev_3): Likewise.
5204 (instantiate_scev_2): Likewise.
5205 (instantiate_scev_1): Likewise.
5206 (instantiate_scev_r): Do not blindly handle N-operand trees.
5207 Do not instantiate array-refs. Handle all constants and invariants.
5208 (instantiate_scev): Make instantiate_below parameter
5209 the edge into the region.
5210 (resolve_mixers): Use the loop preheader edge for the region
5211 parameter to instantiate_scev_r.
5212 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5214 2017-10-13 Richard Biener <rguenther@suse.de>
5216 PR tree-optimization/82525
5217 * graphite-isl-ast-to-gimple.c
5218 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5220 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5221 Fail code generation when we cannot represent the isl integer.
5222 (binary_op_to_tree): Elide modulo operations that are no-ops
5223 in the type we code generate. Remove now superfluous code
5226 2017-10-13 Richard Biener <rguenther@suse.de>
5228 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5229 (scop_detection::harmful_loop_in_region): Remove premature
5230 IV type restriction.
5231 (scop_detection::graphite_can_represent_scev): We can handle
5232 pointer IVs just fine.
5234 2017-10-13 Alan Modra <amodra@gmail.com>
5236 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5237 "Clobbers and Scratch Registers". Add paragraph on
5238 alternative to clobbers for scratch registers and OpenBLAS
5241 2017-10-13 Alan Modra <amodra@gmail.com>
5243 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5244 example of a memory input for a string of known length. Move
5245 commentary out of table. Add a number of new examples
5246 covering array memory inputs.
5248 2017-10-12 Martin Liska <mliska@suse.cz>
5250 PR tree-optimization/82493
5251 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5252 (test_range_functions): New function.
5253 (sbitmap_c_tests): Likewise.
5254 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5255 * selftest.h (sbitmap_c_tests): New function.
5257 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5259 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5261 * config/rs6000/amo.h: Fix spacing issue.
5263 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5266 * config/i386/i386.md (*ashl<mode>3_mask_1,
5267 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5268 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5271 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5273 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5274 (profile_probability): Set max_probability
5275 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5277 * profile-count.c (profile_probability::differs_from_p): Do not
5278 rely on max_probaiblity == 10000
5280 2017-10-12 Jeff Law <law@redhat.com>
5282 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5285 2017-10-12 Martin Sebor <msebor@redhat.com>
5289 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5290 (handle_alias_pairs): Call it.
5291 * common.opt (-Wattribute-alias): New option.
5292 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5293 * doc/invoke.texi (-Wattribute-alias): Document.
5295 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5298 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5300 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5302 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5303 (make_hard_regno_born, make_hard_regno_dead): Update
5304 bb_killed_pseudos and bb_gen_pseudos.
5306 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5308 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5310 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5312 * config/alpha/alpha.c (alpha_split_conditional_move):
5313 Use std::swap instead of manually swapping.
5314 (alpha_stdarg_optimize_hook): Ditto.
5315 (alpha_canonicalize_comparison): Ditto.
5317 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5319 * tree-loop-distribution.c (struct builtin_info): New struct.
5320 (struct partition): Refactor fields into struct builtin_info.
5321 (partition_free): Free struct builtin_info.
5322 (build_size_arg_loc, build_addr_arg_loc): Delete.
5323 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5324 information from struct builtin_info.
5325 (find_single_drs): New function refactored from classify_partition.
5326 Also moved builtin validity checks to this function.
5327 (compute_access_range, alloc_builtin): New functions.
5328 (classify_builtin_st, classify_builtin_ldst): New functions.
5329 (classify_partition): Refactor code into functions find_single_drs,
5330 classify_builtin_st and classify_builtin_ldst.
5331 (distribute_loop): Don't do runtime alias check when distributing
5333 (find_seed_stmts_for_distribution): New function.
5334 (pass_loop_distribution::execute): Refactor code finding seed
5335 stmts into above function. Support distribution for the innermost
5336 two-level loop nest. Adjust dump information.
5338 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5340 * tree-loop-distribution.c: Adjust the general comment.
5341 (NUM_PARTITION_THRESHOLD): New macro.
5342 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
5343 (classify_partition): Skip builtin pattern of loop nest's inner loop.
5344 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
5345 in call to build_partition_graph.
5346 (finalize_partitions): New parameter. Make loop distribution more
5347 conservative by fusing more partitions.
5348 (distribute_loop): Don't do runtime alias check in case of loop nest
5350 (find_seed_stmts_for_distribution): New function.
5351 (prepare_perfect_loop_nest): New function.
5352 (pass_loop_distribution::execute): Refactor code finding seed stmts
5353 and loop nest into above functions. Support loop nest distribution.
5354 Adjust dump information accordingly.
5356 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5358 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
5359 and set PTYPE_SEQUENTIAL for merged partition.
5361 2017-10-12 Richard Biener <rguenther@suse.de>
5363 PR tree-optimization/69728
5365 2017-09-19 Richard Biener <rguenther@suse.de>
5367 PR tree-optimization/69728
5368 * graphite-sese-to-poly.c (schedule_error): New global.
5369 (add_loop_schedule): Handle empty domain by failing the
5371 (build_original_schedule): Handle schedule_error.
5373 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
5374 domain by returning an unchanged schedule.
5376 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5378 * genrecog.c (validate_pattern): For VEC_SELECT verify that
5379 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
5381 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
5383 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
5386 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5389 * expr.c (store_field): Don't optimize away bitsize == 0 store
5390 from CALL_EXPR with addressable return type.
5392 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5394 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
5395 * config/rs6000/rs6000.md (sel): Delete mode attribute.
5396 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
5397 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
5398 TARGET_ISEL instead of TARGET_ISEL<sel>.
5400 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
5402 * config/rs6000/rs6000.c
5403 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
5405 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5407 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
5409 (reg_or_cint_operand): Fix comment.
5410 (reg_or_zero_operand): New predicate.
5411 * config/rs6000/rs6000-protos.h (output_isel): Delete.
5412 * config/rs6000/rs6000.c (output_isel): Delete.
5413 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
5414 instead of reg_or_cint_operand. Output instruction directly (not via
5416 (isel_unsigned_<mode>): Ditto.
5417 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
5418 gpc_reg_operand. Add an instruction alternative for this. Output
5419 instruction directly.
5420 (*isel_reversed_unsigned_<mode>): Ditto.
5422 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5424 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
5425 (TARGET_CANONICALIZE_COMPARISON): Define.
5427 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
5430 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
5431 Check whether the dest is REG before adding REG_EQUIV note.
5433 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5436 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5438 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5439 (make_hard_regno_born, make_hard_regno_dead): Update
5440 bb_killed_pseudos and bb_gen_pseudos.
5442 2017-10-11 Nathan Sidwell <nathan@acm.org>
5444 * incpath.h (enum incpath_kind): Name enum, prefix values.
5445 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
5446 * incpath.c (heads, tails): Use INC_MAX.
5447 (add_env_var_paths, add_standard_paths): Use incpath_kind.
5448 (merge_include_chains, split_quote_chain,
5449 register_include_chains): Update incpath_kind names.
5450 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
5451 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
5453 (add_framework_path, darwin_register_objc_includes): Likewise.
5454 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
5456 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5458 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5459 Do not use float_operator operator predicate.
5460 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5461 * config/i386/predicates.md (float_operator): Remove predicate.
5463 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5465 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
5466 (*jccxf_i387): Ditto.
5467 (*jcc<mode>_i387): Ditto.
5468 (*jccu<mode>_i387): Ditto.
5469 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
5470 (*jcc_*_i387 splitters): Remove.
5471 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
5472 * config/i386/i386.c (ix86_split_fp_branch): Remove.
5473 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
5476 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5478 * profile-count.h (slow_safe_scale_64bit): New function.
5479 (safe_scale_64bit): New inline.
5480 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
5481 * profile-count.c: Include wide-int.h
5482 (slow_safe_scale_64bit): New.
5484 2017-10-11 Nathan Sidwell <nathan@acm.org>
5486 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
5487 HAS_DECL_ASSEMBLER_NAME_P.
5488 * gimple-expr.c (gimple_decl_printable_name: Check
5489 HAS_DECL_ASSEMBLER_NAME_P too.
5490 * ipa-utils.h (type_in_anonymous_namespace_p): Check
5491 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
5492 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
5493 * passes.c (rest_of_decl_compilation): Check
5494 HAS_DECL_ASSEMBLER_NAME_P too.
5495 * recog.c (verify_changes): Likewise.
5496 * tree-pretty-print.c (dump_decl_name): Likewise.
5497 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
5499 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
5500 (SET_DECL_ASSEMBLER_NAME): Use it.
5501 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
5502 (COPY_DECL_ASSEMBLER_NAME): Likewise.
5503 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
5505 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5507 * config.gcc (i386, x86_64): Add extra objects.
5508 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
5509 (ix86_min_insn_size): Declare.
5510 (ix86_issue_rate): Declare.
5511 (ix86_adjust_cost): Declare.
5512 (ia32_multipass_dfa_lookahead): Declare.
5513 (ix86_macro_fusion_p): Declare.
5514 (ix86_macro_fusion_pair_p): Declare.
5515 (ix86_bd_has_dispatch): Declare.
5516 (ix86_bd_do_dispatch): Declare.
5517 (ix86_core2i7_init_hooks): Declare.
5518 (ix86_atom_sched_reorder): Declare.
5519 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
5520 (COSTS_N_BYTES): Move to x86-tune-costs.h.
5521 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
5522 (rip_relative_addr_p): Rename to ...
5523 (ix86_rip_relative_addr_p): ... this one; export.
5524 (memory_address_length): Update.
5525 (ix86_issue_rate): Move to x86-tune-sched.c.
5526 (ix86_flags_dependent): Move to x86-tune-sched.c.
5527 (ix86_agi_dependent): Move to x86-tune-sched.c.
5528 (exact_dependency_1): Move to x86-tune-sched.c.
5529 (exact_store_load_dependency): Move to x86-tune-sched.c.
5530 (ix86_adjust_cost): Move to x86-tune-sched.c.
5531 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
5532 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
5533 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
5534 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
5535 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
5536 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
5537 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
5538 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
5539 (min_insn_size): Rename to ...
5540 (ix86_min_insn_size): ... this one; export.
5541 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
5542 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
5543 (core2i7_first_cycle_multipass_backtrack): Move to
5544 x86-tune-sched-core.c.
5545 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
5546 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
5547 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
5548 (ix86_avoid_jump_mispredicts): Update.
5549 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
5550 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
5551 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
5552 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
5553 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
5554 (MAX_INSN): Move to ix86-tune-sched-bd.c.
5555 (MAX_IMM): Move to ix86-tune-sched-bd.c.
5556 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
5557 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
5558 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
5559 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
5560 (MAX_STORE): Move to ix86-tune-sched-bd.c.
5561 (BIG): Move to ix86-tune-sched-bd.c.
5562 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
5563 (enum insn_path): Move to ix86-tune-sched-bd.c.
5564 (get_mem_group): Move to ix86-tune-sched-bd.c.
5565 (is_cmp): Move to ix86-tune-sched-bd.c.
5566 (dispatch_violation): Move to ix86-tune-sched-bd.c.
5567 (is_branch): Move to ix86-tune-sched-bd.c.
5568 (is_prefetch): Move to ix86-tune-sched-bd.c.
5569 (init_window): Move to ix86-tune-sched-bd.c.
5570 (allocate_window): Move to ix86-tune-sched-bd.c.
5571 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
5572 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
5573 (process_end_window): Move to ix86-tune-sched-bd.c.
5574 (allocate_next_window): Move to ix86-tune-sched-bd.c.
5575 (find_constant): Move to ix86-tune-sched-bd.c.
5576 (get_num_immediates): Move to ix86-tune-sched-bd.c.
5577 (has_immediate): Move to ix86-tune-sched-bd.c.
5578 (get_insn_path): Move to ix86-tune-sched-bd.c.
5579 (get_insn_group): Move to ix86-tune-sched-bd.c.
5580 (count_num_restricted): Move to ix86-tune-sched-bd.c.
5581 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
5582 (add_insn_window): Move to ix86-tune-sched-bd.c.
5583 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
5584 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
5585 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
5586 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
5587 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
5588 (do_dispatch): Move to ix86-tune-sched-bd.c.
5589 (has_dispatch): Move to ix86-tune-sched-bd.c.
5590 * config/i386/t-i386: Add new object files.
5591 * config/i386/x86-tune-costs.h: New file.
5592 * config/i386/x86-tune-sched-atom.c: New file.
5593 * config/i386/x86-tune-sched-bd.c: New file.
5594 * config/i386/x86-tune-sched-core.c: New file.
5595 * config/i386/x86-tune-sched.c: New file.
5597 2017-10-11 Liu Hao <lh_mouse@126.com>
5599 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
5600 the generic version below instead.
5601 (should_colorize): Recognize Windows consoles as terminals
5603 * pretty-print.c [__MINGW32__] (write_all): New function.
5604 [__MINGW32__] (find_esc_head): Likewise.
5605 [__MINGW32__] (find_esc_terminator): Likewise.
5606 [__MINGW32__] (eat_esc_sequence): Likewise.
5607 [__MINGW32__] (mingw_ansi_fputs): New function that handles
5609 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
5612 2017-10-11 Richard Biener <rguenther@suse.de>
5614 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
5615 Properly call analyze_scalar_evolution with the loop of the stmt.
5617 2017-10-11 Richard Biener <rguenther@suse.de>
5619 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
5620 * tree-core.h (tree_base): Add chrec_var union member.
5621 * tree.h (CHREC_VAR): Remove.
5622 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
5623 * tree-chrec.h (build_polynomial_chrec): Adjust.
5624 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
5625 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
5627 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
5629 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
5630 * match.pd: ... here.
5631 ((T) X == (T) Y): Relax condition.
5633 2017-10-11 Bin Cheng <bin.cheng@arm.com>
5635 PR tree-optimization/82472
5636 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
5638 (break_alias_scc_partitions): Update postorder number.
5640 2017-10-11 Martin Liska <mliska@suse.cz>
5643 * opts.c (parse_no_sanitize_attribute): Do not use error_value
5645 * opts.h (parse_no_sanitize_attribute): Remove last argument.
5647 2017-10-11 Martin Liska <mliska@suse.cz>
5649 * print-rtl.c (print_insn): Move declaration of idbuf
5650 to same scope as name.
5652 2017-10-11 Martin Liska <mliska@suse.cz>
5657 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5660 2017-10-11 Martin Liska <mliska@suse.cz>
5663 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5666 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
5668 * config.gcc (default_gnu_indirect_function): Default to yes for
5669 arm*-*-linux* with glibc.
5671 2017-10-11 Richard Biener <rguenther@suse.de>
5673 * tree-scalar-evolution.c (get_scalar_evolution): Handle
5674 default-defs and types we do not want to analyze.
5675 (interpret_loop_phi): Replace unreachable code with an assert.
5676 (compute_scalar_evolution_in_loop): Remove and inline ...
5677 (analyze_scalar_evolution_1): ... here, replacing condition with
5678 what makes the intent clearer. Remove handling of cases
5679 get_scalar_evolution now handles.
5681 2017-10-10 Jim Wilson <wilson@tuliptree.org>
5683 PR rtl-optimization/81434
5684 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
5685 comment for main loop. In sched_group_found if, also add checks for
5686 pass and min_cost_group.
5688 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
5690 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
5691 (rs6000_insn_cost): New function.
5692 * config/rs6000/rs6000.md (cost): New attribute.
5694 2017-10-10 Jakub Jelinek <jakub@redhat.com>
5695 H.J. Lu <hongjiu.lu@intel.com>
5699 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
5700 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
5701 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
5702 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
5703 to OPTION_MASK_ISA_AVX512VL - builtins that have both
5704 OPTION_MASK_ISA_MMX and some other bit set require both
5705 mmx and the ISAs without the mmx bit.
5706 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
5707 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
5708 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
5709 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
5710 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
5711 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
5712 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
5713 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
5714 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
5715 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
5716 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
5717 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
5718 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
5719 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
5720 Add OPTION_MASK_ISA_MMX.
5722 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
5724 * config.gcc (armv7*-*-freebsd*): New target.
5725 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
5727 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
5729 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
5730 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
5733 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
5735 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
5736 a template parameter.
5737 (WIDE_INT_REF_FOR): Update accordingly.
5738 * tree.h (wi::int_traits <const_tree>): Delete.
5739 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
5740 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
5741 (wi::tree_to_wide_ref): New typedef.
5742 (wi::to_wide): New function.
5743 * calls.c (get_size_range): Use wi::to_wide when operating on
5745 * cgraph.c (cgraph_node::create_thunk): Likewise.
5746 * config/i386/i386.c (ix86_data_alignment): Likewise.
5747 (ix86_local_alignment): Likewise.
5748 * dbxout.c (stabstr_O): Likewise.
5749 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
5750 * expr.c (const_vector_from_tree): Likewise.
5751 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
5752 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
5753 (fold_negate_expr_1, int_const_binop_1, const_binop)
5754 (fold_convert_const_int_from_real, optimize_bit_field_compare)
5755 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
5756 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
5757 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
5758 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
5759 (fold_not_const, round_up_loc): Likewise.
5760 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
5761 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
5762 (alloca_call_type): Likewise.
5763 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
5764 * godump.c (go_output_typedef): Likewise.
5765 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
5766 * internal-fn.c (get_min_precision): Likewise.
5767 * ipa-cp.c (ipcp_store_vr_results): Likewise.
5768 * ipa-polymorphic-call.c
5769 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
5770 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
5771 (ipa_modify_call_arguments): Likewise.
5772 * match.pd: Likewise.
5773 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
5774 * print-tree.c (print_node_brief, print_node): Likewise.
5775 * stmt.c (expand_case): Likewise.
5776 * stor-layout.c (layout_type): Likewise.
5777 * tree-affine.c (tree_to_aff_combination): Likewise.
5778 * tree-cfg.c (group_case_labels_stmt): Likewise.
5779 * tree-data-ref.c (dr_analyze_indices): Likewise.
5780 (prune_runtime_alias_test_list): Likewise.
5781 * tree-dump.c (dequeue_and_dump): Likewise.
5782 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
5783 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
5784 * tree-pretty-print.c (dump_generic_node): Likewise.
5785 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
5786 (simple_iv_with_niters): Likewise.
5787 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
5788 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
5789 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
5790 * tree-ssa-loop-niter.c (split_to_var_and_offset)
5791 (refine_value_range_using_guard, number_of_iterations_ne_max)
5792 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
5793 (get_cst_init_from_scev, record_nonwrapping_iv)
5794 (scev_var_range_cant_overflow): Likewise.
5795 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
5796 * tree-ssa-pre.c (compute_avail): Likewise.
5797 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
5798 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
5799 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
5800 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
5801 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
5802 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
5803 (dump_case_nodes, try_switch_expansion): Likewise.
5804 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
5805 (vect_do_peeling): Likewise.
5806 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
5807 * tree-vect-stmts.c (vectorizable_load): Likewise.
5808 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
5809 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
5810 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
5811 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
5812 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
5813 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
5814 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
5815 (evrp_dom_walker::before_dom_children): Likewise.
5816 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
5817 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
5818 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
5819 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
5820 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
5821 (get_range_pos_neg): Likewise.
5822 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
5823 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
5824 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
5825 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
5826 * config/avr/avr.c (avr_fold_builtin): Likewise.
5827 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
5828 * config/msp430/msp430.c (msp430_attr): Likewise.
5829 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
5830 * config/powerpcspe/powerpcspe-c.c
5831 (altivec_resolve_overloaded_builtin): Likewise.
5832 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
5833 (rs6000_expand_ternop_builtin): Likewise.
5834 * config/rs6000/rs6000-c.c
5835 (altivec_resolve_overloaded_builtin): Likewise.
5836 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
5837 (rs6000_expand_ternop_builtin): Likewise.
5838 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
5840 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5842 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
5843 when copying loop nest with only one inner loop.
5845 2017-10-10 Richard Biener <rguenther@suse.de>
5847 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
5848 blocks if SCEV is active.
5849 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
5851 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
5852 (scev_initialize): Assert we are not yet initialized.
5854 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5856 * tree-loop-distribution.c (generate_loops_for_partition): Remove
5857 inner loop's exit stmt by making it always exit the loop, otherwise
5858 we would generate an infinite empty loop.
5860 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5862 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
5863 renaming variables in new preheader if it's deleted.
5865 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5867 * tree-loop-distribution.c (struct partition): Remove unused field
5868 loops of the structure.
5869 (partition_alloc, partition_free): Ditto.
5870 (build_rdg_partition_for_vertex): Ditto.
5872 2017-10-09 Jeff Law <law@redhat.com>
5874 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
5875 return type to match prototype and documentation.
5877 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5879 * config/rs6000/rs6000.c (processor_costs): Move to ...
5880 * config/rs6000/rs6000.h: ... here.
5881 (rs6000_cost): Declare.
5883 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
5885 * except.c (setjmp_fn): New global variable.
5886 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
5887 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
5888 if DONT_USE_BUILTIN_SETJMP is defined.
5890 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5892 * target.def (insn_cost): New hook.
5893 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
5894 * doc/tm.texi: Regenerate.
5895 * rtlanal.c (insn_cost): Use the new hook.
5897 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5899 * combine.c (combine_validate_cost): Compute the new insn_cost,
5900 not just pattern_cost.
5901 (try_combine): Adjust comment.
5903 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5905 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
5907 * combine.c (uid_insn_cost): Adjust comment.
5908 (combine_validate_cost): Adjust comment. Use pattern_cost instead
5910 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
5911 * dse.c (find_shift_sequence): Ditto.
5912 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
5913 (bb_valid_for_noce_process_p): Use pattern_cost.
5914 * rtl.h (insn_rtx_cost): Delete.
5915 (pattern_cost): New prototype.
5916 (insn_cost): New prototype.
5917 * rtlanal.c (insn_rtx_cost): Rename to...
5918 (pattern_cost): ... this.
5921 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
5923 * config/i386/i386.md (*jcc_2): Remove insn pattern.
5924 (*jcc<mode>_0_r_i387): Ditto.
5925 (*jccxf_r_i387): Ditto.
5926 (*jcc<mode>_r_i387): Ditto.
5927 (*jccu<mode>_r_i387): Ditto.
5928 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
5929 (*jcc): Rename from *jcc_1.
5931 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5933 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
5934 deferred rescans after the lvx/stvx recombination pre-pass.
5936 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
5938 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
5939 memory operation instruction support.
5940 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
5941 (rs6000-ibm-aix[789]*): Likewise.
5942 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
5943 Document new functions.
5945 2017-10-09 Richard Biener <rguenther@suse.de>
5947 PR tree-optimization/82397
5948 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
5949 equality only for semantically equal trees.
5951 2017-10-09 Richard Biener <rguenther@suse.de>
5953 PR tree-optimization/82449
5954 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
5955 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
5956 allow constant addresses.
5957 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
5960 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5962 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
5965 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5968 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
5971 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5974 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
5976 2017-10-09 Jakub Jelinek <jakub@redhat.com>
5979 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
5980 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
5982 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
5984 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
5985 (WI_BINARY_PREDICATE_RESULT): Likewise.
5986 (wi::binary_traits::operator_result): New type.
5987 (wi::binary_traits::predicate_result): Likewise.
5988 (generic_wide_int::operator~, unary generic_wide_int::operator-)
5989 (generic_wide_int::operator==, generic_wide_int::operator!=)
5990 (generic_wide_int::operator&, generic_wide_int::and_not)
5991 (generic_wide_int::operator|, generic_wide_int::or_not)
5992 (generic_wide_int::operator^, generic_wide_int::operator+
5993 (binary generic_wide_int::operator-, generic_wide_int::operator*):
5995 (operator~, unary operator-, operator==, operator!=, operator&)
5996 (operator|, operator^, operator+, binary operator-, operator*): New
5998 * expr.c (get_inner_reference): Use wi::bit_and_not.
5999 * fold-const.c (fold_binary_loc): Likewise.
6000 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6001 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6002 (bit_value_binop): Likewise.
6003 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6004 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6005 (extract_range_from_binary_expr_1): Likewise.
6006 (masked_increment): Likewise.
6007 (simplify_bit_ops_using_ranges): Likewise.
6009 2017-10-09 Martin Jambor <mjambor@suse.cz>
6012 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6013 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6014 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6016 (hsa_fixup_mov_insn_type): New function.
6017 (hsa_op_with_type::get_in_type): Use it.
6018 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6019 immediates in an assert.
6020 (hsa_op_with_type::extend_int_to_32bit): New method.
6021 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6022 types. Convert to dest type if necessary.
6023 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6024 (reg_for_gimple_ssa): Pass false as min32int to
6025 hsa_type_for_scalar_tree_type.
6026 (gen_hsa_addr): Fixup type when creating addresable temporary.
6027 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6028 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6029 necessary. Call hsa_fixup_mov_insn_type.
6030 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6031 extend operands and convert to dest type if necessary.
6032 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6033 to dest type if necessary.
6034 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6035 if conversion nt necessary and size matches.
6036 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6037 to dest type if necessary.
6038 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6039 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6041 (gen_hsa_clrsb): Likewise.
6042 (gen_hsa_ffs): Likewise.
6043 (gen_hsa_divmod): Extend operands and convert to dest type if
6045 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6047 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6049 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6050 Remove empty default arguments. Use a brace block as output
6052 (conditional return): Ditto.
6054 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6055 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6057 (group_ending_nop): Ditto.
6058 (doloop_end): Ditto.
6059 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6060 (splitters for those): Ditto.
6062 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6064 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6065 a conditional jump (and the compare for it) so that pc_rtx is the
6067 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6068 for the deleted and renamed ctr<mode>_internal[234] patterns.
6069 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6070 Delete second conditional return pattern.
6071 (ctr<mode>_internal2): Delete this second bdnz pattern.
6072 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6073 (ctr<mode>_internal4): Delete this second bdz pattern.
6075 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6077 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6078 (always_initialized_rtx_for_ssa_name_p): New predicate.
6079 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6080 (finish_out_of_ssa): Free new field of SA.
6081 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6082 * tree-ssa-coalesce.c: Include tree-ssa.h.
6083 (get_parm_default_def_partitions): Remove extern keyword.
6084 (get_undefined_value_partitions): New function.
6085 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6086 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6088 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6090 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6092 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6094 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6096 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6097 for targets that preffer 128bit.
6099 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6101 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6103 2017-10-08 Olivier Hainque <hainque@adacore.com>
6105 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6106 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6108 (thumb_set_return_address): Likewise.
6110 2017-10-08 Olivier Hainque <hainque@adacore.com>
6112 * common/config/arm/arm-common.c (arm_except_unwind_info):
6113 Handle DWARF2_UNWIND_INFO.
6115 2017-10-07 Michael Collison <michael.collison@arm.com>
6117 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6120 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6122 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6123 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6124 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6125 defined, force the creation of a new block for a dispatch label.
6127 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6129 * invoke.texi (Wsuggest-attribute=cold): Document.
6130 * common.opt (Wsuggest-attribute=cold): New
6131 * ipa-pure-const.c (warn_function_cold): New function.
6132 * predict.c (compute_function_frequency): Use it.
6133 * predict.h (warn_function_cold): Declare.
6135 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6137 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6140 2017-10-06 Martin Liska <mliska@suse.cz>
6142 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6143 keyword for member functions.
6144 (struct sanopt_tree_couple): New struct.
6145 (struct sanopt_tree_couple_hash): New function.
6146 (struct sanopt_ctx): Add new hash_map.
6147 (has_dominating_ubsan_ptr_check): New function.
6148 (record_ubsan_ptr_check_stmt): Likewise.
6149 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6150 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6151 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6153 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6156 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6157 aarch64_simd_valid_immediate on CONST_VECTORs.
6158 (aarch64_reg_or_bic_imm): Likewise.
6160 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6162 PR rtl-optimization/82396
6163 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6165 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6167 * graphite-dependences.c (scop_get_reads): Move code to...
6168 (scop_get_must_writes): Move code to...
6169 (scop_get_may_writes): Move code to...
6170 (scop_get_reads_and_writes): ... here.
6171 (scop_get_dependences): Call scop_get_reads_and_writes.
6173 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6175 PR tree-optimization/82434
6176 * fold-const.h (can_native_encode_type_p,
6177 can_native_encode_string_p): Remove.
6178 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6179 don't encode anything, just return what would be otherwise returned.
6180 (native_encode_fixed, native_encode_complex, native_encode_vector):
6182 (native_encode_string): Likewise. Inline by hand
6183 can_native_encode_string_p.
6184 (can_native_encode_type_p): Remove.
6185 (can_native_encode_string_p): Remove.
6186 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6187 STRING_CSTs using can_native_encode_string_p, test all
6188 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6189 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6190 argument from native_encode_expr.
6191 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6192 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6195 2017-10-06 Richard Biener <rguenther@suse.de>
6197 PR tree-optimization/82397
6198 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6199 operand_equal_p but rely on data_ref_compare_tree for detecting
6201 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6202 to match up with dr_group_sort_cmp.
6204 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6207 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6209 * config/s390/s390-builtin-types.def: Regenerate.
6211 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6214 * config/s390/s390-builtin-types.def: Regenerate.
6215 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6216 Change flag from B_VXE to B_VX.
6217 (s390_vec_min_dbl): Remove B_VXE flag.
6219 2017-10-06 Richard Biener <rguenther@suse.de>
6221 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6222 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6223 translate_isl_ast_to_gimple::is_valid_rename,
6224 translate_isl_ast_to_gimple::get_rename,
6225 translate_isl_ast_to_gimple::get_def_bb_for_const,
6226 translate_isl_ast_to_gimple::get_new_name,
6227 translate_isl_ast_to_gimple::collect_all_ssa_names,
6228 translate_isl_ast_to_gimple::copy_loop_phi_args,
6229 translate_isl_ast_to_gimple::collect_all_ssa_names,
6230 translate_isl_ast_to_gimple::copy_loop_phi_args,
6231 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6232 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6233 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6234 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6235 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6236 translate_isl_ast_to_gimple::copy_cond_phi_args,
6237 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6238 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6239 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6240 translate_isl_ast_to_gimple::rename_uses,
6241 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6242 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6243 (set_rename_for_each_def): Likewise.
6244 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6245 here. Handle rewriting SCEV analyzable uses here.
6246 (copy_bb_and_scalar_dependences): Generate code for PHI
6248 (graphite_regenerate_ast_isl): Adjust.
6249 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6250 (add_write, add_read): New functions.
6251 (build_cross_bb_scalars_def): Use it and simplify.
6252 (build_cross_bb_scalars_use): Likewise.
6253 (graphite_find_cross_bb_scalar_vars): Inline into...
6254 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6255 simulating out-of-SSA. Compute liveout and add dependencies.
6256 (build_scops): Force an empty entry block.
6257 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6259 (sese_build_liveouts): Declare.
6260 (sese_trivially_empty_bb_p): Likewise.
6261 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6262 compute liveout and debug_liveout.
6263 (sese_bad_liveouts_use): Remove.
6264 (sese_reset_debug_liveouts_bb): Likewise.
6265 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6266 (sese_build_liveouts): Build liveout and debug_liveout and store
6268 (new_sese_info): Adjust.
6269 (free_sese_info): Likewise.
6270 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6271 do not build liveout here.
6272 (move_sese_in_condition): Adjust region entry.
6273 (scev_analyzable_p): Match up with chrec_apply requirements.
6274 (sese_trivially_empty_bb_p): New.
6275 * tree-into-ssa.c (get_reaching_def): Properly support generating
6276 default-defs for incremental rewrite of anonymous names.
6278 2017-10-06 Richard Biener <rguenther@suse.de>
6280 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6281 precision do not perform modulo reduction.
6283 2017-10-06 Richard Biener <rguenther@suse.de>
6285 PR tree-optimization/82436
6286 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6287 conservatively choose the vectorization factor when checking
6288 whether we can perform the required load permutation.
6289 (vect_transform_slp_perm_load): Assert when we may not fail.
6291 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6293 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6294 message for incompatible -msdata=* and -mcall-* options.
6296 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6298 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6299 rate for post-reload scheduling.
6301 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6303 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6305 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6307 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6308 to improve monte carlo in scimark.
6310 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6312 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6313 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6314 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6315 pentium4_cost, nocona_cost): Set reassociation width to 1.
6316 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6317 width to 2 for fp operations and 1 otherwise.
6318 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6320 (atom_cost): Set reassociation width to 2.
6321 (slm_cost, generic_cost): Set fp reassociation width
6322 to 2 and 1 otherwise.
6323 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6324 (core_cost): Set fp reassociation width to 4 and vector to 2.
6325 (ix86_reassociation_width): Rewrite using cost table; special case
6326 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6327 and TARGET_AVX128_OPTIMAL.
6328 * config/i386/i386.h (processor_costs): Add
6329 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6330 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6331 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6332 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6333 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6334 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6336 2017-10-05 Nathan Sidwell <nathan@acm.org>
6338 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6340 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6342 * config/arm/arm.c (arm_test_fpu_data): New.
6343 (arm_run_selftests): Call arm_test_fpu_data.
6345 2017-10-04 Nathan Sidwell <nathan@acm.org>
6347 * toplev.c (toplev::main): Remove excess parens on pretty_printer
6349 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
6351 2017-10-04 Sudakshina Das <sudi.das@arm.com>
6353 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
6354 check type for aarch64_simd_valid_immediate.
6355 (aarch64_output_simd_mov_immediate): Update prototype.
6356 (aarch64_simd_valid_immediate): Update prototype.
6357 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
6358 support for ORR-immediate.
6359 (and<mode>3): modified pattern to add support for BIC-immediate.
6360 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
6361 now checks for valid immediate for BIC and ORR based on new enum
6363 (aarch64_output_simd_mov_immediate): Function now used to output
6364 BIC/ORR imm as well based on new enum argument.
6365 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6367 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6368 (aarch64_reg_or_bic_imm): Likewise.
6370 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6372 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
6373 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
6375 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6379 PR rtl-optimization/82396
6380 * haifa-sched.c (autopref_multipass_init): Simplify
6382 (autopref_rank_data): Simplify sort order.
6383 * sched-int.h (autopref_multipass_data_): Remove
6384 multi_mem_insn_p, min_offset and max_offset.
6386 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6388 * doc/sourcebuild.texi: Document vect_peeling_profitable.
6390 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6392 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
6395 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6397 * doc/sourcebuild.texi: Document vect_long_mult.
6399 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
6401 PR tree-optimization/82413
6402 * fold-const.c (build_range_check): Use widest_int when comparing
6403 the maximum ETYPE value with HIGH.
6405 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6407 PR rtl-optimization/82396
6408 * haifa-sched.c (autopref_multipass_init): Simplify
6410 (autopref_rank_data): Simplify sort order.
6411 * sched-int.h (autopref_multipass_data_): Remove
6412 multi_mem_insn_p, min_offset and max_offset.
6414 2017-10-04 Jakub Jelinek <jakub@redhat.com>
6416 PR tree-optimization/82381
6417 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
6418 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
6421 PR tree-optimization/82374
6422 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
6423 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
6424 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
6425 current_function_decl to the new decl.
6427 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6429 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
6430 helper macro for IEEE float128 hardware built-in functions.
6431 (SQRTF128_ODD): Add built-in functions with the round-to-odd
6433 (TRUNCF128_ODD): Likewise.
6434 (ADDF128_ODD): Likewise.
6435 (SUBF128_ODD): Likewise.
6436 (MULF128_ODD): Likewise.
6437 (DIVF128_ODD): Likewise.
6438 (FMAF128_ODD): Likewise.
6439 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
6440 UNSPEC_TRUNC_ROUND_TO_ODD.
6441 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
6442 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
6443 floating point round to odd instructions.
6444 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
6445 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
6446 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
6447 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
6448 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
6449 (trunc<mode>sf2_hw): Change the truncate with round to odd
6450 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
6451 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
6452 to odd hardware instructions.
6453 (sub<mode>3_odd): Likewise.
6454 (mul<mode>3_odd): Likewise.
6455 (div<mode>3_odd): Likewise.
6456 (sqrt<mode>2_odd): Likewise.
6457 (fma<mode>4_odd): Likewise.
6458 (fms<mode>4_odd): Likewise.
6459 (nfma<mode>4_odd): Likewise.
6460 (nfms<mode>4_odd): Likewise.
6461 (trunc<mode>df2_odd): Change the truncate with round to odd
6462 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
6464 * doc/extend.texi (PowerPC built-in functions): Update documentation
6465 for existing IEEE float128-bit built-in functions. Add built-in
6466 functions that generate the IEEE 128-bit floating point round to
6469 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
6471 PR rtl-optimization/77729
6472 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
6473 to (X&(C1&~C2))|C2 transformations.
6475 2017-10-03 Martin Jambor <mjambor@suse.cz>
6477 PR tree-optimization/82363
6478 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
6479 mismatch, mark lacc written regardless of racc.
6481 2017-10-03 Jakub Jelinek <jakub@redhat.com>
6483 PR tree-optimization/82381
6484 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
6485 stmt_to_insert nor wheather SSA_NAMEs are default defs.
6486 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
6487 fallthrough into reassoc_stmt_dominates_stmt_p.
6490 * combine.c (combine_instructions): Don't combine in unreachable
6493 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6496 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
6497 function to not use the have_cpu variable. Do not set cpu_index,
6498 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
6500 (rs6000_valid_attribute_p): Remove duplicate initializations of
6501 old_optimize and func_optimize.
6502 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
6503 (rs6000_activate_target_options): Make global.
6504 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
6507 2017-10-02 Jakub Jelinek <jakub@redhat.com>
6509 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
6510 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
6511 Set *poffset to 0 and *psize and *pmax_size to -1 if
6512 *poffset + *psize overflows in HOST_WIDE_INT.
6514 PR tree-optimization/82387
6515 PR tree-optimization/82388
6516 PR tree-optimization/82389
6517 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
6518 instead of live_bytes non-NULL.
6520 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
6523 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
6526 2017-10-02 Richard Biener <rguenther@suse.de>
6528 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
6529 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
6530 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
6532 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6534 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
6535 requested precision matches the type's.
6536 * calls.c (alloc_max_size): Calculate the new candidate size as
6537 a widest_int and use wi::to_widest when comparing it with the
6538 current candidate size.
6539 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
6540 zero rather than integer_zero_node.
6541 * match.pd: Check for a no-op conversion before using wi::add
6542 rather than after. Use tree_to_uhwi when summing small shift
6543 counts into an unsigned int.
6545 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6546 Alan Hayward <alan.hayward@arm.com>
6547 David Sherwood <david.sherwood@arm.com>
6550 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
6551 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
6552 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
6553 POINTER_AND_FP_REGS.
6555 2017-10-02 Richard Biener <rguenther@suse.de>
6557 PR tree-optimization/82355
6558 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6559 a mapping for the enclosing loop but avoid generating one for
6561 (copy_bb_and_scalar_dependences): Remove premature codegen
6562 error on PHIs in blocks duplicated into multiple places.
6563 * graphite-scop-detection.c
6564 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6565 in the region use it as loop and nest to analyze the DR in.
6566 (try_generate_gimple_bb): Likewise.
6567 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6568 (add_loop_constraints): For blocks in a loop not in the region
6569 create a dimension with a single iteration.
6570 * sese.h (gbb_loop_at_index): Remove assert.
6572 2017-10-01 Kevin Buettner <kevinb@redhat.com>
6574 * omp-expand.c (adjust_context_scope): New function.
6575 (expand_parallel_call): Call adjust_context_scope.
6577 2017-10-01 Jeff Law <law@redhat.com>
6579 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
6580 dom_opt_dom_walker class with direct access to private members.
6581 Add comments. Call test_for_singularity.
6582 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
6583 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
6584 m_dummy_cond anymore.
6585 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
6587 (pass_dominator:execute): Build the dummy_cond here and pass it
6588 to the dom_opt_dom_walker ctor.
6589 (test_for_singularity): New function.
6591 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
6592 Maya Rashish <coypu@sdf.org>
6594 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
6595 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
6596 (alpha*-*-netbsd*) Use nbsd_tm_file.
6597 (arm*-*-netbsdelf*) Likewise.
6598 (i[34567]86-*-netbsdelf*) Likewise.
6599 (x86_64-*-netbsd*) Likewise.
6600 (mips*-*-netbsd*) Likewise.
6601 (powerpc-*-netbsd*) Likewise.
6602 (sh*-*-netbsd*) Likewise.
6603 (sparc-*-netbsdelf*) Likewise.
6604 (sparc64-*-netbsd*) Likewise.
6605 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
6607 (vax-*-netbsdelf*) Likewise.
6608 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
6609 (UINT_FAST8_TYPE) Likewise.
6610 (INT_FAST16_TYPE) Check CHAR_FAST16.
6611 (UINT_FAST16_TYPE) Likewise.
6613 2017-09-30 Jakub Jelinek <jakub@redhat.com>
6616 * config/i386/i386.md
6617 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
6618 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
6619 *divmodsi4_zext_2): New define_insn_and_split.
6620 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
6621 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
6622 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
6623 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
6624 New define_insn_and_split.
6625 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
6626 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
6627 operands[1] having DImode when mode is SImode.
6629 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
6630 always SImode for DIV and MOD in REG_EQUAL notes.
6632 2017-09-29 Yury Gribov <tetra2005@gmail.com>
6635 * match.pd: Fix handling of NaNs in pattern.
6637 2017-09-29 Jeff Law <law@redhat.com>
6639 * sbitmap.c (bitmap_bit_in_range_p): New function.
6640 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
6641 * tree-ssa-dse.c (live_bytes_read): New function.
6642 (dse_classify_store): Ignore reads of dead bytes.
6644 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
6645 typos and whitespace errors.
6646 * config/i386/predicates.md (address_no_seg_operand): Likewise.
6647 * config/s390/s390.c (s390_emit_prologue): Likewise.
6649 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6652 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
6653 with a symbol for LRA.
6655 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6657 PR rtl-optimization/82338
6658 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
6660 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
6662 * genmodes.c (calc_wider_mode): Suppress qsort macro.
6663 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
6664 (qsort_chk): Declare.
6665 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
6666 (qsort_chk): New function.
6668 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6670 PR tree-optimization/82337
6671 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
6672 phi definition if the PHI result appears in an abnormal PHI.
6673 (find_basis_for_base_expr): Don't record a basis if the LHS of the
6674 basis appears in an abnormal PHI.
6676 2017-09-29 Richard Biener <rguenther@suse.de>
6678 * graphite-isl-ast-to-gimple.c
6679 (translate_isl_ast_to_gimple::set_codegen_error): New function.
6680 (binary_op_to_tree): Use it.
6681 (get_rename_from_scev): Likewise.
6682 (copy_loop_phi_nodes): Likewise.
6683 (copy_bb_and_scalar_dependences): Likewise.
6684 (translate_pending_phi_nodes): Likewise.
6686 2017-09-29 Jakub Jelinek <jakub@redhat.com>
6689 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
6690 for movabsq $(i32 << shift), r64.
6692 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
6694 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
6695 index when encoding %esp as %rsp to avoid 0x67 prefix.
6697 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
6699 * config/i386/i386.md (*movsf_internal, *movdf_internal):
6700 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
6702 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
6704 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
6705 Extensions with more than 16 double VFP registers.
6706 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
6707 to_clear_mask and all code related to it. Replace the remaining
6708 entry by a sbitmap and adapt code accordingly.
6710 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
6712 * brig-builtins.def: Change pure attributes to const.
6714 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6716 * config.gcc (default_gnu_indirect_function): Default to yes for
6717 sparc*-*-linux* with glibc.
6719 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6721 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
6722 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
6723 when creating .init_array and .fini_array sections with priority
6726 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
6729 * config/aarch64/aarch64.c
6730 (aarch64_builtin_support_vector_misalignment): Always return false
6731 when misalignment is unknown.
6733 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
6735 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
6736 this function to return false if the definition used by the swap
6737 instruction is artificial, or if the memory address from which the
6738 constant value is loaded is not represented by a base address held
6739 in a register or if the base address register is a frame or stack
6740 pointer. Additionally, return false if the base address of the
6741 loaded constant is a SYMBOL_REF but is not considered to be a
6743 (replace_swapped_load_constant): New function.
6744 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
6745 loaded constant vector with a load of a swapped constant vector.
6747 2017-09-27 Carl Love <cel@us.ibm.com>
6749 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
6750 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
6751 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
6754 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
6756 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
6757 first, always call autopref_rank_data otherwise.
6759 2017-09-27 Richard Biener <rguenther@suse.de>
6761 * graphite-scop-detection.c (find_scop_parameters): Move
6762 loop bound handling ...
6763 (gather_bbs::before_dom_children): ... here, avoiding the need
6764 to build scop_info->loop_nest.
6765 (record_loop_in_sese): Remove.
6766 * sese.h (sese_info_t::loop_nest): Remove.
6767 * sese.c (new_sese_info): Do not allocate loop_nest.
6768 (free_sese_info): Do not free loop_nest.
6770 2017-09-27 Jakub Jelinek <jakub@redhat.com>
6773 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
6774 lhs from calls if the lhs has addressable type.
6776 2017-09-27 Richard Biener <rguenther@suse.de>
6778 * graphite.h (scop::max_alias_set): New member.
6779 * graphite-scop-detection.c: Remove references to non-existing
6780 --param in comments.
6781 (build_alias_sets): Record the maximum alias set used for drs.
6782 (build_scops): Support zero as unlimited for
6783 --param graphite-max-arrays-per-scop.
6784 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
6786 (build_poly_sr_1): ... here. Compute alias set based on the
6787 maximum alias set used for drs rather than
6788 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
6790 2017-09-27 Richard Biener <rguenther@suse.de>
6792 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
6793 --param loop-block-tile-size=0 to disable tiling.
6795 2017-09-27 Richard Biener <rguenther@suse.de>
6797 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
6798 (graphite-max-nb-scop-params): Document special value zero.
6799 * domwalk.h (dom_walker::STOP): New symbolical constant.
6800 (dom_walker::dom_walker): Add optional parameter for bb to
6802 (dom_walker::~dom_walker): Declare.
6803 (dom_walker::before_dom_children): Document STOP return value.
6804 (dom_walker::m_user_bb_to_rpo): New member.
6805 (dom_walker::m_bb_to_rpo): Likewise.
6806 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
6807 mapping here if not provided by the user.
6808 (dom_walker::~dom_walker): Free bb to RPO mapping if not
6809 provided by the user.
6810 (dom_walker::STOP): Define.
6811 (dom_walker::walk): Do not compute bb to RPO mapping here.
6812 Support STOP return value from before_dom_children to stop
6814 * graphite-optimize-isl.c (optimize_isl): If the schedule
6815 is the same still generate code if -fgraphite-identity
6816 or -floop-parallelize-all are given.
6817 * graphite-scop-detection.c: Include cfganal.h.
6818 (gather_bbs::gather_bbs): Get and pass through bb to RPO
6820 (gather_bbs::before_dom_children): Return STOP for BBs
6822 (build_scops): Compute bb to RPO mapping and pass it to
6823 the domwalk. Treat --param graphite-max-nb-scop-params=0
6824 as not limiting the number of params.
6825 * graphite.c (graphite_initialize): Remove limit on the
6826 number of basic-blocks in a function.
6827 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
6828 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
6829 default value of 10.
6831 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6833 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
6834 Adjust code to eliminate needing to do the shift right 32-bits
6835 operation after XSCVDPSPN.
6837 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6839 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
6840 ((X / Y) != 0 -> X >= Y): Likewise.
6842 2017-09-26 Carl Love <cel@us.ibm.com>
6844 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
6845 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
6846 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
6847 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
6848 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
6849 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
6850 definitions and overloading.
6851 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
6852 statement for P9V_BUILTIN_XST_LEN_R.
6853 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
6854 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
6855 define_expand and define_insn for the instructions and builtins.
6856 * doc/extend.texi: Update the built-in documentation file for the new
6858 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
6859 define_insn for the instructions
6861 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
6864 * gcc/config/netbsd-protos.h: New file.
6865 * gcc/config/netbsd.c: New file.
6866 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
6867 * gcc/config/t-netbsd: New file.
6868 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
6869 (tmake_file) Add t-netbsd.
6870 (extra_objs) Add netbsd.o.
6872 2017-09-26 Janus Weil <janus@gcc.gnu.org>
6876 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
6878 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6880 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
6881 sign extension from a vector register to a GPR by doing a 32-bit
6882 direct move and then an EXTSW.
6883 (extendsi<mode>2 splitter): Likewise.
6884 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
6885 right or vector extract after doing XSCVDPSPN. Use
6886 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
6888 (movdi_from_sf_zero_ext): Likewise.
6889 (reload_gpr_from_vsxsf): Likewise.
6890 (p8_mfvsrd_4_disf): Delete, no longer used.
6891 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
6892 then needing to move the SFmode to a GPR to use the XSCVDPSP
6893 instruction instead of FRSP and XSCVDPSPN.
6894 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
6895 it is adjacent to the other XSCVSPDP insns.
6896 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
6897 SFmode to be in traditional Altivec registers.
6898 (vsx_xscvdpspn): Eliminate useless alternative constraint.
6899 (vsx_xscvspdpn): Likewise.
6900 (vsx_xscvspdpn_scalar): Likewise.
6902 2017-09-26 Martin Jambor <mjambor@suse.cz>
6904 * tree-sra.c (compare_access_positions): Put integral types first,
6905 stabilize sorting of integral types, remove conditions putting
6906 non-full-precision integers last.
6907 (sort_and_splice_var_accesses): Disable scalarization if a
6908 non-integert would be represented by a non-full-precision integer.
6910 2017-09-26 Joseph Myers <joseph@codesourcery.com>
6912 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
6913 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
6914 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
6915 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
6916 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
6917 conditionals inside the function instead of around it. Call
6918 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
6919 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
6921 2017-09-26 Richard Biener <rguenther@suse.de>
6923 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
6925 (scop_detection::build_scop_breadth): ... this. Removed.
6926 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
6927 (scop_detection::harmful_stmt_in_bb): Likewise.
6928 (scop_detection::graphite_can_represent_stmt): Likewise.
6929 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
6930 (scop_detection::can_represent_loop): Remove recursion, fold in ...
6931 (scop_detection::can_represent_loop_1): ... this. Removed.
6932 (scop_detection::harmful_loop_in_region): Simplify after inlining
6933 the above and remove more quadraticness.
6934 (build_scops): Adjust.
6935 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
6938 2017-09-26 Jakub Jelinek <jakub@redhat.com>
6941 * config/i386/i386.c (ix86_print_operand_address_as): Only test
6942 REGNO (base) == SP_REG if base is a REG.
6945 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
6946 if it is different SSA_NAME.
6947 (optimize_range_tests_cmp_bitwise): New function.
6948 (optimize_range_tests): Call it.
6950 2017-09-26 Richard Biener <rguenther@suse.de>
6952 PR tree-optimization/82321
6953 * graphite.c (canonicalize_loop_closed_ssa): Properly check
6954 for the def being inside the loop.
6956 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6958 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
6960 * config/s390/s390-builtins.def: Fix constraint on op4.
6962 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6964 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
6965 independent expanders.
6966 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
6967 ("vec_ordered", "vec_unordered"): New expanders.
6969 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6971 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
6974 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6976 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
6977 vec_unpacks_lo_v16qi.
6978 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
6980 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6982 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
6983 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
6984 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
6986 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6988 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
6990 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
6992 ("*vec_slb<mode>"): New insn pattern.
6993 ("vec_shr_<mode>"): New expander.
6994 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
6995 and force the shift count operand to V16QImode.
6996 ("vec_srb<mode>"): Set shift count mode to V16QI.
6998 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7000 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7001 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7002 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7004 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7007 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7009 2017-09-26 Richard Biener <rguenther@suse.de>
7011 PR tree-optimization/82320
7012 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7015 2017-09-25 Jeff Law <law@redhat.com>
7017 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7018 prototype for new argument.
7019 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7020 mostly extracted from rs6000_emit_allocate_stack.
7021 (rs6000_emit_probe_stack_range_stack_clash): New function.
7022 (rs6000_emit_allocate_stack): Call
7023 rs6000_emit_probe_stack_range_stack_clash as needed.
7024 (rs6000_emit_probe_stack_range): Add additional argument
7025 to call to gen_probe_stack_range{si,di}.
7026 (output_probe_stack_range): New.
7027 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7028 (output_probe_stack_range_stack_clash): New.
7029 (rs6000_emit_prologue): Emit notes into dump file as requested.
7030 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7031 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7032 Add additional operand and pass it to output_probe_stack_range.
7034 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7036 PR tree-optimization/82163
7037 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7038 (checking_verify_loop_closed_ssa): New parameter.
7039 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7040 (check_loop_closed_ssa_stmt): Delete.
7041 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7042 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7043 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7046 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7048 * brig-builtins.def: Treat HSAIL barrier builtins as
7049 setjmp/longjump style functions.
7051 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7053 * target.def (constant_alignment): New hook.
7054 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7055 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7056 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7057 * doc/tm.texi: Regenerate.
7058 * targhooks.h (default_constant_alignment): Declare.
7059 (constant_alignment_word_strings): Likewise.
7060 * targhooks.c (default_constant_alignment): New function.
7061 (constant_alignment_word_strings): Likewise.
7062 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7063 instead of CONSTANT_ALIGNMENT.
7064 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7065 (force_const_mem): Likewise.
7066 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7067 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7068 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7069 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7070 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7072 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7073 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7074 constant_alignment_word_strings.
7075 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7076 (CONSTANT_ALIGNMENT): Likewise.
7077 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7078 (arm_constant_alignment): New function.
7079 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7080 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7081 constant_alignment_word_strings.
7082 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7083 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7084 constant_alignment_word_strings.
7085 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7086 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7087 (cris_constant_alignment): New function.
7088 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7089 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7090 (epiphany_constant_alignment): New function.
7091 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7092 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7093 constant_alignment_word_strings.
7094 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7095 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7096 constant_alignment_word_strings.
7097 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7098 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7099 constant_alignment_word_strings.
7100 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7101 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7102 * config/i386/i386.c (ix86_constant_alignment): Make static.
7103 Use the same interface as the target hook.
7104 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7105 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7106 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7107 constant_alignment_word_strings.
7108 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7109 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7110 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7111 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7112 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7113 constant_alignment_word_strings.
7114 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7115 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7116 constant_alignment_word_strings.
7117 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7118 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7119 constant_alignment_word_strings.
7120 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7121 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7123 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7124 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7125 * config/mips/mips.c (mips_constant_alignment): New function.
7126 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7127 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7128 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7129 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7130 (mmix_constant_alignment): Make static. Use the same interface
7132 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7133 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7134 constant_alignment_word_strings.
7135 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7136 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7137 constant_alignment_word_strings.
7138 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7139 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7140 constant_alignment_word_strings.
7141 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7142 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7143 (rs6000_constant_alignment): New function.
7144 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7145 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7146 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7147 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7148 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7149 (rs6000_constant_alignment): New function.
7150 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7151 * config/s390/s390.c (s390_constant_alignment): New function.
7152 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7153 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7154 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7155 constant_alignment_word_strings.
7156 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7157 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7158 (sparc_constant_alignment): New function.
7159 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7160 * config/spu/spu.c (spu_constant_alignment): New function.
7161 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7162 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7163 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7164 constant_alignment_word_strings.
7165 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7166 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7167 constant_alignment_word_strings.
7168 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7169 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7170 constant_alignment_word_strings.
7171 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7172 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7173 (visium_constant_alignment): New function.
7174 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7175 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7176 (xtensa_constant_alignment): New function.
7177 * system.h (CONSTANT_ALIGNMENT): Poison.
7179 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7181 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7182 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7183 (rs6000_builtin_valid_without_lhs): New helper function.
7184 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7185 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7187 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7189 * target.h (vec_perm_indices): Use unsigned short rather than
7191 (auto_vec_perm_indices): Likewise.
7192 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7193 Use unsigned int rather than unsigned char.
7194 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7196 2017-09-25 Richard Biener <rguenther@suse.de>
7198 * cfgloop.h (sort_sibling_loops): Declare.
7199 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7200 (sort_sibling_loops): New function sorting the sibling loop list
7202 * graphite.c (graphite_transform_loops): Sort sibling loops.
7204 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7206 * target.def (vec_perm_const_ok): Change sel parameter to
7208 * optabs-query.c (can_vec_perm_p): Update accordingly.
7209 * doc/tm.texi: Regenerate.
7210 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7211 auto_vec_perm_indices and remove separate nelt field.
7212 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7213 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7214 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7215 (aarch64_expand_vec_perm_const): Update accordingly.
7216 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7217 to vec_perm_indices.
7218 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7219 auto_vec_perm_indices and remove separate nelt field.
7220 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7221 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7222 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7224 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7225 to vec_perm_indices.
7226 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7227 sel to vec_perm_indices.
7228 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7229 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7230 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7232 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7235 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7238 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7239 on the FUNCTION_DECL function context if it has a DIE that is a
7242 2017-09-25 Richard Biener <rguenther@suse.de>
7244 PR tree-optimization/82285
7245 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7248 2017-09-25 Tom de Vries <tom@codesourcery.com>
7252 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7255 2017-09-25 Richard Biener <rguenther@suse.de>
7257 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7258 ISL errors other than isl_error_quota happen. Dump if the
7259 schedule is the same.
7260 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7261 errors instead of aborting inside ISL.
7263 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7266 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7267 of libgcc_eh for m64.
7268 * config/i386/darwin64.h: Likewise.
7270 2017-09-25 Richard Biener <rguenther@suse.de>
7273 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7274 attribute for incomplete types nor twice for complete ones.
7276 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7279 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7280 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7283 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7286 * config/i386/i386.opt (mprefer-avx256): Use
7287 ix86_target_flags variable.
7288 * config/i386/i386.c (ix86_target_string): Move
7289 -mprefer-avx256 to flag2_opts.
7291 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7294 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7295 and x != -1 | y != -1 into (x & y) != -1.
7297 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7299 * config.gcc: Add new case statement to set
7300 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7301 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7302 s390x-*-linux* case statements. Added aarch64 to the list of
7303 supported architectures.
7305 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7307 PR tree-optimization/82289
7308 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7309 STMT_VINFO_RELEVANT_P.
7311 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7312 Alan Hayward <alan.hayward@arm.com>
7313 David Sherwood <david.sherwood@arm.com>
7315 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7316 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7317 (extract_range_from_binary_expr_1): Don't call
7318 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7320 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7321 Alan Hayward <alan.hayward@arm.com>
7322 David Sherwood <david.sherwood@arm.com>
7324 * target.def (preferred_vector_alignment): New hook.
7325 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7327 * doc/tm.texi: Regenerate.
7328 * targhooks.h (default_preferred_vector_alignment): Declare.
7329 * targhooks.c (default_preferred_vector_alignment): New function.
7330 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7332 (DR_TARGET_ALIGNMENT): New macro.
7333 (aligned_access_p): Update commentary.
7334 (vect_known_alignment_in_bytes): New function.
7335 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7337 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7338 Calculate the misalignment based on the target alignment rather than
7340 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7341 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7342 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
7343 the target alignment, rather than masking the element misalignment
7344 with the number of elements in a vector. Also use the target
7345 alignment when calculating the maximum number of peels.
7346 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
7347 instead of TYPE_ALIGN_UNIT.
7348 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
7349 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
7350 (vect_create_addr_base_for_vector_ref): Update call accordingly.
7351 (vect_create_data_ref_ptr): Likewise.
7352 (vect_setup_realignment): Realign by ANDing with
7353 -DR_TARGET_MISALIGNMENT.
7354 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
7355 the number of peels based on DR_TARGET_ALIGNMENT.
7356 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
7357 with the guaranteed alignment boundary when deciding whether
7359 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
7360 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
7361 (ensure_base_align): Remove stmt_info parameter. Get the
7362 target base alignment from DR_TARGET_ALIGNMENT.
7363 (vectorizable_store): Update call accordingly. Interpret
7364 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7366 (vectorizable_load): Likewise.
7368 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7369 Alan Hayward <alan.hayward@arm.com>
7370 David Sherwood <david.sherwood@arm.com>
7372 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
7373 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
7374 (vect_enhance_data_refs_alignment): Likewise.
7376 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7378 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7379 error. Only quit immediately if parsing is complete.
7380 (BEGIN): Initialize fatal_err and parse_done.
7381 (begin fpu, end fpu): Check number of arguments.
7382 (begin arch, end arch): Likewise.
7383 (begin cpu, end cpu): Likewise.
7384 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7385 (optalias): Likewise.
7387 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7389 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7390 * config/arm/arm-isa.h: Delete. Move definitions to ...
7391 * arm-cpus.in: ... here. Use new feature and fgroup values.
7392 * config/arm/arm.c (arm_option_override): Use lower case for feature
7394 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7395 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7396 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7398 (print_isa_bits_for): New function.
7399 (gen_isa): New function.
7400 (gen_comm_data): Use print_isa_bits_for.
7401 (define feature): New keyword.
7402 (define fgroup): New keyword.
7403 * config/arm/t-arm (TM_H): Remove.
7404 (GTM_H): Add arm-isa.h.
7405 (arm-isa.h): Add rule to generate file.
7406 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7407 case for feature bit names.
7409 2017-09-22 Richard Biener <rguenther@suse.de>
7411 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
7413 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
7414 print of no dependency loops ...
7415 * graphite.c (graphite_transform_loops): ... here.
7416 (canonicalize_loop_closed_ssa_form): Work from inner to outer
7418 (same_close_phi_node, remove_duplicate_close_phi,
7419 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
7420 (canonicalize_loop_closed_ssa): ... here and simplify.
7421 * graphite-optimize-isl.c: Include tree-vectorizer.h.
7422 (optimize_isl): Use dump_printf_loc to tell when we stopped
7423 optimizing because of an ISL timeout.
7425 2017-09-22 Richard Biener <rguenther@suse.de>
7427 PR tree-optimization/82291
7428 * tree-if-conv.c (predicate_mem_writes): Make sure to
7429 remove writes in blocks predicated with false.
7431 2017-09-22 Richard Biener <rguenther@suse.de>
7433 * sese.c: Include cfganal.h.
7434 (if_region_set_false_region): Remove.
7435 (create_if_region_on_edge): Likewise.
7436 (move_sese_in_condition): Re-implement without destroying
7439 2017-09-22 Richard Biener <rguenther@suse.de>
7441 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
7442 Verify both BBs contain loop PHI nodes before dispatching to
7444 (graphite_regenerate_ast_isl): Do not recompute dominators,
7445 do not verify three times. Restructure for clarity.
7446 * graphite-scop-detection.c (same_close_phi_node,
7447 remove_duplicate_close_phi, make_close_phi_nodes_unique,
7448 defined_in_loop_p, canonicalize_loop_closed_ssa,
7449 canonicalize_loop_closed_ssa_form): Simplify, remove excess
7450 checking and SSA rewrite, move to ...
7451 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
7452 (graphite_initialize): Do not pass in ctx, do not reset the
7453 SCEV cache, compute only dominators.
7454 (graphite_transform_loops): Allocate ISL ctx after
7455 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
7456 Maintain post-dominators only around build_scops.
7457 * sese.c (if_region_set_false_region): Make static. Free
7458 and recompute dominators.
7459 (move_sese_in_condition): Assert we don't get called with
7460 post-dominators computed.
7461 * sese.h (if_region_set_false_region): Remove.
7463 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
7465 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
7466 mode attribute for TARGET_AVX512VL.
7468 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
7470 * config/i386/i386.opt (mprefer-avx256): New option.
7471 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
7473 (ix86_preferred_simd_mode): Return 256-bit AVX modes
7474 for TARGET_PREFER_AVX256.
7475 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
7477 2017-09-21 Jeff Law <law@redhat.com>
7479 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
7480 Fix dump output if the only stack space is for pushed registers.
7482 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7484 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
7487 2017-09-21 Martin Sebor <msebor@redhat.com>
7490 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
7491 code (in C++) or code that triggers warnings.
7493 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
7495 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
7497 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
7499 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
7500 * sched-rgn.c: Ditto.
7501 * sel-sched-ir.c: Ditto.
7503 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
7505 * toplev.h (set_random_seed): Adjust return type.
7506 * toplev.c (init_local_tick): Move eager initialization of random_seed
7507 to get_random_seed. Adjust comment.
7508 (init_random_seed): Inline to get_random_seed, delete.
7509 (get_random_seed): Initialize random_seed lazily.
7510 (set_random_seed): Do not return previous value.
7511 (print_switch_value): Do not call get_random_seed.
7513 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
7515 * cgraph.c (delete_function_version): New, broken out from...
7516 (cgraph_node::delete_function_version): ...here. Rename to
7517 cgraph_node::delete_function_version_by_decl. Update all uses.
7518 (cgraph_node::remove): Call delete_function_version.
7520 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7523 * tree-inline.c (expand_call_inline): Emit clobber stmts for
7524 VAR_DECLs to which addressable non-volatile parameters are mapped
7525 and for id->retvar after the return value assignment. Clear
7526 id->retval and id->retbnd after inlining.
7528 2017-09-21 Richard Biener <rguenther@suse.de>
7530 PR tree-optimization/82276
7531 PR tree-optimization/82244
7532 * tree-vrp.c (build_assert_expr_for): Set
7533 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
7535 (remove_range_assertions): Revert earlier change.
7537 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
7540 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
7542 2017-09-21 Richard Biener <rguenther@suse.de>
7544 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7545 Restore valid IL after code generation errors.
7546 * graphite.c (graphite_transform_loops): Diagnose code
7547 generation issues as MSG_MISSED_OPTIMIZATION and continue
7548 with processing SCOPs.
7550 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7551 Alan Hayward <alan.hayward@arm.com>
7552 David Sherwood <david.sherwood@arm.com>
7554 * calls.c (compute_argument_addresses): Use simplify_gen_binary
7555 rather than choosing between plus_constant and gen_rtx_<CODE>.
7556 * expr.c (emit_push_insn): Likewise.
7557 (expand_expr_real_2): Likewise.
7559 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7560 Alan Hayward <alan.hayward@arm.com>
7561 David Sherwood <david.sherwood@arm.com>
7563 * loop-unroll.c (split_iv): Call copy_rtx on the step.
7565 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7566 Alan Hayward <alan.hayward@arm.com>
7567 David Sherwood <david.sherwood@arm.com>
7569 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
7570 calling tree_to_uhwi.
7572 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7573 Alan Hayward <alan.hayward@arm.com>
7574 David Sherwood <david.sherwood@arm.com>
7576 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
7577 INTEGER_CST rather than a negative test for ADDR_EXPR.
7579 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7580 Alan Hayward <alan.hayward@arm.com>
7581 David Sherwood <david.sherwood@arm.com>
7583 * tree-vrp.c (extract_range_from_binary_expr_1): Check
7584 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
7586 2017-09-21 Richard Biener <rguenther@suse.de>
7588 PR tree-optimization/71351
7589 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
7590 graphite_create_new_loop_guard): Remove, fold remaining parts
7592 (translate_isl_ast_node_for): ... here and simplify.
7594 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7597 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
7598 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
7599 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
7600 alternative always use QI mode, for -Os imov (=R,R) alternative
7601 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
7604 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7605 Jeff Law <law@redhat.com>
7607 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
7608 (allocate_stack_space): New function, partially extracted from
7610 (s390_emit_prologue): Track offset to most recent stack probe.
7611 Code to allocate space moved into allocate_stack_space.
7612 Dump actions when no stack is allocated.
7613 (s390_prologue_plus_offset): New function.
7614 (s390_emit_stack_probe): Likewise.
7616 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
7618 * common.opt (Wa, Wl, Wp, g, gz=): Add
7620 (gno-column-info): Remove.
7621 (gcolumn-info): Drop RejectNegative.
7623 (gno-record-gcc-switches): Remove.
7624 (grecord-gcc-switches): Drop RejectNegative.
7625 (gno-split-dwarf): Remove.
7626 (gsplit-dwarf): Drop RejectNegative.
7627 (gno-strict-dwarf): Remove.
7628 (gstrict-dwarf): Drop RejectNegative.
7629 * config/darwin.opt (gfull, gused): Add RejectNegative.
7630 * dwarf2out.c (gen_producer_string): Drop
7631 gno-record-gcc-switches handler.
7632 * optc-gen.awk: Add g to prefixes with negative forms.
7633 * opts-common.c (remapping_prefix_p): New.
7634 (find_opt): Check it.
7635 (generate_canonical_option): Test g prefix.
7636 (option_map): Add -gno- mapping.
7637 (add_misspelling_candidates): Check remapping_prefix_p.
7639 2017-09-20 Jeff Law <law@redhat.com>
7641 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
7642 thinko in stack clash protection support.
7644 * explow.c (compute_stack_clash_protection_loop_data): Use
7645 CONST_INT_P instead of explicit test. Verify object is a
7646 CONST_INT_P before looking at INTVAL.
7647 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
7648 instead of explicit test.
7650 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7653 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
7654 address instead of to r1 and r11.
7656 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
7658 * config.gcc: Support "knm".
7659 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
7660 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7662 * config/i386/i386.c (m_KNM): Define.
7663 (processor_target_table): Add "knm".
7665 (ix86_option_override_internal): Add "knm".
7666 (ix86_issue_rate): Add PROCESSOR_KNM.
7667 (ix86_adjust_cost): Ditto.
7668 (ia32_multipass_dfa_lookahead): Ditto.
7669 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
7670 (fold_builtin_cpu): Add M_INTEL_KNM.
7671 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
7672 (processor_type): Add PROCESSOR_KNM.
7673 * config/i386/x86-tune.def: Add m_KNM.
7674 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
7676 2017-09-20 Richard Biener <rguenther@suse.de>
7678 PR tree-optimization/80213
7679 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
7680 are allowed in empty BBs as well.
7681 (canonicalize_loop_closed_ssa): Also look for other complex
7683 (scop_detection::get_sese): Include the loop-closed PHI block
7685 (scop_detection::merge_sese): Remove code adding extra blocks.
7686 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
7687 (build_scops): Assert the final returned scop is invalid.
7689 2017-09-20 Richard Biener <rguenther@suse.de>
7691 PR tree-optimization/82264
7692 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
7694 (vn_phi_lookup): Likewise.
7695 (vn_phi_insert): Likewise.
7697 2017-09-20 Jakub Jelinek <jakub@redhat.com>
7699 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
7700 that fits into uhwi or shwi, add DW_AT_const_value regardless
7701 of early_dwarf without going through RTL, using add_AT_unsigned
7704 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
7705 (DEBUG_LTO_ABBREV_SECTION): Likewise.
7706 (DEBUG_LTO_MACINFO_SECTION): Likewise.
7707 (DEBUG_MACRO_SECTION): Likewise.
7708 (DEBUG_LTO_MACRO_SECTION): Likewise.
7709 (DEBUG_STR_DWO_SECTION): Likewise.
7710 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
7711 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
7712 (DEBUG_LTO_DWO_LINE_SECTION): Define.
7713 (DEBUG_LTO_LINE_STR_SECTION): Define.
7714 (init_sections_and_labels): Initialize debug_line_str_section
7715 variable. Initialize debug_loc_section for -gdwarf-5 to
7716 DEBUG_LOCLISTS_SECTION. Formatting fixes.
7718 2017-09-20 Richard Biener <rguenther@suse.de>
7720 * graphite-sese-to-poly.c (extract_affine): Properly handle
7721 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
7723 2017-09-20 Richard Biener <rguenther@suse.de>
7725 PR tree-optimization/81373
7726 * graphite-scop-detection.c (build_cross_bb_scalars_def):
7727 Force SESE live-out defs to be handled even if they are
7730 2017-09-19 Jeff Law <law@redhat.com>
7732 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
7733 nothing for stack adjustments with REG_STACK_CHECK.
7734 * sched-deps.c (parse_add_or_inc): Reject insns with
7735 REG_STACK_CHECK from dependency breaking.
7736 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
7737 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
7738 * reg-notes.def (STACK_CHECK): New note.
7740 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
7741 (ix86_expand_prologue): Dump stack clash info as needed.
7742 Call ix86_adjust_stack_and_probe_stack_clash as needed.
7744 * function.c (dump_stack_clash_frame_info): New function.
7745 * function.h (dump_stack_clash_frame_info): Prototype.
7746 (enum stack_clash_probes): New enum.
7748 * config/alpha/alpha.c (alpha_expand_prologue): Also check
7749 flag_stack_clash_protection.
7750 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
7751 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
7752 (arm_frame_pointer_required): Likewise.
7753 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
7754 (ia64_expand_prologue): Likewise.
7755 * config/mips/mips.c (mips_expand_prologue): Likewise.
7756 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
7757 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7758 (sparc_flat_expand_prologue): Likewise.
7759 * config/spu/spu.c (spu_expand_prologue): Likewise.
7761 * explow.c: Include "params.h".
7762 (anti_adjust_stack_and_probe_stack_clash): New function.
7763 (get_stack_check_protect): Likewise.
7764 (compute_stack_clash_protection_loop_data): Likewise.
7765 (emit_stack_clash_protection_loop_start): Likewise.
7766 (emit_stack_clash_protection_loop_end): Likewise.
7767 (allocate_dynamic_stack_space): Use get_stack_check_protect.
7768 Use anti_adjust_stack_and_probe_stack_clash.
7769 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
7770 (emit_stack_clash_protection_loop_start): Likewise.
7771 (emit_stack_clash_protection_loop_end): Likewise.
7772 * rtl.h (get_stack_check_protect): Prototype.
7773 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
7774 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
7775 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
7777 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
7779 * doc/tm.texi: Rebuilt.
7780 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
7781 get_stack_check_protect.
7782 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
7783 * config/arm/arm.c (arm_expand_prologue): Likewise.
7784 (arm_frame_pointer_required): Likewise.
7785 * config/i386/i386.c (ix86_expand_prologue): Likewise.
7786 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
7787 * config/mips/mips.c (mips_expand_prologue): Likewise.
7788 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
7789 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
7790 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7791 (sparc_flat_expand_prologue): Likewise.
7793 * common.opt (-fstack-clash-protection): New option.
7794 * flag-types.h (enum stack_check_type): Note difference between
7795 -fstack-check= and -fstack-clash-protection.
7796 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
7797 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
7798 * toplev.c (process_options): Issue warnings/errors for cases
7799 not handled with -fstack-clash-protection.
7800 * doc/invoke.texi (-fstack-clash-protection): Document new option.
7801 (-fstack-check): Note additional problem with -fstack-check=generic.
7802 Note that -fstack-check is primarily for Ada and refer users
7803 to -fstack-clash-protection for stack-clash-protection.
7804 Document new params for stack clash protection.
7806 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
7808 * config/i386/i386.c (ix86_split_long_move): Do not handle
7809 address used for LEA in a special way.
7811 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
7813 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
7815 2017-09-19 Martin Sebor <msebor@redhat.com>
7818 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
7819 of incompatible types.
7821 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
7823 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7824 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
7825 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7826 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
7828 2017-09-19 Richard Biener <rguenther@suse.de>
7830 PR tree-optimization/82244
7831 * tree-vrp.c (remove_range_assertions): Do not propagate
7832 a constant to abnormals but replace the assert with a copy.
7834 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
7836 PR rtl-optimization/57878
7837 PR rtl-optimization/68988
7838 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
7839 avoidance test involving non_reload_pseudos. Move frequency test
7840 below the general fragmentation avoidance test.
7842 2017-09-19 Richard Biener <rguenther@suse.de>
7844 PR tree-optimization/69728
7845 * graphite-sese-to-poly.c (schedule_error): New global.
7846 (add_loop_schedule): Handle empty domain by failing the
7848 (build_original_schedule): Handle schedule_error.
7850 2017-09-19 Richard Biener <rguenther@suse.de>
7852 * graphite-scop-detection.c (scop_detection::can_represent_loop):
7853 Do not iterate to sibling loops but only to siblings of inner
7856 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
7859 * config/m68k/m68k.md (moveq feeding equality comparison): Check
7860 that the registers are different.
7862 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
7864 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
7865 to processor_model and "amdfam17h" to arch_names_table.
7866 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
7868 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7871 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
7873 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7874 Alan Hayward <alan.hayward@arm.com>
7875 David Sherwood <david.sherwood@arm.com>
7877 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
7879 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
7881 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
7882 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
7883 vect_schedule_slp_instance.
7884 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
7885 Update call to vect_slp_analyze_node_operations. Simplify return
7887 (vect_slp_analyze_bb_1): Update call accordingly.
7888 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
7889 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
7890 (vect_schedule_slp): Update call accordingly.
7892 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7893 Alan Hayward <alan.hayward@arm.com>
7894 David Sherwood <david.sherwood@arm.com>
7896 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
7897 with types that aren't in fact scalar.
7899 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7901 * tree-vect-slp.c (vect_record_max_nunits): New function,
7903 (vect_build_slp_tree_1): ...here.
7904 (vect_build_slp_tree_2): Call it for phis too.
7906 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7908 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
7909 to vect_get_vec_def_for_operand when getting the mask operand.
7911 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7912 Alan Hayward <alan.hayward@arm.com>
7913 David Sherwood <david.sherwood@arm.com>
7915 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
7918 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7919 Alan Hayward <alan.hayward@arm.com>
7920 David Sherwood <david.sherwood@arm.com>
7922 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
7923 calculation for vector booleans.
7925 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7926 Alan Hayward <alan.hayward@arm.com>
7927 David Sherwood <david.sherwood@arm.com>
7929 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
7931 (vect_transform_stmt): ...here.
7932 (vect_analyze_stmt): Use it instead of calling
7933 vectorizable_live_operation directly.
7935 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
7937 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
7938 non-SIMT targets in acc vector loops.
7940 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
7942 * configure.ac: Add arc and check if assembler supports gdwarf2.
7943 * configure: Regenerate.
7945 2017-09-18 Richard Biener <rguenther@suse.de>
7947 PR tree-optimization/82220
7948 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
7949 epilogue niters from the min_profitable_iters compute.
7951 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7954 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
7955 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
7956 (ix86_init_pic_reg): Revert 2017-09-01 changes.
7958 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
7961 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
7962 switching to a new text section.
7964 2017-09-18 Richard Biener <rguenther@suse.de>
7966 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
7968 (build_alias_set): Reject aliases with no access function.
7970 2017-09-18 Richard Biener <rguenther@suse.de>
7972 PR tree-optimization/79622
7973 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
7975 (build_cross_bb_scalars_use): Likewise.
7977 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
7979 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
7981 2017-09-18 Alan Modra <amodra@gmail.com>
7984 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
7985 stack_pointer_rtx for count 0. Update comments. Break up
7986 large rtl expression.
7988 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
7990 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
7991 Increase to 20 bytes.
7992 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
7993 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
7994 or avx version of the stub.
7996 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
7999 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8000 compute the minimum stack alignment. Also update preferred stack
8001 boundary for leaf functions.
8003 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8005 PR tree-optimization/82228
8006 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8009 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8011 * common/config/nds32/nds32-common.c
8012 (nds32_option_optimization_table): Refine formatting.
8013 (nds32_option_optimization_table): Use -fsched-pressure and
8014 -fomit-frame-pointer for specific optimization level.
8016 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8018 * config/nds32/nds32.c: Refine formatting and comments.
8019 * config/nds32/nds32.h: Likewise.
8020 * config/nds32/nds32.md: Likewise.
8021 * config/nds32/nds32-cost.c: Likewise.
8022 * config/nds32/nds32-isr.c: Likewise.
8023 * config/nds32/nds32-md-auxiliary.c: Likewise.
8024 * config/nds32/nds32-multiple.md: Likewise.
8025 * config/nds32/nds32-predicates.c: Likewise.
8027 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8028 Jakub Jelinek <jakub@redhat.com>
8030 Add support for -std=c++2a.
8031 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8033 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8035 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8038 * doc/extend.texi (Common Function Attributes): Add
8039 references to ARM, AArch64, and S/390 specific attributes.
8040 (Function Specific Option Pragmas): Add AArch64 and S/390
8041 to list of back ends that support the target pragma.
8043 2017-09-15 Nathan Sidwell <nathan@acm.org>
8045 * doc/standards.texi: Fix C++17 description. Update URLs for
8048 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8050 * common.opt (Wcast-align=strict): New warning option.
8051 * doc/invoke.texi: Document -Wcast-align=strict.
8053 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8055 * cgraph.h (cgraph_thunk_info): Add comments.
8056 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8057 assert for VIRTUAL_* arguments stricter.
8059 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8061 PR tree-optimization/71026
8062 * match.pd: Move RDIV patterns from fold-const.c
8063 * fold-const.c (distribute_real_division): Removed.
8064 (fold_binary_loc): Remove calls to distribute_real_divison.
8066 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8068 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8069 c++1z and gnu++1z as deprecated. Change other references to
8070 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8071 Change -Wc++1z-compat to -Wc++17-compat.
8072 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8073 * dwarf2out.c (highest_c_language): Handle C++17.
8074 (gen_compile_unit_die): Likewise.
8076 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8078 PR rtl-optimization/82192
8079 * combine.c (make_extraction): Don't look through non-paradoxical
8080 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8083 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8084 Alan Hayward <alan.hayward@arm.com>
8085 David Sherwood <david.sherwood@arm.com>
8087 * target.def (function_arg_offset): New hook.
8088 * targhooks.h (default_function_arg_offset): Declare.
8089 * targhooks.c (default_function_arg_offset): New function.
8090 * function.c (locate_and_pad_parm): Use
8091 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8092 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8093 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8094 * doc/tm.texi: Regenerate.
8095 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8096 * config/spu/spu.c (spu_function_arg_offset): New function.
8097 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8098 * system.h (FUNCTION_ARG_OFFSET): Poison.
8100 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8101 Alan Hayard <alan.hayward@arm.com>
8102 David Sherwood <david.sherwood@arm.com>
8104 * target.def (truly_noop_truncation): New hook.
8105 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8106 than TRULY_NOOP_TRUNCATION.
8107 * hooks.h (hook_bool_uint_uint_true): Declare.
8108 * hooks.c (hook_bool_uint_uint_true): New function.
8109 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8110 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8111 * doc/tm.texi: Regenerate.
8112 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8113 rather than TRULY_NOOP_TRUNCATION in comments.
8114 (simplify_comparison): Likewise.
8115 (record_truncated_value): Likewise.
8116 * expmed.c (extract_bit_field_1): Likewise.
8117 (extract_split_bit_field): Likewise.
8118 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8119 instead of TRULY_NOOP_TRUNCATION.
8120 * function.c (assign_parm_setup_block): Likewise.
8121 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8122 * rtlhooks.c: Include target.h.
8123 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8124 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8125 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8126 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8127 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8128 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8129 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8130 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8131 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8132 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8133 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8134 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8135 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8136 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8137 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8138 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8139 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8140 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8141 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8142 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8143 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8144 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8145 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8146 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8147 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8148 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8149 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8150 rather than TRULY_NOOP_TRUNCATION in comments.
8151 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8152 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8153 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8154 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8155 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8156 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8157 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8158 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8159 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8160 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8161 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8162 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8163 rather than TRULY_NOOP_TRUNCATION in comments.
8164 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8165 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8166 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8167 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8168 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8169 TRULY_NOOP_TRUNCATION condition.
8170 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8171 (TRULY_NOOP_TRUNCATION): Delete.
8172 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8173 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8174 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8175 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8176 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8177 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8178 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8179 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8180 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8181 rather than TRULY_NOOP_TRUNCATION in comments.
8182 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8183 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8184 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8185 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8186 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8187 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8189 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8192 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8193 (*cmp_ior): Likewise.
8194 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8195 (*ior_scc_scc_cmp): Likewise.
8196 (*and_scc_scc): Likewise.
8197 (*and_scc_scc_cmp): Likewise.
8199 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8200 Alan Hayard <alan.hayward@arm.com>
8201 David Sherwood <david.sherwood@arm.com>
8203 * target.def (can_change_mode_class): New hook.
8204 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8205 (hard_regno_nregs): Likewise.
8206 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8207 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8208 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8209 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8210 (LOAD_EXTEND_OP): Update accordingly.
8211 * doc/tm.texi: Regenerate.
8212 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8213 CANNOT_CHANGE_MODE_CLASS.
8214 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8215 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8216 * combine.c (simplify_set): Update accordingly.
8217 * emit-rtl.c (validate_subreg): Likewise.
8218 * recog.c (general_operand): Likewise.
8219 * regcprop.c (mode_change_ok): Likewise.
8220 * reload1.c (choose_reload_regs): Likewise.
8221 (inherit_piecemeal_p): Likewise.
8222 * rtlanal.c (simplify_subreg_regno): Likewise.
8223 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8224 instead of CANNOT_CHANGE_MODE_CLASS.
8225 (reload_cse_simplify_operands): Likewise.
8226 * reload.c (push_reload): Use targetm.can_change_mode_class
8227 instead of CANNOT_CHANGE_MODE_CLASS.
8228 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8229 REG_CANNOT_CHANGE_MODE_P.
8230 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8231 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8232 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8233 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8234 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8235 (arm_can_change_mode_class): New function.
8236 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8237 than CANNOT_CHANGE_MODE_CLASS in comments.
8238 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8239 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8240 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8241 (ix86_can_change_mode_class): ...this new function, inverting the
8242 sense of the return value.
8243 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8244 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8245 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8246 (ia64_can_change_mode_class): New function.
8247 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8248 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8249 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8250 (m32c_can_change_mode_class): ...this new function, inverting the
8251 sense of the return value.
8252 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8253 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8254 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8255 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8256 (mips_can_change_mode_class): ...this new function, inverting the
8257 sense of the return value.
8258 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8259 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8260 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8261 (msp430_can_change_mode_class): New function.
8262 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8263 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8264 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8265 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8266 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8267 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8268 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8269 (pa_cannot_change_mode_class): Replace with...
8270 (pa_can_change_mode_class): ...this new function, inverting the
8271 sense of the return value.
8272 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8273 than CANNOT_CHANGE_MODE_CLASS in comments.
8274 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8275 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8276 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8277 (pdp11_cannot_change_mode_class): Replace with...
8278 (pdp11_can_change_mode_class): ...this new function, inverting the
8279 sense of the return value.
8280 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8281 * config/powerpcspe/powerpcspe-protos.h
8282 (rs6000_cannot_change_mode_class_ptr): Delete.
8283 * config/powerpcspe/powerpcspe.c
8284 (rs6000_cannot_change_mode_class_ptr): Delete.
8285 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8286 (rs6000_option_override_internal): Assign to
8287 targetm.can_change_mode_class instead of
8288 rs6000_cannot_change_mode_class_ptr.
8289 (rs6000_cannot_change_mode_class): Replace with...
8290 (rs6000_can_change_mode_class): ...this new function, inverting the
8291 sense of the return value.
8292 (rs6000_debug_cannot_change_mode_class): Replace with...
8293 (rs6000_debug_can_change_mode_class): ...this new function.
8294 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8295 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8296 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8297 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8298 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8300 * config/rs6000/rs6000.c (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/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8311 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8312 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8313 (s390_can_change_mode_class): ...this new function, inverting the
8314 sense of the return value.
8315 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8316 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8317 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8318 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8319 (sh_cannot_change_mode_class): Replace with...
8320 (sh_can_change_mode_class): ...this new function, inverting the
8321 sense of the return value.
8322 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8323 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8324 (sparc_can_change_mode_class): New function.
8325 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8326 * config/spu/spu.c (spu_can_change_mode_class): New function.
8327 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8328 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8329 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8330 (visium_can_change_mode_class): New function.
8331 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8333 2017-09-15 Richard Biener <rguenther@suse.de>
8335 PR tree-optimization/82217
8336 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8337 but not undefined case.
8339 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8342 * postreload.c (reload_cse_simplify_operands): Skip
8343 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
8345 2017-09-15 Richard Biener <rguenther@suse.de>
8347 PR tree-optimization/68823
8348 * graphite-scop-detection.c (build_alias_set): If we have a
8349 possible dependence check whether we can handle them by just
8350 looking at the DRs DR_ACCESS_FNs.
8351 (build_scops): If build_alias_set fails, fail the SCOP.
8353 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
8355 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
8356 to support float128 built-in functions that require the ISA 3.0
8358 (BU_FLOAT128_3_HW): Likewise.
8359 (SQRTF128): Add support for the IEEE 128-bit square root and fma
8361 (FMAF128): Likewise.
8363 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8364 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8365 floating point instructions.
8366 (rs6000_invalid_builtin): Likewise.
8367 (rs6000_builtin_mask_names): Likewise.
8368 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8369 (RS6000_BTM_FLOAT128_HW): Likewise.
8370 (RS6000_BTM_COMMON): Likewise.
8371 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
8373 * doc/extend.texi (RS/6000 built-in functions): Document the
8374 IEEE 128-bit floating point square root and fused multiply-add
8377 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
8379 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
8380 reg (r2) isn't in the set of registers defined in the prologue.
8382 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8383 Alan Hayward <alan.hayward@arm.com>
8384 David Sherwood <david.sherwood@arm.com>
8386 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
8387 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
8388 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
8389 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
8390 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
8392 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
8393 max_vectorization_factor.
8394 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
8396 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8397 Alan Hayward <alan.hayward@arm.com>
8398 David Sherwood <david.sherwood@arm.com>
8400 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
8401 (vect_worthwhile_without_simd_p): Declare.
8402 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
8403 (vectorizable_reduction): Use it.
8404 * tree-vect-stmts.c (vectorizable_shift): Likewise.
8405 (vectorizable_operation): Likewise.
8407 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8408 Alan Hayward <alan.hayward@arm.com>
8409 David Sherwood <david.sherwood@arm.com>
8411 * tree-vectorizer.h (vect_get_num_copies): New function.
8412 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
8413 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8414 (vectorizable_induction): Likewise.
8415 (vectorizable_live_operation): Likewise.
8416 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8417 (vectorizable_bswap): Likewise.
8418 (vectorizable_call): Likewise.
8419 (vectorizable_conversion): Likewise.
8420 (vectorizable_assignment): Likewise.
8421 (vectorizable_shift): Likewise.
8422 (vectorizable_operation): Likewise.
8423 (vectorizable_store): Likewise.
8424 (vectorizable_load): Likewise.
8425 (vectorizable_condition): Likewise.
8426 (vectorizable_comparison): Likewise.
8427 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
8429 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8430 Alan Hayward <alan.hayward@arm.com>
8431 David Sherwood <david.sherwood@arm.com>
8433 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
8434 of vect_init_vector.
8436 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8437 Alan Hayward <alan.hayward@arm.com>
8438 David Sherwood <david.sherwood@arm.com>
8440 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
8441 an inline wrapper that provides a location.
8442 (gimple_build_vector): Likewise.
8443 * gimple-fold.c (gimple_build_vector_from_val): New function.
8444 (gimple_build_vector): Likewise.
8445 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
8446 functions to build the initial value. Always return a gimple value.
8447 (get_initial_defs_for_reduction): Likewise. Only compute
8449 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
8450 vect_init_vector on the results from get_initial_def(s)_for_reduction.
8451 (vectorizable_induction): Use gimple_build_vector rather than
8454 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8455 Alan Hayward <alan.hayward@arm.com>
8456 David Sherwood <david.sherwood@arm.com>
8458 * target.h (vec_perm_indices): New typedef.
8459 (auto_vec_perm_indices): Likewise.
8460 * optabs-query.h: Include target.h
8461 (can_vec_perm_p): Take a vec_perm_indices *.
8462 * optabs-query.c (can_vec_perm_p): Likewise.
8463 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
8464 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8465 * tree-vect-generic.c (lower_vec_perm): Likewise.
8466 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
8467 (vect_grouped_load_supported): Likewise.
8468 (vect_shift_permute_load_chain): Likewise.
8469 (vect_permute_store_chain): Use auto_vec_perm_indices.
8470 (vect_permute_load_chain): Likewise.
8471 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
8472 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
8473 Update uses of can_vec_perm_p.
8474 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
8475 mode with a number of elements. Take a vec_perm_indices *.
8476 (vect_create_epilog_for_reduction): Update accordingly.
8477 Use auto_vec_perm_indices.
8478 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
8479 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
8480 (vect_transform_slp_perm_load): Likewise.
8481 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
8482 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
8483 (vect_gen_perm_mask_checked): Likewise.
8484 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
8485 (vect_gen_perm_mask_checked): Likewise.
8486 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
8487 (vectorizable_store): Likewise.
8488 (vectorizable_load): Likewise.
8489 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
8490 (vectorizable_bswap): Likewise.
8492 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8493 Alan Hayward <alan.hayward@arm.com>
8494 David Sherwood <david.sherwood@arm.com>
8496 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
8497 * tree.c (build_vector): Likewise.
8498 (build_vector_from_ctor): Update accordingly.
8499 (build_vector_from_val): Likewise.
8500 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8501 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8502 * tree-vect-generic.c (add_rshift): Likewise.
8503 (expand_vector_divmod): Likewise.
8504 (optimize_vector_constructor): Likewise.
8505 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
8506 (vect_transform_slp_perm_load): Likewise.
8507 (vect_schedule_slp_instance): Likewise.
8508 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
8509 (vectorizable_call): Likewise.
8510 (vect_gen_perm_mask_any): Likewise. Add elements in order.
8511 * expmed.c (make_tree): Likewise.
8512 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
8513 a vector passed to build_vector.
8514 (fold_convert_const): Likewise.
8515 (exact_inverse): Likewise.
8516 (fold_ternary_loc): Likewise.
8517 (fold_relational_const): Likewise.
8518 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
8519 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
8520 (const_unop): Likewise. Store the reduction accumulator in a
8521 variable rather than an array.
8522 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
8523 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
8524 the new vector, rather than constructing it after the input arrays.
8525 (native_interpret_vector): Use auto_vec<tree> when building
8526 a vector passed to build_vector. Add elements in order.
8527 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
8528 auto_vec<tree> when building a vector passed to build_vector.
8529 (vect_create_epilog_for_reduction): Likewise.
8530 (vectorizable_induction): Likewise.
8531 (get_initial_def_for_reduction): Likewise. Fix indentation of
8533 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
8535 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
8536 passed to build_vector.
8538 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8539 Alan Hayward <alan.hayward@arm.com>
8540 David Sherwood <david.sherwood@arm.com>
8542 * tree-core.h (tree_base::u): Add an "nelts" field.
8543 (tree_vector): Use VECTOR_CST_NELTS as the length.
8544 * tree.c (tree_size): Likewise.
8545 (make_vector): Initialize VECTOR_CST_NELTS.
8546 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
8547 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
8548 TYPE_VECTOR_SUBPARTS.
8549 * expr.c (const_vector_mask_from_tree): Consistently use "units"
8550 as the number of units, setting it from VECTOR_CST_NELTS.
8551 (const_vector_from_tree): Likewise.
8552 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
8553 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
8554 (fold_negate_expr_1): Likewise.
8555 (fold_convert_const): Likewise.
8556 (const_binop): Likewise. Differentiate the number of output and
8558 (const_unop): Likewise.
8559 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
8560 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
8561 in cases that did the opposite.
8563 2017-09-14 Richard Biener <rguenther@suse.de>
8565 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
8568 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
8570 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
8572 2017-09-14 Jakub Jelinek <jakub@redhat.com>
8575 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
8576 if and where to split a bb, except for splitting before debug insn
8577 sequences followed by non-label real insn. Delete debug insns
8578 in between basic blocks.
8580 * combine.c (make_compound_operation_int): Formatting fixes.
8582 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
8583 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8584 * config/netbsd.h (LINK_EH_SPEC): Likewise.
8585 * config/sol2.h (LINK_EH_SPEC): Likewise.
8586 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8587 * config/s390/linux.h (LINK_SPEC): Likewise.
8588 * config/freebsd.h (LINK_EH_SPEC): Likewise.
8589 * config/openbsd.h (LINK_EH_SPEC): Likewise.
8590 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8591 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
8592 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
8593 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8594 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
8595 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
8597 2017-09-13 Jakub Jelinek <jakub@redhat.com>
8599 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
8601 (ENDFILE_LINUX_SPEC): Likewise.
8602 (LINK_EH_SPEC): Likewise.
8603 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
8604 (LINK_OS_LINUX_SPEC32): Likewise.
8605 (LINK_OS_LINUX_SPEC64): Likewise.
8606 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
8607 (LINK_OS_LINUX_SPEC): Likewise.
8609 2017-09-13 Martin Liska <mliska@suse.cz>
8612 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
8613 CASE_HIGH is NULL_TREE.
8615 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8616 Alan Hayward <alan.hayward@arm.com>
8617 David Sherwood <david.sherwood@arm.com>
8619 * target.def (secondary_memory_needed): New hook.
8620 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
8621 instead of SECONDARY_MEMORY_NEEDED.
8622 (secondary_memory_needed_mode): Likewise.
8623 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
8624 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
8625 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
8626 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
8627 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
8628 * doc/tm.texi: Regenerate.
8629 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
8630 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
8631 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8632 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
8633 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
8634 * config/i386/i386.c (inline_secondary_memory_needed): Put the
8635 mode argument first and change the reg_class arguments to reg_class_t.
8636 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
8637 Make static. Update the call to inline_secondary_memory_needed.
8638 (ix86_register_move_cost): Update the call to
8639 inline_secondary_memory_needed.
8640 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8641 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
8643 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
8644 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8646 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
8647 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
8648 * config/mips/mips.c (mips_secondary_memory_needed): Make static
8649 and match hook interface. Add comment from mips.h.
8650 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8651 * config/mmix/mmix.md (truncdfsf2): Refer to
8652 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8654 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
8655 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
8656 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8657 (pa_secondary_memory_needed): New function.
8658 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
8659 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
8660 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8661 (pdp11_secondary_memory_needed): Make static and match hook interface.
8662 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
8663 * config/powerpcspe/powerpcspe-protos.h
8664 (rs6000_secondary_memory_needed_ptr): Delete.
8665 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
8667 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8668 (rs6000_option_override_internal): Assign to
8669 targetm.secondary_memory_needed rather than
8670 rs6000_secondary_memory_needed_ptr.
8671 (rs6000_secondary_memory_needed): Match hook interface.
8672 (rs6000_debug_secondary_memory_needed): Likewise.
8673 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
8674 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
8675 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
8676 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8677 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
8678 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
8680 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
8681 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8682 (rs6000_option_override_internal): Assign to
8683 targetm.secondary_memory_needed rather than
8684 rs6000_secondary_memory_needed_ptr.
8685 (rs6000_secondary_memory_needed): Match hook interface.
8686 (rs6000_debug_secondary_memory_needed): Likewise.
8687 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
8688 * config/s390/s390.c (s390_secondary_memory_needed): New function.
8689 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8690 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
8691 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8692 (sparc_secondary_memory_needed): New function.
8693 * lra-constraints.c (check_and_process_move): Refer to
8694 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8696 (curr_insn_transform): Likewise.
8697 (process_alt_operands): Use targetm.secondary_memory_needed
8698 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8699 (check_secondary_memory_needed_p): Likewise.
8700 (choose_split_class): Likewise.
8701 * reload.c: Unconditionally include code that was previously
8702 conditional on SECONDARY_MEMORY_NEEDED.
8703 (push_secondary_reload): Use targetm.secondary_memory_needed
8704 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8705 (push_reload): Likewise.
8706 * reload1.c: Unconditionally include code that was previously
8707 conditional on SECONDARY_MEMORY_NEEDED.
8708 (choose_reload_regs): Use targetm.secondary_memory_needed
8709 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8710 (gen_reload): Likewise.
8711 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
8713 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8714 Alan Hayward <alan.hayward@arm.com>
8715 David Sherwood <david.sherwood@arm.com>
8717 * target.def (secondary_memory_needed_mode): New hook:
8718 * targhooks.c (default_secondary_memory_needed_mode): Declare.
8719 * targhooks.h (default_secondary_memory_needed_mode): New function.
8720 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
8721 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
8722 * doc/tm.texi: Regenerate.
8723 * lra-constraints.c (check_and_process_move): Use
8724 targetm.secondary_memory_needed_mode instead of
8725 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
8726 (curr_insn_transform): Likewise.
8727 * reload.c (get_secondary_mem): Likewise.
8728 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8729 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
8731 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8732 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8733 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
8734 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8735 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
8737 * config/powerpcspe/powerpcspe-protos.h
8738 (rs6000_secondary_memory_needed_mode): Delete.
8739 * config/powerpcspe/powerpcspe.c
8740 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8741 (rs6000_secondary_memory_needed_mode): Make static.
8742 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8743 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
8745 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8747 (rs6000_secondary_memory_needed_mode): Make static.
8748 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8749 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
8750 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8751 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8752 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8754 (sparc_secondary_memory_needed_mode): New function.
8755 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
8757 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
8759 * config/aarch64/constraints.md (Umq): New constraint.
8760 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
8762 (mov<mode>): Update condition.
8764 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8766 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
8767 when bitposition is the same.
8769 2017-09-13 Richard Biener <rguenther@suse.de>
8771 * dwarf2out.c (output_die_symbol): Remove.
8772 (output_die): Do not output a DIEs symbol.
8774 2017-09-13 Richard Biener <rguenther@suse.de>
8777 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
8778 default-def to avoid breaking iterator update with the weird
8779 interaction with cgraph_update_edges_for_call_stmt_node.
8781 2017-09-13 Richard Biener <rguenther@suse.de>
8783 * tree-cfg.c (verify_gimple_assign_binary): Add verification
8784 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
8785 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
8786 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
8788 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
8790 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
8791 Disable pc relative literal load irrespective of
8792 TARGET_FIX_ERR_A53_84341 for default.
8794 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
8796 * config/sparc/sparc.c (output_return): Output the source location of
8797 the insn in the delay slot, if any.
8798 (output_sibcall): Likewise.
8800 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
8803 * common.opt (-static-pie): New alias.
8804 (shared): Negate static-pie.
8805 (-no-pie): Update help text.
8807 (static-pie): New option.
8808 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
8809 -static-pie support.
8810 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
8811 (LINK_EH_SPEC): Likewise.
8812 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8813 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8814 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8815 * gcc.c (LINK_COMMAND_SPEC): Likewise.
8816 (init_gcc_specs): Likewise.
8817 (init_spec): Likewise.
8818 (display_help): Update help message for -pie.
8819 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
8822 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
8824 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
8825 (movdi_aarch64): Likewise.
8826 (movti_aarch64): Likewise.
8828 2017-09-12 Simon Wright <simon@pushface.org>
8831 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
8832 calculation of the minor version, always output as 0.
8834 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8837 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8838 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
8839 on it early, rather than manual conversion late. For
8840 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
8841 instead of performing manual conversion.
8843 2017-09-12 Carl Love <cel@us.ibm.com>
8845 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
8846 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
8848 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8849 VMULOSW): Add definitions.
8850 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8851 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8852 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8853 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8854 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8856 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8858 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
8860 (movti_aarch64): Likewise.
8861 (movdf_aarch64): Likewise.
8862 (movtf_aarch64): Likewise.
8863 (load_pairdi): Likewise.
8864 (store_pairdi): Likewise.
8865 (load_pairdf): Likewise.
8866 (store_pairdf): Likewise.
8867 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
8868 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
8869 (ldr_got_small_<mode>): Likewise.
8870 (ldr_got_small_28k_<mode>): Likewise.
8871 (ldr_got_tiny): Likewise.
8872 * config/aarch64/iterators.md (ldst_sz): New.
8873 (ldpstp_sz): Likewise.
8874 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
8876 (thunderx_load): Split load_8 to load_16.
8877 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
8879 (thunderx2t99_storepair_basic): Split store_8 to store_16.
8880 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
8881 (xgene1_store_pair): Split store_8 to store_16.
8882 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
8883 (falkor_st_0_st_sd): Split store_8 to store_16.
8885 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8887 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
8888 and store1/2/3/4 to store_4/8/12/16.
8889 * config/aarch64/aarch64.md: Update for rename.
8890 * config/arm/arm.md: Likewise.: Likewise.
8891 * config/arm/arm.c: Likewise.
8892 * config/arm/thumb1.md: Likewise.
8893 * config/arm/thumb2.md: Likewise.
8894 * config/arm/vfp.md: Likewise.
8895 * config/arm/arm-generic.md: Likewise.
8896 * config/arm/arm1020e.md: Likewise.
8897 * config/arm/arm1026ejs.md: Likewise.
8898 * config/arm/arm1136jfs.md: Likewise.
8899 * config/arm/arm926ejs.md: Likewise.
8900 * config/arm/cortex-a15.md: Likewise.
8901 * config/arm/cortex-a17.md: Likewise.
8902 * config/arm/cortex-a5.md: Likewise.
8903 * config/arm/cortex-a53.md: Likewise.
8904 * config/arm/cortex-a57.md: Likewise.
8905 * config/arm/cortex-a7.md: Likewise.
8906 * config/arm/cortex-a8.md: Likewise.
8907 * config/arm/cortex-a9.md: Likewise.
8908 * config/arm/cortex-m4.md: Likewise.
8909 * config/arm/cortex-m7.md: Likewise.
8910 * config/arm/cortex-r4.md: Likewise.
8911 * config/arm/exynos-m1.md: Likewise.
8912 * config/arm/fa526.md: Likewise.
8913 * config/arm/fa606te.md: Likewise.
8914 * config/arm/fa626te.md: Likewise.
8915 * config/arm/fa726te.md: Likewise.
8916 * config/arm/fmp626.md: Likewise.
8917 * config/arm/iwmmxt.md: Likewise.
8918 * config/arm/ldmstm.md: Likewise.
8919 * config/arm/marvell-pj4.md: Likewise.
8920 * config/arm/xgene1.md: Likewise.
8921 * config/aarch64/thunderx.md: Likewise.
8922 * config/aarch64/thunderx2t99.md: Likewise.
8923 * config/aarch64/falkor.md: Likewise.
8925 2017-09-12 Martin Liska <mliska@suse.cz>
8927 * attribs.c (private_lookup_attribute): New function.
8928 * attribs.h (private_lookup_attribute): Declared here.
8929 (lookup_attribute): Called from this place.
8931 2017-09-12 Richard Biener <rguenther@suse.de>
8933 PR tree-optimization/82157
8934 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
8935 stmts with side-effects.
8937 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8938 Alan Hayward <alan.hayward@arm.com>
8939 David Sherwood <david.sherwood@arm.com>
8941 * target.def (hard_regno_nregs): New hook.
8942 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
8943 * targhooks.h (default_hard_regno_nregs): Declare.
8944 * targhooks.c (default_hard_regno_nregs): New function.
8945 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
8946 (TARGET_HARD_REGNO_NREGS): ...this hook.
8947 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
8948 (CLASS_MAX_NREGS): Likewise.
8949 * doc/tm.texi: Regenerate.
8950 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
8951 instead of HARD_REGNO_NREGS.
8952 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
8953 HARD_REGNO_NREGS in the comment.
8954 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
8955 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
8956 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
8957 Return an unsigned int.
8958 (TARGET_HARD_REGNO_NREGS): Redefine.
8959 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
8960 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
8961 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8962 (arc_hard_regno_nregs): New function.
8963 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
8964 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
8965 (arm_hard_regno_nregs): New function.
8966 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
8967 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
8968 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
8969 (TARGET_HARD_REGNO_NREGS): Redefine.
8970 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
8971 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
8972 (HARD_REGNO_NREGS): Delete.
8973 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
8974 (cr16_hard_regno_nregs): New function.
8975 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
8976 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
8977 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
8978 (cris_hard_regno_nregs): New function.
8979 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
8980 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
8981 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8982 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
8983 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
8984 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
8985 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
8986 (frv_hard_regno_nregs): Make static. Take and return an
8988 (frv_class_max_nregs): Remove outdated copy of documentation.
8989 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
8990 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
8991 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
8992 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
8993 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
8994 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
8995 (TARGET_HARD_REGNO_NREGS): Redefine.
8996 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
8997 (CLASS_MAX_NREGS): Update comment.
8998 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
8999 (ia64_hard_regno_nregs): New function.
9000 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9001 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9002 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9003 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9004 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9006 (m32c_hard_regno_nregs): Likewise. Make static.
9007 (TARGET_HARD_REGNO_NREGS): Redefine.
9008 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9009 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9010 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9011 (m68k_hard_regno_nregs): New function.
9012 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9013 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9014 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9015 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9016 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9017 Take and return an unsigned int.
9018 (TARGET_HARD_REGNO_NREGS): Redefine.
9019 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9020 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9021 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9022 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9023 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9024 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9025 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9026 (msp430_hard_regno_nregs): Make static. Take and return an
9028 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9029 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9030 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9031 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9032 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9033 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9034 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9035 (TARGET_HARD_REGNO_NREGS): Redefine.
9036 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9037 (PA_HARD_REGNO_NREGS): ...this.
9038 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9039 (PA_HARD_REGNO_NREGS): ...this.
9040 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9041 (pa_hard_regno_nregs): New function.
9042 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9043 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9044 (pdp11_hard_regno_nregs): New function.
9045 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9046 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9047 (rs6000_hard_regno_nregs_hook): New function.
9048 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9049 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9050 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9051 Take and return an unsigned int. Move earlier in file.
9052 (TARGET_HARD_REGNO_NREGS): Redefine.
9053 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9054 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9055 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9056 (rl78_hard_regno_nregs): Make static. Take and return an
9058 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9059 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9060 (rs6000_hard_regno_nregs_hook): New function.
9061 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9062 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9063 (TARGET_HARD_REGNO_NREGS): Redefine.
9064 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9065 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9066 instead of HARD_REGNO_NREGS.
9067 (s390_hard_regno_nregs): New function.
9068 (s390_hard_regno_mode_ok): Add comment from s390.h.
9069 (TARGET_HARD_REGNO_NREGS): Redefine.
9070 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9071 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9072 (sh_hard_regno_nregs): New function.
9073 (sh_pass_in_reg_p): Use it.
9074 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9075 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9076 (sparc_hard_regno_nregs): New function.
9077 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9078 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9079 (spu_function_arg_advance): Use it, supplying a valid register number.
9080 (TARGET_HARD_REGNO_NREGS): Redefine.
9081 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9082 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9083 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9084 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9085 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9086 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9087 (CLASS_MAX_NREGS): Remove copy of old documentation.
9088 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9089 (visium_hard_regno_nregs): New function.
9090 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9091 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9092 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9093 xtensa_hard_regno_nregs): New function.
9094 * system.h (HARD_REGNO_NREGS): Poison.
9096 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9098 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9099 hard_regno_nregs instead of HARD_REGNO_NREGS.
9100 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9101 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9102 (c6x_expand_epilogue): Likewise.
9103 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9104 (frv_read_iacc_argument): Likewise.
9105 * config/sh/sh.c: Include regs.h.
9106 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9107 (regs_used): Likewise.
9108 (output_stack_adjust): Likewise.
9109 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9110 * expmed.c: Include regs.h.
9111 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9112 * ree.c: Include regs.h.
9113 (combine_reaching_defs): Use hard_regno_nregs instead of
9115 (add_removable_extension): Likewise.
9117 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9119 * regs.h (hard_regno_nregs): Turn into a function.
9120 (end_hard_regno): Update accordingly.
9121 * caller-save.c (setup_save_areas): Likewise.
9122 (save_call_clobbered_regs): Likewise.
9123 (replace_reg_with_saved_mem): Likewise.
9124 (insert_restore): Likewise.
9125 (insert_save): Likewise.
9126 * combine.c (can_change_dest_mode): Likewise.
9127 (move_deaths): Likewise.
9128 (distribute_notes): Likewise.
9129 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9130 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9131 (rs6000_split_multireg_move): Likewise.
9132 (rs6000_register_move_cost): Likewise.
9133 (rs6000_memory_move_cost): Likewise.
9134 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9135 (rs6000_split_multireg_move): Likewise.
9136 (rs6000_register_move_cost): Likewise.
9137 (rs6000_memory_move_cost): Likewise.
9138 * cselib.c (cselib_reset_table): Likewise.
9139 (cselib_lookup_1): Likewise.
9140 * emit-rtl.c (set_mode_and_regno): Likewise.
9141 * function.c (aggregate_value_p): Likewise.
9142 * ira-color.c (setup_profitable_hard_regs): Likewise.
9143 (check_hard_reg_p): Likewise.
9144 (calculate_saved_nregs): Likewise.
9145 (assign_hard_reg): Likewise.
9146 (improve_allocation): Likewise.
9147 (calculate_spill_cost): Likewise.
9148 * ira-emit.c (modify_move_list): Likewise.
9149 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9150 (ira_hard_reg_in_set_p): Likewise.
9151 * ira.c (setup_reg_mode_hard_regset): Likewise.
9152 (clarify_prohibited_class_mode_regs): Likewise.
9153 (check_allocation): Likewise.
9154 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9155 (lra_setup_reg_renumber): Likewise.
9156 (setup_try_hard_regno_pseudos): Likewise.
9157 (spill_for): Likewise.
9158 (assign_hard_regno): Likewise.
9159 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9160 * lra-constraints.c (in_class_p): Likewise.
9161 (lra_constraint_offset): Likewise.
9162 (simplify_operand_subreg): Likewise.
9163 (lra_constraints): Likewise.
9164 (split_reg): Likewise.
9165 (split_if_necessary): Likewise.
9166 (invariant_p): Likewise.
9167 (inherit_in_ebb): Likewise.
9168 * lra-lives.c (process_bb_lives): Likewise.
9169 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9170 (get_hard_regs): Likewise.
9171 (do_remat): Likewise.
9172 * lra-spills.c (assign_spill_hard_regs): Likewise.
9173 * mode-switching.c (create_pre_exit): Likewise.
9174 * postreload.c (reload_combine_recognize_pattern): Likewise.
9175 * recog.c (peep2_find_free_register): Likewise.
9176 * regcprop.c (kill_value_regno): Likewise.
9177 (set_value_regno): Likewise.
9178 (copy_value): Likewise.
9179 (maybe_mode_change): Likewise.
9180 (find_oldest_value_reg): Likewise.
9181 (copyprop_hardreg_forward_1): Likewise.
9182 * regrename.c (check_new_reg_p): Likewise.
9183 (regrename_do_replace): Likewise.
9184 * reload.c (push_reload): Likewise.
9185 (combine_reloads): Likewise.
9186 (find_dummy_reload): Likewise.
9187 (operands_match_p): Likewise.
9188 (find_reloads): Likewise.
9189 (find_equiv_reg): Likewise.
9190 (reload_adjust_reg_for_mode): Likewise.
9191 * reload1.c (count_pseudo): Likewise.
9192 (count_spilled_pseudo): Likewise.
9193 (find_reg): Likewise.
9194 (clear_reload_reg_in_use): Likewise.
9195 (free_for_value_p): Likewise.
9196 (allocate_reload_reg): Likewise.
9197 (choose_reload_regs): Likewise.
9198 (reload_adjust_reg_for_temp): Likewise.
9199 (emit_reload_insns): Likewise.
9200 (delete_output_reload): Likewise.
9201 * rtlanal.c (subreg_get_info): Likewise.
9202 * sched-deps.c (sched_analyze_reg): Likewise.
9203 * sel-sched.c (init_regs_for_mode): Likewise.
9204 (mark_unavailable_hard_regs): Likewise.
9205 (choose_best_reg_1): Likewise.
9206 (verify_target_availability): Likewise.
9207 * valtrack.c (dead_debug_insert_temp): Likewise.
9208 * var-tracking.c (track_loc_p): Likewise.
9209 (emit_note_insn_var_location): Likewise.
9210 * varasm.c (make_decl_rtl): Likewise.
9211 * reginfo.c (choose_hard_reg_mode): Likewise.
9212 (init_reg_modes_target): Refer directly to
9213 this_target_regs->x_hard_regno_nregs.
9215 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9217 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9218 instead of hard_regno_nregs.
9220 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9222 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9223 end_hard_regno instead of hard_regno_nregs.
9224 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9225 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9226 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9227 * ira-color.c (improve_allocation): Likewise.
9228 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9229 * lra-lives.c (mark_regno_live): Likewise.
9230 (mark_regno_dead): Likewise.
9231 * lra-remat.c (operand_to_remat): Likewise.
9232 * lra.c (collect_non_operand_hard_regs): Likewise.
9233 * postreload.c (reload_combine_note_store): Likewise.
9234 (move2add_valid_value_p): Likewise.
9235 * reload.c (regno_clobbered_p): Likewise.
9237 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9239 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9241 * config/v850/v850.c (v850_reorg): Likewise.
9242 * reload.c (refers_to_regno_for_reload_p): Likewise.
9243 (find_equiv_reg): Likewise.
9244 * reload1.c (reload_reg_reaches_end_p): Likewise.
9246 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9248 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9250 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9251 * config/arm/arm.c (output_move_neon): Likewise.
9252 (arm_attr_length_move_neon): Likewise.
9253 (neon_split_vcombine): Likewise.
9254 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9255 (c6x_mark_reg_written): Likewise.
9256 (c6x_dwarf_register_span): Likewise.
9257 * config/i386/i386.c (ix86_save_reg): Likewise.
9258 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9259 (rws_access_reg): Likewise.
9260 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9261 * mode-switching.c (create_pre_exit): Likewise.
9262 * ree.c (combine_reaching_defs): Likewise.
9263 (add_removable_extension): Likewise.
9264 * regcprop.c (find_oldest_value_reg): Likewise.
9265 (copyprop_hardreg_forward_1): Likewise.
9266 * reload.c (reload_inner_reg_of_subreg): Likewise.
9267 (push_reload): Likewise.
9268 (combine_reloads): Likewise.
9269 (find_dummy_reload): Likewise.
9270 (reload_adjust_reg_for_mode): Likewise.
9271 * reload1.c (find_reload_regs): Likewise.
9272 (forget_old_reloads_1): Likewise.
9273 (reload_reg_free_for_value_p): Likewise.
9274 (reload_adjust_reg_for_temp): Likewise.
9275 (emit_reload_insns): Likewise.
9276 (delete_output_reload): Likewise.
9277 * sel-sched.c (choose_best_reg_1): Likewise.
9278 (choose_best_pseudo_reg): Likewise.
9280 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9281 Alan Hayward <alan.hayward@arm.com>
9282 David Sherwood <david.sherwood@arm.com>
9284 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9285 * target.def (slow_unaligned_access): New hook.
9286 * targhooks.h (default_slow_unaligned_access): Declare.
9287 * targhooks.c (default_slow_unaligned_access): New function.
9288 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9289 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9290 * doc/tm.texi: Regenerate.
9291 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9292 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9293 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9295 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9296 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9298 (rs6000_slow_unaligned_access): New function.
9299 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9300 (expand_block_compare): Likewise.
9301 (expand_strn_compare): Likewise.
9302 (rs6000_rtx_costs): Likewise.
9303 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9304 (riscv_slow_unaligned_access): Likewise.
9305 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9306 (riscv_slow_unaligned_access_p): ...this and make static.
9307 (riscv_option_override): Update accordingly.
9308 (riscv_slow_unaligned_access): New function.
9309 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9310 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9311 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9312 (rs6000_slow_unaligned_access): New function.
9313 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9314 (rs6000_rtx_costs): Likewise.
9315 * config/rs6000/rs6000-string.c (expand_block_compare)
9316 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9317 of SLOW_UNALIGNED_ACCESS.
9318 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9319 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9320 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9321 of SLOW_UNALIGNED_ACCESS.
9322 * expmed.c (simple_mem_bitfield_p): Likewise.
9323 * expr.c (alignment_for_piecewise_move): Likewise.
9324 (emit_group_load_1): Likewise.
9325 (emit_group_store): Likewise.
9326 (copy_blkmode_from_reg): Likewise.
9327 (emit_push_insn): Likewise.
9328 (expand_assignment): Likewise.
9329 (store_field): Likewise.
9330 (expand_expr_real_1): Likewise.
9331 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9332 * lra-constraints.c (simplify_operand_subreg): Likewise.
9333 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9334 * gimple-ssa-store-merging.c: Likewise in block comment at start
9336 * tree-ssa-strlen.c: Include target.h.
9337 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9338 of SLOW_UNALIGNED_ACCESS.
9339 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9341 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9343 PR rtl-optimization/82185
9344 * expmed.c (emit_store_flag_int): Only test tem if it has been
9347 2017-09-12 Richard Biener <rguenther@suse.de>
9350 * match.pd ((FTYPE) N CMP CST): Fix typo.
9352 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
9354 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
9356 (mips_near_type_p): Add 'short_call' attribute as a synonym
9358 * doc/extend.texi (short_call): Document new function attribute.
9360 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9363 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9364 assertion check that in the condition.
9365 (get_atomic_generic_size): Likewise. Before testing if parameter
9366 has pointer type, if it has array type, call for C++
9367 default_conversion to perform array-to-pointer conversion.
9369 2017-09-12 Richard Biener <rguenther@suse.de>
9371 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
9372 for operations we cannot scalarize.
9374 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
9376 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
9377 vectors heap vectors. Clean up comments.
9378 Make visited_bbs a reference.
9379 (profitable_jump_thread_path): Make GC
9380 vectors heap vectors. Clean up comments.
9382 (convert_and_register_jump_thread_path): Make GC vectors heap
9384 (check_subpath_and_update_thread_path): Same. Clean up comments.
9385 Make visited_bbs a reference.
9386 (handle_phi): Abstract common code to to
9387 register_jump_thread_path_if_profitable.
9388 Rename VAR_BB to DEF_BB.
9390 Make GC vectors heap vectors.
9391 Make visited_bbs a reference.
9392 (handle_assignment): Same.
9393 (register_jump_thread_path_if_profitable): New.
9394 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
9396 Make GC vectors heap vectors. Clean up comments.
9397 Make visited_bbs a reference.
9398 (find_jump_threads_backwards): Make visited_bbs live in the stack.
9399 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
9402 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
9405 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
9406 words of E_DImode object are reachable by xtensa_uimm8x4 access.
9408 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
9410 Revert r251800 and r251799.
9412 2017-09-11 Martin Jambor <mjambor@suse.cz>
9415 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
9416 arguments in advance.
9417 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
9418 use it to find predecessor edges.
9419 (naive_outof_ssa): Collect vector of predecessors.
9421 2017-09-08 Jason Merrill <jason@redhat.com>
9423 PR c++/70029 - ICE with ref-qualifier and -flto
9424 * langhooks.h (struct lang_hooks_for_types): Add
9425 copy_lang_qualifiers.
9426 * attribs.c (build_type_attribute_qual_variant): Use it.
9427 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
9429 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
9430 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
9432 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
9435 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
9436 (*mulsi3_sp64): New instruction.
9437 (mulsi3): New expander.
9439 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
9441 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
9443 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9445 * sancov.c: Include memmodel.h.
9447 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
9450 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
9453 2017-09-07 Carl Love <cel@us.ibm.com>
9455 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
9456 the sldi instruction.
9458 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
9460 * sancov.c: Include tm_p.h.
9462 2017-09-07 Jakub Jelinek <jakub@redhat.com>
9465 * output.h (switch_to_other_text_partition): New declaration.
9466 * varasm.c (switch_to_other_text_partition): New function.
9467 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
9468 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
9469 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
9470 to the other text partition before emitting LCL label and switch back
9471 after emitting the word after it.
9473 2017-09-07 Richard Biener <rguenther@suse.de>
9475 * passes.def (pass_split_crit_edges): Remove instance before PRE.
9476 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
9477 critical edges here, after loop init.
9478 (pass_data_pre): Remove PROP_no_crit_edges flags.
9479 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
9480 for valueization of call args to avoid leaking VN_TOP.
9481 (visit_use): Assert we do not visit default defs.
9482 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
9483 Use error_mark_node to more easily detect leaking VN_TOP.
9484 All default-defs are varying, not VN_TOP. Mark them visited.
9485 (run_scc_vn): Make code match comment.
9487 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9489 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
9490 OPTION_MASK_FLOAT128_KEYWORD.
9491 (POWERPC_MASKS): Likewise.
9492 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
9493 support for the -mfloat128-type option, and make -mfloat128
9494 default on PowerPC Linux systems. Define or undefine
9495 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
9496 Define __float128 to be __ieee128 if IEEE 128-bit support is
9497 enabled, or undefine it.
9498 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
9499 Delete defining __FLOAT128_TYPE__.
9500 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
9501 -mfloat128-type option and make -mfloat128 default on PowerPC
9503 (TARGET_FLOAT128_TYPE): Likewise.
9504 (-mfloat128-type): Likewise.
9505 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9506 Delete the -mfloat128-type option and make -mfloat128 default on
9507 PowerPC Linux systems. Always use __ieee128 to be the keyword for
9508 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
9509 128-bit floating point is enabled. Change tests from using
9510 -mfloat128-type to -mfloat128.
9511 (rs6000_mangle_type): Use the correct mangling for the __float128
9512 type even if normal long double is restricted to 64-bits.
9513 (floatn_mode): Enable the _Float128 type by default on VSX Linux
9515 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
9516 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
9517 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
9519 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
9520 documentation for -mfloat128.
9522 2017-09-06 Olivier Hainque <hainque@adacore.com>
9524 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
9526 2017-09-06 Wish Wu <wishwu007@gmail.com>
9527 Jakub Jelinek <jakub@redhat.com>
9529 * asan.c (initialize_sanitizer_builtins): Add
9530 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
9531 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
9532 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
9533 BT_FN_VOID_UINT64_PTR variables.
9534 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
9535 (BT_FN_VOID_UINT16_UINT16): Likewise.
9536 (BT_FN_VOID_UINT32_UINT32): Likewise.
9537 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
9538 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
9539 (BT_FN_VOID_UINT64_PTR): Likewise.
9540 * common.opt (flag_sanitize_coverage): New variable.
9541 (fsanitize-coverage=trace-pc): Remove.
9542 (fsanitize-coverage=): Add.
9543 * flag-types.h (enum sanitize_coverage_code): New enum.
9544 * fold-const.c (fold_range_test): Disable non-short-circuit
9545 optimization if flag_sanitize_coverage.
9546 (fold_truth_andor): Likewise.
9547 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
9548 * opts.c (COVERAGE_SANITIZER_OPT): Define.
9549 (coverage_sanitizer_opts): New array.
9550 (get_closest_sanitizer_option): Add OPTS argument, handle also
9551 OPT_fsanitize_coverage_.
9552 (parse_sanitizer_options): Adjusted to also handle
9553 OPT_fsanitize_coverage_.
9554 (common_handle_option): Add OPT_fsanitize_coverage_.
9555 * sancov.c (instrument_comparison, instrument_switch): New function.
9556 (sancov_pass): Add trace-cmp support.
9557 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
9558 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
9559 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
9560 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
9561 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
9562 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
9563 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
9564 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
9565 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
9566 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
9568 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9570 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9571 error. Only quit immediately if parsing is complete.
9572 (BEGIN): Initialize fatal_err and parse_done.
9573 (begin fpu, end fpu): Check number of arguments.
9574 (begin arch, end arch): Likewise.
9575 (begin cpu, end cpu): Likewise.
9576 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9577 (optalias): Likewise.
9579 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9581 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9582 * config/arm/arm-isa.h: Delete. Move definitions to ...
9583 * arm-cpus.in: ... here. Use new feature and fgroup values.
9584 * config/arm/arm.c (arm_option_override): Use lower case for feature
9586 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9587 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9588 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9590 (print_isa_bits_for): New function.
9591 (gen_isa): New function.
9592 (gen_comm_data): Use print_isa_bits_for.
9593 (define feature): New keyword.
9594 (define fgroup): New keyword.
9595 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
9596 (arm-isa.h): Add rule to generate file.
9597 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9598 case for feature bit names.
9600 2017-09-06 Richard Biener <rguenther@suse.de>
9602 * tree-ssa-pre.c (NECESSARY): Remove.
9603 (create_expression_by_pieces): Do not touch pass-local flags.
9604 (insert_into_preds_of_block): Likewise.
9605 (do_pre_regular_insertion): Likewise.
9606 (eliminate_insert): Likewise.
9607 (eliminate_dom_walker::before_dom_children): Likewise.
9608 (fini_eliminate): Do not look at inserted_exprs.
9609 (mark_operand_necessary): Remove.
9610 (remove_dead_inserted_code): Replace with simple work-list
9611 algorithm based on inserted_exprs and SSA uses.
9612 (pass_pre::execute): Re-order fini_eliminate and
9613 remove_dead_inserted_code.
9615 2017-09-06 Olivier Hainque <hainque@adacore.com>
9617 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
9618 for VxWorks 7. Adjust surrounding comments.
9620 2017-09-06 Richard Biener <rguenther@suse.de>
9622 * gimple-ssa-strength-reduction.c
9623 (find_candidates_dom_walker::before_dom_children): Also allow
9626 2017-09-06 Richard Biener <rguenther@suse.de>
9628 PR tree-optimization/82108
9629 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
9630 for gap in the non-permutation SLP case.
9632 2017-09-06 Martin Jambor <mjambor@suse.cz>
9634 PR tree-optimization/82078
9635 * tree-sra.c (sort_and_splice_var_accesses): Move call to
9636 add_access_to_work_queue...
9637 (build_accesses_from_assign): ...here.
9638 (propagate_all_subaccesses): Make sure racc is the group
9639 representative, if there is one.
9641 2017-09-06 Jakub Jelinek <jakub@redhat.com>
9644 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
9647 2017-09-06 Richard Biener <rguenther@suse.de>
9649 * gimple-ssa-strength-reduction.c
9650 (find_candidates_dom_walker::before_doom_children): Use a
9651 type and not a mode check.
9653 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
9656 * config/arm/predicates.md (arm_general_adddi_operand): Create new
9658 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
9660 2017-09-05 Jeff Law <law@redhat.com>
9662 PR tree-optimization/64910
9663 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
9664 cases where we have 3 or more operands.
9666 2017-09-05 Jakub Jelinek <jakub@redhat.com>
9669 * omp-low.c (lower_omp_for): Recompute tree invariant if
9670 gimple_omp_for_initial/final is ADDR_EXPR.
9673 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
9674 into gimple val before gimplification fo the COND_EXPR.
9676 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
9678 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
9679 REGION_COPY argument.
9680 (thread_through_all_blocks): Remove unused argument to
9681 duplicate_thread_path.
9683 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9684 Alan Hayward <alan.hayward@arm.com>
9685 David Sherwood <david.sherwood@arm.com>
9687 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
9688 Take a scalar_mode rather than a machine_mode.
9689 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9690 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
9691 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9692 (aarch64_gen_adjusted_ldpstp): Likewise.
9693 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
9695 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9696 Alan Hayward <alan.hayward@arm.com>
9697 David Sherwood <david.sherwood@arm.com>
9699 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
9700 Take a scalar_int_mode instead of a machine_mode.
9701 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9702 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9703 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9704 (aarch64_simd_attr_length_rglist): Delete.
9705 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
9706 a scalar_int_mode instead of a machine_mode.
9707 (aarch64_add_offset): Likewise.
9708 (aarch64_internal_mov_immediate): Likewise
9709 (aarch64_add_constant_internal): Likewise.
9710 (aarch64_add_constant): Likewise.
9711 (aarch64_movw_imm): Likewise.
9712 (aarch64_rtx_arith_op_extract_p): Likewise.
9713 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9714 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9715 Remove assert that the mode isn't a vector.
9716 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9717 (aarch64_expand_mov_immediate): Update calls after above changes.
9718 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
9719 (aarch64_and_bitmask_imm): Check for scalar integer modes.
9720 (aarch64_move_imm): Likewise.
9721 (aarch64_can_const_movi_rtx_p): Likewise.
9722 (aarch64_strip_extend): Likewise.
9723 (aarch64_extr_rtx_p): Likewise.
9724 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
9725 a CONST_INT when the mode parameter is VOIDmode.
9726 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
9728 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9730 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
9731 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9732 (bitwise_type_for_mode): Update accordingly.
9734 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9736 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
9737 * stor-layout.c (mode_for_size_tree): Likewise.
9738 (mode_for_array): Update accordingly.
9739 (layout_decl): Likewise.
9740 (compute_record_mode): Likewise. Only set the mode once.
9742 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9744 * target.def (get_mask_mode): Change return type to opt_mode.
9746 * doc/tm.texi: Regenerate.
9747 * targhooks.h (default_get_mask_mode): Return an opt_mode.
9748 * targhooks.c (default_get_mask_mode): Likewise.
9749 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
9750 * optabs-query.c (can_vec_mask_load_store_p): Update use of
9751 targetm.get_mask_mode.
9752 * tree.c (build_truth_vector_type): Likewise.
9754 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9756 * machmode.h (mode_for_vector): Return an opt_mode.
9757 * stor-layout.c (mode_for_vector): Likewise.
9758 (mode_for_int_vector): Update accordingly.
9759 (layout_type): Likewise.
9760 * config/i386/i386.c (emit_memmov): Likewise.
9761 (ix86_expand_set_or_movmem): Likewise.
9762 (ix86_expand_vector_init): Likewise.
9763 (ix86_get_mask_mode): Likewise.
9764 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
9766 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
9767 * expmed.c (extract_bit_field_1): Likewise.
9768 * expr.c (expand_expr_real_2): Likewise.
9769 * optabs-query.c (can_vec_perm_p): Likewise.
9770 (can_vec_mask_load_store_p): Likewise.
9771 * optabs.c (expand_vec_perm): Likewise.
9772 * targhooks.c (default_get_mask_mode): Likewise.
9773 * tree-vect-stmts.c (vectorizable_store): Likewise.
9774 (vectorizable_load): Likewise.
9775 (get_vectype_for_scalar_type_and_size): Likewise.
9777 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9779 * machmode.h (mode_for_int_vector): New function.
9780 * stor-layout.c (mode_for_int_vector): Likewise.
9781 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
9782 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
9783 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
9784 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
9785 (s390_expand_vcond): Likewise.
9787 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9789 * machmode.h (opt_machine_mode): New type.
9790 (opt_mode<T>): Allow construction from anything that can be
9792 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
9793 (mode_for_size): Return an opt_machine_mode.
9794 * stor-layout.c (mode_for_size): Likewise.
9795 (mode_for_size_tree): Update call accordingly.
9796 (bitwise_mode_for_mode): Likewise.
9797 (make_fract_type): Likewise.
9798 (make_accum_type): Likewise.
9799 * caller-save.c (replace_reg_with_saved_mem): Update call
9801 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9802 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9803 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9804 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9805 * expmed.c (extract_bit_field_1): Likewise.
9806 * reload.c (get_secondary_mem): Likewise.
9807 * varasm.c (assemble_integer): Likewise.
9808 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
9811 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9813 * machmode.h (decimal_float_mode_for_size): New function.
9814 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
9815 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
9816 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
9817 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
9818 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
9819 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
9821 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9823 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
9824 (get_builtin_sync_mode): Likewise.
9825 (expand_ifn_atomic_compare_exchange): Likewise.
9826 (expand_builtin_atomic_clear): Likewise.
9827 (expand_builtin_atomic_test_and_set): Likewise.
9828 (fold_builtin_atomic_always_lock_free): Likewise.
9829 * calls.c (compute_argument_addresses): Likewise.
9830 (emit_library_call_value_1): Likewise.
9831 (store_one_arg): Likewise.
9832 * combine.c (combine_instructions): Likewise.
9833 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
9834 * config/arm/arm.c (arm_function_value): Likewise.
9835 (aapcs_allocate_return_reg): Likewise.
9836 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
9837 * config/i386/i386.c (construct_container): Likewise.
9838 (ix86_gimplify_va_arg): Likewise.
9839 (ix86_expand_sse_cmp): Likewise.
9840 (emit_memmov): Likewise.
9841 (emit_memset): Likewise.
9842 (expand_small_movmem_or_setmem): Likewise.
9843 (ix86_expand_pextr): Likewise.
9844 (ix86_expand_pinsr): Likewise.
9845 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
9846 * config/microblaze/microblaze.c (microblaze_block_move_straight):
9848 * config/mips/mips.c (mips_function_value_1) Likewise.
9849 (mips_block_move_straight): Likewise.
9850 (mips_expand_ins_as_unaligned_store): Likewise.
9851 * config/powerpcspe/powerpcspe.c
9852 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9853 (rs6000_darwin64_record_arg_flush): Likewise.
9854 * config/rs6000/rs6000.c
9855 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9856 (rs6000_darwin64_record_arg_flush): Likewise.
9857 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
9858 (sparc_function_value_1): Likewise.
9859 * config/spu/spu.c (adjust_operand): Likewise.
9860 (spu_emit_branch_or_set): Likewise.
9861 (arith_immediate_p): Likewise.
9862 * emit-rtl.c (gen_lowpart_common): Likewise.
9863 * expr.c (expand_expr_real_1): Likewise.
9864 * function.c (assign_parm_setup_block): Likewise.
9865 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9866 * reload1.c (alter_reg): Likewise.
9867 * stor-layout.c (mode_for_vector): Likewise.
9868 (layout_type): Likewise.
9870 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9872 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
9873 (spu_convert_move): Likewise.
9874 * lower-subreg.c (resolve_simple_move): Likewise.
9876 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9879 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
9880 define_insn to a define_expand.
9881 (altivec_vsum2sws_direct): New define_insn.
9882 (altivec_vsumsws): Convert from a define_insn to a define_expand.
9884 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9886 * config/arm/arm.c (arm_option_params_internal): Improve setting of
9889 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
9894 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
9895 realign stack if stack alignment needed is less than incoming
9898 2017-09-05 Marek Polacek <polacek@redhat.com>
9901 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
9904 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9906 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
9908 2017-09-05 Richard Biener <rguenther@suse.de>
9910 PR tree-optimization/82084
9911 * fold-const.c (can_native_encode_string_p): Handle wide characters.
9913 2017-09-05 Richard Biener <rguenther@suse.de>
9915 PR tree-optimization/82102
9916 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
9918 2017-09-05 Martin Liska <mliska@suse.cz>
9920 PR tree-optimization/82032
9921 * tree-cfg.c (generate_range_test): New function.
9922 * tree-cfg.h (generate_range_test): Declared here.
9923 * tree-cfgcleanup.c (convert_single_case_switch): New function.
9924 (cleanup_control_expr_graph): Use it.
9925 * tree-switch-conversion.c (try_switch_expansion): Remove
9927 (emit_case_nodes): Use generate_range_test.
9929 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9932 * config/i386/i386.md (*<btsc><mode>_mask): Add
9933 TARGET_USE_BT to insn constraint.
9934 (*btr<mode>_mask): Ditto.
9936 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
9938 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
9939 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
9941 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
9944 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
9945 TARGET_NEON and TARGET_IWMMXT.
9946 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
9947 TARGET_NEON and TARGET_IWMMXT.
9948 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
9950 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9952 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
9953 (ix86_rewrite_tls_address): Ditto.
9954 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
9955 (ix86_rewrite_tls_address_1): Ditto.
9956 (ix86_rewrite_tls_address): Ditto.
9957 * config/i386/predicates.md (tls_address_pattern): New predicate.
9958 * config/i386/i386.md (TLS address splitter): New splitter.
9960 2017-09-04 Richard Biener <rguenther@suse.de>
9962 PR tree-optimization/82084
9963 * fold-const.h (can_native_encode_string_p): Declare.
9964 * fold-const.c (can_native_encode_string_p): Factor out from ...
9965 (native_encode_string): ... here.
9966 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
9967 vectorizing stores from constants we later cannot handle.
9969 2017-09-04 Marek Polacek <polacek@redhat.com>
9972 * doc/invoke.texi: Update -Wtautological-compare documentation.
9974 2017-09-04 Jeff Law <law@redhat.com>
9976 PR tree-optimization/64910
9977 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
9978 swap the first and last operand if the last is a constant.
9980 2017-09-04 Marek Polacek <polacek@redhat.com>
9983 * convert.c (do_narrow): When sanitizing signed integer overflows,
9984 bail out for signed types.
9985 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
9987 2017-09-04 Richard Biener <rguenther@suse.de>
9989 PR tree-optimization/82060
9990 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9991 Move devirtualization after stmt folding and before EH/AB/noreturn
9992 cleanup to get the stmt refs canonicalized. Use a bool instead
9993 of gimple_modified_p since that doesn't work for NOPs. Schedule
9994 NOPs generated by folding for removal.
9996 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9997 Alan Hayward <alan.hayward@arm.com>
9998 David Sherwood <david.sherwood@arm.com>
10000 * coretypes.h (pad_direction): New enum.
10001 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10002 (FUNCTION_ARG_PADDING): Likewise.
10003 * target.def (function_arg_padding): New hook.
10004 * targhooks.h (default_function_arg_padding): Declare.
10005 * targhooks.c (default_function_arg_padding): New function.
10006 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10007 (TARGET_FUNCTION_ARG_PADDING): ...this.
10008 * doc/tm.texi: Regenerate.
10009 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10010 instead of direction.
10011 (compute_argument_addresses): Likewise.
10012 (load_register_parameters): Likewise.
10013 (emit_library_call_value_1): Likewise.
10014 (store_one_arg): Use targetm.calls.function_arg_padding instead
10015 of FUNCTION_ARG_PADDING.
10016 (must_pass_in_stack_var_size_or_pad): Likewise.
10017 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10018 (emit_group_store): Likewise.
10019 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10020 instead of FUNCTION_ARG_PADDING.
10021 (emit_push_insn): Likewise, and propagate enum change throughout
10023 * function.h (direction): Delete.
10024 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10026 * function.c (assign_parm_find_stack_rtl): Likewise.
10027 (assign_parm_setup_block_p): Likewise.
10028 (assign_parm_setup_block): Likewise.
10029 (gimplify_parameters): Likewise.
10030 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10031 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10033 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10034 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10035 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10036 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10037 (aarch64_function_arg_padding): ...this new function.
10038 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10039 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10040 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10041 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10042 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10043 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10044 (arm_pad_arg_upward): Replace with...
10045 (arm_function_arg_padding): ...this new function.
10046 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10048 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10049 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10050 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10051 (ia64_hpux_function_arg_padding): Replace with...
10052 (ia64_function_arg_padding): ...this new function. Use pad_direction
10053 instead of direction. Check for TARGET_HPUX.
10054 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10055 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10056 (iq2000_function_arg_padding): New function.
10057 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10058 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10059 (mips_function_arg_padding): ...this new function.
10060 (mips_pad_reg_upward): Update accordingly.
10061 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10062 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10063 targetm.calls.function_arg_padding.
10064 (FUNCTION_ARG_PADDING): Delete.
10065 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10066 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10067 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10068 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10069 (nios2_block_reg_padding): Return pad_direction instead of direction.
10070 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10071 instead of direction.
10072 (nios2_function_arg_padding): Likewise. Make static.
10073 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10074 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10075 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10076 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10077 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10078 (pa_function_arg_padding): Make static. Return pad_direction instead
10080 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10081 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10082 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10083 instead of direction. Use targetm.calls.function_arg_padding.
10084 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10085 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10086 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10087 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10088 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10090 (function_arg_padding): Rename to...
10091 (rs6000_function_arg_padding): ...this. Make static. Return
10092 pad_direction instead of direction.
10093 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10094 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10095 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10096 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10097 instead of direction. Use targetm.calls.function_arg_padding.
10098 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10099 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10100 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10101 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10102 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10103 (function_arg_padding): Rename to...
10104 (rs6000_function_arg_padding): ...this. Make static. Return
10105 pad_direction instead of direction.
10106 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10107 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10108 * config/s390/s390.c (s390_function_arg_padding): New function.
10109 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10110 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10111 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10112 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10113 (function_arg_padding): Rename to...
10114 (sparc_function_arg_padding): ...this. Make static. Return
10115 pad_direction instead of direction.
10116 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10117 * config/spu/spu.c (spu_function_arg_padding): New function.
10118 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10119 * system.h (FUNCTION_ARG_PADDING): Poison.
10121 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10122 Alan Hayward <alan.hayward@arm.com>
10123 David Sherwood <david.sherwood@arm.com>
10125 * target.def (modes_tieable_p): New hook.
10126 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10127 (TARGET_MODES_TIEABLE_P): ...this.
10128 * doc/tm.texi.in: Regenerate.
10129 * hooks.h (hook_bool_mode_mode_true): Declare.
10130 * hooks.c (hook_bool_mode_mode_true): New function.
10131 * combine.c (subst): Use targetm.modes_tieable_p instead of
10133 * dse.c (find_shift_sequence): Likewise.
10134 * expmed.c (extract_low_bits): Likewise.
10135 * lower-subreg.c: Include target.h.
10136 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10138 * rtlanal.c (rtx_cost): Likewise.
10139 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10140 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10141 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10142 (TARGET_MODES_TIEABLE_P): Redefine.
10143 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10144 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10145 (TARGET_MODES_TIEABLE_P): Redefine.
10146 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10147 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10148 (arc_modes_tieable_p): New function.
10149 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10150 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10151 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10152 (arm_modes_tieable_p): Make static.
10153 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10154 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10155 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10156 (TARGET_MODES_TIEABLE_P): Redefine.
10157 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10158 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10159 (TARGET_MODES_TIEABLE_P): Redefine.
10160 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10161 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10162 (cr16_modes_tieable_p): New function.
10163 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10164 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10165 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10166 (TRULY_NOOP_TRUNCATION): Update comment.
10167 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10168 (TRULY_NOOP_TRUNCATION): Update comment.
10169 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10170 (frv_modes_tieable_p): New function.
10171 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10172 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10173 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10174 (TARGET_MODES_TIEABLE_P): Redefine.
10175 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10176 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10177 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10178 (TARGET_MODES_TIEABLE_P): Redefine.
10179 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10180 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10181 (ia64_modes_tieable_p): New function.
10182 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10183 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10184 (iq2000_modes_tieable_p): New function.
10185 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10186 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10187 (lm32_modes_tieable_p): New function.
10188 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10189 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10190 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10191 (TARGET_MODES_TIEABLE_P): Redefine.
10192 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10193 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10194 (m32r_modes_tieable_p): New function.
10195 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10196 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10197 (m68k_modes_tieable_p): New function.
10198 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10199 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10200 (mcore_modes_tieable_p): New function.
10201 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10202 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10204 (TARGET_MODES_TIEABLE_P): Redefine.
10205 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10206 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10207 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10208 (TARGET_MODES_TIEABLE_P): Redefine.
10209 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10210 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10211 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10212 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10213 (mn10300_modes_tieable_p): ...this and make static.
10214 (TARGET_MODES_TIEABLE_P): Redefine.
10215 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10216 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10217 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10218 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10219 (msp430_modes_tieable_p): Make static.
10220 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10221 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10222 (TARGET_MODES_TIEABLE_P): Redefine.
10223 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10224 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10225 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10226 (TARGET_MODES_TIEABLE_P): Redefine.
10227 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10228 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10229 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10230 (TARGET_MODES_TIEABLE_P): Redefine.
10231 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10232 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10233 (pdp11_modes_tieable_p): New function.
10234 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10235 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10236 (rs6000_modes_tieable_p): New function.
10237 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10238 * config/powerpcspe/powerpcspe.md: Update comment.
10239 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10240 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10241 (TARGET_MODES_TIEABLE_P): Redefine.
10242 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10243 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10244 (rl78_modes_tieable_p): New function.
10245 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10246 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10247 (rs6000_modes_tieable_p): New function.
10248 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10249 * config/rs6000/rs6000.md: Update comment.
10250 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10251 * config/rx/rx.c (rx_modes_tieable_p): New function.
10252 (TARGET_MODES_TIEABLE_P): Redefine.
10253 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10254 * config/s390/s390.c (s390_modes_tieable_p): New function.
10255 (TARGET_MODES_TIEABLE_P): Redefine.
10256 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10257 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10258 (sh_modes_tieable_p): New function.
10259 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10260 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10261 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10262 (sparc_modes_tieable_p): Make static.
10263 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10264 * config/spu/spu.c (spu_modes_tieable_p): New function.
10265 (TARGET_MODES_TIEABLE_P): Redefine.
10266 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10267 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10268 (TARGET_MODES_TIEABLE_P): Redefine.
10269 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10270 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10271 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10272 * config/v850/v850.c (v850_modes_tieable_p): New function.
10273 (TARGET_MODES_TIEABLE_P): Redefine.
10274 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10275 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10276 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10277 (visium_modes_tieable_p): New function.
10278 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10279 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10280 (xtensa_modes_tieable_p): New function.
10281 * system.h (MODES_TIEABLE_P): Poison.
10283 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10284 Alan Hayward <alan.hayward@arm.com>
10285 David Sherwood <david.sherwood@arm.com>
10287 * target.def (hard_regno_mode_ok): New hook.
10288 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10289 (TARGET_HARD_REGNO_MODE_OK): ...this.
10290 * doc/tm.texi.in: Regenerate.
10291 * hooks.h (hook_bool_uint_mode_true): Declare.
10292 * hooks.c (hook_bool_uint_mode_true): New function.
10293 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10294 HARD_REGNO_MODE_OK.
10295 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10296 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10297 instead of HARD_REGNO_MODE_OK.
10298 * caller-save.c: Include target.h.
10299 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10300 HARD_REGNO_MODE_OK.
10301 * combine.c (can_combine_p): Likewise.
10302 (combinable_i3pat): Likewise.
10303 (can_change_dest_mode): Likewise.
10304 * expr.c (init_expr_target): Likewise.
10305 (convert_move): Likewise.
10306 (convert_modes): Likewise.
10307 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10308 (setup_prohibited_mode_move_regs): Likewise.
10309 * ira.h (target_ira): Likewise.
10310 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10311 * lra-constraints.c (process_alt_operands): Likewise.
10312 (split_reg): Likewise.
10313 * recog.c (peep2_find_free_register): Likewise.
10314 * ree.c (combine_reaching_defs): Likewise.
10315 * regcprop.c (maybe_mode_change): Likewise.
10316 * reginfo.c (init_reg_sets_1): Likewise.
10317 (choose_hard_reg_mode): Likewise.
10318 (simplifiable_subregs): Likewise.
10319 * regrename.c (check_new_reg_p): Likewise.
10320 * reload.c (find_valid_class): Likewise.
10321 (find_valid_class_1): Likewise.
10322 (reload_inner_reg_of_subreg): Likewise.
10323 (push_reload): Likewise.
10324 (combine_reloads): Likewise.
10325 (find_dummy_reload): Likewise.
10326 (find_reloads): Likewise.
10327 * reload1.c (find_reg): Likewise.
10328 (set_reload_reg): Likewise.
10329 (allocate_reload_reg): Likewise.
10330 (choose_reload_regs): Likewise.
10331 (reload_adjust_reg_for_temp): Likewise.
10332 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10333 (simplify_subreg_regno): Likewise.
10334 * sel-sched.c (init_regs_for_mode): Likewise.
10335 * varasm.c (make_decl_rtl): Likewise.
10336 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10337 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10338 HARD_REGNO_MODE_OK.
10339 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10340 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10341 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10342 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
10343 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
10344 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10345 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
10346 (arc_mode_class): Delete.
10347 (HARD_REGNO_MODE_OK): Delete.
10348 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10349 (arc_hard_regno_mode_ok): Rename old array to...
10350 (arc_hard_regno_mode_ok_modes): ...this.
10351 (arc_conditional_register_usage): Update accordingly.
10352 (arc_mode_class): Make static.
10353 (arc_hard_regno_mode_ok): New function.
10354 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
10355 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
10356 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10357 (arm_hard_regno_mode_ok): Make static.
10358 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
10359 HARD_REGNO_MODE_OK.
10360 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
10361 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
10362 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10364 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10365 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10366 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10367 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10368 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10369 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10370 * config/bfin/predicates.md (valid_reg_operand): Use
10371 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
10372 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
10373 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
10374 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10375 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
10376 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
10377 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10378 (cr16_hard_regno_mode_ok): Make static and return a bool.
10379 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
10380 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10381 (cris_hard_regno_mode_ok): New function.
10382 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
10383 (epiphany_mode_class): Delete.
10384 (HARD_REGNO_MODE_OK): Delete.
10385 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
10386 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10387 (hard_regno_mode_ok): Rename to...
10388 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
10389 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
10390 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
10391 HARD_REGNO_MODE_OK.
10392 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
10393 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
10394 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10395 (frv_hard_regno_mode_ok): Make static and return a bool.
10396 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
10397 HARD_REGNO_MODE_OK.
10398 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
10399 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
10400 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
10401 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
10403 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10404 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
10405 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
10406 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
10408 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10409 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
10410 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10411 (ia64_hard_regno_mode_ok): New function.
10412 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
10413 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10414 (iq2000_hard_regno_mode_ok): New function.
10415 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
10416 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10417 (lm32_hard_regno_mode_ok): New function.
10418 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
10419 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
10420 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
10421 instead of HARD_REGNO_MODE_OK.
10422 (m32c_hard_regno_ok): Rename to...
10423 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
10424 (m32c_cannot_change_mode_class): Update accordingly.
10425 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10426 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
10427 (m32r_mode_class): Delete.
10428 (HARD_REGNO_MODE_OK): Delete.
10429 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10430 (m32r_hard_regno_mode_ok): Rename to...
10431 (m32r_hard_regno_modes): ...this.
10432 (m32r_mode_class): Make static.
10433 (m32r_hard_regno_mode_ok): New function.
10434 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
10435 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
10436 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10437 (m68k_hard_regno_mode_ok): Make static.
10438 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
10439 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10440 (mcore_hard_regno_mode_ok): New function.
10441 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
10442 (HARD_REGNO_MODE_OK): Delete.
10443 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
10445 (microblaze_hard_regno_mode_ok_p): ...this and make static.
10446 (microblaze_hard_regno_mode_ok): New function.
10447 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10448 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
10449 (mips_hard_regno_mode_ok): Delete.
10450 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
10451 (mips_hard_regno_mode_ok_p): ...this and make static.
10452 (mips_hard_regno_mode_ok_p): Rename to...
10453 (mips_hard_regno_mode_ok_uncached): ...this.
10454 (mips_hard_regno_mode_ok): New function.
10455 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
10456 of HARD_REGNO_MODE_OK.
10457 (mips_option_override): Update after above name changes.
10458 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10459 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
10460 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
10461 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
10462 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
10463 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10464 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
10465 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
10466 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
10467 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10468 (msp430_hard_regno_mode_ok): Make static and return a bool.
10469 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
10470 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
10471 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
10473 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10474 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
10475 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
10476 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
10477 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
10478 (PA_HARD_REGNO_MODE_OK): ...this
10479 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
10480 (PA_HARD_REGNO_MODE_OK): ...this.
10481 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10482 (pa_hard_regno_mode_ok): New function.
10483 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
10484 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10485 (pdp11_hard_regno_mode_ok): New function.
10486 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
10487 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
10489 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
10491 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10492 (rs6000_hard_regno_mode_ok): Rename to...
10493 (rs6000_hard_regno_mode_ok_uncached): ...this.
10494 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10495 (rs6000_hard_regno_mode_ok): New function.
10496 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
10497 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
10498 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
10499 (riscv_hard_regno_mode_ok): ...this and make static.
10500 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10501 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
10502 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
10503 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10504 (rl78_hard_regno_mode_ok): Make static and return bool.
10505 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
10506 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
10508 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
10509 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10510 (rs6000_hard_regno_mode_ok): Rename to...
10511 (rs6000_hard_regno_mode_ok_uncached): ...this.
10512 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10513 (rs6000_hard_regno_mode_ok): New function.
10514 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
10515 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
10516 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10517 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
10518 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
10519 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
10520 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10521 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
10522 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
10523 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10524 (sh_hard_regno_mode_ok): Make static.
10525 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
10526 instead of HARD_REGNO_MODE_OK.
10527 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
10528 (sparc_mode_class): Delete.
10529 (HARD_REGNO_MODE_OK): Delete.
10530 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10531 (hard_regno_mode_classes): Make static.
10532 (sparc_mode_class): Likewise.
10533 (sparc_hard_regno_mode_ok): New function.
10534 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
10535 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
10536 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
10538 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10539 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
10540 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
10541 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
10542 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
10543 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10544 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
10545 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
10546 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10547 (visium_hard_regno_mode_ok): New function.
10548 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
10549 instead of HARD_REGNO_MODE_OK.
10550 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
10551 (HARD_REGNO_MODE_OK): Delete.
10552 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
10553 (xtensa_hard_regno_mode_ok_p): ...this and make static.
10554 (xtensa_option_override): Update accordingly.
10555 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10556 (xtensa_hard_regno_mode_ok): New function.
10557 * system.h (HARD_REGNO_MODE_OK): Poison.
10559 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10560 Alan Hayward <alan.hayward@arm.com>
10561 David Sherwood <david.sherwood@arm.com>
10563 * target.def (hard_regno_call_part_clobbered): New hook.
10564 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
10565 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
10566 * doc/tm.texi: Regenerate.
10567 * hooks.h (hook_bool_uint_mode_false): Declare.
10568 * hooks.c (hook_bool_uint_mode_false): New function.
10569 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10570 * cselib.c (cselib_process_insn): Use
10571 targetm.hard_regno_call_part_clobbered instead of
10572 HARD_REGNO_CALL_PART_CLOBBERED.
10573 * ira-conflicts.c (ira_build_conflicts): Likewise.
10574 * ira-costs.c (ira_tune_allocno_costs): Likewise.
10575 * lra-constraints.c (need_for_call_save_p): Likewise.
10576 * lra-lives.c: Include target.h.
10577 (check_pseudos_live_through_calls): Use
10578 targetm.hard_regno_call_part_clobbered instead of
10579 HARD_REGNO_CALL_PART_CLOBBERED.
10580 * regcprop.c: Include target.h.
10581 (copyprop_hardreg_forward_1): Use
10582 targetm.hard_regno_call_part_clobbered instead of
10583 HARD_REGNO_CALL_PART_CLOBBERED.
10584 * reginfo.c (choose_hard_reg_mode): Likewise.
10585 * regrename.c (check_new_reg_p): Likewise.
10586 * reload.c (find_equiv_reg): Likewise.
10587 * reload1.c (emit_reload_insns): Likewise.
10588 * sched-deps.c (deps_analyze_insn): Likewise.
10589 * sel-sched.c (init_regs_for_mode): Likewise.
10590 (mark_unavailable_hard_regs): Likewise.
10591 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
10592 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10593 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
10595 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10596 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10597 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
10599 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
10601 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10602 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10603 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
10605 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10606 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10607 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
10609 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10610 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
10612 * config/powerpcspe/powerpcspe.c
10613 (rs6000_hard_regno_call_part_clobbered): New function.
10614 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10615 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10616 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
10618 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10619 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10620 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
10622 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10623 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10624 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
10626 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10627 Alan Hayward <alan.hayward@arm.com>
10628 David Sherwood <david.sherwood@arm.com>
10630 * rtl.h (subreg_memory_offset): Declare.
10631 * emit-rtl.c (subreg_memory_offset): New function.
10632 * expmed.c (store_bit_field_1): Use it.
10633 * expr.c (undefined_operand_subword_p): Likewise.
10634 * simplify-rtx.c (simplify_subreg): Likewise.
10636 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
10638 PR rtl-optimization/57448
10641 * optabs.c (expand_atomic_load): Place compiler memory barriers if
10642 using atomic_load pattern.
10643 (expand_atomic_store): Likewise.
10645 2017-09-04 Jakub Jelinek <jakub@redhat.com>
10648 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
10649 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
10650 handling. Use replace_call_with_value with NULL instead of
10651 gsi_replace, unlink_stmt_vdef and release_defs.
10653 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
10656 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
10658 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10661 * rtl.h (emit_library_call_value_1): Declare.
10662 (emit_library_call): Replace declaration with a series of overloads.
10663 Remove the parameter count argument.
10664 (emit_library_call_value): Likewise.
10665 * calls.c (emit_library_call_value_1): Make global. Replace varargs
10666 with an "rtx_mode_t *".
10667 (emit_library_call_value): Delete.
10668 (emit_library_call): Likewise.
10669 * asan.c (asan_emit_stack_protection): Update calls accordingly.
10670 (asan_emit_allocas_unpoison): Likewise.
10671 * builtins.c (expand_builtin_powi): Likewise.
10672 (expand_asan_emit_allocas_unpoison): Likewise.
10673 * cfgexpand.c (expand_main_function): Likewise.
10674 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
10675 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
10676 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
10677 * config/arm/arm.c (arm_trampoline_init): Likewise.
10678 (arm_call_tls_get_addr): Likewise.
10679 (arm_expand_divmod_libfunc): Likewise.
10680 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
10681 (smulsi3_highpart): Likewise.
10682 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
10683 (c6x_expand_compare): Likewise.
10684 (c6x_expand_movmem): Likewise.
10685 * config/frv/frv.c (frv_trampoline_init): Likewise.
10686 * config/i386/i386.c (ix86_trampoline_init): Likewise.
10687 (ix86_expand_divmod_libfunc): Likewise.
10688 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
10689 (ia64_expand_compare): Likewise.
10690 (ia64_profile_hook): Likewise.
10691 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
10692 (nonlocal_goto): Likewise.
10693 (restore_stack_nonlocal): Likewise.
10694 * config/m32r/m32r.c (block_move_call): Likewise.
10695 (m32r_trampoline_init): Likewise.
10696 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
10697 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
10698 (m68k_call_m68k_read_tp): Likewise.
10699 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
10700 (microblaze_expand_divide): Likewise.
10701 * config/mips/mips.h (mips_args): Likewise.
10702 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
10703 (MIPS_ICACHE_SYNC): Likewise.
10704 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
10705 (nios2_trampoline_init): Likewise.
10706 * config/pa/pa.c (hppa_tls_call): Likewise.
10707 (pa_trampoline_init): Likewise.
10708 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
10709 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
10710 (expand_strn_compare): Likewise.
10711 (rs6000_generate_compare): Likewise.
10712 (rs6000_expand_float128_convert): Likewise.
10713 (output_profile_hook): Likewise.
10714 (rs6000_trampoline_init): Likewise.
10715 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
10716 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
10717 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
10718 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
10719 (rs6000_generate_compare): Likewise.
10720 (rs6000_expand_float128_convert): Likewise.
10721 (output_profile_hook): Likewise.
10722 (rs6000_trampoline_init): Likewise.
10723 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
10724 * config/sh/sh.c (sh_trampoline_init): Likewise.
10725 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
10726 (sparc_emit_float_lib_cmp): Likewise.
10727 (sparc32_initialize_trampoline): Likewise.
10728 (sparc64_initialize_trampoline): Likewise.
10729 (sparc_profile_hook): Likewise.
10730 * config/spu/spu.c (ea_load_store): Likewise.
10731 * config/spu/spu.md (floatunssidf2): Likewise.
10732 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
10733 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
10734 * config/visium/visium.c (expand_block_move_4): Likewise.
10735 (expand_block_move_2): Likewise.
10736 (expand_block_move_1): Likewise.
10737 (expand_block_set_4): Likewise.
10738 (expand_block_set_2): Likewise.
10739 (expand_block_set_1): Likewise.
10740 (visium_trampoline_init): Likewise.
10741 (visium_profile_hook): Likewise.
10742 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
10743 (xtensa_setup_frame_addresses): Likewise.
10744 (xtensa_trampoline_init): Likewise.
10745 * except.c (sjlj_emit_function_enter): Likewise.
10746 (sjlj_emit_function_exit): Likewise.
10747 * explow.c (allocate_dynamic_stack_space): Likewise.
10748 (probe_stack_range): Likewise.
10749 * expr.c (convert_mode_scalar): Likewise.
10750 * optabs.c (expand_binop): Likewise.
10751 (expand_twoval_binop_libfunc): Likewise.
10752 (expand_unop): Likewise.
10753 (prepare_cmp_insn): Likewise.
10754 (prepare_float_lib_cmp): Likewise.
10755 (expand_float): Likewise.
10756 (expand_fix): Likewise.
10757 (expand_fixed_convert): Likewise.
10758 (maybe_emit_sync_lock_test_and_set): Likewise.
10759 (expand_atomic_compare_and_swap): Likewise.
10760 (expand_mem_thread_fence): Likewise.
10761 (expand_atomic_fetch_op): Likewise.
10763 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
10765 * doc/generic.texi (OpenACC): Adjust URL.
10766 * doc/invoke.texi (C Dialect Options): Ditto.
10768 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
10770 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
10771 predicate for operand 1. Add (m,<S>) constraint.
10772 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
10773 Prevent memory operand 1 with register operand 2.
10775 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
10777 PR rtl-optimization/82024
10778 * combine.c (try_combine): If the combination result is a PARALLEL,
10779 and we only need to retain the SET in there that would be placed
10780 at I2, check that we can place that at I3 instead, before doing so.
10782 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10785 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
10787 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
10788 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
10791 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
10792 Jeff Law <law@redhat.com>
10794 * varasm.c (bss_initializer_p): Do not put constants into .bss
10795 (categorize_decl_for_section): Handle bss_initializer_p returning
10796 false when DECL_INITIAL is NULL.
10798 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10801 * config/s390/s390.c (s390_can_inline_p): New function.
10803 2017-09-01 Jeff Law <law@redhat.com>
10805 PR tree-optimization/82052
10806 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
10807 Always initialize the returned slot after a hash table miss
10808 when INSERT is true.
10810 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
10812 * config/s390/s390.md (mem_signal_fence): Remove.
10813 * doc/md.texi (mem_signal_fence): Remove.
10814 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
10816 * target-insns.def (mem_signal_fence): Remove.
10818 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10821 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
10824 * asan.c (create_odr_indicator): Strip name encoding from assembler
10825 name before appending it after __odr_asan_.
10827 2017-09-01 Martin Liska <mliska@suse.cz>
10829 PR tree-optimization/82059
10830 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
10831 frequency only when an edge is redirected.
10833 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10835 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
10836 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
10837 (arc_conditional_register_usage): Remove ARC600 lp_count
10839 (arc_file_start): Emit Tag_ARC_CPU_variation.
10840 (arc_can_use_doloop_p): New conditions to use ZOLs.
10841 (hwloop_fail): New function.
10842 (hwloop_optimize): Likewise.
10843 (hwloop_pattern_reg): Likewise.
10844 (arc_doloop_hooks): New struct, to be used with reorg_loops.
10845 (arc_reorg_loops): New function, calls reorg_loops.
10846 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
10847 (arc600_corereg_hazard): Remove ZOL checking, case handled by
10849 (arc_loop_hazard): Remove function, functionality moved into
10851 (arc_hazard): Remove arc_loop_hazard call.
10852 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
10853 into hwloop_optimize.
10854 (arc_label_align): Remove ZOL handling.
10855 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
10856 * config/arc/arc.md (doloop_begin): Remove pattern.
10857 (doloop_begin_i): Likewise.
10858 (doloop_end_i): Likewise.
10859 (doloop_fallback): Likewise.
10860 (doloop_fallback_m): Likewise.
10861 (doloop_end): Reimplement expand.
10862 (arc_lp): New pattern for LP instruction.
10863 (loop_end): New pattern.
10864 (loop_fail): Likewise.
10865 (decrement_and_branch_until_zero): Likewise.
10866 * config/arc/arc.opt (mlpc-width): New option.
10867 * doc/invoke.texi (mlpc-width): Document option.
10869 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10871 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
10872 (arc_ccfsm_advance): Fix checking for delay slots.
10873 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
10875 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10877 * config/arc/arc.md (movqi_insn): Add stores to save constant long
10879 (movhi_insn): Update store instruction constraint which are saving
10880 6-bit short immediates.
10881 (movsi_insn): Consider also short scaled load operations.
10882 (zero_extendhisi2_i): Use Usd constraint instead of T.
10883 (extendhisi2_i): Add q constraint.
10884 (arc_clzsi2): Add type and length attributes.
10885 (arc_ctzsi2): Likewise.
10886 * config/arc/constraints.md (Usc): Update constraint, the
10887 assembler can parse two relocations for a single instruction.
10889 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10891 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
10892 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
10894 2017-08-31 Olivier Hainque <hainque@adacore.com>
10896 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
10897 match as powerpc-wrs-vxworks*.
10899 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
10901 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
10902 register constraint for by-element operand.
10903 (aarch64_mls_elt_merge<mode>): Likewise.
10905 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10907 * config/arc/arc.c (arc_can_follow_jump): Check for short
10910 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10912 * config.gcc: Use g.opt for arc.
10913 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
10914 functionality moved to ...
10915 (legitimate_scaled_address_p): New function, ...here.
10916 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
10917 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
10918 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
10920 (arc_override_options): Handle G option.
10921 (arc_output_pic_addr_const): Correct function definition.
10922 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
10923 (arc_decl_anon_ns_mem_p): Delete.
10924 (arc_in_small_data_p): Overhaul this function to take into
10925 consideration the value given via G option.
10926 (arc_rewrite_small_data_1): Renamed and corrected old
10927 arc_rewrite_small_data function.
10928 (arc_rewrite_small_data): New function.
10929 (small_data_pattern): Don't use pic_offset_table_rtx.
10930 * config/arc/arc.h (CC1_SPEC): Recognize G option.
10931 * config/arc/simdext.md (movmisalignv2hi): Use
10932 prepare_move_operands function.
10934 (movmisalign*): Likewise.
10935 * doc/invoke.texi (ARC options): Document -G option.
10937 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10939 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
10941 * config/arc/arc.c (arc_print_operand): Output scalled address for
10942 sdata whenever is possible.
10943 (arc_in_small_data_p): Allow sdata for 64bit datum when double
10944 load/stores are available.
10945 (compact_sda_memory_operand): Check for the alignment required by
10946 code density instructions.
10947 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
10949 * config/arc/constraints.md (Usd): Update constraint.
10950 (Us0): New constraint.
10951 (Usc): Update constraint.
10953 2017-08-31 Richard Biener <rguenther@suse.de>
10955 PR middle-end/82054
10956 * dwarf2out.c (dwarf2out_early_global_decl): Process each
10957 function only once.
10959 2017-08-31 Tamar Christina <tamar.christina@arm.com>
10961 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
10962 Resize type_signature.
10964 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10965 Alan Hayward <alan.hayward@arm.com>
10966 David Sherwood <david.sherwood@arm.com>
10968 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
10969 subregs whose inner modes can be stored in GPRs.
10970 (aarch64_classify_index): Likewise.
10972 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10973 Alan Hayward <alan.hayward@arm.com>
10974 David Sherwood <david.sherwood@arm.com>
10976 * config/aarch64/iterators.md (V_cmp_result): Rename to...
10977 (V_INT_EQUIV): ...this.
10978 (v_cmp_result): Rename to...
10979 (v_int_equiv): ...this.
10980 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
10981 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
10982 (copysign<mode>3): Likewise.
10983 (aarch64_simd_bsl<mode>_internal): Likewise.
10984 (aarch64_simd_bsl<mode>): Likewise.
10985 (vec_cmp<mode><mode>): Likewise.
10986 (vcond<mode><mode>): Likewise.
10987 (vcond<v_cmp_mixed><mode>): Likewise.
10988 (vcondu<mode><v_cmp_mixed>): Likewise.
10989 (aarch64_cm<optab><mode>): Likewise.
10990 (aarch64_cmtst<mode>): Likewise.
10991 (aarch64_fac<optab><mode>): Likewise.
10992 (vec_perm_const<mode>): Likewise.
10993 (vcond_mask_<mode><v_cmp_result>): Rename to...
10994 (vcond_mask_<mode><v_int_equiv>): ...this.
10995 (vec_cmp<mode><v_cmp_result>): Rename to...
10996 (vec_cmp<mode><v_int_equiv>): ...this.
10998 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10999 Alan Hayward <alan.hayward@arm.com>
11000 David Sherwood <david.sherwood@arm.com>
11002 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11004 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11005 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11006 (UNSPEC_LD4_DREG): New unspecs.
11007 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11008 (aarch64_ld2<mode>_dreg_be): Replace with...
11009 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11011 (aarch64_ld3<mode>_dreg_le)
11012 (aarch64_ld3<mode>_dreg_be): Replace with...
11013 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11015 (aarch64_ld4<mode>_dreg_le)
11016 (aarch64_ld4<mode>_dreg_be): Replace with...
11017 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11020 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11022 PR tree-optimization/81987
11023 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11024 insert an initializer in a location not dominated by the stride
11027 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11029 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11030 on the entire header of the finally block in the fallthru case.
11032 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11034 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11036 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11038 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11039 rs6000_emit_move_from_cr and call renamed function.
11040 (rs6000_emit_prologue): Call renamed functions.
11041 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11042 movesi_from_cr, remove volatile CRs.
11044 2017-08-30 Jon Beniston <jon@beniston.com>
11045 Richard Biener <rguenther@suse.de>
11047 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11048 of VECTOR_MODE_P check.
11049 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11050 element vector types.
11052 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11054 * df.h (df_read_modify_subreg_p): Remove in favor of...
11055 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11056 const_rtx instead of an rtx.
11057 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11058 * df-problems.c (df_word_lr_mark_ref): Likewise.
11059 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11060 (mark_pseudo_reg_dead): Likewise.
11061 (mark_ref_dead): Likewise.
11062 * reginfo.c (init_subregs_of_mode): Likewise.
11063 * sched-deps.c (sched_analyze_1): Likewise.
11064 * df-scan.c (df_def_record_1): Likewise.
11065 (df_uses_record): Likewise.
11066 (df_read_modify_subreg_p): Remove in favor of...
11067 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11068 const_rtx instead of an rtx.
11070 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11071 Alan Hayward <alan.hayward@arm.com>
11072 David Sherwood <david.sherwood@arm.com>
11074 * rtl.h (partial_subreg_p): New function.
11075 * caller-save.c (save_call_clobbered_regs): Use it.
11076 * calls.c (expand_call): Likewise.
11077 * combine.c (combinable_i3pat): Likewise.
11078 (simplify_set): Likewise.
11079 (make_extraction): Likewise.
11080 (make_compound_operation_int): Likewise.
11081 (gen_lowpart_or_truncate): Likewise.
11082 (force_to_mode): Likewise.
11083 (make_field_assignment): Likewise.
11084 (reg_truncated_to_mode): Likewise.
11085 (record_truncated_value): Likewise.
11086 (move_deaths): Likewise.
11087 * cse.c (record_jump_cond): Likewise.
11088 (cse_insn): Likewise.
11089 * cselib.c (cselib_lookup_1): Likewise.
11090 * expmed.c (extract_bit_field_using_extv): Likewise.
11091 * function.c (assign_parm_setup_reg): Likewise.
11092 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11093 * ira-build.c (create_insn_allocnos): Likewise.
11094 * lra-coalesce.c (merge_pseudos): Likewise.
11095 * lra-constraints.c (match_reload): Likewise.
11096 (simplify_operand_subreg): Likewise.
11097 (curr_insn_transform): Likewise.
11098 * lra-lives.c (process_bb_lives): Likewise.
11099 * lra.c (new_insn_reg): Likewise.
11100 (lra_substitute_pseudo): Likewise.
11101 * regcprop.c (mode_change_ok): Likewise.
11102 (maybe_mode_change): Likewise.
11103 (copyprop_hardreg_forward_1): Likewise.
11104 * reload.c (push_reload): Likewise.
11105 (find_reloads): Likewise.
11106 (find_reloads_subreg_address): Likewise.
11107 * reload1.c (alter_reg): Likewise.
11108 (eliminate_regs_1): Likewise.
11109 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11111 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11113 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11114 back to if statements, including unpack.
11116 2017-08-30 Martin Liska <mliska@suse.cz>
11118 PR inline-asm/82001
11119 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11121 (func_checker::compare_asm_inputs_outputs): ... this function.
11122 (func_checker::compare_gimple_asm): Use the function to compare
11123 also ASM constrains.
11124 * ipa-icf-gimple.h: Rename the function.
11126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11127 Alan Hayward <alan.hayward@arm.com>
11128 David Sherwood <david.sherwood@arm.com>
11130 * coretypes.h (complex_mode): New type.
11131 * gdbhooks.py (build_pretty_printer): Handle it.
11132 * machmode.h (complex_mode): New class.
11133 (complex_mode::includes_p): New function.
11134 (is_complex_int_mode): Likewise.
11135 (is_complex_float_mode): Likewise.
11136 * genmodes.c (get_mode_class): Handle complex mode classes.
11137 * function.c (expand_function_end): Use is_complex_int_mode.
11139 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11140 Alan Hayward <alan.hayward@arm.com>
11141 David Sherwood <david.sherwood@arm.com>
11143 * coretypes.h (scalar_mode_pod): New typedef.
11144 * gdbhooks.py (build_pretty_printer): Handle it.
11145 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11146 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11147 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11148 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11149 as_a <scalar_mode>.
11151 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11152 Alan Hayward <alan.hayward@arm.com>
11153 David Sherwood <david.sherwood@arm.com>
11155 * machmode.h (mode_for_vector): Take a scalar_mode instead
11157 * stor-layout.c (mode_for_vector): Likewise.
11158 * explow.c (promote_mode): Use as_a <scalar_mode>.
11159 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11161 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11162 Alan Hayward <alan.hayward@arm.com>
11163 David Sherwood <david.sherwood@arm.com>
11165 * target.def (preferred_simd_mode): Take a scalar_mode
11166 instead of a machine_mode.
11167 * targhooks.h (default_preferred_simd_mode): Likewise.
11168 * targhooks.c (default_preferred_simd_mode): Likewise.
11169 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11170 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11171 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11172 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11173 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11174 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11175 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11176 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11178 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11179 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11180 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11181 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11182 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11183 * doc/tm.texi: Regenerate.
11184 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11187 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11188 Alan Hayward <alan.hayward@arm.com>
11189 David Sherwood <david.sherwood@arm.com>
11191 * target.def (scalar_mode_supported_p): Take a scalar_mode
11192 instead of a machine_mode.
11193 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11194 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11195 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11196 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11197 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11198 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11199 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11200 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11201 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11202 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11203 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11204 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11205 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11206 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11208 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11209 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11210 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11211 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11212 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11214 * doc/tm.texi: Regenerate.
11216 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11217 Alan Hayward <alan.hayward@arm.com>
11218 David Sherwood <david.sherwood@arm.com>
11220 * coretypes.h (opt_scalar_mode): New typedef.
11221 * gdbhooks.py (build_pretty_printers): Handle it.
11222 * machmode.h (mode_iterator::get_2xwider): Add overload for
11224 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11226 * expr.c (convert_mode_scalar): Likewise.
11227 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11228 * optabs.c (expand_float): Likewise.
11229 (expand_fix): Likewise.
11230 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
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 * optabs.c (expand_float): Explicitly check for scalars before
11237 using a branching expansion.
11238 (expand_fix): Likewise.
11240 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11241 Alan Hayward <alan.hayward@arm.com>
11242 David Sherwood <david.sherwood@arm.com>
11244 * expr.c (convert_mode): Split scalar handling out into...
11245 (convert_mode_scalar): ...this new function. Treat the modes
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 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11254 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): 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 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11261 rather than a machine_mode.
11262 (fixed_from_string): Likewise.
11263 (fixed_convert): Likewise.
11264 (fixed_convert_from_int): Likewise.
11265 (fixed_convert_from_real): Likewise.
11266 (real_convert_from_fixed): Likewise.
11267 * fixed-value.c (fixed_from_double_int): Likewise.
11268 (fixed_from_string): Likewise.
11269 (fixed_convert): Likewise.
11270 (fixed_convert_from_int): Likewise.
11271 (fixed_convert_from_real): Likewise.
11272 (real_convert_from_fixed): Likewise.
11273 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
11275 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11276 Alan Hayward <alan.hayward@arm.com>
11277 David Sherwood <david.sherwood@arm.com>
11279 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11280 of separate mode class checks. Do not allow vector modes here.
11281 (immed_wide_int_const): Use as_a <scalar_mode>.
11282 * explow.c (trunc_int_for_mode): Likewise.
11283 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11284 (wi::shwi): Likewise.
11285 (wi::min_value): Likewise.
11286 (wi::max_value): Likewise.
11287 * dwarf2out.c (loc_descriptor): Likewise.
11288 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11289 for CONST_WIDE_INT.
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 * tree.h (SCALAR_TYPE_MODE): New macro.
11296 * expr.c (expand_expr_addr_expr_1): Use it.
11297 (expand_expr_real_2): Likewise.
11298 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11299 (fold_convert_const_fixed_from_int): Likewise.
11300 (fold_convert_const_fixed_from_real): Likewise.
11301 (native_encode_fixed): Likewise
11302 (native_encode_complex): Likewise
11303 (native_encode_vector): Likewise.
11304 (native_interpret_fixed): Likewise.
11305 (native_interpret_real): Likewise.
11306 (native_interpret_complex): Likewise.
11307 (native_interpret_vector): Likewise.
11308 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11309 (simd_clone_adjust_argument_types): Likewise.
11310 (simd_clone_init_simd_arrays): Likewise.
11311 (simd_clone_adjust): Likewise.
11312 * stor-layout.c (layout_type): Likewise.
11313 * tree.c (build_minus_one_cst): Likewise.
11314 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11315 * tree-inline.c (estimate_move_cost): Likewise.
11316 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11317 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11318 (vectorizable_reduction): Likewise.
11319 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11320 (vect_recog_mixed_size_cond_pattern): Likewise.
11321 (check_bool_pattern): Likewise.
11322 (adjust_bool_pattern): Likewise.
11323 (search_type_for_mask_1): Likewise.
11324 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11325 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11326 (vectorizable_load): Likewise.
11327 (vectorizable_store): Likewise.
11328 * ubsan.c (ubsan_encode_value): Likewise.
11329 * varasm.c (output_constant): Likewise.
11331 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11332 Alan Hayward <alan.hayward@arm.com>
11333 David Sherwood <david.sherwood@arm.com>
11335 * coretypes.h (scalar_mode): New class.
11336 * machmode.h (scalar_mode): Likewise.
11337 (scalar_mode::includes_p): New function.
11338 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11339 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11340 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11341 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11342 * expmed.c (store_bit_field_1): Likewise.
11343 (extract_bit_field_1): Likewise.
11344 * expr.c (write_complex_part): Likewise.
11345 (read_complex_part): Likewise.
11346 (emit_move_complex_push): Likewise.
11347 (expand_expr_real_2): Likewise.
11348 * function.c (assign_parm_setup_reg): Likewise.
11349 (assign_parms_unsplit_complex): Likewise.
11350 * optabs.c (expand_binop): Likewise.
11351 * rtlanal.c (subreg_get_info): Likewise.
11352 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11353 * varasm.c (output_constant_pool_2): Likewise.
11355 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11356 Alan Hayward <alan.hayward@arm.com>
11357 David Sherwood <david.sherwood@arm.com>
11359 * expmed.c (extract_high_half): Use scalar_int_mode and remove
11361 (expmed_mult_highpart_optab): Likewise.
11362 (expmed_mult_highpart): Likewise.
11364 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11365 Alan Hayward <alan.hayward@arm.com>
11366 David Sherwood <david.sherwood@arm.com>
11368 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11369 instead of a machine_mode.
11370 (builtin_memset_read_str): Likewise.
11371 * builtins.c (c_readstr): Likewise.
11372 (builtin_memcpy_read_str): Likewise.
11373 (builtin_strncpy_read_str): Likewise.
11374 (builtin_memset_read_str): Likewise.
11375 (builtin_memset_gen_str): Likewise.
11376 (expand_builtin_signbit): Use scalar_int_mode for local variables.
11377 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
11378 instead of a machine_mode.
11379 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
11381 (make_extraction): Likewise.
11382 (try_widen_shift_mode): Take and return scalar_int_modes instead
11384 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
11385 a scalar_int_mode instead of a machine_mode.
11386 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
11387 (avr_addr_space_pointer_mode): Likewise.
11388 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
11389 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
11390 (msp430_unwind_word_mode): Likewise.
11391 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
11392 (spu_addr_space_pointer_mode): Likewise.
11393 (spu_addr_space_address_mode): Likewise.
11394 (spu_libgcc_cmp_return_mode): Likewise.
11395 (spu_libgcc_shift_count_mode): Likewise.
11396 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
11397 (rl78_addr_space_pointer_mode): Likewise.
11398 (fl78_unwind_word_mode): Likewise.
11399 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
11401 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
11402 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
11403 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
11404 (mips_valid_pointer_mode): Likewise.
11405 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
11406 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
11407 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
11409 (ft32_addr_space_address_mode): Likewise.
11410 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
11411 scalar_int_mode instead of a machine_mode.
11412 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
11414 (m32c_addr_space_address_mode): Likewise.
11415 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
11416 (rs6000_eh_return_filter_mode): Likewise.
11417 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
11418 (rs6000_eh_return_filter_mode): Likewise.
11419 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
11420 (s390_libgcc_shift_count_mode): Likewise.
11421 (s390_unwind_word_mode): Likewise.
11422 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
11424 * target.def (mode_rep_extended): Likewise.
11425 (valid_pointer_mode): Likewise.
11426 (addr_space.valid_pointer_mode): Likewise.
11427 (eh_return_filter_mode): Return a scalar_int_mode rather than
11429 (libgcc_cmp_return_mode): Likewise.
11430 (libgcc_shift_count_mode): Likewise.
11431 (unwind_word_mode): Likewise.
11432 (addr_space.pointer_mode): Likewise.
11433 (addr_space.address_mode): Likewise.
11434 * doc/tm.texi: Regenerate.
11435 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
11437 (do_jump): Use scalar_int_mode for local variables.
11438 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
11439 rather than a machine_mode.
11440 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
11441 (scompare_loc_descriptor_wide): Likewise.
11442 (scompare_loc_descriptor_narrow): Likewise.
11443 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
11445 * except.c (sjlj_emit_dispatch_table): Likewise.
11446 (expand_builtin_eh_copy_values): Likewise.
11447 * explow.c (convert_memory_address_addr_space_1): Likewise.
11448 Take a scalar_int_mode rather than a machine_mode.
11449 (convert_memory_address_addr_space): Take a scalar_int_mode rather
11450 than a machine_mode.
11451 (memory_address_addr_space): Use scalar_int_mode for local variables.
11452 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
11453 rather than a machine_mode.
11454 * expmed.c (mask_rtx): Likewise.
11455 (init_expmed_one_conv): Likewise.
11456 (expand_mult_highpart_adjust): Likewise.
11457 (extract_high_half): Likewise.
11458 (expmed_mult_highpart_optab): Likewise.
11459 (expmed_mult_highpart): Likewise.
11460 (expand_smod_pow2): Likewise.
11461 (expand_sdiv_pow2): Likewise.
11462 (emit_store_flag_int): Likewise.
11463 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
11465 (extract_low_bits): Likewise.
11466 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
11468 * expr.c (pieces_addr::adjust): Likewise.
11469 (can_store_by_pieces): Likewise.
11470 (store_by_pieces): Likewise.
11471 (clear_by_pieces_1): Likewise.
11472 (expand_expr_addr_expr_1): Likewise.
11473 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
11474 (expand_expr_real_1): Likewise.
11475 (try_casesi): Likewise.
11476 * final.c (shorten_branches): Likewise.
11477 * fold-const.c (fold_convert_const_int_from_fixed): Change the
11478 type of "mode" to machine_mode.
11479 * internal-fn.c (expand_arith_overflow_result_store): Take a
11480 scalar_int_mode rather than a machine_mode.
11481 (expand_mul_overflow): Use scalar_int_mode for local variables.
11482 * loop-doloop.c (doloop_modify): Likewise.
11483 (doloop_optimize): Likewise.
11484 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
11485 than a machine_mode.
11486 (expand_doubleword_shift_condmove): Likewise.
11487 (expand_doubleword_shift): Likewise.
11488 (expand_doubleword_clz): Likewise.
11489 (expand_doubleword_popcount): Likewise.
11490 (expand_doubleword_parity): Likewise.
11491 (expand_absneg_bit): Use scalar_int_mode for local variables.
11492 (prepare_float_lib_cmp): Likewise.
11493 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
11494 rather than a machine_mode.
11495 (convert_memory_address_addr_space): Likewise.
11496 (get_mode_bounds): Likewise.
11497 (get_address_mode): Return a scalar_int_mode rather than a
11499 * rtlanal.c (get_address_mode): Likewise.
11500 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
11501 than a machine_mode.
11502 * targhooks.c (default_mode_rep_extended): Likewise.
11503 (default_valid_pointer_mode): Likewise.
11504 (default_addr_space_valid_pointer_mode): Likewise.
11505 (default_eh_return_filter_mode): Return a scalar_int_mode rather
11506 than a machine_mode.
11507 (default_libgcc_cmp_return_mode): Likewise.
11508 (default_libgcc_shift_count_mode): Likewise.
11509 (default_unwind_word_mode): Likewise.
11510 (default_addr_space_pointer_mode): Likewise.
11511 (default_addr_space_address_mode): Likewise.
11512 * targhooks.h (default_eh_return_filter_mode): Likewise.
11513 (default_libgcc_cmp_return_mode): Likewise.
11514 (default_libgcc_shift_count_mode): Likewise.
11515 (default_unwind_word_mode): Likewise.
11516 (default_addr_space_pointer_mode): Likewise.
11517 (default_addr_space_address_mode): Likewise.
11518 (default_mode_rep_extended): Take a scalar_int_mode rather than
11520 (default_valid_pointer_mode): Likewise.
11521 (default_addr_space_valid_pointer_mode): Likewise.
11522 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
11524 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
11525 rather than a machine_mode.
11526 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
11527 for local variables.
11528 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11529 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
11530 than a machine_mode.
11532 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11533 Alan Hayward <alan.hayward@arm.com>
11534 David Sherwood <david.sherwood@arm.com>
11536 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
11537 the mode argument to scalar_int_mode.
11538 (do_jump_by_parts_zero_rtx): Likewise.
11539 (do_jump_by_parts_equality_rtx): Likewise.
11540 (do_jump_by_parts_greater): Take a mode argument.
11541 (do_jump_by_parts_equality): Likewise.
11542 (do_jump_1): Update calls accordingly.
11544 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11545 Alan Hayward <alan.hayward@arm.com>
11546 David Sherwood <david.sherwood@arm.com>
11548 * is-a.h (safe_dyn_cast): New function.
11549 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
11550 (jump_table_for_label): Likewise.
11551 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
11552 instead of an rtx_insn *.
11553 (shorten_branches): Use dyn_cast instead of LABEL_P and
11554 JUMP_TABLE_DATA_P. Use jump_table_for_label and
11555 rtx_jump_table_data::get_data_mode.
11556 (final_scan_insn): Likewise.
11558 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11559 Alan Hayward <alan.hayward@arm.com>
11560 David Sherwood <david.sherwood@arm.com>
11562 * combine.c (try_combine): Use is_a <scalar_int_mode> when
11563 trying to combine a full-register integer set with a subreg
11566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11567 Alan Hayward <alan.hayward@arm.com>
11568 David Sherwood <david.sherwood@arm.com>
11570 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
11571 that is always either address_mode or pointer_mode.
11573 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11574 Alan Hayward <alan.hayward@arm.com>
11575 David Sherwood <david.sherwood@arm.com>
11577 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
11578 when the two are known to be equal.
11580 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11581 Alan Hayward <alan.hayward@arm.com>
11582 David Sherwood <david.sherwood@arm.com>
11584 * simplify-rtx.c (simplify_const_unary_operation): Use
11585 is_a <scalar_int_mode> instead of checking for a nonzero
11586 precision. Forcibly convert op_mode to a scalar_int_mode
11587 in that case. More clearly differentiate the operand and
11588 result modes and use the former when deciding what the value
11589 of a count-bits operation should be. Use is_int_mode instead
11590 of checking for a MODE_INT. Remove redundant check for whether
11591 this mode has a zero precision.
11593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11594 Alan Hayward <alan.hayward@arm.com>
11595 David Sherwood <david.sherwood@arm.com>
11597 * optabs.c (widen_leading): Change the type of the mode argument
11598 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
11599 (widen_bswap): Likewise.
11600 (expand_parity): Likewise.
11601 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
11602 (expand_ffs): Likewise.
11603 (epand_unop): Check for scalar integer modes before calling the
11606 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11607 Alan Hayward <alan.hayward@arm.com>
11608 David Sherwood <david.sherwood@arm.com>
11610 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
11612 (expand_expr_real_1): Update call accordingly.
11614 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11615 Alan Hayward <alan.hayward@arm.com>
11616 David Sherwood <david.sherwood@arm.com>
11618 * expmed.c (store_bit_field_using_insv): Add op0_mode and
11619 value_mode arguments. Use scalar_int_mode internally.
11620 (store_bit_field_1): Rename the new integer mode from imode
11621 to op0_mode and use it instead of GET_MODE (op0). Update calls
11622 to store_split_bit_field, store_bit_field_using_insv and
11623 store_fixed_bit_field.
11624 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
11625 Use scalar_int_mode internally. Use a bit count rather than a mode
11626 when calculating the largest bit size for get_best_mode.
11627 Update calls to store_split_bit_field and store_fixed_bit_field_1.
11628 (store_fixed_bit_field_1): Add mode and value_mode arguments.
11629 Remove assertion that OP0 has a scalar integer mode.
11630 (store_split_bit_field): Add op0_mode and value_mode arguments.
11631 Update calls to extract_fixed_bit_field.
11632 (extract_bit_field_using_extv): Add an op0_mode argument.
11633 Use scalar_int_mode internally.
11634 (extract_bit_field_1): Rename the new integer mode from imode to
11635 op0_mode and use it instead of GET_MODE (op0). Update calls to
11636 extract_split_bit_field, extract_bit_field_using_extv and
11637 extract_fixed_bit_field.
11638 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
11639 to extract_split_bit_field and extract_fixed_bit_field_1.
11640 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
11641 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
11642 on the target mode.
11643 (extract_split_bit_field): Add an op0_mode argument. Update call
11644 to extract_fixed_bit_field.
11646 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11647 Alan Hayward <alan.hayward@arm.com>
11648 David Sherwood <david.sherwood@arm.com>
11650 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
11651 * explow.c (hard_function_value): Likewise.
11652 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
11653 convert_to_mode call outside the loop.
11654 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
11655 for the mode iterator. Require the mode specified by max_pieces
11657 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
11659 (copy_blkmode_to_reg): Likewise.
11660 (set_storage_via_setmem): Likewise.
11661 * optabs.c (prepare_cmp_insn): Likewise.
11662 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11663 * stor-layout.c (finish_bitfield_representative): Likewise.
11665 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11666 Alan Hayward <alan.hayward@arm.com>
11667 David Sherwood <david.sherwood@arm.com>
11669 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
11670 * expr.c (convert_move): Use them.
11671 (convert_modes): Likewise.
11672 (store_expr_with_bounds): Likewise.
11674 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11675 Alan Hayward <alan.hayward@arm.com>
11676 David Sherwood <david.sherwood@arm.com>
11678 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
11679 parameter for the mode of "x". Remove the "known_x", "known_mode"
11680 and "known_ret" arguments. Change the type of the mode argument
11681 to scalar_int_mode.
11682 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
11683 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
11684 (reg_num_sign_bit_copies_for_combine): Likewise.
11685 * rtlanal.c (nonzero_bits1): Likewise.
11686 (num_sign_bit_copies1): Likewise.
11687 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
11688 (reg_num_sign_bit_copies_general): Likewise.
11689 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
11690 (reg_nonzero_bits_general): Likewise.
11692 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11693 Alan Hayward <alan.hayward@arm.com>
11694 David Sherwood <david.sherwood@arm.com>
11696 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
11697 than in subroutines. Return 1 for non-integer modes.
11698 (cached_num_sign_bit_copies): Change the type of the mode parameter
11699 to scalar_int_mode.
11700 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
11701 classes. Handle CONST_INT_P first and then check whether X also
11702 has a scalar integer mode. Check the same thing for inner registers
11703 of a SUBREG and for values that are being extended or truncated.
11705 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11706 Alan Hayward <alan.hayward@arm.com>
11707 David Sherwood <david.sherwood@arm.com>
11709 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
11710 in subroutines. Return the mode mask for non-integer modes.
11711 (cached_nonzero_bits): Change the type of the mode parameter
11712 to scalar_int_mode.
11713 (nonzero_bits1): Likewise. Remove early exit for other mode
11714 classes. Handle CONST_INT_P first and then check whether X
11715 also has a scalar integer mode.
11717 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11718 Alan Hayward <alan.hayward@arm.com>
11719 David Sherwood <david.sherwood@arm.com>
11721 * expr.c (widest_int_mode_for_size): Make the comment match the code.
11722 Return a scalar_int_mode and assert that the size is greater than
11724 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
11725 (op_by_pieces_d::op_by_pieces_d): Likewise.
11726 (op_by_pieces_d::run): Likewise.
11727 (can_store_by_pieces): Likewise.
11729 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11730 Alan Hayward <alan.hayward@arm.com>
11731 David Sherwood <david.sherwood@arm.com>
11733 * combine.c (extract_left_shift): Add a mode argument and update
11735 (make_compound_operation_int): Change the type of the mode parameter
11736 to scalar_int_mode and update the call to extract_left_shift.
11738 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11739 Alan Hayward <alan.hayward@arm.com>
11740 David Sherwood <david.sherwood@arm.com>
11742 * combine.c (simplify_and_const_int): Change the type of the mode
11743 parameter to scalar_int_mode.
11744 (simplify_and_const_int_1): Likewise. Update recursive call.
11746 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11747 Alan Hayward <alan.hayward@arm.com>
11748 David Sherwood <david.sherwood@arm.com>
11750 * combine.c (simplify_compare_const): Check that the mode is a
11751 scalar_int_mode (rather than VOIDmode) before testing its
11753 (simplify_comparison): Move COMPARISON_P handling out of the
11754 loop and restrict the latter part of the loop to scalar_int_modes.
11755 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
11756 and when considering SUBREG_REGs. Use is_int_mode instead of
11757 checking GET_MODE_CLASS against MODE_INT.
11759 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11760 Alan Hayward <alan.hayward@arm.com>
11761 David Sherwood <david.sherwood@arm.com>
11763 * combine.c (try_widen_shift_mode): Move check for equal modes to...
11764 (simplify_shift_const_1): ...here. Use scalar_int_mode for
11765 shift_unit_mode and for modes involved in scalar shifts.
11767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11768 Alan Hayward <alan.hayward@arm.com>
11769 David Sherwood <david.sherwood@arm.com>
11771 * combine.c (force_int_to_mode): New function, split out from...
11772 (force_to_mode): ...here. Keep xmode up-to-date and use it
11773 instead of GET_MODE (x).
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 * optabs-query.h (extraction_insn::struct_mode): Change type to
11780 opt_scalar_int_mode and update comment.
11781 (extraction_insn::field_mode): Change type to scalar_int_mode.
11782 (extraction_insn::pos_mode): Likewise.
11783 * combine.c (make_extraction): Update accordingly.
11784 * optabs-query.c (get_traditional_extraction_insn): Likewise.
11785 (get_optab_extraction_insn): Likewise.
11786 * recog.c (simplify_while_replacing): Likewise.
11787 * expmed.c (narrow_bit_field_mem): Change the type of the mode
11788 parameter to opt_scalar_int_mode.
11790 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11791 Alan Hayward <alan.hayward@arm.com>
11792 David Sherwood <david.sherwood@arm.com>
11794 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
11795 to a scalar_int_mode instead of a machine_mode.
11796 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
11797 (get_best_mode): Return a boolean and use a pointer argument to store
11798 the selected mode. Replace the limit mode parameter with a bit limit.
11799 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
11800 for the values returned by bit_field_mode_iterator::next_mode.
11801 (store_bit_field): Update call to get_best_mode.
11802 (store_fixed_bit_field): Likewise.
11803 (extract_fixed_bit_field): Likewise.
11804 * expr.c (optimize_bitfield_assignment_op): Likewise.
11805 * fold-const.c (optimize_bit_field_compare): Likewise.
11806 (fold_truth_andor_1): Likewise.
11807 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
11808 Update for new type of m_mode.
11809 (get_best_mode): As above.
11811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11812 Alan Hayward <alan.hayward@arm.com>
11813 David Sherwood <david.sherwood@arm.com>
11815 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
11816 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
11817 (store_bit_field): Check is_a <scalar_int_mode> before calling
11818 strict_volatile_bitfield_p.
11819 (extract_bit_field): Likewise.
11821 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11822 Alan Hayward <alan.hayward@arm.com>
11823 David Sherwood <david.sherwood@arm.com>
11825 * target.def (cstore_mode): Return a scalar_int_mode.
11826 * doc/tm.texi: Regenerate.
11827 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
11828 * targhooks.h (default_cstore_mode): Likewise.
11829 * targhooks.c (default_cstore_mode): Likewise, using a forced
11831 * expmed.c (emit_cstore): Expect the target of the cstore to be
11834 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11835 Alan Hayward <alan.hayward@arm.com>
11836 David Sherwood <david.sherwood@arm.com>
11838 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
11840 (niter_desc): Likewise mode.
11841 (iv_analyze): Add a mode parameter.
11843 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
11844 and change its type to scalar_int_mode.
11845 * loop-iv.c: Update commentary at head of file.
11846 (iv_constant): Pass the mode paraeter before the rtx it describes
11847 and change its type to scalar_int_mode. Remove VOIDmode handling.
11848 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
11849 (iv_extend): Likewise.
11850 (shorten_into_mode): Likewise.
11851 (iv_add): Use scalar_int_mode.
11852 (iv_mult): Likewise.
11853 (iv_shift): Likewise.
11854 (canonicalize_iv_subregs): Likewise.
11855 (get_biv_step_1): Pass the outer_mode parameter before the rtx
11856 it describes and change its mode to scalar_int_mode. Also change
11857 the type of the returned inner_mode to scalar_int_mode.
11858 (get_biv_step): Likewise, turning outer_mode from a pointer
11859 into a direct parameter. Update call to get_biv_step_1.
11860 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
11861 iv_constant and get_biv_step.
11862 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
11863 and change its type to scalar_int_mode. Don't initialise iv->mode
11864 to VOIDmode and remove later checks for its still being VOIDmode.
11865 Update calls to iv_analyze_op and iv_analyze_expr. Check
11866 is_a <scalar_int_mode> when changing the mode under consideration.
11867 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
11868 Update call to iv_analyze_expr.
11869 (iv_analyze_op): Add a mode parameter. Reject subregs whose
11870 inner register is not also a scalar_int_mode. Update call to
11872 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
11873 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
11874 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
11875 separate mode class checks. Update calls to iv_analyze. Remove
11876 fix-up of VOIDmodes after iv_analyze_biv.
11877 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
11878 don't have a scalar_int_mode. Update call to biv_p.
11880 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11881 Alan Hayward <alan.hayward@arm.com>
11882 David Sherwood <david.sherwood@arm.com>
11884 * cfgexpand.c (convert_debug_memory_address): Use
11885 as_a <scalar_int_mode>.
11886 * combine.c (expand_compound_operation): Likewise.
11887 (make_extraction): Likewise.
11888 (change_zero_ext): Likewise.
11889 (simplify_comparison): Likewise.
11890 * cse.c (cse_insn): Likewise.
11891 * dwarf2out.c (minmax_loc_descriptor): Likewise.
11892 (mem_loc_descriptor): Likewise.
11893 (loc_descriptor): Likewise.
11894 * expmed.c (init_expmed_one_mode): Likewise.
11895 (synth_mult): Likewise.
11896 (emit_store_flag_1): Likewise.
11897 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
11898 of a comparison with size.
11899 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
11900 (reduce_to_bit_field_precision): Likewise.
11901 * function.c (expand_function_end): Likewise.
11902 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
11903 * loop-doloop.c (doloop_modify): Likewise.
11904 * optabs.c (expand_binop): Likewise.
11905 (expand_unop): Likewise.
11906 (expand_copysign_absneg): Likewise.
11907 (prepare_cmp_insn): Likewise.
11908 (maybe_legitimize_operand): Likewise.
11909 * recog.c (const_scalar_int_operand): Likewise.
11910 * rtlanal.c (get_address_mode): Likewise.
11911 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11912 (simplify_cond_clz_ctz): Likewise.
11913 * tree-nested.c (get_nl_goto_field): Likewise.
11914 * tree.c (build_vector_type_for_mode): Likewise.
11915 * var-tracking.c (use_narrower_mode): Likewise.
11917 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11918 Alan Hayward <alan.hayward@arm.com>
11919 David Sherwood <david.sherwood@arm.com>
11921 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
11922 * builtins.c (expand_builtin_signbit): Use it.
11923 * cfgexpand.c (expand_debug_expr): Likewise.
11924 * dojump.c (do_jump): Likewise.
11925 (do_compare_and_jump): Likewise.
11926 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
11927 * expmed.c (make_tree): Likewise.
11928 * expr.c (expand_expr_real_2): Likewise.
11929 (expand_expr_real_1): Likewise.
11930 (try_casesi): Likewise.
11931 * fold-const-call.c (fold_const_call_ss): Likewise.
11932 * fold-const.c (unextend): Likewise.
11933 (extract_muldiv_1): Likewise.
11934 (fold_single_bit_test): Likewise.
11935 (native_encode_int): Likewise.
11936 (native_encode_string): Likewise.
11937 (native_interpret_int): Likewise.
11938 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
11939 * internal-fn.c (expand_addsub_overflow): Likewise.
11940 (expand_neg_overflow): Likewise.
11941 (expand_mul_overflow): Likewise.
11942 (expand_arith_overflow): Likewise.
11943 * match.pd: Likewise.
11944 * stor-layout.c (layout_type): Likewise.
11945 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11946 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
11947 * tree-ssanames.c (get_range_info): Likewise.
11948 * tree-switch-conversion.c (array_value_type) Likewise.
11949 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
11950 (vect_recog_divmod_pattern): Likewise.
11951 (vect_recog_mixed_size_cond_pattern): Likewise.
11952 * tree-vrp.c (extract_range_basic): Likewise.
11953 (simplify_float_conversion_using_ranges): Likewise.
11954 * tree.c (int_fits_type_p): Likewise.
11955 * ubsan.c (instrument_bool_enum_load): Likewise.
11956 * varasm.c (mergeable_string_section): Likewise.
11957 (narrowing_initializer_constant_valid_p): Likewise.
11958 (output_constant): Likewise.
11960 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11961 Alan Hayward <alan.hayward@arm.com>
11962 David Sherwood <david.sherwood@arm.com>
11964 * machmode.h (NARROWEST_INT_MODE): New macro.
11965 * expr.c (alignment_for_piecewise_move): Use it instead of
11966 GET_CLASS_NARROWEST_MODE (MODE_INT).
11967 (push_block): Likewise.
11968 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
11970 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11972 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11973 Alan Hayward <alan.hayward@arm.com>
11974 David Sherwood <david.sherwood@arm.com>
11976 * postreload.c (move2add_valid_value_p): Change the type of the
11977 mode parameter to scalar_int_mode.
11978 (move2add_use_add2_insn): Add a mode parameter and use it instead
11980 (move2add_use_add3_insn): Likewise.
11981 (reload_cse_move2add): Update accordingly.
11983 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11984 Alan Hayward <alan.hayward@arm.com>
11985 David Sherwood <david.sherwood@arm.com>
11987 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
11989 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
11990 * optabs.c (expand_unop): Likewise.
11992 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11993 Alan Hayward <alan.hayward@arm.com>
11994 David Sherwood <david.sherwood@arm.com>
11996 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
11997 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
11998 (popcount_loc_descriptor): Likewise.
11999 (bswap_loc_descriptor): Likewise.
12000 (rotate_loc_descriptor): Likewise.
12001 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12002 calling the functions above.
12004 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12005 Alan Hayward <alan.hayward@arm.com>
12006 David Sherwood <david.sherwood@arm.com>
12008 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12010 (try_combine): Likewise.
12011 (simplify_if_then_else): Likewise.
12012 * cse.c (cse_insn): Likewise.
12013 * dwarf2out.c (mem_loc_descriptor): Likewise.
12014 * emit-rtl.c (gen_lowpart_common): Likewise.
12015 * simplify-rtx.c (simplify_truncation): Likewise.
12016 (simplify_binary_operation_1): Likewise.
12017 (simplify_const_relational_operation): Likewise.
12018 (simplify_ternary_operation): Likewise.
12019 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
12021 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12022 Alan Hayward <alan.hayward@arm.com>
12023 David Sherwood <david.sherwood@arm.com>
12025 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12026 * reload.c (push_reload): Likewise.
12027 (find_reloads): Likewise.
12029 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12030 Alan Hayward <alan.hayward@arm.com>
12031 David Sherwood <david.sherwood@arm.com>
12033 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12034 (make_compound_operation_int): Likewise.
12035 (change_zero_ext): Likewise.
12036 * expr.c (convert_move): Likewise.
12037 (convert_modes): Likewise.
12038 * fwprop.c (forward_propagate_subreg): Likewise.
12039 * loop-iv.c (get_biv_step_1): Likewise.
12040 * optabs.c (widen_operand): Likewise.
12041 * postreload.c (move2add_valid_value_p): Likewise.
12042 * recog.c (simplify_while_replacing): Likewise.
12043 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12044 (simplify_binary_operation_1): Likewise. Remove redundant
12045 mode equality check.
12047 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12048 Alan Hayward <alan.hayward@arm.com>
12049 David Sherwood <david.sherwood@arm.com>
12051 * combine.c (combine_simplify_rtx): Add checks for
12052 is_a <scalar_int_mode>.
12053 (simplify_if_then_else): Likewise.
12054 (make_field_assignment): Likewise.
12055 (simplify_comparison): Likewise.
12056 * ifcvt.c (noce_try_bitop): Likewise.
12057 * loop-invariant.c (canonicalize_address_mult): Likewise.
12058 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12060 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12061 Alan Hayward <alan.hayward@arm.com>
12062 David Sherwood <david.sherwood@arm.com>
12064 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12065 is_a <scalar_int_mode> instead of != BLKmode.
12067 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12068 Alan Hayward <alan.hayward@arm.com>
12069 David Sherwood <david.sherwood@arm.com>
12071 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12072 instead of != VOIDmode.
12073 * combine.c (if_then_else_cond): Likewise.
12074 (change_zero_ext): Likewise.
12075 * dwarf2out.c (mem_loc_descriptor): Likewise.
12076 (loc_descriptor): Likewise.
12077 * rtlanal.c (canonicalize_condition): Likewise.
12078 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12080 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12081 Alan Hayward <alan.hayward@arm.com>
12082 David Sherwood <david.sherwood@arm.com>
12084 * simplify-rtx.c (simplify_binary_operation_1): Use
12085 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12087 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12088 Alan Hayward <alan.hayward@arm.com>
12089 David Sherwood <david.sherwood@arm.com>
12091 * wide-int.h (int_traits<unsigned char>) New class.
12092 (int_traits<unsigned short>) Likewise.
12093 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12094 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12096 * combine.c (set_nonzero_bits_and_sign_copies): Use
12097 is_a <scalar_int_mode>.
12098 (find_split_point): Likewise.
12099 (combine_simplify_rtx): Likewise.
12100 (simplify_logical): Likewise.
12101 (expand_compound_operation): Likewise.
12102 (expand_field_assignment): Likewise.
12103 (make_compound_operation): Likewise.
12104 (extended_count): Likewise.
12105 (change_zero_ext): Likewise.
12106 (simplify_comparison): Likewise.
12107 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12108 (ucompare_loc_descriptor): Likewise.
12109 (minmax_loc_descriptor): Likewise.
12110 (mem_loc_descriptor): Likewise.
12111 (loc_descriptor): Likewise.
12112 * expmed.c (init_expmed_one_mode): Likewise.
12113 * lra-constraints.c (lra_constraint_offset): Likewise.
12114 * optabs.c (prepare_libcall_arg): Likewise.
12115 * postreload.c (move2add_note_store): Likewise.
12116 * reload.c (operands_match_p): Likewise.
12117 * rtl.h (load_extend_op): Likewise.
12118 * rtlhooks.c (gen_lowpart_general): Likewise.
12119 * simplify-rtx.c (simplify_truncation): Likewise.
12120 (simplify_unary_operation_1): Likewise.
12121 (simplify_binary_operation_1): Likewise.
12122 (simplify_const_binary_operation): Likewise.
12123 (simplify_const_relational_operation): Likewise.
12124 (simplify_subreg): Likewise.
12125 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12126 * var-tracking.c (adjust_mems): Likewise.
12127 (prepare_call_arguments): Likewise.
12129 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12130 Alan Hayward <alan.hayward@arm.com>
12131 David Sherwood <david.sherwood@arm.com>
12133 * machmode.h (is_int_mode): New fuction.
12134 * combine.c (find_split_point): Use it.
12135 (combine_simplify_rtx): Likewise.
12136 (simplify_if_then_else): Likewise.
12137 (simplify_set): Likewise.
12138 (simplify_shift_const_1): Likewise.
12139 (simplify_comparison): Likewise.
12140 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12141 * cse.c (notreg_cost): Likewise.
12142 (cse_insn): Likewise.
12143 * cselib.c (cselib_lookup_1): Likewise.
12144 * dojump.c (do_jump_1): Likewise.
12145 (do_compare_rtx_and_jump): Likewise.
12146 * dse.c (get_call_args): Likewise.
12147 * dwarf2out.c (rtl_for_decl_init): Likewise.
12148 (native_encode_initializer): Likewise.
12149 * expmed.c (emit_store_flag_1): Likewise.
12150 (emit_store_flag): Likewise.
12151 * expr.c (convert_modes): Likewise.
12152 (store_field): Likewise.
12153 (expand_expr_real_1): Likewise.
12154 * fold-const.c (fold_read_from_constant_string): Likewise.
12155 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12156 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12157 * optabs.c (expand_binop): Likewise.
12158 (expand_unop): Likewise.
12159 (expand_abs_nojump): Likewise.
12160 (expand_one_cmpl_abs_nojump): Likewise.
12161 * simplify-rtx.c (mode_signbit_p): Likewise.
12162 (val_signbit_p): Likewise.
12163 (val_signbit_known_set_p): Likewise.
12164 (val_signbit_known_clear_p): Likewise.
12165 (simplify_relational_operation_1): Likewise.
12166 * tree.c (vector_type_mode): Likewise.
12168 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12169 Alan Hayward <alan.hayward@arm.com>
12170 David Sherwood <david.sherwood@arm.com>
12172 * machmode.h (smallest_mode_for_size): Fix formatting.
12173 (smallest_int_mode_for_size): New function.
12174 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12175 instead of smallest_mode_for_size.
12176 * combine.c (make_extraction): Likewise.
12177 * config/arc/arc.c (arc_expand_movmem): Likewise.
12178 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12179 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12180 * config/s390/s390.c (s390_expand_insv): Likewise.
12181 * config/sparc/sparc.c (assign_int_registers): Likewise.
12182 * config/spu/spu.c (spu_function_value): Likewise.
12183 (spu_function_arg): Likewise.
12184 * coverage.c (get_gcov_type): Likewise.
12185 (get_gcov_unsigned_t): Likewise.
12186 * dse.c (find_shift_sequence): Likewise.
12187 * expmed.c (store_bit_field_1): Likewise.
12188 * expr.c (convert_move): Likewise.
12189 (store_field): Likewise.
12190 * internal-fn.c (expand_arith_overflow): Likewise.
12191 * optabs-query.c (get_best_extraction_insn): Likewise.
12192 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12193 * stor-layout.c (layout_type): Likewise.
12194 (initialize_sizetypes): Likewise.
12195 * targhooks.c (default_get_mask_mode): Likewise.
12196 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12198 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12199 Alan Hayward <alan.hayward@arm.com>
12200 David Sherwood <david.sherwood@arm.com>
12202 * machmode.h (opt_mode::else_blk): New function.
12203 (int_mode_for_mode): Declare.
12204 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12205 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12207 * cfgexpand.c (expand_debug_expr): Likewise.
12208 * combine.c (gen_lowpart_or_truncate): Likewise.
12209 (gen_lowpart_for_combine): Likewise.
12210 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12211 * config/avr/avr.c (avr_to_int_mode): Likewise.
12212 (avr_out_plus_1): Likewise.
12213 (avr_out_plus): Likewise.
12214 (avr_out_round): Likewise.
12215 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12216 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12217 (s390_expand_vcond): Likewise.
12218 * config/spu/spu.c (spu_split_immediate): Likewise.
12219 (spu_expand_mov): Likewise.
12220 * dse.c (get_stored_val): Likewise.
12221 * expmed.c (store_bit_field_1): Likewise.
12222 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12224 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12225 (extract_low_bits): Likewise.
12226 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12227 handling rather than repeating the check.
12228 (emit_group_store): Likewise.
12229 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12230 * optabs.c (expand_absneg_bit): Likewise.
12231 (expand_copysign_absneg): Likewise.
12232 (expand_copysign_bit): Likewise.
12233 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12234 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12235 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12236 * var-tracking.c (prepare_call_arguments): Likewise.
12237 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12238 int_mode_for_mode instead of mode_for_size.
12239 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12242 Alan Hayward <alan.hayward@arm.com>
12243 David Sherwood <david.sherwood@arm.com>
12245 * machmode.h (int_mode_for_size): New function.
12246 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12247 instead of mode_for_size.
12248 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12250 * combine.c (expand_field_assignment): Use int_mode_for_size
12251 instead of mode_for_size.
12252 (make_extraction): Likewise.
12253 (simplify_shift_const_1): Likewise.
12254 (simplify_comparison): Likewise.
12255 * dojump.c (do_jump): Likewise.
12256 * dwarf2out.c (mem_loc_descriptor): Likewise.
12257 * emit-rtl.c (init_derived_machine_modes): Likewise.
12258 * expmed.c (flip_storage_order): Likewise.
12259 (convert_extracted_bit_field): Likewise.
12260 * expr.c (copy_blkmode_from_reg): Likewise.
12261 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12262 * internal-fn.c (expand_mul_overflow): Likewise.
12263 * lower-subreg.c (simple_move): Likewise.
12264 * optabs-libfuncs.c (init_optabs): Likewise.
12265 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12266 * tree.c (vector_type_mode): Likewise.
12267 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12268 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12269 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12270 * tree-vect-stmts.c (vectorizable_load): Likewise.
12271 (vectorizable_store): Likewise.
12273 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12274 Alan Hayward <alan.hayward@arm.com>
12275 David Sherwood <david.sherwood@arm.com>
12277 * coretypes.h (pod_mode): New type.
12278 (scalar_int_mode_pod): New typedef.
12279 * machmode.h (pod_mode): New class.
12280 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12281 * genmodes.c (emit_mode_int_n): Update accordingly.
12282 * lower-subreg.h (target_lower_subreg): Change type to
12283 scalar_int_mode_pod.
12284 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12285 scalar_int_mode_pod.
12287 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12288 Alan Hayward <alan.hayward@arm.com>
12289 David Sherwood <david.sherwood@arm.com>
12291 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12292 machine_mode to scalar_int_mode.
12293 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12294 (rs6000_option_override_internal): Remove cast to int.
12295 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12296 machine_mode to scalar_int_mode.
12297 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12298 (rs6000_option_override_internal): Remove cast to int.
12299 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12300 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12302 * config/s390/s390.c (s390_expand_builtin): Likewise.
12303 * coretypes.h (scalar_int_mode): New type.
12304 (opt_scalar_int_mode): New typedef.
12305 * machmode.h (scalar_int_mode): New class.
12306 (scalar_int_mode::includes_p): New function.
12307 (byte_mode): Change type to scalar_int_mode.
12308 (word_mode): Likewise.
12309 (ptr_mode): Likewise.
12310 * emit-rtl.c (byte_mode): Likewise.
12311 (word_mode): Likewise.
12312 (ptr_mode): Likewise.
12313 (init_derived_machine_modes): Update accordingly.
12314 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12315 and MODE_PARTIAL_INT.
12316 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12317 opt_scalar_int_mode.
12319 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12320 Alan Hayward <alan.hayward@arm.com>
12321 David Sherwood <david.sherwood@arm.com>
12323 * target.def (libgcc_floating_mode_supported_p): Take a
12325 * doc/tm.texi: Regenerate.
12326 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12328 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12329 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12332 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12333 Alan Hayward <alan.hayward@arm.com>
12334 David Sherwood <david.sherwood@arm.com>
12336 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12337 * doc/tm.texi: Regenerate.
12338 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12339 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12340 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12341 * targhooks.h (default_floatn_mode): Likewise.
12342 * targhooks.c (default_floatn_mode): Likewise.
12343 * tree.c (build_common_tree_nodes): Update accordingly.
12345 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12346 Alan Hayward <alan.hayward@arm.com>
12347 David Sherwood <david.sherwood@arm.com>
12349 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
12350 (mode_iterator::iterate_p): Likewise.
12351 (mode_iterator::get_wider): Likewise.
12352 * expr.c (init_expr_target): Use opt_scalar_float_mode.
12354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12355 Alan Hayward <alan.hayward@arm.com>
12356 David Sherwood <david.sherwood@arm.com>
12358 * coretypes.h (opt_scalar_float_mode): New typedef.
12359 * machmode.h (float_mode_for_size): New function.
12360 * emit-rtl.c (double_mode): Delete.
12361 (init_emit_once): Use float_mode_for_size.
12362 * stor-layout.c (layout_type): Likewise.
12363 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
12365 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12366 Alan Hayward <alan.hayward@arm.com>
12367 David Sherwood <david.sherwood@arm.com>
12369 * output.h (assemble_real): Take a scalar_float_mode.
12370 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
12371 * config/arm/arm.md (consttable_4): Likewise.
12372 (consttable_8): Likewise.
12373 (consttable_16): Likewise.
12374 * config/mips/mips.md (consttable_float): Likewise.
12375 * config/s390/s390.c (s390_output_pool_entry): Likewise.
12376 * varasm.c (assemble_real): Take a scalar_float_mode.
12377 (output_constant_pool_2): Update accordingly.
12378 (output_constant): Likewise.
12380 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12381 Alan Hayward <alan.hayward@arm.com>
12382 David Sherwood <david.sherwood@arm.com>
12384 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
12385 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
12386 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
12387 (native_encode_real): Likewise.
12388 (native_interpret_real): Likewise.
12389 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
12390 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12392 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12393 Alan Hayward <alan.hayward@arm.com>
12394 David Sherwood <david.sherwood@arm.com>
12396 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
12397 <scalar_float_mode>. Simplify.
12398 (gen_extend_conv_libfunc): Likewise.
12400 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12401 Alan Hayward <alan.hayward@arm.com>
12402 David Sherwood <david.sherwood@arm.com>
12404 * coretypes.h (scalar_float_mode): New type.
12405 * machmode.h (mode_traits::from_int): Use machine_mode if
12406 USE_ENUM_MODES is defined.
12407 (is_a): New function.
12409 (dyn_cast): Likewise.
12410 (scalar_float_mode): New class.
12411 (scalar_float_mode::includes_p): New function.
12412 (is_float_mode): Likewise.
12413 * gdbhooks.py (MachineModePrinter): New class.
12414 (build_pretty_printer): Use it for scalar_float_mode.
12415 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
12416 (format_helper::format_helper): Turn into a template.
12417 * genmodes.c (get_mode_class): New function.
12418 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
12419 or machine_mode if none.
12420 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
12421 as_a <scalar_float_mode>.
12422 * dwarf2out.c (mem_loc_descriptor): Likewise.
12423 (insert_float): Likewise.
12424 (add_const_value_attribute): Likewise.
12425 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12426 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
12427 (expand_unop): Update accordingly.
12428 (expand_abs_nojump): Likewise.
12429 (expand_copysign_absneg): Take a scalar_float_mode.
12430 (expand_copysign_bit): Likewise.
12431 (expand_copysign): Update accordingly.
12433 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12434 Alan Hayward <alan.hayward@arm.com>
12435 David Sherwood <david.sherwood@arm.com>
12437 * coretypes.h (opt_mode): New class.
12438 * machmode.h (opt_mode): Likewise.
12439 (opt_mode::else_void): New function.
12440 (opt_mode::require): Likewise.
12441 (opt_mode::exists): Likewise.
12442 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
12443 (GET_MODE_2XWIDER_MODE): Likewise.
12444 (mode_iterator::get_wider): Update accordingly.
12445 (mode_iterator::get_2xwider): Likewise.
12446 (mode_iterator::get_known_wider): Likewise, turning into a template.
12447 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
12448 forcing a wider mode to exist.
12449 * config/cr16/cr16.h (LONG_REG_P): Likewise.
12450 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12451 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
12452 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
12453 * lower-subreg.c (init_lower_subreg): Likewise.
12454 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
12455 on the final iteration.
12456 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
12457 a wider mode exists before asking for a move pattern.
12458 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
12459 forcing a wider mode to exist.
12460 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
12461 returning false if no such mode exists.
12462 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
12463 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
12464 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
12465 Avoid checking for a MODE_INT if we already know the mode is not a
12467 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
12468 forcing a wider mode to exist.
12469 (expmed_mult_highpart_optab): Likewise.
12470 (expmed_mult_highpart): Likewise.
12471 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
12473 * lto-streamer-in.c (lto_input_mode_table): Likewise.
12474 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
12475 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12476 * internal-fn.c (expand_mul_overflow): Update use of
12477 GET_MODE_2XWIDER_MODE.
12478 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12479 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
12480 GET_MODE_WIDER_MODE.
12481 (convert_plusminus_to_widen): Likewise.
12482 * tree-switch-conversion.c (array_value_type): Likewise.
12483 * var-tracking.c (emit_note_insn_var_location): Likewise.
12484 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12485 Return false inside rather than outside the loop if no wider mode
12487 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
12488 and GET_MODE_2XWIDER_MODE
12489 (can_compare_p): Use else_void.
12490 * gdbhooks.py (OptMachineModePrinter): New class.
12491 (build_pretty_printer): Use it for opt_mode.
12493 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12494 Alan Hayward <alan.hayward@arm.com>
12495 David Sherwood <david.sherwood@arm.com>
12497 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
12498 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
12500 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12501 Alan Hayward <alan.hayward@arm.com>
12502 David Sherwood <david.sherwood@arm.com>
12504 * machmode.h (mode_traits): New structure.
12505 (get_narrowest_mode): New function.
12506 (mode_iterator::start): Likewise.
12507 (mode_iterator::iterate_p): Likewise.
12508 (mode_iterator::get_wider): Likewise.
12509 (mode_iterator::get_known_wider): Likewise.
12510 (mode_iterator::get_2xwider): Likewise.
12511 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
12512 (FOR_EACH_MODE): Likewise.
12513 (FOR_EACH_MODE_FROM): Likewise.
12514 (FOR_EACH_MODE_UNTIL): Likewise.
12515 (FOR_EACH_WIDER_MODE): Likewise.
12516 (FOR_EACH_2XWIDER_MODE): Likewise.
12517 * builtins.c (expand_builtin_strlen): Use new mode iterators.
12518 * combine.c (simplify_comparison): Likewise
12519 * config/i386/i386.c (type_natural_mode): Likewise.
12520 * cse.c (cse_insn): Likewise.
12521 * dse.c (find_shift_sequence): Likewise.
12522 * emit-rtl.c (init_derived_machine_modes): Likewise.
12523 (init_emit_once): Likewise.
12524 * explow.c (hard_function_value): Likewise.
12525 * expmed.c (extract_fixed_bit_field_1): Likewise.
12526 (extract_bit_field_1): Likewise.
12527 (expand_divmod): Likewise.
12528 (emit_store_flag_1): Likewise.
12529 * expr.c (init_expr_target): Likewise.
12530 (convert_move): Likewise.
12531 (alignment_for_piecewise_move): Likewise.
12532 (widest_int_mode_for_size): Likewise.
12533 (emit_block_move_via_movmem): Likewise.
12534 (copy_blkmode_to_reg): Likewise.
12535 (set_storage_via_setmem): Likewise.
12536 (compress_float_constant): Likewise.
12537 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12538 * optabs-query.c (get_best_extraction_insn): Likewise.
12539 * optabs.c (expand_binop): Likewise.
12540 (expand_twoval_unop): Likewise.
12541 (expand_twoval_binop): Likewise.
12542 (widen_leading): Likewise.
12543 (widen_bswap): Likewise.
12544 (expand_parity): Likewise.
12545 (expand_unop): Likewise.
12546 (prepare_cmp_insn): Likewise.
12547 (prepare_float_lib_cmp): Likewise.
12548 (expand_float): Likewise.
12549 (expand_fix): Likewise.
12550 (expand_sfix_optab): Likewise.
12551 * postreload.c (move2add_use_add2_insn): Likewise.
12552 * reg-stack.c (reg_to_stack): Likewise.
12553 * reginfo.c (choose_hard_reg_mode): Likewise.
12554 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12555 * stor-layout.c (mode_for_size): Likewise.
12556 (smallest_mode_for_size): Likewise.
12557 (mode_for_vector): Likewise.
12558 (finish_bitfield_representative): Likewise.
12559 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
12560 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
12561 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12562 * var-tracking.c (prepare_call_arguments): Likewise.
12564 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12565 Alan Hayward <alan.hayward@arm.com>
12566 David Sherwood <david.sherwood@arm.com>
12568 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
12569 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
12570 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
12571 * machmode.h (mode_size): Move earlier in file.
12572 (mode_precision): Likewise.
12573 (mode_inner): Likewise.
12574 (mode_nunits): Likewise.
12575 (mode_unit_size): Likewise.
12576 (unit_unit_precision): Likewise.
12577 (mode_wider): Likewise.
12578 (mode_2xwider): Likewise.
12579 (machine_mode): New class.
12580 (mode_to_bytes): New function.
12581 (mode_to_bits): Likewise.
12582 (mode_to_precision): Likewise.
12583 (mode_to_inner): Likewise.
12584 (mode_to_unit_size): Likewise.
12585 (mode_to_unit_precision): Likewise.
12586 (mode_to_nunits): Likewise.
12587 (GET_MODE_SIZE): Use mode_to_bytes.
12588 (GET_MODE_BITSIZE): Use mode_to_bits.
12589 (GET_MODE_PRECISION): Use mode_to_precision.
12590 (GET_MODE_INNER): Use mode_to_inner.
12591 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
12592 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
12593 (GET_MODE_NUNITS): Use mode_to_nunits.
12594 * system.h (ALWAYS_INLINE): New macro.
12595 * config/powerpcspe/powerpcspe-c.c
12596 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
12597 int for arg1_mode and arg2_mode.
12599 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12600 Alan Hayward <alan.hayward@arm.com>
12601 David Sherwood <david.sherwood@arm.com>
12603 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12604 Prefix mode names with E_ in case statements.
12605 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12606 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
12607 (aarch64_split_simd_move): Likewise.
12608 (aarch64_gen_storewb_pair): Likewise.
12609 (aarch64_gen_loadwb_pair): Likewise.
12610 (aarch64_gen_store_pair): Likewise.
12611 (aarch64_gen_load_pair): Likewise.
12612 (aarch64_get_condition_code_1): Likewise.
12613 (aarch64_constant_pool_reload_icode): Likewise.
12614 (get_rsqrte_type): Likewise.
12615 (get_rsqrts_type): Likewise.
12616 (get_recpe_type): Likewise.
12617 (get_recps_type): Likewise.
12618 (aarch64_gimplify_va_arg_expr): Likewise.
12619 (aarch64_simd_container_mode): Likewise.
12620 (aarch64_emit_load_exclusive): Likewise.
12621 (aarch64_emit_store_exclusive): Likewise.
12622 (aarch64_expand_compare_and_swap): Likewise.
12623 (aarch64_gen_atomic_cas): Likewise.
12624 (aarch64_emit_bic): Likewise.
12625 (aarch64_emit_atomic_swap): Likewise.
12626 (aarch64_emit_atomic_load_op): Likewise.
12627 (aarch64_evpc_trn): Likewise.
12628 (aarch64_evpc_uzp): Likewise.
12629 (aarch64_evpc_zip): Likewise.
12630 (aarch64_evpc_ext): Likewise.
12631 (aarch64_evpc_rev): Likewise.
12632 (aarch64_evpc_dup): Likewise.
12633 (aarch64_gen_ccmp_first): Likewise.
12634 (aarch64_gen_ccmp_next): Likewise.
12635 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12636 (alpha_emit_xfloating_libcall): Likewise.
12637 (emit_insxl): Likewise.
12638 (alpha_arg_type): Likewise.
12639 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
12640 (arc_preferred_simd_mode): Likewise.
12641 (arc_secondary_reload): Likewise.
12642 (get_arc_condition_code): Likewise.
12643 (arc_print_operand): Likewise.
12644 (arc_legitimate_constant_p): Likewise.
12645 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12646 * config/arc/arc.md (casesi_load): Likewise.
12647 (casesi_compact_jump): Likewise.
12648 * config/arc/predicates.md (proper_comparison_operator): Likewise.
12649 (cc_use_register): Likewise.
12650 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12651 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12652 (arm_init_iwmmxt_builtins): Likewise.
12653 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
12654 (neon_expand_vector_init): Likewise.
12655 (arm_attr_length_move_neon): Likewise.
12656 (maybe_get_arm_condition_code): Likewise.
12657 (arm_emit_vector_const): Likewise.
12658 (arm_preferred_simd_mode): Likewise.
12659 (arm_output_iwmmxt_tinsr): Likewise.
12660 (thumb1_output_casesi): Likewise.
12661 (thumb2_output_casesi): Likewise.
12662 (arm_emit_load_exclusive): Likewise.
12663 (arm_emit_store_exclusive): Likewise.
12664 (arm_expand_compare_and_swap): Likewise.
12665 (arm_evpc_neon_vuzp): Likewise.
12666 (arm_evpc_neon_vzip): Likewise.
12667 (arm_evpc_neon_vrev): Likewise.
12668 (arm_evpc_neon_vtrn): Likewise.
12669 (arm_evpc_neon_vext): Likewise.
12670 (arm_validize_comparison): Likewise.
12671 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
12672 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
12673 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
12674 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
12675 (c6x_preferred_simd_mode): Likewise.
12676 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
12677 (epiphany_rtx_costs): Likewise.
12678 * config/epiphany/predicates.md (proper_comparison_operator):
12680 * config/frv/frv.c (condexec_memory_operand): Likewise.
12681 (frv_emit_move): Likewise.
12682 (output_move_single): Likewise.
12683 (output_condmove_single): Likewise.
12684 (frv_hard_regno_mode_ok): Likewise.
12685 (frv_matching_accg_mode): Likewise.
12686 * config/h8300/h8300.c (split_adds_subs): Likewise.
12687 (h8300_rtx_costs): Likewise.
12688 (h8300_print_operand): Likewise.
12689 (compute_mov_length): Likewise.
12690 (output_logical_op): Likewise.
12691 (compute_logical_op_length): Likewise.
12692 (compute_logical_op_cc): Likewise.
12693 (h8300_shift_needs_scratch_p): Likewise.
12694 (output_a_shift): Likewise.
12695 (compute_a_shift_length): Likewise.
12696 (compute_a_shift_cc): Likewise.
12697 (expand_a_rotate): Likewise.
12698 (output_a_rotate): Likewise.
12699 * config/i386/i386.c (classify_argument): Likewise.
12700 (function_arg_advance_32): Likewise.
12701 (function_arg_32): Likewise.
12702 (function_arg_64): Likewise.
12703 (function_value_64): Likewise.
12704 (ix86_gimplify_va_arg): Likewise.
12705 (ix86_legitimate_constant_p): Likewise.
12706 (put_condition_code): Likewise.
12707 (split_double_mode): Likewise.
12708 (ix86_avx256_split_vector_move_misalign): Likewise.
12709 (ix86_expand_vector_logical_operator): Likewise.
12710 (ix86_split_idivmod): Likewise.
12711 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
12712 (ix86_build_const_vector): Likewise.
12713 (ix86_build_signbit_mask): Likewise.
12714 (ix86_match_ccmode): Likewise.
12715 (ix86_cc_modes_compatible): Likewise.
12716 (ix86_expand_branch): Likewise.
12717 (ix86_expand_sse_cmp): Likewise.
12718 (ix86_expand_sse_movcc): Likewise.
12719 (ix86_expand_int_sse_cmp): Likewise.
12720 (ix86_expand_vec_perm_vpermi2): Likewise.
12721 (ix86_expand_vec_perm): Likewise.
12722 (ix86_expand_sse_unpack): Likewise.
12723 (ix86_expand_int_addcc): Likewise.
12724 (ix86_split_to_parts): Likewise.
12725 (ix86_vectorize_builtin_gather): Likewise.
12726 (ix86_vectorize_builtin_scatter): Likewise.
12727 (avx_vpermilp_parallel): Likewise.
12728 (inline_memory_move_cost): Likewise.
12729 (ix86_tieable_integer_mode_p): Likewise.
12730 (x86_maybe_negate_const_int): Likewise.
12731 (ix86_expand_vector_init_duplicate): Likewise.
12732 (ix86_expand_vector_init_one_nonzero): Likewise.
12733 (ix86_expand_vector_init_one_var): Likewise.
12734 (ix86_expand_vector_init_concat): Likewise.
12735 (ix86_expand_vector_init_interleave): Likewise.
12736 (ix86_expand_vector_init_general): Likewise.
12737 (ix86_expand_vector_set): Likewise.
12738 (ix86_expand_vector_extract): Likewise.
12739 (emit_reduc_half): Likewise.
12740 (ix86_emit_i387_round): Likewise.
12741 (ix86_mangle_type): Likewise.
12742 (ix86_expand_round_sse4): Likewise.
12743 (expand_vec_perm_blend): Likewise.
12744 (canonicalize_vector_int_perm): Likewise.
12745 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
12746 (expand_vec_perm_1): Likewise.
12747 (expand_vec_perm_interleave3): Likewise.
12748 (expand_vec_perm_even_odd_pack): Likewise.
12749 (expand_vec_perm_even_odd_1): Likewise.
12750 (expand_vec_perm_broadcast_1): Likewise.
12751 (ix86_vectorize_vec_perm_const_ok): Likewise.
12752 (ix86_expand_vecop_qihi): Likewise.
12753 (ix86_expand_mul_widen_hilo): Likewise.
12754 (ix86_expand_sse2_abs): Likewise.
12755 (ix86_expand_pextr): Likewise.
12756 (ix86_expand_pinsr): Likewise.
12757 (ix86_preferred_simd_mode): Likewise.
12758 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
12759 * config/i386/sse.md (*andnot<mode>3): Likewise.
12760 (<mask_codefor><code><mode>3<mask_name>): Likewise.
12761 (*<code><mode>3): Likewise.
12762 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
12763 (ia64_expand_atomic_op): Likewise.
12764 (ia64_arg_type): Likewise.
12765 (ia64_mode_to_int): Likewise.
12766 (ia64_scalar_mode_supported_p): Likewise.
12767 (ia64_vector_mode_supported_p): Likewise.
12768 (expand_vec_perm_broadcast): Likewise.
12769 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
12770 (iq2000_function_arg_advance): Likewise.
12771 (iq2000_function_arg): Likewise.
12772 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
12773 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
12774 (m68k_libcall_value): Likewise.
12775 (m68k_function_value): Likewise.
12776 (sched_attr_op_type): Likewise.
12777 * config/mcore/mcore.c (mcore_output_move): Likewise.
12778 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
12780 (microblaze_function_arg): Likewise.
12781 * config/mips/mips.c (mips16_build_call_stub): Likewise.
12782 (mips_print_operand): Likewise.
12783 (mips_mode_ok_for_mov_fmt_p): Likewise.
12784 (mips_vector_mode_supported_p): Likewise.
12785 (mips_preferred_simd_mode): Likewise.
12786 (mips_expand_vpc_loongson_even_odd): Likewise.
12787 (mips_expand_vec_unpack): Likewise.
12788 (mips_expand_vi_broadcast): Likewise.
12789 (mips_expand_vector_init): Likewise.
12790 (mips_expand_vec_reduc): Likewise.
12791 (mips_expand_msa_cmp): Likewise.
12792 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
12793 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
12794 (cc_flags_for_mode): Likewise.
12795 * config/msp430/msp430.c (msp430_print_operand): Likewise.
12796 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
12797 (nds32_output_casesi_pc_relative): Likewise.
12798 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12799 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
12800 (nvptx_gen_unpack): Likewise.
12801 (nvptx_gen_pack): Likewise.
12802 (nvptx_gen_shuffle): Likewise.
12803 (nvptx_gen_wcast): Likewise.
12804 (nvptx_preferred_simd_mode): Likewise.
12805 * config/pa/pa.c (pa_secondary_reload): Likewise.
12806 * config/pa/predicates.md (base14_operand): Likewise.
12807 * config/powerpcspe/powerpcspe-c.c
12808 (altivec_resolve_overloaded_builtin): Likewise.
12809 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
12811 (rs6000_preferred_simd_mode): Likewise.
12812 (output_vec_const_move): Likewise.
12813 (rs6000_expand_vector_extract): Likewise.
12814 (rs6000_split_vec_extract_var): Likewise.
12815 (reg_offset_addressing_ok_p): Likewise.
12816 (rs6000_legitimate_offset_address_p): Likewise.
12817 (rs6000_legitimize_address): Likewise.
12818 (rs6000_emit_set_const): Likewise.
12819 (rs6000_const_vec): Likewise.
12820 (rs6000_emit_move): Likewise.
12821 (spe_build_register_parallel): Likewise.
12822 (rs6000_darwin64_record_arg_recurse): Likewise.
12823 (swap_selector_for_mode): Likewise.
12824 (spe_init_builtins): Likewise.
12825 (paired_init_builtins): Likewise.
12826 (altivec_init_builtins): Likewise.
12827 (do_load_for_compare): Likewise.
12828 (rs6000_generate_compare): Likewise.
12829 (rs6000_expand_float128_convert): Likewise.
12830 (emit_load_locked): Likewise.
12831 (emit_store_conditional): Likewise.
12832 (rs6000_output_function_epilogue): Likewise.
12833 (rs6000_handle_altivec_attribute): Likewise.
12834 (rs6000_function_value): Likewise.
12835 (emit_fusion_gpr_load): Likewise.
12836 (emit_fusion_p9_load): Likewise.
12837 (emit_fusion_p9_store): Likewise.
12838 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
12839 (fusion_gpr_mem_load): Likewise.
12840 (fusion_addis_mem_combo_load): Likewise.
12841 (fusion_addis_mem_combo_store): Likewise.
12842 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
12843 (fusion_gpr_mem_load): Likewise.
12844 (fusion_addis_mem_combo_load): Likewise.
12845 (fusion_addis_mem_combo_store): Likewise.
12846 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12848 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
12849 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12850 (rs6000_preferred_simd_mode): Likewise.
12851 (output_vec_const_move): Likewise.
12852 (rs6000_expand_vector_extract): Likewise.
12853 (rs6000_split_vec_extract_var): Likewise.
12854 (reg_offset_addressing_ok_p): Likewise.
12855 (rs6000_legitimate_offset_address_p): Likewise.
12856 (rs6000_legitimize_address): Likewise.
12857 (rs6000_emit_set_const): Likewise.
12858 (rs6000_const_vec): Likewise.
12859 (rs6000_emit_move): Likewise.
12860 (rs6000_darwin64_record_arg_recurse): Likewise.
12861 (swap_selector_for_mode): Likewise.
12862 (paired_init_builtins): Likewise.
12863 (altivec_init_builtins): Likewise.
12864 (rs6000_expand_float128_convert): Likewise.
12865 (emit_load_locked): Likewise.
12866 (emit_store_conditional): Likewise.
12867 (rs6000_output_function_epilogue): Likewise.
12868 (rs6000_handle_altivec_attribute): Likewise.
12869 (rs6000_function_value): Likewise.
12870 (emit_fusion_gpr_load): Likewise.
12871 (emit_fusion_p9_load): Likewise.
12872 (emit_fusion_p9_store): Likewise.
12873 * config/rx/rx.c (rx_gen_move_template): Likewise.
12874 (flags_from_mode): Likewise.
12875 * config/s390/predicates.md (s390_alc_comparison): Likewise.
12876 (s390_slb_comparison): Likewise.
12877 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
12878 (s390_vector_mode_supported_p): Likewise.
12879 (s390_cc_modes_compatible): Likewise.
12880 (s390_match_ccmode_set): Likewise.
12881 (s390_canonicalize_comparison): Likewise.
12882 (s390_emit_compare_and_swap): Likewise.
12883 (s390_branch_condition_mask): Likewise.
12884 (s390_rtx_costs): Likewise.
12885 (s390_secondary_reload): Likewise.
12886 (__SECONDARY_RELOAD_CASE): Likewise.
12887 (s390_expand_cs): Likewise.
12888 (s390_preferred_simd_mode): Likewise.
12889 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
12890 * config/sh/sh.c (sh_print_operand): Likewise.
12891 (dump_table): Likewise.
12892 (sh_secondary_reload): Likewise.
12893 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12894 * config/sh/sh.md (casesi_worker_1): Likewise.
12895 (casesi_worker_2): Likewise.
12896 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
12897 (fcc_comparison_operator): Likewise.
12898 * config/sparc/sparc.c (sparc_expand_move): Likewise.
12899 (emit_soft_tfmode_cvt): Likewise.
12900 (sparc_preferred_simd_mode): Likewise.
12901 (output_cbranch): Likewise.
12902 (sparc_print_operand): Likewise.
12903 (sparc_expand_vec_perm_bmask): Likewise.
12904 (vector_init_bshuffle): Likewise.
12905 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
12906 (spu_vector_mode_supported_p): Likewise.
12907 (spu_expand_insv): Likewise.
12908 (spu_emit_branch_or_set): Likewise.
12909 (spu_handle_vector_attribute): Likewise.
12910 (spu_builtin_splats): Likewise.
12911 (spu_builtin_extract): Likewise.
12912 (spu_builtin_promote): Likewise.
12913 (spu_expand_sign_extend): Likewise.
12914 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
12915 (tilegx_simd_int): Likewise.
12916 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
12917 (tilepro_simd_int): Likewise.
12918 * config/v850/v850.c (const_double_split): Likewise.
12919 (v850_print_operand): Likewise.
12920 (ep_memory_offset): Likewise.
12921 * config/vax/vax.c (vax_rtx_costs): Likewise.
12922 (vax_output_int_move): Likewise.
12923 (vax_output_int_add): Likewise.
12924 (vax_output_int_subtract): Likewise.
12925 * config/visium/predicates.md (visium_branch_operator): Likewise.
12926 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
12927 (visium_print_operand_address): Likewise.
12928 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12929 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
12930 (xtensa_expand_conditional_branch): Likewise.
12931 (xtensa_copy_incoming_a7): Likewise.
12932 (xtensa_output_literal): Likewise.
12933 * dfp.c (decimal_real_maxval): Likewise.
12934 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12937 Alan Hayward <alan.hayward@arm.com>
12938 David Sherwood <david.sherwood@arm.com>
12940 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
12941 (mode_nunits_inline): Likewise.
12942 (mode_inner_inline): Likewise.
12943 (mode_unit_size_inline): Likewise.
12944 (mode_unit_precision_inline): Likewise.
12945 (emit_insn_modes_h): Likewise. Also emit a #define of the
12947 (emit_mode_wider): Add an E_ prefix to mode names.
12948 (emit_mode_complex): Likewise.
12949 (emit_mode_inner): Likewise.
12950 (emit_mode_adjustments): Likewise.
12951 (emit_mode_int_n): Likewise.
12952 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
12953 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
12954 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
12955 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
12956 (CRC32_BUILTIN, ENTRY): Likewise.
12957 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
12958 (aarch64_pop_regs): Likewise.
12959 (aarch64_process_components): Likewise.
12960 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
12961 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
12962 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
12963 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
12964 * config/arm/arm.c (arm_init_libfuncs): Likewise.
12965 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
12967 * config/i386/i386-builtin.def (pcmpestr): Likewise.
12968 (pcmpistr): Likewise.
12969 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
12970 * config/mmix/mmix.c (mmix_output_condition): Likewise.
12971 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12973 * config/rl78/rl78.c (mduc_regs): Likewise.
12974 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12975 (htm_expand_builtin): Likewise.
12976 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
12977 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
12978 * config/xtensa/xtensa.c (print_operand): Likewise.
12979 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
12980 (NUM_MODE_VECTOR_INT): Likewise.
12981 * genoutput.c (null_operand): Likewise.
12982 (output_operand_data): Likewise.
12983 * genrecog.c (print_parameter_value): Likewise.
12984 * lra.c (debug_operand_data): Likewise.
12986 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12988 * dwarf2out.c (scompare_loc_descriptor_wide)
12989 (scompare_loc_descriptor_narrow): New functions, split out from...
12990 (scompare_loc_descriptor): ...here.
12991 * expmed.c (emit_store_flag_int): New function, split out from...
12992 (emit_store_flag): ...here.
12994 2017-08-30 Richard Biener <rguenther@suse.de>
12996 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
12997 (dwarf2out_early_finish): Move setting of AT_pubnames from
12998 early debug output to early finish.
13000 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13002 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13003 and -mdata-region to the assembler.
13005 2017-08-30 Richard Biener <rguenther@suse.de>
13007 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13009 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13010 (dwarf2out_early_global_decl): Only generate a DIE for the
13011 abstract origin if it doesn't already exist or is a declaration DIE.
13012 (resolve_addr): Do not add the linkage name twice when
13013 generating a stub DIE for the DW_TAG_GNU_call_site target.
13015 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13017 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13018 Use machine_mode rather than int for arg1_mode.
13020 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13023 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13024 that the second argument of the built-in functions to unpack
13025 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13026 switch statement instead a lot of if statements.
13027 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13028 Allow 64-bit values to be in Altivec registers as well as
13029 traditional floating point registers.
13030 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13032 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13034 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13035 MAX_REGS_PER_ADDRESS == 1.
13037 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13039 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13040 * config/i386/i386.c (ix86_option_override_internal): Simplify
13041 setting of opts->x_flag_entry.
13043 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13044 Jakub Jelinek <jakub@redhat.com>
13045 Richard Biener <rguenther@suse.de>
13047 PR tree-optimization/81503
13048 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13049 folded constant fits in the target type; reorder tests for clarity.
13051 2017-08-29 Martin Liska <mliska@suse.cz>
13053 * passes.def: Include pass_lower_switch.
13054 * stmt.c (dump_case_nodes): Remove and move to
13055 tree-switch-conversion.
13056 (case_values_threshold): Likewise.
13057 (expand_switch_as_decision_tree_p): Likewise.
13058 (emit_case_decision_tree): Likewise.
13059 (expand_case): Likewise.
13060 (balance_case_nodes): Likewise.
13061 (node_has_low_bound): Likewise.
13062 (node_has_high_bound): Likewise.
13063 (node_is_bounded): Likewise.
13064 (emit_case_nodes): Likewise.
13065 (struct simple_case_node): New struct.
13066 (add_case_node): Remove.
13067 (emit_case_dispatch_table): Use vector instead of case_list.
13068 (reset_out_edges_aux): Remove.
13069 (compute_cases_per_edge): Likewise.
13070 (expand_case): Build list of simple_case_node.
13071 (expand_sjlj_dispatch_table): Use it.
13072 * tree-switch-conversion.c (struct case_node): Moved from
13073 stmt.c and adjusted.
13074 (emit_case_nodes): Likewise.
13075 (node_has_low_bound): Likewise.
13076 (node_has_high_bound): Likewise.
13077 (node_is_bounded): Likewise.
13078 (case_values_threshold): Likewise.
13079 (reset_out_edges_aux): Likewise.
13080 (compute_cases_per_edge): Likewise.
13081 (add_case_node): Likewise.
13082 (dump_case_nodes): Likewise.
13083 (balance_case_nodes): Likewise.
13084 (expand_switch_as_decision_tree_p): Likewise.
13085 (emit_jump): Likewise.
13086 (emit_case_decision_tree): Likewise.
13087 (try_switch_expansion): Likewise.
13088 (do_jump_if_equal): Likewise.
13089 (emit_cmp_and_jump_insns): Likewise.
13090 (fix_phi_operands_for_edge): New function.
13091 (record_phi_operand_mapping): Likewise.
13092 (class pass_lower_switch): New pass.
13093 (pass_lower_switch::execute): New function.
13094 (make_pass_lower_switch): Likewise.
13095 (conditional_probability):
13096 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13097 * tree-pass.h: Add make_pass_lower_switch.
13099 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13102 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13105 2017-08-29 Richard Biener <rguenther@suse.de>
13107 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13108 we do not add a DW_AT_inline attribute twice.
13109 (gen_subprogram_die): Remove code setting DW_AT_inline on
13110 DECL_ABSTRACT_P nodes.
13112 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13114 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13115 calls to internal functions.
13116 (gimplify_modify_expr): Likewise.
13117 * tree-call-cdce.c (use_internal_fn): Likewise.
13118 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13119 (convert_to_divmod): Set the nothrow flag.
13120 * tree-if-conv.c (predicate_mem_writes): Likewise.
13121 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13122 (vectorizable_call): Likewise.
13123 (vectorizable_store): Likewise.
13124 (vectorizable_load): Likewise.
13125 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13126 (vect_recog_mask_conversion_pattern): Likewise.
13128 2017-08-29 Martin Liska <mliska@suse.cz>
13131 * gcc.c (driver_handle_option): Add new argument.
13132 * opts-common.c (handle_option): Pass
13133 target_option_override_hook.
13134 * opts-global.c (lang_handle_option): Add new option.
13135 (set_default_handlers): Add new argument.
13136 (decode_options): Likewise.
13137 * opts.c (target_handle_option): Likewise.
13138 (common_handle_option): Call target_option_override_hook.
13139 * opts.h (struct cl_option_handler_func): Add hook for
13140 target option override.
13141 (struct cl_option_handlers): Likewise.
13142 (set_default_handlers): Add new argument.
13143 (decode_options): Likewise.
13144 (common_handle_option): Likewise.
13145 (target_handle_option): Likewise.
13146 * toplev.c (toplev::main): Pass targetm.target_option.override
13149 2017-08-29 Richard Biener <rguenther@suse.de>
13150 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13152 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13153 life to the active subtree.
13155 2017-08-28 Jeff Law <law@redhat.com>
13157 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13158 derive_equivalences.
13159 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13160 Code moved into....
13161 (edge_info::derive_equivalences): New private member function
13163 * tree-ssa-dom.c (class edge_info): Changed from a struct
13164 to a class. Add ctor/dtor, methods and data members.
13165 (edge_info::edge_info): Renamed from allocate_edge_info.
13166 Initialize additional members.
13167 (edge_info::~edge_info): New.
13168 (free_dom_edge_info): Delete the edge info.
13169 (record_edge_info): Use new class & associated member functions.
13170 Tighten forms for testing for edge equivalences.
13171 (record_temporary_equivalences): Iterate over the simple
13172 equivalences rather than assuming there's only one per edge.
13173 (cprop_into_successor_phis): Iterate over the simple
13174 equivalences rather than assuming there's only one per edge.
13175 (optimize_stmt): Use operand_equal_p rather than pointer
13176 equality for mini-DSE code.
13178 2017-08-28 Nathan Sidwell <nathan@acm.org>
13180 * gcc.c (execute): Fold SIGPIPE handling into switch
13181 statement. Adjust internal error message.
13183 2017-08-28 Richard Biener <rguenther@suse.de>
13186 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13187 Do nothing for removed DIEs.
13189 2017-08-28 Richard Biener <rguenther@suse.de>
13191 PR tree-optimization/81977
13192 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13195 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13198 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13199 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13200 using targetm.gen_mem_thread_fence.
13202 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13205 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13206 predicate to register_operand. Reorder operands.
13207 (*btr<mode>): Ditto.
13208 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13209 (*btr<mode>_mask): Ditto.
13211 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13213 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13214 * config/rs6000/xmmintrin.h: New file.
13215 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13217 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13220 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13221 parameter and_insn and return it.
13222 (recombine_lvx_pattern): Insert a copy to ensure availability of
13223 the base register of the copied masking operation at the point of
13224 the instruction replacement.
13225 (recombine_stvx_pattern): Likewise.
13227 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13229 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13230 undocumented switches.
13231 (-mpower9-dform-vector): Likewise.
13232 (-mpower9-dform): Likewise.
13233 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13234 comments to delete references to -mpower9-dform* switches.
13235 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13236 Delete reference to -mpower9-dform* switches, test for
13237 -mpower9-vector instead.
13238 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13239 (OTHER_P9_VECTOR_MASKS): Likewise.
13240 (POWERPC_MASKS): Likewise.
13241 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13242 tests against -mpower9-dform* to -mpower9-vector. Delete code
13243 that checked for -mpower9-dform* consistancy with other options.
13244 Add test for -mpower9-misc to enable other power9 switches.
13245 (rs6000_init_hard_regno_mode_ok): Likewise.
13246 (rs6000_option_override_internal): Likewise.
13247 (rs6000_emit_prologue): Likewise.
13248 (rs6000_emit_epilogue): Likewise.
13249 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13250 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13251 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13253 (emit_fusion_p9_store): Likewise.
13254 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13255 resetting these macros if the assembler does not support ISA 3.0
13257 (TARGET_P9_DFORM_VECTOR): Likewise.
13258 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13259 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13261 2017-08-25 Alan Modra <amodra@gmail.com>
13263 PR rtl-optimization/81747
13264 * cse.c (cse_extended_basic_block): Don't attempt to record
13265 equivalences for degenerate conditional branches that branch
13266 to their fall-through.
13268 2017-08-24 Martin Sebor <msebor@redhat.com>
13270 PR middle-end/81908
13271 * gimple-fold.c (size_must_be_zero_p): New function.
13272 (gimple_fold_builtin_memory_op): Call it.
13274 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13276 * config/rs6000/mm_malloc.h: New file.
13278 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13280 PR tree-optimization/81913
13281 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13282 analysis when either IVs in condition can wrap.
13284 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13286 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13287 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13289 2017-08-24 Richard Biener <rguenther@suse.de>
13292 * targhooks.c (default_target_can_inline_p): Properly
13293 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13294 is present and always compare.
13295 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13296 infer -mfpmath=sse from TARGET_SSE_P.
13297 (ix86_can_inline_p): Properly use target_option_default_node when
13298 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13300 2017-08-24 Richard Biener <rguenther@suse.de>
13303 * dwarf2out.c (output_die): Handle flag_generate_offload like
13305 (output_comp_unit): Likewise.
13306 (gen_array_type_die): Likewise.
13307 (dwarf2out_early_finish): Likewise.
13308 (note_variable_value_in_expr): Likewise.
13309 (dwarf2out_finish): Likewise. Adjust assert.
13310 * cgraphunit.c (symbol_table::compile): Move setting of
13311 flag_generate_offload earlier ...
13312 (symbol_table::finalize_compilation_unit): ... here, before
13313 early debug finalization.
13315 2017-08-24 Richard Biener <rguenther@suse.de>
13317 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13318 and ipa-fnsummary.h.
13319 (ix86_can_inline_p): When ix86_fpmath flags do not match
13320 check whether the callee uses FP math at all.
13322 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13324 PR middle-end/81931
13325 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13326 instead of TYPE_PRECISION.
13328 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13329 Alan Hayward <alan.hayward@arm.com>
13330 David Sherwood <david.sherwood@arm.com>
13332 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13334 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13336 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13338 (ix86_frame::stack_realign_allocate): New field.
13339 (struct machine_frame_state): Modify comments.
13340 (machine_frame_state::sp_realigned_fp_end): New field.
13341 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13342 layout calculation.
13343 (sp_valid_at): Add assertion to assure no attempt to access invalid
13344 offset of a realigned stack.
13345 (fp_valid_at): Likewise.
13346 (choose_baseaddr): Modify comments.
13347 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
13348 ix86_expand_prologue.
13349 (ix86_expand_prologue): Modify stack realignment and allocation.
13350 (ix86_expand_epilogue): Modify comments.
13351 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
13352 avx2_runtime, avx512f, and avx512f_runtime.
13354 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
13356 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
13357 (mstackrealign): Do not init to -1.
13358 * config/i386/i386.c (ix86_option_override_internal):
13359 Check opts_set, not opts when setting default value of
13360 opts->x_ix86_force_align_arg_pointer.
13362 2017-08-23 Richard Biener <rguenther@suse.de>
13364 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13365 lang_hooks.decl_printable_name.
13366 * print-rtl-function.c (print_rtx_function): Likewise.
13367 * tree-pretty-print.c (dump_function_header): Likewise.
13369 2017-08-23 Richard Biener <rguenther@suse.de>
13372 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
13373 -g0 at compile-time.
13375 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13377 PR middle-end/19706
13378 * doc/sourcebuild.texi (Other hardware attributes):
13381 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13383 PR middle-end/19706
13384 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
13385 Fix single-use check.
13387 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
13389 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
13391 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13393 * doc/install.texi: Modify to add more details on running selected
13396 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13398 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
13399 is combined with -mabi=ms.
13400 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
13403 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13405 PR tree-optimization/81488
13406 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
13407 and cached_basis fields.
13408 (MAX_SPREAD): New constant.
13409 (alloc_cand_and_find_basis): Initialize new fields.
13410 (clear_visited): New function.
13411 (create_phi_basis_1): Rename from create_phi_basis, set visited
13412 and cached_basis fields.
13413 (create_phi_basis): New wrapper function.
13414 (phi_add_costs_1): Rename from phi_add_costs, add spread
13415 parameter, set visited field, short-circuit when limits reached.
13416 (phi_add_costs): New wrapper function.
13417 (record_phi_increments_1): Rename from record_phi_increments, set
13419 (record_phi_increments): New wrapper function.
13420 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
13421 (phi_incr_cost): New wrapper function.
13422 (all_phi_incrs_profitable_1): Rename from
13423 all_phi_incrs_profitable, set visited field.
13424 (all_phi_incrs_profitable): New wrapper function.
13426 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13427 Alan Hayward <alan.hayward@arm.com>
13428 David Sherwood <david.sherwood@arm.com>
13430 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
13431 that takes the outer and inner modes.
13432 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
13433 comparison as the canonical test for a paradoxical subreg.
13434 * combine.c (simplify_set): Use paradoxical_subreg_p.
13435 (make_extraction): Likewise.
13436 (force_to_mode): Likewise.
13437 (rtx_equal_for_field_assignment_p): Likewise.
13438 (gen_lowpart_for_combine): Likewise.
13439 (simplify_comparison): Likewise.
13440 * cse.c (equiv_constant): Likewise.
13441 * expmed.c (store_bit_field_1): Likewise.
13442 * final.c (alter_subreg): Likewise.
13443 * fwprop.c (propagate_rtx): Likewise.
13444 (forward_propagate_subreg): Likewise.
13445 * ira-conflicts.c (ira_build_conflicts): Likewise.
13446 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
13447 * lra-constraints.c (curr_insn_transform): Likewise.
13448 (split_reg): Likewise.
13449 * lra-eliminations.c (move_plus_up): Likewise.
13450 (lra_eliminate_regs_1): Likewise.
13451 * recog.c (general_operand): Likewise.
13452 * ree.c (combine_reaching_defs): Likewise.
13453 * reload.c (push_reload): Likewise.
13454 (find_reloads): Likewise.
13455 * reload1.c (elimination_effects): Likewise.
13456 (compute_reload_subreg_offset): Likewise.
13457 (choose_reload_regs): Likewise.
13458 * rtlanal.c (subreg_lsb_1): Likewise.
13459 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13460 (simplify_subreg): Likewise.
13461 * var-tracking.c (track_loc_p): Likewise.
13462 * emit-rtl.c (byte_lowpart_offset): Likewise.
13463 (paradoxical_subreg_p): Delete out-of-line definition.
13465 2017-08-22 Jeff Law <law@redhat.com>
13467 PR tree-optimization/81741
13468 PR tree-optimization/71947
13469 * tree-ssa-dom.c: Include tree-inline.h.
13470 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
13471 equivalences if one is more expensive to compute than the other.
13472 * tree-ssa-scopedtables.h (class const_or_copies): Make
13473 record_const_or_copy_raw method private.
13474 (class avail_exprs_stack): New method simplify_binary_operation.
13475 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
13476 avail_exprs_stack::simplify_binary_operation as needed.
13477 (avail_exprs_stack::simplify_binary_operation): New function.
13479 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
13481 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
13482 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
13483 (DOT_SYMBOLS): Likewise.
13484 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
13485 (RELOCATABLE_NEEDS_FIXUP): Likewise.
13486 (RS6000_ABI_NAME): Likewise.
13487 (TARGET_CMODEL): Likewise.
13488 (TOC_SECTION_ASM_OP): Likewise.
13489 (SET_CMODEL): New macro.
13490 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
13492 2017-08-22 Richard Biener <rguenther@suse.de>
13494 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
13495 to allow for free-lang-data replacements similar to verify_type_variant.
13497 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13498 Alan Hayward <alan.hayward@arm.com>
13499 David Sherwood <david.sherwood@arm.com>
13501 * config/aarch64/aarch64.md (casesi): Use DImode rather than
13502 VOIDmode for the LABEL_REF.
13504 2017-08-22 Richard Biener <rguenther@suse.de>
13506 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
13508 2017-08-22 Richard Biener <rguenther@suse.de>
13510 * common.opt (feliminate-dwarf2-dups): Ignore.
13511 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
13512 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
13513 same_die_p_wrap, compute_section_prefix,
13514 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
13515 (comdat_symbol_id, comdat_symbol_number): Likewise.
13516 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
13518 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
13519 (output_die): Mark unreachable path unreachable.
13520 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
13521 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
13522 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
13523 (dwarf2out_early_finish): Likewise.
13525 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
13527 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
13529 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
13532 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
13533 not VAR_P. Filter attribute warnings with OPT_Wattributes.
13534 (avr_attribute_table) <io, io_low, address>: Initialize
13535 .decl_required with true.
13537 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
13539 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
13540 undocumented debugging options.
13541 (-mvsx-scalar-double): Likewise.
13542 (-mallow-df-permute): Likewise.
13543 (-mvectorize-builtins): Likewise.
13544 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13545 (rs6000_builtin_vectorized_function): Likewise.
13546 (rs6000_builtin_md_vectorized_function): Likewise.
13547 (rs6000_opt_vars): Likewise.
13549 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
13552 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
13553 (*btrq_imm): Rename from *btrq.
13554 (*btcq_imm): Rename from *btcq.
13555 (btsc): New code attribute.
13556 (*<btsc><mode>): New insn pattern.
13557 (*btr<mode>): Ditto.
13558 (*<btsc><mode>_mask): New insn_and_split pattern.
13559 (*btr<mode>_mask): Ditto.
13561 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13562 Alan Hayward <alan.hayward@arm.com>
13563 David Sherwood <david.sherwood@arm.com>
13565 * function.c (pad_below): Simplify padding calculation.
13567 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13568 Alan Hayward <alan.hayward@arm.com>
13569 David Sherwood <david.sherwood@arm.com>
13571 * target.def (function_prologue): Remove frame size argument.
13572 (function_epilogue): Likewise.
13573 * doc/tm.texi: Regenerate.
13574 * final.c (final_start_function): Update call to function_prologue.
13575 (final_end_function): Update call to function_epilogue.
13576 (default_function_pro_epilogue): Remove frame size argument.
13577 * output.h (default_function_pro_epilogue): Likewise.
13578 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
13579 (arm_output_function_prologue): Likewise.
13580 * config/frv/frv.c (frv_function_prologue): Likewise.
13581 (frv_function_epilogue): Likewise.
13582 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
13583 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
13584 (ia64_output_function_epilogue): Likewise.
13585 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
13586 (m32r_output_function_epilogue): Likewise.
13587 * config/microblaze/microblaze.c (microblaze_function_prologue)
13588 (microblaze_function_epilogue): Likewise.
13589 * config/mips/mips.c (mips_output_function_prologue): Likewise.
13590 (mips_output_function_epilogue): Likewise.
13591 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
13592 (mmix_target_asm_function_epilogue): Likewise.
13593 * config/msp430/msp430.c (msp430_start_function): Likewise.
13594 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
13595 (nds32_asm_function_epilogue): Likewise.
13596 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
13597 * config/pa/pa.c (pa_output_function_prologue): Likewise.
13598 (pa_output_function_epilogue): Likewise.
13599 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
13600 (rs6000_output_function_epilogue): Likewise.
13601 * config/rl78/rl78.c (rl78_start_function): Likewise.
13602 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
13603 (rs6000_output_function_epilogue): Likewise.
13604 * config/rx/rx.c (rx_output_function_prologue): Likewise.
13605 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
13606 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
13607 (sparc_asm_function_epilogue): Likewise.
13609 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13611 * tree.h (type_has_mode_precision_p): New function.
13612 * convert.c (convert_to_integer_1): Use it.
13613 * expr.c (expand_expr_real_2): Likewise.
13614 (expand_expr_real_1): Likewise.
13615 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
13616 * match.pd: Likewise.
13617 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
13618 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
13619 * tree-tailcall.c (process_assignment): Likewise.
13620 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13621 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
13622 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
13623 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13624 (vectorizable_assignment): Likewise.
13625 (vectorizable_shift): Likewise.
13626 (vectorizable_operation): Likewise.
13627 * tree-vrp.c (register_edge_assert_for_2): Likewise.
13629 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
13631 * match.pd: Add pow (C, x) simplification.
13633 2017-08-21 Richard Biener <rguenther@suse.de>
13635 PR tree-optimization/81900
13636 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
13637 for blocks with abnormal predecessors.
13638 (compute_antic): Do not set visited flag prematurely.
13640 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
13643 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
13645 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13647 * stor-layout.h (vector_type_mode): Move to...
13648 * tree.h (vector_type_mode): ...here.
13649 * stor-layout.c (vector_type_mode): Move to...
13650 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
13652 2017-08-21 Richard Biener <rguenther@suse.de>
13654 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
13655 register_external_die hooks.
13656 (debug_false_tree_charstarstar_uhwistar): Declare.
13657 (debug_nothing_tree_charstar_uhwi): Likewise.
13658 * debug.c (do_nothing_debug_hooks): Adjust.
13659 (debug_false_tree_charstarstar_uhwistar): New do nothing.
13660 (debug_nothing_tree_charstar_uhwi): Likewise.
13661 * dbxout.c (dbx_debug_hooks): Adjust.
13662 (xcoff_debug_hooks): Likewise.
13663 * sdbout.c (sdb_debug_hooks): Likewise.
13664 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13665 * dwarf2out.c (macinfo_label_base): New global.
13666 (dwarf2out_register_external_die): New function for the
13667 register_external_die hook.
13668 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
13669 (dwarf2_debug_hooks): Use them.
13670 (dwarf2_lineno_debug_hooks): Adjust.
13671 (struct die_struct): Add with_offset flag.
13672 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
13673 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
13674 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
13675 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
13676 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
13677 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
13678 defining section names for the early LTO debug variants.
13679 (reset_indirect_string): New helper.
13680 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
13681 (print_dw_val): Add support for offsetted symbol references.
13682 (get_ultimate_context): Split out from is_cxx.
13683 (is_cxx): Use get_ultimate_context.
13684 (is_fortran): Add decl overload.
13685 (compute_comp_unit_symbol): Split out worker from
13686 compute_section_prefix.
13687 (compute_section_prefix): Call compute_comp_unit_symbol and
13688 set comdat_type_p here.
13689 (output_die): Skip DIE symbol output for the LTO added one.
13690 Handle DIE symbol references with offset.
13691 (output_comp_unit): Guard section name mangling properly.
13692 For LTO debug sections emit a symbol at the section beginning
13693 which we use to refer to its DIEs.
13694 (add_abstract_origin_attribute): For DIEs registered via
13695 dwarf2out_register_external_die directly refer to the early
13696 DIE rather than indirectly through the shadow one we created.
13697 Remove obsolete call to dwarf2out_abstract_function for
13698 non-function/block origins.
13699 (gen_array_type_die): When generating early LTO debug do
13700 not emit DW_AT_string_length.
13701 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
13702 late when in LTO. As suggested place a gcc_unreachable for
13703 the DECL_ABSTRACT_P case.
13704 (gen_subprogram_die): Avoid another specification DIE
13705 for early built declarations/definitions for the late LTO case.
13706 (gen_variable_die): Add type references for late duplicated VLA dies
13708 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
13709 we have the abstract instance already.
13710 (process_scope_var): Adjust decl DIE contexts in LTO which
13711 first puts them in limbo.
13712 (gen_decl_die): Do not generate type DIEs late apart from
13713 types for VLAs or for decls we do not yet have a DIE. Do not
13714 call dwarf2out_abstract_function late.
13715 (dwarf2out_early_global_decl): Make sure to create DIEs
13716 for abstract instances of a decl first.
13717 (dwarf2out_late_global_decl): Adjust comment.
13718 (output_macinfo_op): With multiple macro sections use
13719 macinfo_label_base to distinguish labels.
13720 (output_macinfo): Likewise. Update macinfo_label_base.
13721 Pass in the line info label.
13722 (note_variable_value_in_expr): When generating LTO resolve
13723 all variable values here by generating DIEs as needed.
13724 (init_sections_and_labels): Add early LTO debug flag parameter
13725 and generate different sections and names if set. Add generation
13726 counter for the labels so we can have multiple of them.
13727 (reset_dies): Helper to allow DIEs to be output multiple times.
13728 (dwarf2out_finish): When outputting DIEs to the fat part of an
13729 LTO object first reset DIEs.
13730 (dwarf2out_early_finish): Output early DIEs when generating LTO.
13731 (modified_type_die): Check for decl_ultimate_origin being self
13733 (gen_type_die_with_usage): Likewise.
13734 (gen_typedef_die): Allow decl_ultimate_origin being self.
13735 (set_decl_abstract_flags): Remove.
13736 (set_block_abstract_flags): Likewise.
13737 (dwarf2out_abstract_function): Treat the early generated DIEs
13738 as the abstract copy and only add DW_AT_inline and
13739 DW_AT_artificial here and call set_decl_origin_self.
13740 If the DIE has an abstract origin don't do anything.
13741 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
13742 if we have none yet (Go fails to build one, PR78628).
13743 (variably_modified_type_p): Prevent endless recursion for Ada
13744 cyclic pointer types.
13745 * lto-streamer-in.c: Include debug.h.
13746 (dref_queue): New global.
13747 (lto_read_tree_1): Stream in DIE references.
13748 (lto_input_tree): Register DIE references.
13749 (input_function): Stream DECL_DEBUG_ARGS.
13750 * lto-streamer-out.c: Include debug.h.
13751 (lto_write_tree_1): Output DIE references.
13752 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
13753 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
13754 (output_function): Stream DECL_DEBUG_ARGS.
13755 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
13756 Stream DECL_ABSTRACT_ORIGIN.
13757 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
13758 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
13759 DECL_CONTEXT for file-scope decls.
13760 * lto-streamer.h (struct dref_entry): Declare.
13761 (dref_queue): Likewise.
13762 * cfgexpand.c (pass_expand::execute): Do not call the
13763 outlining_inline_function hook here.
13764 * lto-wrapper.c (debug_obj): New global.
13765 (tool_cleanup): Unlink it if required.
13766 (debug_objcopy): New function.
13767 (run_gcc): Handle early debug sections in the IL files by
13768 extracting them to separate files, partially linkin them and
13769 feeding the result back as result to the linker.
13770 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
13771 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
13772 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
13773 sections into a separate segment.
13774 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
13776 (darwin_asm_dwarf_section): Likewise.
13777 (darwin_asm_output_dwarf_offset): Likewise.
13778 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
13780 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13781 Alan Hayward <alan.hayward@arm.com>
13782 David Sherwood <david.sherwood@arm.com>
13784 * read-md.h (md_reader::record_potential_iterator_use): Replace
13785 pointer argument with an rtx and an index.
13786 * read-rtl.c (iterator_group::apply_iterator): Likewise.
13787 (apply_mode_iterator): Likewise.
13788 (apply_code_iterator): Likewise.
13789 (apply_int_iterator): Likewise.
13790 (apply_subst_iterator): Likewise.
13791 (record_iterator_use): Likewise.
13792 (record_attribute_use): Likewise.
13793 (md_reader::record_potential_iterator_use): Likewise. Update calls
13794 to record_iterator_use and apply_iterator.
13795 (iterator_use): Replace ptr with x and index.
13796 (attribute_use): Likewise.
13797 (apply_attribute_uses): Update calls to apply_iterator.
13798 (apply_iterators): Likewise. Update initialization of iterator_use.
13799 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
13800 and record_potential_iterator_use.
13801 (rtx_reader::read_rtx_operand): Likewise.
13803 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13804 Alan Hayward <alan.hayward@arm.com>
13805 David Sherwood <david.sherwood@arm.com>
13807 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
13810 2017-08-21 Richard Biener <rguenther@suse.de>
13812 PR middle-end/81884
13813 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
13814 at struct end conservatively when comparing common bases.
13816 2017-08-21 Richard Biener <rguenther@suse.de>
13818 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
13819 (mem_ref_in_stmt): Remove.
13820 (determine_max_movement): Use ref index to get at the reference.
13821 (invariantness_dom_walker::before_dom_children): Deal with
13822 lim data already initialized.
13823 (gather_mem_refs_stmt): Initialize lim data and record ref index.
13825 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13827 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
13828 (TARGET_ISA_ROUND): Ditto.
13829 (TARGET_ROUND): Ditto.
13830 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
13831 * config/i386/i386.md: Ditto.
13832 * config/i386/sse.md: Ditto.
13833 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
13834 with OPTION_MASK_ISA_SSE4_1.
13836 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13839 * doc/extend.texi (x86 Built-in Functions): Correct the name of
13840 __builtin_ia32_lzcnt_u16.
13842 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
13845 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
13846 (rs6000_set_current_function): Rewrite function to use it.
13848 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
13851 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
13852 and TYPE_WARN_IF_NOT_ALIGN.
13853 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
13854 (handle_warn_if_not_align): New.
13855 (place_union_field): Call handle_warn_if_not_align.
13856 (place_field): Call handle_warn_if_not_align.
13857 Copy TYPE_WARN_IF_NOT_ALIGN.
13858 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
13859 (layout_type): Likewise.
13860 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
13862 (tree_decl_common): Add warn_if_not_align.
13863 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
13864 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
13865 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
13866 (DECL_WARN_IF_NOT_ALIGN): Likewise.
13867 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
13868 * doc/extend.texi: Document warn_if_not_aligned attribute.
13869 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
13871 2017-08-17 Martin Liska <mliska@suse.cz>
13874 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
13875 (get_data_dependence): Use it as pointer type.
13876 (distribute_loop): Likewise.
13878 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13880 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
13881 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
13882 (p8_vmrgow_<mode>_direct): New define_insn.
13883 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
13884 handle endianness for vmrgew and vmrgow permute patterns.
13886 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13888 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
13889 * config/rs6000/rs6000-cpus.def: Remove comment.
13890 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
13891 (POWERPC_MASKS): Likewise.
13892 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
13893 use of TARGET_VSX_TIMODE.
13894 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
13895 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
13896 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
13897 (rs6000_option_override_internal): Remove dead code.
13898 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
13899 (rs6000_legitimize_reload_address): Likewise.
13900 (rs6000_legitimate_address_p): Likewise.
13901 (rs6000_opt_masks): Delete "vsx-timode".
13902 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
13903 from function comment.
13904 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
13905 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
13906 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
13908 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
13909 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
13910 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
13911 (VSX_TI): Likewise.
13913 (define_peephole2): Likewise.
13915 2017-08-17 Martin Sebor <msebor@redhat.com>
13918 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
13919 past the end of an array.
13920 (test_pp_format): Add test cases.
13922 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
13924 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
13925 missing ECF_NOTHROW flags.
13927 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13930 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
13931 operands residing in integer registers.
13932 (*vsx_le_perm_load_<mode>): Likewise.
13933 (*vsx_le_perm_store_<mode>): Likewise.
13934 (define_peephole2): Add peepholes to optimize the above.
13936 2017-08-17 Marek Polacek <polacek@redhat.com>
13938 PR middle-end/81814
13939 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
13940 to mimic what shorten_compare did. Change the return type to bool.
13941 (fold_cond_expr_with_comparison): Update call to
13942 operand_equal_for_comparison_p.
13943 (fold_ternary_loc): Likewise.
13945 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
13947 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
13949 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
13951 2017-08-17 Richard Biener <rguenther@suse.de>
13953 * tree-ssa-structalias.c (solve_graph): When propagating
13954 to successors update the graphs succ edges and avoid duplicate work.
13956 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
13959 * config/i386/i386.c (ix86_option_override_internal): Save target
13960 specific options after ix86_stack_protector_guard_reg was changed.
13962 2017-08-17 Richard Biener <rguenther@suse.de>
13964 PR tree-optimization/81827
13965 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
13966 (new_var_info): Initialize it conservatively.
13967 (get_call_vi): Mark register vars.
13968 (new_scalar_tmp_constraint_exp): Likewise.
13969 (handle_rhs_call): Likewise.
13970 (handle_const_call): Likewise.
13971 (create_function_info_for): Likewise.
13972 (solve_constraints): Sort varinfos to separate register from
13973 non-register vars to pack points-to solution bitmaps during
13976 2017-08-17 Marek Polacek <polacek@redhat.com>
13978 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
13980 2017-08-17 Richard Biener <rguenther@suse.de>
13982 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
13983 to true when overflow is undefined and we saturated the result.
13985 2017-08-17 Alan Modra <amodra@gmail.com>
13988 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
13989 Don't use store multiple if only one reg needs saving.
13990 (interesting_frame_related_regno): New function.
13991 (rs6000_frame_related): Don't emit frame info for regs that
13993 (rs6000_emit_epilogue): Likewise.
13995 2017-08-16 Nathan Sidwell <nathan@acm.org>
13997 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
13998 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
13999 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14000 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14001 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14002 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14003 (verify_type): Adjust for TYPE_BINFO move.
14004 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14005 process TYPE_BINFO directly.
14006 (hash_tree): Likewise.
14007 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14009 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14012 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14014 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14016 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14019 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14020 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14021 Add ix86_binary_operator_ok to insn constraint.
14022 (*iordi_1_bts): Ditto.
14023 (*xordi_1_btc): Ditto.
14024 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14025 Update corresponding peephole2 pattern.
14029 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14031 PR tree-optimization/81832
14032 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14033 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14035 2017-08-16 Marek Polacek <polacek@redhat.com>
14038 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14039 regarding size_zero_node.
14041 2017-08-16 Martin Liska <mliska@suse.cz>
14044 * config.gcc: Respect previously set extra_objs in case
14047 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14049 PR tree-optimization/81835
14050 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14051 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14052 not depend on the phi.
14054 2017-08-16 Alan Modra <amodra@gmail.com>
14056 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14059 2017-08-16 Alan Modra <amodra@gmail.com>
14061 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14062 (save_reg_p): ..into this. Update all callers.
14063 (first_reg_to_save): Simplify.
14065 2017-08-16 Alan Modra <amodra@gmail.com>
14067 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14070 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14074 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14075 constant count if that count is less than 32.
14077 2017-08-15 Nathan Sidwell <nathan@acm.org>
14079 * gcc.c (execute): Emit friendlier message if inferior is killed
14080 by an external cause.
14082 2017-08-15 Richard Biener <rguenther@suse.de>
14084 PR tree-optimization/81790
14085 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14086 CONSTRUCTORs from simplifying and VN.
14088 2017-08-14 Martin Sebor <msebor@redhat.com>
14090 * builtin-attrs.def: Add comments.
14092 2017-08-14 Martin Sebor <msebor@redhat.com>
14095 * doc/extend.texi (attribute nonstring): Document new attribute.
14097 2017-08-14 Martin Sebor <msebor@redhat.com>
14100 * tree-diagnostic.c (default_tree_printer): Handle %G.
14101 * gimple-pretty-print.h (percent_G_format): Declare new function.
14102 * gimple-pretty-print.c (percent_G_format): Define.
14103 * tree-pretty-print.c (percent_K_format): Add argument.
14105 2017-08-14 Martin Sebor <msebor@redhat.com>
14107 PR translation/79998
14108 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14109 Remove a stray space.
14111 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14114 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14115 (*iordi_1_bts): Ditto.
14116 (*xordi_1_btc): Ditto.
14118 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14121 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14122 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14124 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14125 quoted strings, and make more translator-friendly.
14126 (darwin_rs6000_override_options): Likewise.
14127 (rs6000_option_override_internal): Likewise.
14128 (rs6000_return_in_memory): Fix overlong line.
14129 (init_cmulative_args): Use quoted strings, and make more
14130 translator-friendly.
14131 (rs6000_pass_by_reference): Fix overlong line.
14132 (def_builtin): Use quoted strings.
14133 (altivec_expand_predicate_builtin): Use quoted strings, and make
14134 more translator-friendly.
14135 (htm_expand_builtin): Use quoted strings.
14136 (cpu_expand_builtin): Use quoted strings, and make more
14137 translator-friendly.
14138 (altivec_expand_builtin): Likewise.
14139 (paired_expand_predicate_builtin): Likewise.
14140 (rs6000_invalid_builtin): Likewise.
14141 (builtin_function_type): Use quoted strings.
14142 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14143 more translator-friendly.
14144 (rs6000_trampoline_init): Likewise.
14145 (rs6000_handle_altivec_attribute): Likewise.
14146 (rs6000_inner_target_options): Use quoted strings.
14147 (rs6000_disable_incompatible_switches): Likewise.
14148 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14149 strings, and make more translator-friendly.
14150 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14152 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14154 PR tree-optimization/81799
14155 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14156 cond_expr to simple gimple operand.
14158 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14160 PR middle-end/46932
14161 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14163 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14167 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14168 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14169 TARGET_GASISR_PROLOGUES.
14170 * config/avr/avr.c (avr_option_override): Same.
14171 (avr_pass_pre_proep::execute): Same.
14173 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14176 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14177 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14179 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14181 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14182 define_insn_and_split. Split to a memory load from 0 in
14183 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14184 using PTR mode iterator.
14185 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14186 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14187 (*add_tp_<mode>): Redefine as define_insn_and_split.
14188 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14189 address space. Merge with *add_tp_x32 using PTR mode iterator.
14190 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14191 Split to an add with a memory load from 0 in
14192 DEFAULT_TLS_SEG_REG address space.
14194 2017-08-12 Andrew Pinski <apinski@cavium.com>
14196 * config/aarch64/aarch64-option-extensions.def (rdma):
14197 Fix feature string to what Linux prints out in /proc/cpuinfo.
14199 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14202 * dwarf2out.c (modified_type_die): For C typedef types that have
14203 an ultimate origin, process the ultimate origin instead of the
14205 (gen_typedef_die): Assert that input DECLs have no ultimate
14207 (gen_type_die_with_usage): For typedef variants that have an
14208 ultimate origin, just call gen_decl_die on the original DECL.
14209 (process_scope_var): Avoid creating DIEs for local typedefs and
14210 concrete static variables.
14212 2017-08-12 Alan Modra <amodra@gmail.com>
14216 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14217 match gnu-user.h startfile.
14218 (ENDFILE_LINUX_SPEC): Similarly.
14220 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14223 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14225 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14227 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14228 * config/aarch64/aarch64.md (mov<mode>): Change.
14229 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14230 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14231 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14233 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14235 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14236 for storage order barriers.
14238 2017-08-11 Martin Liska <mliska@suse.cz>
14241 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14242 variables of void type.
14244 2017-08-11 Martin Liska <mliska@suse.cz>
14246 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14247 TARGET_SUPPORTS_ALIASES.
14248 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14249 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14250 (optimize_weakref): Likewise.
14251 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14252 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14253 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14255 2017-08-11 Martin Liska <mliska@suse.cz>
14258 * config/i386/i386.c (make_resolver_func): Do complete
14259 refactoring of the function.
14261 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14264 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14265 * config/i386/i386.c (ix86_stack_protect_guard): Use
14266 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14267 * doc/invoke.texi (x86 Options): Document
14268 -mstack-protector-guard-symbol= option.
14270 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14272 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14273 * config/i386/i386.c (ix86_split_stack_guard): New function.
14274 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14275 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14276 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14277 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14278 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14279 (split_stack_space_check): Call ix86_split_stack_guard.
14281 2017-08-10 Martin Sebor <msebor@redhat.com>
14283 * print-tree.c (print_node): Print location using the established
14285 Replace spaces with colons.
14286 (debug_raw, debug): Ditto.
14288 2017-08-10 Martin Sebor <msebor@redhat.com>
14291 * pretty-print.c (pp_format): Correct the handling of %s precision.
14293 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14296 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14298 (ix86_finalize_stack_frame_flags): This. Also clear
14299 frame_pointer_needed if -fno-omit-frame-pointer is used without
14301 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14302 with ix86_finalize_stack_frame_flags.
14303 (ix86_expand_epilogue): Likewise.
14304 (ix86_expand_split_stack_prologue): Likewise.
14305 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14307 2017-08-10 Martin Liska <mliska@suse.cz>
14310 * c-attribs.c (handle_target_attribute):
14311 Report warning for an empty string argument of target attribute.
14313 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14316 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14318 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14319 or DECL_NONLOCAL labels.
14320 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14321 or DECL_NONLOCAL labels here.
14323 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14325 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14326 to indicate when early gimple folding has been disabled.
14327 (rs6000_gimple_fold_builtin): Add debug content.
14328 (rs6000_invalid_builtin): Fix whitespace.
14329 (rs6000_expand_builtin): Fix whitespace.
14330 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14332 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14335 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14336 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14338 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14340 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14341 (qdf24xx): Likewise.
14342 * config/aarch64/aarch64.md: Include falkor.md.
14343 * config/aarch64/falkor.md: New.
14345 2017-08-09 Marek Polacek <polacek@redhat.com>
14348 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
14349 * diagnostic.c (emit_diagnostic): Add a comment.
14350 (emit_diagnostic_valist): New function.
14352 2017-08-09 Marek Polacek <polacek@redhat.com>
14355 * input.c (make_location): New overload.
14356 * input.h (make_location): Declare.
14358 2017-08-08 Alan Modra <amodra@gmail.com>
14359 H.J. Lu <hongjiu.lu@intel.com>
14362 * gcc.c (NO_PIE_SPEC): Delete.
14363 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14365 (LINK_PIE_SPEC): ..to here.
14366 (LINK_COMMAND_SPEC): Support -no-pie.
14367 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14368 chain of crtbegin*.o selection, update for PIE_SPEC changes and
14370 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
14371 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
14372 (ENDFILE_CRTEND_SPEC): Similarly.
14374 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
14377 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
14378 (mstack-protector-guard-offset=): Ditto.
14379 * config/i386/i386.c (ix86_option_override): Handle
14380 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
14382 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
14383 ix86_stack_protect_guard_offset variables.
14384 (TARGET_STACK_PROTECT_GUARD): Always define.
14385 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
14386 and -mstack-protector-guard-offset= options.
14388 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14390 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
14391 boundary case for the last candidate.
14393 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14395 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
14396 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
14398 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14400 PR middle-end/19706
14401 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
14402 * config/aarch64/aarch64-builtins.c
14403 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
14404 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
14405 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
14407 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14408 Andrew Pinski <pinskia@gmail.com>
14410 PR middle-end/19706
14411 * internal-fn.def (XORSIGN): New.
14412 * optabs.def (xorsign_optab): New.
14413 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
14414 (convert_expand_mult_copysign): New.
14415 (pass_optimize_widening_mul::execute): Call
14416 convert_expand_mult_copysign.
14418 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14420 PR tree-optimization/81354
14421 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
14422 Insert on edges rather than explicitly creating landing pads.
14423 (analyze_candidates_and_replace): Commit edge inserts.
14425 2017-08-08 Richard Biener <rguenther@suse.de>
14427 PR middle-end/81719
14428 * tree-ssa-loop-niter.c: Include tree-dfa.h.
14429 (expand_simple_operations): Also look through ADDR_EXPRs with
14430 MEM_REF bases treating them as POINTER_PLUS_EXPR.
14432 2017-08-08 Richard Biener <rguenther@suse.de>
14434 PR tree-optimization/81723
14435 * tree-vect-slp.c (struct bst_traits): New hash traits.
14436 (bst_fail): New global.
14437 (vect_build_slp_tree_2): New worker, split out from ...
14438 (vect_build_slp_tree): ... this now wrapping it with using
14439 bst_fail set to cache SLP tree build fails. Properly handle
14441 (vect_analyze_slp_instance): Allocate and free bst_fail.
14443 2017-08-08 Martin Liska <mliska@suse.cz>
14446 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
14447 LABEL_DECLs that can be from a different function.
14449 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14451 PR tree-optimization/81744
14452 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
14453 loop's number of iterations.
14455 2017-08-08 Martin Liska <mliska@suse.cz>
14457 * asan.c: Include header files.
14458 * attribs.c (build_decl_attribute_variant): New function moved
14460 (build_type_attribute_qual_variant): Likewise.
14461 (cmp_attrib_identifiers): Likewise.
14462 (simple_cst_list_equal): Likewise.
14463 (omp_declare_simd_clauses_equal): Likewise.
14464 (attribute_value_equal): Likewise.
14465 (comp_type_attributes): Likewise.
14466 (build_type_attribute_variant): Likewise.
14467 (lookup_ident_attribute): Likewise.
14468 (remove_attribute): Likewise.
14469 (merge_attributes): Likewise.
14470 (merge_type_attributes): Likewise.
14471 (merge_decl_attributes): Likewise.
14472 (merge_dllimport_decl_attributes): Likewise.
14473 (handle_dll_attribute): Likewise.
14474 (attribute_list_equal): Likewise.
14475 (attribute_list_contained): Likewise.
14476 * attribs.h (lookup_attribute): New function moved from tree.[ch].
14477 (lookup_attribute_by_prefix): Likewise.
14478 * bb-reorder.c: Include header files.
14479 * builtins.c: Likewise.
14480 * calls.c: Likewise.
14481 * cfgexpand.c: Likewise.
14482 * cgraph.c: Likewise.
14483 * cgraphunit.c: Likewise.
14484 * convert.c: Likewise.
14485 * dwarf2out.c: Likewise.
14486 * final.c: Likewise.
14487 * fold-const.c: Likewise.
14488 * function.c: Likewise.
14489 * gimple-expr.c: Likewise.
14490 * gimple-fold.c: Likewise.
14491 * gimple-pretty-print.c: Likewise.
14492 * gimple.c: Likewise.
14493 * gimplify.c: Likewise.
14494 * hsa-common.c: Likewise.
14495 * hsa-gen.c: Likewise.
14496 * internal-fn.c: Likewise.
14497 * ipa-chkp.c: Likewise.
14498 * ipa-cp.c: Likewise.
14499 * ipa-devirt.c: Likewise.
14500 * ipa-fnsummary.c: Likewise.
14501 * ipa-inline.c: Likewise.
14502 * ipa-visibility.c: Likewise.
14504 * lto-cgraph.c: Likewise.
14505 * omp-expand.c: Likewise.
14506 * omp-general.c: Likewise.
14507 * omp-low.c: Likewise.
14508 * omp-offload.c: Likewise.
14509 * omp-simd-clone.c: Likewise.
14510 * opts-global.c: Likewise.
14511 * passes.c: Likewise.
14512 * predict.c: Likewise.
14513 * sancov.c: Likewise.
14514 * sanopt.c: Likewise.
14515 * symtab.c: Likewise.
14516 * toplev.c: Likewise.
14517 * trans-mem.c: Likewise.
14518 * tree-chkp.c: Likewise.
14519 * tree-eh.c: Likewise.
14520 * tree-into-ssa.c: Likewise.
14521 * tree-object-size.c: Likewise.
14522 * tree-parloops.c: Likewise.
14523 * tree-profile.c: Likewise.
14524 * tree-ssa-ccp.c: Likewise.
14525 * tree-ssa-live.c: Likewise.
14526 * tree-ssa-loop.c: Likewise.
14527 * tree-ssa-sccvn.c: Likewise.
14528 * tree-ssa-structalias.c: Likewise.
14529 * tree-ssa.c: Likewise.
14530 * tree-streamer-in.c: Likewise.
14531 * tree-vectorizer.c: Likewise.
14532 * tree-vrp.c: Likewise.
14533 * tsan.c: Likewise.
14534 * ubsan.c: Likewise.
14535 * varasm.c: Likewise.
14536 * varpool.c: Likewise.
14537 * tree.c: Remove functions moved to attribs.[ch].
14538 * tree.h: Likewise.
14539 * config/aarch64/aarch64.c: Add attrs.h header file.
14540 * config/alpha/alpha.c: Likewise.
14541 * config/arc/arc.c: Likewise.
14542 * config/arm/arm.c: Likewise.
14543 * config/avr/avr.c: Likewise.
14544 * config/bfin/bfin.c: Likewise.
14545 * config/c6x/c6x.c: Likewise.
14546 * config/cr16/cr16.c: Likewise.
14547 * config/cris/cris.c: Likewise.
14548 * config/darwin.c: Likewise.
14549 * config/epiphany/epiphany.c: Likewise.
14550 * config/fr30/fr30.c: Likewise.
14551 * config/frv/frv.c: Likewise.
14552 * config/ft32/ft32.c: Likewise.
14553 * config/h8300/h8300.c: Likewise.
14554 * config/i386/winnt.c: Likewise.
14555 * config/ia64/ia64.c: Likewise.
14556 * config/iq2000/iq2000.c: Likewise.
14557 * config/lm32/lm32.c: Likewise.
14558 * config/m32c/m32c.c: Likewise.
14559 * config/m32r/m32r.c: Likewise.
14560 * config/m68k/m68k.c: Likewise.
14561 * config/mcore/mcore.c: Likewise.
14562 * config/microblaze/microblaze.c: Likewise.
14563 * config/mips/mips.c: Likewise.
14564 * config/mmix/mmix.c: Likewise.
14565 * config/mn10300/mn10300.c: Likewise.
14566 * config/moxie/moxie.c: Likewise.
14567 * config/msp430/msp430.c: Likewise.
14568 * config/nds32/nds32-isr.c: Likewise.
14569 * config/nds32/nds32.c: Likewise.
14570 * config/nios2/nios2.c: Likewise.
14571 * config/nvptx/nvptx.c: Likewise.
14572 * config/pa/pa.c: Likewise.
14573 * config/pdp11/pdp11.c: Likewise.
14574 * config/powerpcspe/powerpcspe.c: Likewise.
14575 * config/riscv/riscv.c: Likewise.
14576 * config/rl78/rl78.c: Likewise.
14577 * config/rx/rx.c: Likewise.
14578 * config/s390/s390.c: Likewise.
14579 * config/sh/sh.c: Likewise.
14580 * config/sol2.c: Likewise.
14581 * config/sparc/sparc.c: Likewise.
14582 * config/spu/spu.c: Likewise.
14583 * config/stormy16/stormy16.c: Likewise.
14584 * config/tilegx/tilegx.c: Likewise.
14585 * config/tilepro/tilepro.c: Likewise.
14586 * config/v850/v850.c: Likewise.
14587 * config/vax/vax.c: Likewise.
14588 * config/visium/visium.c: Likewise.
14589 * config/xtensa/xtensa.c: Likewise.
14591 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14594 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
14595 constraints since the -mupper-regs-* switches have been
14597 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
14598 into a vector from a double word element that was extracted from
14599 another vector, and eliminate extra XXPERMDI instructions.
14600 (vsx_concat_<mode>_2): Likewise.
14601 (vsx_concat_<mode>_3): Likewise.
14602 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
14603 concat to allow optimizing inserts from previous extracts.
14605 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
14607 * config/i386/i386.c (ix86_stack_protect_guard): Generate
14608 memory reference to a SSP offset in TLS address space.
14609 (ix86_print_operand) <case '@'>: Remove.
14610 (ix86_print_operand_punct_valid_p): Remove '@' code.
14611 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
14612 UNSPEC_SP_TLS_TEST.
14613 (stack_tls_protect_set_<mode>): Remove.
14614 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
14615 (stack_tls_protect_test_<mode>): Remove.
14616 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
14618 2017-08-07 Olivier Hainque <hainque@adacore.com>
14621 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
14623 2017-08-07 Douglas Rupp <rupp@adacore.com>
14625 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
14626 variable was referenced as multidir in command.
14628 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14631 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
14634 2017-08-07 Martin Liska <mliska@suse.cz>
14636 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
14637 * config/rl78/rl78.c: Add include of attribs.h.
14638 * config/sh/sh.c: Likewise.
14639 * config/v850/v850.c: Likewise.
14641 2017-08-07 Tom de Vries <tom@codesourcery.com>
14643 PR middle-end/78266
14644 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
14646 2017-08-07 Martin Liska <mliska@suse.cz>
14648 * config/mips/mips.c: Include attribs.h.
14650 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
14653 * doc/invoke.texi: Document change in behvaior for -Ofast for
14656 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
14658 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
14660 (aarch64_pop_regs): Likewise.
14661 (aarch64_gen_load_pair): Likewise.
14662 (aarch64_save_callee_saves): Likewise.
14663 (aarch64_restore_callee_saves): Likewise.
14665 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14667 * config/i386/i386.c: Revert the last change.
14669 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14672 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14674 (ix86_finalize_stack_frame_flags): This. Also clear
14675 frame_pointer_needed if -fno-omit-frame-pointer is used without
14677 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14678 with ix86_finalize_stack_frame_flags.
14679 (ix86_expand_epilogue): Likewise.
14680 (ix86_expand_split_stack_prologue): Likewise.
14682 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14685 * config/i386/i386.c (get_builtin_code_for_version): Set priority
14686 to P_AES for Westmere.
14688 2017-08-07 Jonathan Yong <10walls@gmail.com>
14690 * config/i386/mingw.opt (fset-stack-executable): Removed.
14691 * config/i386/cygming.opt (fset-stack-executable): Moved
14693 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
14695 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
14697 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
14699 2017-08-07 Marek Polacek <polacek@redhat.com>
14701 PR middle-end/81737
14702 * fold-const.c (fold_indirect_ref_1): Check type_domain.
14704 2017-08-07 Martin Liska <mliska@suse.cz>
14706 * attribs.h (canonicalize_attr_name): New function.
14707 (cmp_attribs): Move from c-format.c and adjusted.
14708 (is_attribute_p): Moved from tree.h.
14709 * tree-inline.c: Add new includes.
14710 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
14711 (private_is_attribute_p): Remove.
14712 (private_lookup_attribute): Likewise.
14713 (private_lookup_attribute_by_prefix): Simplify.
14714 (remove_attribute): Use is_attribute_p.
14715 * tree.h: Remove removed declarations.
14717 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14719 PR middle-end/81698
14720 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
14721 instead of computing it in the function. Formatting fix.
14722 (expand_case): Don't rely on default_edge being the first edge,
14723 clear it if removing it, pass default_edge to
14724 emit_case_dispatch_table.
14725 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
14728 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
14730 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
14731 insn in the function, emit NOP after the insn.
14733 2017-08-06 Tom de Vries <tom@codesourcery.com>
14735 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
14738 2017-08-06 Tom de Vries <tom@codesourcery.com>
14740 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
14743 2017-08-04 Yury Gribov <tetra2005@gmail.com>
14745 PR tree-optimization/57371
14746 * match.pd: New pattern.
14748 2017-08-04 Marek Polacek <polacek@redhat.com>
14750 PR middle-end/81695
14751 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
14752 perform the computation in offset_int.
14754 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14756 PR tree-optimization/81136
14757 * tree-vectorizer.h: Include tree-hash-traits.h.
14758 (vec_base_alignments): New typedef.
14759 (vec_info): Add a base_alignments field.
14760 (vect_record_base_alignments): Declare.
14761 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
14763 (DR_IS_CONDITIONAL_IN_STMT): New macro.
14764 (create_data_ref): Add an is_conditional_in_stmt argument.
14765 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
14766 the is_conditional_in_stmt field.
14767 (data_ref_loc): Add an is_conditional_in_stmt field.
14768 (get_references_in_stmt): Set the is_conditional_in_stmt field.
14769 (find_data_references_in_stmt): Update call to create_data_ref.
14770 (graphite_find_data_references_in_stmt): Likewise.
14771 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
14772 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
14773 (vect_record_base_alignment): New function.
14774 (vect_record_base_alignments): Likewise.
14775 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
14776 for nested statements even if we fail to compute a misalignment.
14777 Use pooled base alignments for unconditional references.
14778 (vect_find_same_alignment_drs): Compare base addresses instead
14780 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
14781 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
14783 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14785 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
14786 Add an explicit name for the enum. Use auto_vec for slp_instances
14787 and grouped_stores.
14788 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
14790 (_bb_vec_info): Add a constructor and destructor.
14791 (vinfo_for_stmt): Return NULL for uids of -1 as well.
14792 (destroy_loop_vec_info): Delete.
14793 (vect_destroy_datarefs): Likewise.
14794 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
14795 (vec_info::vec_info): New function.
14796 (vec_info::~vec_info): Likewise.
14797 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
14798 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
14799 destroy_loop_vec_info.
14800 * tree-vect-loop.c (new_loop_vec_info): Replace with...
14801 (_loop_vec_info::_loop_vec_info): ...this.
14802 (destroy_loop_vec_info): Replace with...
14803 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
14804 the stmt_vec_infos. Leave handling of vec_info information to its
14805 destructor. Remove explicit vector releases.
14806 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
14807 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
14808 * tree-vect-slp.c (new_bb_vec_info): Replace with...
14809 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
14810 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
14811 (destroy_bb_vec_info): Replace with...
14812 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
14813 information to its destructor.
14814 (vect_slp_analyze_bb_1): Use new and delete instead of
14815 new_bb_vec_info and destroy_bb_vec_info.
14816 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
14819 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14821 * tree-data-ref.h (subscript): Add access_fn field.
14822 (data_dependence_relation): Add could_be_independent_p.
14823 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
14824 (same_access_functions): Move to tree-data-ref.c.
14825 * tree-data-ref.c (ref_contains_union_access_p): New function.
14826 (access_fn_component_p): Likewise.
14827 (access_fn_components_comparable_p): Likewise.
14828 (dr_analyze_indices): Add a reference to access_fn_component_p.
14829 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
14831 (constant_access_functions): Likewise.
14832 (add_other_self_distances): Likewise.
14833 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
14834 (initialize_data_dependence_relation): Use XCNEW and remove
14835 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
14836 of access functions that have the same type. Allow the
14837 subsequence to end with different bases in some circumstances.
14838 Record the chosen access functions in SUB_ACCESS_FN.
14839 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
14840 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
14841 (subscript_dependence_tester_1): Likewise dra and drb.
14842 (build_classic_dist_vector): Update calls accordingly.
14843 (subscript_dependence_tester): Likewise.
14844 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
14845 DDR_COULD_BE_INDEPENDENT_P.
14846 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
14847 comp_alias_ddrs instead of may_alias_ddrs.
14848 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
14850 (vect_analyze_data_ref_dependence): Use it if
14851 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
14852 distance vectors if that fails.
14853 (dependence_distance_ge_vf): New function.
14854 (vect_prune_runtime_alias_test_list): Use it. Don't clear
14855 LOOP_VINFO_MAY_ALIAS_DDRS.
14857 2017-08-04 Richard Biener <rguenther@suse.de>
14859 PR middle-end/81705
14860 * fold-const.c (fold_binary_loc): Properly restrict
14861 minus_var0 && minus_var1 case when associating undefined overflow
14864 2017-08-04 Tom de Vries <tom@codesourcery.com>
14866 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
14868 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14870 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14871 Don't start diagnostic messages with a capital letter.
14872 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14874 (rs6000_invalid_builtin): Likewise.
14875 (rs6000_trampoline_init): Likewise.
14877 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14880 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
14881 after setting changeable df flags.
14883 2017-08-03 Richard Biener <rguenther@suse.de>
14885 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
14886 up if the use is in USE - X.
14888 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
14890 * toplev.c (dumpfile.h): New include.
14891 (internal_error_reentered): New static function. Use it...
14892 (internal_error_function): ...here to handle reentered internal_error.
14894 2017-08-03 Richard Biener <rguenther@suse.de>
14896 PR middle-end/81148
14897 * fold-const.c (split_tree): Add minus_var and minus_con
14898 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
14899 here but always use minus_*.
14900 (associate_trees): Assert we never associate with MINUS_EXPR
14901 and NULL first operand. Do not recurse for PLUS_EXPR operands
14902 when associating as MINUS_EXPR either.
14903 (fold_binary_loc): Track minus_var and minus_con.
14905 2017-08-03 Tom de Vries <tom@codesourcery.com>
14908 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
14909 ACCEL_COMPILER, apply finish_options on
14910 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
14912 2017-08-03 Tom de Vries <tom@codesourcery.com>
14915 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
14916 function_entry_patch_area_size > 0.
14918 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14921 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
14922 instead of 10??LU, perform unit multiplication in wide_int,
14923 don't change alloc_object_size_limit if the limit is larger
14926 PR tree-optimization/81655
14927 PR tree-optimization/81588
14928 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
14929 the case when ranges[i].low and high are 1 for unsigned type with
14932 PR middle-end/81052
14933 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
14934 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
14936 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14938 * tree-vrp.h: Add include guard.
14940 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
14943 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
14944 (ud2): New insn pattern.
14945 * config/i386/i386.c (ix86_expand_epilogue):
14946 For naked functions, generate ud2 instead of trap insn.
14948 2017-08-02 Marek Polacek <polacek@redhat.com>
14951 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
14953 2017-08-02 Tom de Vries <tom@codesourcery.com>
14954 Cesar Philippidis <cesar@codesourcery.com>
14956 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
14957 Add missing edge probabilities.
14959 2017-08-02 Tamar Christina <tamar.christina@arm.com>
14961 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
14962 Correct endianness.
14964 2017-08-02 Jakub Jelinek <jakub@redhat.com>
14966 PR middle-end/79499
14967 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
14968 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
14969 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
14970 prologue_seq sequences - if any.
14972 2017-08-02 Richard Biener <rguenther@suse.de>
14974 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
14975 via vectors if supported, integer extracts via punning if supported
14976 or otherwise vector extracts.
14978 2017-08-02 Richard Biener <rguenther@suse.de>
14980 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
14982 (bitmap_insert_into_set): ... this.
14984 2017-08-02 Richard Biener <rguenther@suse.de>
14986 PR tree-optimization/81633
14988 2015-08-17 Alan Hayward <alan.hayward@arm.com>
14990 PR tree-optimization/71752
14991 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
14993 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
14995 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
14996 (machine_function::call_ms2sysv_pad_out): Remove field.
14997 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
14998 (ix86_compute_frame_layout): Likewise.
15000 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15003 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15004 attribute with interrupt attribute.
15006 2017-08-01 Andrew Pinski <apinski@cavium.com>
15008 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15009 BIT_INSERT_EXPR's operand 1
15010 to see if the types precision matches.
15012 2017-08-01 Martin Liska <mliska@suse.cz>
15014 PR middle-end/70140
15015 * builtins.c (expand_builtin_memcpy_args): Remove.
15016 (expand_builtin_memcpy): Call newly added function
15017 expand_builtin_memory_copy_args.
15018 (expand_builtin_memcpy_with_bounds): Likewise.
15019 (expand_builtin_mempcpy): Remove last argument.
15020 (expand_builtin_mempcpy_with_bounds): Likewise.
15021 (expand_builtin_memory_copy_args): New function created from
15022 expand_builtin_mempcpy_args with small modifications.
15023 (expand_builtin_mempcpy_args): Remove.
15024 (expand_builtin_stpcpy): Remove unused argument.
15025 (expand_builtin): Likewise.
15026 (expand_builtin_with_bounds): Likewise.
15028 2017-08-01 Martin Liska <mliska@suse.cz>
15031 Make mempcpy more optimal (PR middle-end/70140).
15033 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15036 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15037 __builtin_vec_cmpne verify both arguments are compatible vectors
15038 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15039 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15040 move computation of aligned to after checking the argument types.
15044 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15045 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15048 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15049 Doug Rupp <rupp@adacore.com>
15050 Olivier Hainque <hainque@adacore.com>
15052 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15053 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15055 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15056 for TARGET_OS_CPP_BUILTIN.
15057 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15058 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15060 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15061 passing required abi options to the assembler for EABI configurations.
15062 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15063 of .text.hot and .text.unlikely sections for kernel modules when
15064 using ARM style exceptions.
15065 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15066 options. Add EXTRA_CC1_SPEC.
15067 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15069 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15071 (ARM_TARGET2_DWARF_FORMAT): Define.
15072 * config/arm/t-vxworks: Adjust multilib control to removal of the
15073 Diab command line options.
15075 2017-08-01 Martin Liska <mliska@suse.cz>
15077 PR gcov-profile/81561
15078 * gcov.c (unblock): Make unblocking safe as we need to preserve
15079 index correspondence of blocks and block_lists.
15081 2017-08-01 Richard Biener <rguenther@suse.de>
15083 PR tree-optimization/81181
15084 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15085 (compute_antic): ... end of iteration here.
15087 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15089 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15090 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15091 (ftree-slp-vectorize): Likewise.
15092 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15093 can no longer be set independent of flag_tree_loop_vectorize.
15094 * omp-general.c (emp_max_vf): Likewise.
15095 * opts.c (enable_fdo_optimizations): Remove references to
15096 flag_tree_vectorize, these are now implicit.
15097 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15098 and leave it for the options machinery.
15100 2017-08-01 Martin Liska <mliska@suse.cz>
15102 PR middle-end/70140
15103 * builtins.c (expand_builtin_memcpy_args): Remove.
15104 (expand_builtin_memcpy): Call newly added function
15105 expand_builtin_memory_copy_args.
15106 (expand_builtin_memcpy_with_bounds): Likewise.
15107 (expand_builtin_mempcpy): Remove last argument.
15108 (expand_builtin_mempcpy_with_bounds): Likewise.
15109 (expand_builtin_memory_copy_args): New function created from
15110 expand_builtin_mempcpy_args with small modifications.
15111 (expand_builtin_mempcpy_args): Remove.
15112 (expand_builtin_stpcpy): Remove unused argument.
15113 (expand_builtin): Likewise.
15114 (expand_builtin_with_bounds): Likewise.
15116 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15119 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15120 print "ds:" only for immediates in generic address space.
15122 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15125 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15126 (ix86_function_ok_for_sibcall): Return false for naked functions.
15128 2017-08-01 Richard Biener <rguenther@suse.de>
15130 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15131 (compute_antic): Seed worklist with exit block predecessors.
15132 * cfganal.c (dfs_find_deadend): For a cycle return the source
15133 of the edge closing it.
15135 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15137 * config/aarch64/aarch64.c
15138 (aarch64_can_const_movi_rtx_p): Move 0 check.
15140 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15142 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15143 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15145 * match.pd: Ditto in address comparison pattern.
15147 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15149 PR tree-optimization/81627
15150 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15151 closed ssa form for store-store chain.
15153 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15155 PR tree-optimization/81620
15156 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15157 for store-store chain.
15159 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15161 PR tree-optimization/81588
15162 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15163 ranges[i].in_p, invert comparison code ccode. For >/>=,
15164 swap rhs1 and rhs2 and comparison code unconditionally,
15165 for </<= don't do that. Don't swap rhs1/rhs2 again if
15166 ranges[i].in_p, instead invert comparison code ccode if
15167 opcode or oe->rank is BIT_IOR_EXPR.
15170 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15171 a direct optab to conversion optab.
15172 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15173 with GET_MODE_INNER as last argument instead of optab_handler.
15174 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15175 vector extraction if possible and optab is available.
15176 * expr.c (store_constructor): Use convert_optab_handler instead
15177 of optab_handler. Use vector initialization from smaller
15178 vectors if possible and optab is available.
15179 * tree-vect-stmts.c (vectorizable_load): Likewise.
15180 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15181 now have two modes.
15182 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15183 of vec_init from half-sized vectors with the same element mode.
15184 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15185 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15186 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15187 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15188 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15189 after mode in gen_vec_extract* calls.
15190 (vec_extract<mode>): Renamed to ...
15191 (vec_extract<mode><ssescalarmodelower>): ... this.
15192 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15193 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15194 element mode after mode in gen_vec_init* calls.
15195 (VEC_INIT_HALF_MODE): New mode iterator.
15196 (vec_init<mode>): Renamed to ...
15197 (vec_init<mode><ssescalarmodelower>): ... this.
15198 (vec_init<mode><ssehalfvecmodelower>): New expander.
15199 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15200 (vec_extractv2sfsf): ... this.
15201 (vec_initv2sf): Renamed to ...
15202 (vec_initv2sfsf): ... this.
15203 (vec_extractv2si): Renamed to ...
15204 (vec_extractv2sisi): ... this.
15205 (vec_initv2si): Renamed to ...
15206 (vec_initv2sisi): ... this.
15207 (vec_extractv4hi): Renamed to ...
15208 (vec_extractv4hihi): ... this.
15209 (vec_initv4hi): Renamed to ...
15210 (vec_initv4hihi): ... this.
15211 (vec_extractv8qi): Renamed to ...
15212 (vec_extractv8qiqi): ... this.
15213 (vec_initv8qi): Renamed to ...
15214 (vec_initv8qiqi): ... this.
15215 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15216 (vec_init<mode>): Renamed to ...
15217 (vec_init<mode><VEC_base_l>): ... this.
15218 (vec_extract<mode>): Renamed to ...
15219 (vec_extract<mode><VEC_base_l>): ... this.
15220 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15221 (vec_initv2sfsf): ... this.
15222 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15223 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15224 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15225 element mode after mode in gen_vec_init* calls.
15226 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15227 (vec_init<mode><Vel>): ... this.
15228 (vec_extract<mode>): Renamed to ...
15229 (vec_extract<mode><Vel>): ... this.
15230 * config/aarch64/iterators.md (Vel): New mode attribute.
15231 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15232 Add element mode after mode in gen_vec_extract* calls.
15233 * config/s390/vector.md (non_vec_l): New mode attribute.
15234 (vec_extract<mode>): Renamed to ...
15235 (vec_extract<mode><non_vec_l>): ... this.
15236 (vec_init<mode>): Renamed to ...
15237 (vec_init<mode><non_vec_l>): ... this.
15238 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15239 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15241 * config/arm/iterators.md (V_elem_l): New mode attribute.
15242 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15243 (vec_extract<mode><V_elem_l>): ... this.
15244 (vec_extractv2di): Renamed to ...
15245 (vec_extractv2didi): ... this.
15246 (vec_init<mode>): Renamed to ...
15247 (vec_init<mode><V_elem_l>): ... this.
15248 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15249 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15250 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15251 Add element mode after gen_vec_extract* calls.
15252 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15253 (vec_init<mode><unitmode>): ... this.
15254 (vec_extract<mode>): Renamed to ...
15255 (vec_extract<mode><unitmode>): ... this.
15256 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15257 (vec_init<mode><unitmode>): ... this.
15258 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15259 (vec_initv2sfsf): ... this.
15260 (vec_extractv2sf): Renamed to ...
15261 (vec_extractv2sfsf): ... this.
15262 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15263 Add element mode after gen_vec_extract* calls.
15264 * config/mips/mips.md (unitmode): New mode iterator.
15265 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15266 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15267 * config/spu/spu.md (inner_l): New mode attribute.
15268 (vec_init<mode>): Renamed to ...
15269 (vec_init<mode><inner_l>): ... this.
15270 (vec_extract<mode>): Renamed to ...
15271 (vec_extract<mode><inner_l>): ... this.
15272 * config/sparc/sparc.md (veltmode): New mode iterator.
15273 (vec_init<VMALL:mode>): Renamed to ...
15274 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15275 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15276 (vec_initv2sisi): ... this.
15277 (vec_initv2sf): Renamed to ...
15278 (vec_initv2sfsf): ... this.
15279 (vec_extractv2sf): Renamed to ...
15280 (vec_extractv2sfsf): ... this.
15281 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15282 (vec_init<mode>): Renamed to ...
15283 (vec_init<mode><VEC_base_l>): ... this.
15284 (vec_extract<mode>): Renamed to ...
15285 (vec_extract<mode><VEC_base_l>): ... this.
15286 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15287 (vec_initv2sfsf): ... this.
15288 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15289 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15290 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15291 gen_vec_init* calls.
15293 2017-08-01 Richard Biener <rguenther@suse.de>
15295 PR tree-optimization/81297
15296 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15297 TREE_OVERFLOW from INTEGER_CSTs.
15299 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15301 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15303 2017-07-31 Carl Love <cel@us.ibm.com>
15305 * config/rs6000/rs6000-c: Add support for built-in functions
15306 vector signed char vec_xl_be (signed long long, signed char *);
15307 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15308 vector signed int vec_xl_be (signed long long, signed int *);
15309 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15310 vector signed long long vec_xl_be (signed long long, signed long long *);
15311 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15312 vector signed short vec_xl_be (signed long long, signed short *);
15313 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15314 vector double vec_xl_be (signed long long, double *);
15315 vector float vec_xl_be (signed long long, float *);
15316 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15317 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15318 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15320 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15321 (altivec_expand_builtin): Add switch statement to call
15322 altivec_expand_xl_be for each builtin.
15323 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15324 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15325 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15326 __builtin_vsx_le_be_v16qi.
15327 * doc/extend.texi: Update the built-in documentation file for the
15328 new built-in functions.
15330 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15333 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15335 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15337 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15339 * config.gcc: Add z14.
15340 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15341 CPU model numbers for z13s and z14.
15342 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
15344 * config/s390/s390-opts.h (enum processor_type): Rename
15345 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15346 * config/s390/s390.c (processor_table): Add field for CPU name to
15347 be passed to Binutils.
15348 (s390_asm_output_machine_for_arch): Use the new field in
15349 processor_table for Binutils.
15350 (s390_expand_builtin): Replace arch12 with z14.
15351 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15352 (s390_get_sched_attrmask): Likewise.
15353 (s390_get_unit_mask): Likewise.
15354 * config/s390/s390.opt: Add z14 to processor_type enum.
15356 2017-07-31 Martin Jambor <mjambor@suse.cz>
15359 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
15360 regardless of optimization level.
15362 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
15363 Martin Liska <mliska@suse.cz>
15365 * predict.def: Remove old comment and adjust probability.
15366 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15367 PREDICT statements.
15369 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15372 * config/i386/i386.c (ix86_function_naked): New function.
15373 (ix86_can_use_return_insn_p): Return false for naked functions.
15374 (ix86_expand_prologue): Skip prologue for naked functions.
15375 (ix86_expand_epilogue): Skip epilogue for naked functions
15376 and emit trap instruction.
15377 (ix86_warn_func_return): New function.
15378 (ix86_attribute_table): Add "naked" attribute specification.
15379 (TARGET_WARN_FUNC_RETURN): Define.
15380 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
15382 2017-07-31 Martin Liska <mliska@suse.cz>
15384 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
15385 (dump_gimple_bb_header): Always dump BB info.
15386 (pp_cfg_jump): Do not append info about BB when dumping a jump.
15388 2017-07-31 Martin Liska <mliska@suse.cz>
15391 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
15392 also with current_function_decl non-null equality.
15394 2017-07-31 Jakub Jelinek <jakub@redhat.com>
15397 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
15398 change type to the element type, instead add eltype variable and
15399 use it where we are interested in the element type.
15401 PR tree-optimization/81603
15402 * ipa-polymorphic-call.c
15403 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
15404 offset arithmetic in offset_int, bail out if the resulting bit offset
15405 doesn't fit into shwi.
15407 2017-07-31 Martin Liska <mliska@suse.cz>
15409 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
15410 (gimplify_save_expr): Fix comment.
15412 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
15415 * config/i386/i386.c (ix86_function_arg): Update arguments for
15417 (ix86_compute_frame_layout): Set the initial stack offset to
15418 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
15419 INCOMING_FRAME_SP_OFFSET.
15420 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
15421 stack before exception handler returns.
15422 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
15423 the 'ERROR_CODE' for exception handler.
15425 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
15427 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
15428 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
15429 (ASM_OUTPUT_REG_POP): Ditto.
15430 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
15431 instead of asm_fprintf to output pure string.
15433 2017-07-29 Jakub Jelinek <jakub@redhat.com>
15435 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
15436 to imported_module_or_decl hook.
15437 (debug_nothing_tree_tree_tree_bool): Remove.
15438 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
15439 * debug.c (do_nothing_debug_hooks): Use
15440 debug_nothing_tree_tree_tree_bool_bool instead of
15441 debug_nothing_tree_tree_tree_bool.
15442 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15443 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
15444 * sdbout.c (sdb_debug_hooks): Likewise.
15445 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
15446 (gen_namespace_die): Add DW_AT_export_symbols attribute if
15448 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
15449 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
15450 attribute, don't add anything.
15452 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15454 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
15455 (fold_build2_stat_loc): Likewise.
15456 (fold_build3_stat_loc): Likewise.
15457 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
15458 (fold_build1_loc): Remove macro.
15459 (fold_build2_loc): Likewise.
15460 (fold_build3_loc): Likewise.
15462 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15464 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
15465 (gimple_build_debug_bind_source_stat): Likewise.
15466 * gimple.h (gimple_build_debug_bind): Remove macro.
15467 (gimple_build_debug_bind_source): Likewise.
15469 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15471 * bitmap.c (bitmap_alloc): Adjust.
15472 (bitmap_gc_alloc): Likewise.
15473 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
15475 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15477 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
15478 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
15479 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
15480 (bitmap_gc_alloc_stat): Likewise.
15481 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
15483 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15485 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
15486 * rtl.h (shallow_copy_rtx): Remove macro.
15488 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15490 * emit-rtl.c (gen_raw_REG): Adjust.
15491 * gengenrtl.c (gendef): Likewise.
15492 * rtl.c (rtx_alloc_stat): Remove _stat from name.
15493 * rtl.h (rtx_alloc): Remove macro.
15495 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15497 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
15498 (build_tree_list_stat): Likewise.
15499 * tree.h (build_tree_list): Remove macro.
15500 (build_tree_list_vec): Likewise.
15502 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15504 * tree.c (make_vector_stat): Remove _stat from name.
15505 (build_vector_stat): Likewise.
15506 * tree.h (make_vector_stat): Remove macro.
15507 (build_vector_stat): Likewise.
15509 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15511 * tree.h (build_var_debug_value): Remove prototype.
15513 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15515 * tree.c (tree_cons_stat): Remove _stat from name.
15516 * tree.h (tree_cons): Remove macro.
15518 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15520 * tree.c (build_vl_exp_stat): Remove _stat from name.
15521 * tree.h (build_vl_exp): Remove macro.
15523 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15525 * tree.c (build_decl_stat): Remove _stat from name.
15526 * tree.h (build_decl): Remove macro.
15528 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15530 * gimple.c (gimple_build_with_ops_stat): Adjust.
15531 (gimple_alloc_stat): Remove _stat from name.
15532 * gimple.h (gimple_alloc): Remove macro.
15534 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15536 * tree.c (make_tree_vec_stat): Remove _stat from name.
15537 (grow_tree_vec_stat): Likewise.
15538 * tree.h (make_tree_vec_stat): Adjust prototype.
15539 (grow_tree_vec_stat): Likewise.
15540 (make_tree_vec): Remove macro.
15541 (grow_tree_vec): Likewise.
15543 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15545 * fold-const.c (fold_build1_stat_loc): Adjust.
15546 (fold_build2_stat_loc): Likewise.
15547 (fold_build3_stat_loc): Likewise.
15548 * tree.c (build0_stat): Remove _stat from name.
15549 (build1_stat): Likewise.
15550 (build2_stat): Likewise.
15551 (build3_stat): Likewise.
15552 (build4_stat): Likewise.
15553 (build5_stat): Likewise.
15554 * tree.h (build1_loc): Remove macro, and rename _stat function
15556 (build2_loc): Likewise.
15557 (build3_loc): Likewise.
15558 (build4_loc): Likewise.
15559 (build5_loc): Likewise.
15561 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15563 * tree.c (make_int_cst_stat): Remove _stat from name.
15564 * tree.h (make_int_cst_stat): Adjust prototype.
15565 (make_int_cst): Remove macro.
15567 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15569 * tree.c (make_tre_binfo_stat): Remove _stat from name.
15570 * tree.h (make_tree_binfo_stat): Adjust prototype.
15571 (make_tree_binfo): Remove.
15573 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15575 * tree.c (copy_node_stat): Rename to copy_node.
15576 (build_distinct_type_copy): Adjust.
15577 * tree.h (copy_node_stat): Adjust prototype.
15578 (copy_node): Remove macro.
15580 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15582 * tree.c (make_node_stat): rename to make_node.
15583 (build_tree_list_stat): Adjust.
15584 (build0_stat): Likewise.
15585 (build2_stat): Likewise.
15586 (build3_stat): Likewise.
15587 (build4_stat): Likewise.
15588 (build5_stat): Likewise.
15589 (build_decl_stat): Likewise.
15590 * tree.h (make_node_stat): Adjust prototype.
15591 (make_node): remove macro.
15593 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
15595 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
15596 (PPC_FEATURE2_SCV): Likewise.
15597 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
15599 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15601 * config/aarch64/aarch64.c
15602 (aarch64_internal_mov_immediate): Add new special pattern.
15603 * config/aarch64/aarch64.md (*movdi_aarch64):
15604 Add reg/32bit const mov case.
15606 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15607 Richard Sandiford <richard.sandiford@linaro.org>
15609 * config/aarch64/aarch64.md (mov<mode>): Generalize.
15610 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15611 Add integer and movi cases.
15612 (movi-split-hf-df-sf split, fp16): New.
15613 (enabled): Added TARGET_FP_F16INST.
15614 * config/aarch64/iterators.md (GPF_HF): New.
15615 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
15617 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15619 * config/aarch64/aarch64.c
15620 (aarch64_simd_container_mode): Add prototype.
15621 (aarch64_expand_mov_immediate): Add HI support.
15622 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
15623 (aarch64_can_const_movi_rtx_p): New.
15624 (aarch64_preferred_reload_class):
15625 Remove restrictions of using FP registers for certain SIMD operations.
15626 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
15627 (aarch64_valid_floating_const): Add integer move validation.
15628 (aarch64_simd_imm_scalar_p): Remove.
15629 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
15630 (aarch64_legitimate_constant_p): Expand list of supported cases.
15631 * config/aarch64/aarch64-protos.h
15632 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
15633 (aarch64_reinterpret_float_as_int): New.
15634 (aarch64_simd_imm_scalar_p): Remove.
15635 * config/aarch64/constraints.md (Uvi): New.
15636 (Dd): Split into Ds and new Dd.
15637 * config/aarch64/aarch64.md (*movsi_aarch64):
15639 (*movdi_aarch64): Add SIMD mov case.
15641 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15643 * tree-predcom.c: (struct chain): Handle store-store chain in which
15644 stores for elimination only store loop invariant values.
15645 (execute_pred_commoning_chain): Ditto.
15646 (prepare_initializers_chain_store_elim): Ditto.
15647 (prepare_finalizers): Ditto.
15648 (is_inv_store_elimination_chain): New function.
15649 (initialize_root_vars_store_elim_1): New function.
15651 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15653 * tree-predcom.c: Revise general description of the pass.
15654 (enum chain_type): New enum type for store elimination.
15655 (struct chain): New field supporting store elimination.
15656 (struct component): Ditto.
15657 (dump_chain): Dump store-stores chain.
15658 (release_chain): Release resources.
15659 (split_data_refs_to_components): Compute and create component
15660 contains only stores for elimination.
15661 (get_chain_last_ref_at): New function.
15662 (make_invariant_chain): Initialization.
15663 (make_rooted_chain): Specify chain type in parameter and record it.
15664 (add_looparound_copies): Skip for store-stores chain.
15665 (determine_roots_comp): Compute type of chain and pass it to
15667 (initialize_root_vars_store_elim_2): New function.
15668 (finalize_eliminated_stores): New function.
15669 (remove_stmt): Handle store for elimination.
15670 (execute_pred_commoning_chain): Execute predictive commoning on
15671 store-store chains.
15672 (determine_unroll_factor): Skip unroll for store-stores chain.
15673 (prepare_initializers_chain_store_elim): New function.
15674 (prepare_initializers_chain): Hanlde store-store chain.
15675 (prepare_finalizers_chain, prepare_finalizers): New function.
15676 (tree_predictive_commoning_loop): Return integer value indicating
15677 if loop is unrolled or lcssa form is corrupted.
15678 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
15680 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15682 * tree-predcom.c (initialize_root): Delete.
15683 (execute_pred_commoning_chain): Initialize root vars and replace
15684 reference of non-combined chain directly, rather than call above
15687 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15689 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
15690 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
15692 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15694 * tree-predcom.c (struct chain): New field init_seq.
15695 (release_chain): Release init_seq.
15696 (prepare_initializers_chain): Record intialization stmts in above
15698 (insert_init_seqs): New function.
15699 (tree_predictive_commoning_loop): Call insert_init_seqs.
15701 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15703 * tree-predcom.c (determine_roots_comp): Skip trivial components.
15705 2017-07-28 Richard Biener <rguenther@suse.de>
15707 * match.pd: Remove superfluous :c.
15708 * genmatch.c (simplify::id): Add member.
15709 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
15711 (current_id): New global.
15712 (dt_node::parent): Move from ...
15713 (dt_operand::parent): ... here. Add for_id member.
15714 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
15715 (decision_tree::find_node): Relax order requirement when
15716 merging DT_TRUE nodes to ones inbetween the current simplify
15717 and the one we try to merge with. Add diagnostic whenever
15718 we need to enforce pattern order by not merging.
15719 (decision_tree::insert): Set current_id.
15720 (decision_tree::print_node): Dump parent node and for_id.
15721 (parser::last_id): Add member.
15722 (parser::push_simplify): Assign unique id.
15723 (parser::parser): Initialize last_id.
15725 2017-07-28 Martin Liska <mliska@suse.cz>
15728 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
15729 gimple_build_debug_bind.
15731 2017-07-28 Richard Biener <rguenther@suse.de>
15733 PR tree-optimization/81502
15734 * match.pd: Add pattern combining BIT_INSERT_EXPR with
15736 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
15738 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
15739 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
15740 for BIT_FIELD_REF args.
15741 * fold-const.c (make_bit_field_ref): Likewise.
15742 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
15744 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15747 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
15748 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
15749 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
15750 Or it into SANITIZER_UNDEFINED.
15751 * ubsan.c: Include gimple-fold.h and varasm.h.
15752 (ubsan_expand_ptr_ifn): New function.
15753 (instrument_pointer_overflow): New function.
15754 (maybe_instrument_pointer_overflow): New function.
15755 (instrument_object_size): Formatting fix.
15756 (pass_ubsan::execute): Call instrument_pointer_overflow
15757 and maybe_instrument_pointer_overflow.
15758 * internal-fn.c (expand_UBSAN_PTR): New function.
15759 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
15760 * sanitizer.def (__ubsan_handle_pointer_overflow,
15761 __ubsan_handle_pointer_overflow_abort): New builtins.
15762 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
15763 * internal-fn.def (UBSAN_PTR): New internal function.
15764 * opts.c (sanitizer_opts): Add pointer-overflow.
15765 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
15766 * fold-const.c (build_range_check): Compute pointer range check in
15767 integral type if pointer arithmetics would be needed. Formatting
15770 2017-07-28 Martin Liska <mliska@suse.cz>
15773 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
15774 parameters that are of a variable-length.
15776 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
15778 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
15780 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
15781 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
15782 (CRT_CALL_STATIC_FUNCTION): Likewise.
15783 (ASM_DEFAULT_SPEC): New define.
15784 (ASM_SPEC32): Likewise.
15785 (ASM_SPEC64): Likewise.
15786 (ASM_SPEC_COMMON): Likewise.
15787 (ASM_SPEC): Likewise.
15788 (INVALID_64BIT): Likewise.
15789 (LINK_OS_DEFAULT_SPEC): Likewise.
15790 (LINK_OS_SPEC32): Likewise.
15791 (LINK_OS_SPEC64): Likewise.
15792 (POWERPC_LINUX): Likewise.
15793 (PTRDIFF_TYPE): Likewise.
15794 (RESTORE_FP_PREFIX): Likewise.
15795 (RESTORE_FP_SUFFIX): Likewise.
15796 (SAVE_FP_PREFIX): Likewise.
15797 (SAVE_FP_SUFFIX): Likewise.
15798 (SIZE_TYPE): Likewise.
15799 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
15800 (TARGET_64BIT): Likewise.
15801 (TARGET_64BIT): Likewise.
15802 (TARGET_AIX): Likewise.
15803 (WCHAR_TYPE_SIZE): Likewise.
15804 (WCHAR_TYPE): Undefine.
15805 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
15806 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
15807 (CPP_OS_RTEMS_SPEC): Delete.
15808 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
15809 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
15811 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
15813 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15815 PR tree-optimization/81578
15816 * tree-parloops.c (build_new_reduction): Bail out if
15817 reduction_code isn't one of the standard OpenMP reductions.
15818 Move the details printing after that decision.
15820 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15822 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
15823 related to reload_in_progress.
15824 (splat_input_operand): Likewise.
15825 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
15827 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
15829 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
15830 (TARGET_INSTANTIATE_DECLS): Likewise.
15831 (legitimate_indexed_address_p): Delete reload_in_progress code.
15832 (rs6000_debug_legitimate_address_p): Likewise.
15833 (rs6000_eliminate_indexed_memrefs): Likewise.
15834 (rs6000_emit_le_vsx_store): Likewise.
15835 (rs6000_emit_move_si_sf_subreg): Likewise.
15836 (rs6000_emit_move): Likewise.
15837 (register_to_reg_type): Likewise.
15838 (rs6000_pre_atomic_barrier): Likewise.
15839 (rs6000_machopic_legitimize_pic_address): Likewise.
15840 (rs6000_allocate_stack_temp): Likewise.
15841 (rs6000_address_for_fpconvert): Likewise.
15842 (rs6000_address_for_altivec): Likewise.
15843 (rs6000_secondary_memory_needed_rtx): Delete function.
15844 (rs6000_check_sdmode): Likewise.
15845 (rs6000_alloc_sdmode_stack_slot): Likewise.
15846 (rs6000_instantiate_decls): Likewise.
15847 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
15848 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
15849 Delete reload_in_progress.
15850 (*vec_reload_and_plus_<mptrsize>): Likewise.
15851 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
15852 (vsx_div_v2di): Likewise.
15853 (vsx_udiv_v2di): Likewise.
15855 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15857 * config/rs6000/rs6000.opt (mlra): Replace with stub.
15858 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
15859 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
15860 (rs6000_debug_reg_global): Delete print of LRA status.
15861 (rs6000_option_override_internal): Delete dead LRA related code.
15862 (rs6000_lra_p): Delete function.
15863 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
15865 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
15867 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
15868 * config/riscv/rtems.h: New file.
15870 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15871 Sudakshina Das <sudi.das@arm.com>
15873 * config/aarch64/aarch64.md
15874 (define_split for and<mode>3nr_compare): Move
15875 non aarch64_logical_operand to a register.
15876 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
15877 register immediate operand to a register.
15878 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
15880 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15882 PR middle-end/81564
15883 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
15885 2017-07-27 Richard Biener <rguenther@suse.de>
15887 PR tree-optimization/81573
15888 PR tree-optimization/81494
15889 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
15890 multi defuse cycle case.
15892 2017-07-27 Richard Biener <rguenther@suse.de>
15894 PR tree-optimization/81571
15895 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
15898 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
15900 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
15901 earlier and only if MASK_FPU is set. Adjust formatting.
15903 2017-07-27 Martin Liska <mliska@suse.cz>
15905 * opt-functions.awk: Add validation of value of Init.
15906 * optc-gen.awk: Pass new argument.
15908 2017-07-27 Martin Liska <mliska@suse.cz>
15910 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
15911 Fix wrong condition.
15913 2017-07-27 Martin Liska <mliska@suse.cz>
15915 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
15916 BBs and edges seen by autoFDO.
15918 2017-07-27 Richard Biener <rguenther@suse.de>
15920 PR tree-optimization/81502
15921 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
15922 with incompatible but same sized type.
15923 (execute_update_addresses_taken): Likewise.
15925 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
15927 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
15928 flag_tree_loop_vectorize rather than flag_tree_vectorize.
15930 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15933 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
15934 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
15935 Change s_operand to memory_operand.
15937 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
15939 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
15940 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
15941 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
15942 Emit instructions rather than returning an expression. Handle TFmode
15943 and KFmode by casting to TImode.
15944 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
15945 (rs6000_emit_le_vsx_store): Likewise.
15946 * config/rs6000/vsx.md (VSX_TI): New iterator.
15947 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
15948 (*vsx_le_undo_permute_<mode>): Likewise.
15949 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
15950 emit the split sequence.
15951 (*vsx_le_perm_store_<mode>): Likewise.
15953 2017-07-27 Jakub Jelinek <jakub@redhat.com>
15955 PR tree-optimization/81555
15956 PR tree-optimization/81556
15957 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
15958 if true, force CHANGED for the recursive invocation.
15959 (reassociate_bb): Remember original length of ops array, pass
15960 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
15962 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
15963 attributes for noipa attribute. For naked attribute use
15964 lookup_attribute first before lookup_attribute_spec.
15965 * final.c (rest_of_handle_final): Disable IPA RA for functions with
15967 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
15968 for functions with noipa attribute.
15969 (cgraph_externally_visible_p): Return true for functions with noipa
15971 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
15972 for functions with noipa attribute.
15973 * doc/extend.texi: Document noipa function attribute.
15974 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
15975 also for functions with noipa attribute.
15976 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
15978 2017-07-26 Andrew Pinski <apinski@cavium.com>
15980 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
15981 vec_unalign_load_cost and vec_unalign_store_cost.
15983 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
15985 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
15986 -mvsx-small-integer option.
15987 (ISA_3_0_MASKS_IEEE): Likewise.
15988 (OTHER_VSX_VECTOR_MASKS): Likewise.
15989 (POWERPC_MASKS): Likewise.
15990 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
15991 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
15992 code, only testing for DImode being allowed in non-VSX floating
15994 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
15995 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
15997 (rs6000_option_override_internal): Delete -mvsx-small-integer.
15998 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
15999 TARGET_P8_VECTOR test.
16000 (rs6000_secondary_reload_simple_move): Likewise.
16001 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16002 since TARGET_P9_VECTOR was already tested.
16003 (rs6000_opt_masks): Remove -mvsx-small-integer.
16004 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16005 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16007 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16008 test for TARGET_VEXTRACTUB was used, and that uses
16010 (p9 extract splitter): Likewise.
16011 (vsx_extract_<mode>_di_p9): Likewise.
16012 (vsx_extract_<mode>_store_p9): Likewise.
16013 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16014 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16015 the elimination of TARGET_VSX_SMALL_INTEGER.
16016 (vsx_extract_<mode>_p8): Likewise.
16017 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16018 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16019 (vsx_set_<mode>_p9): Likewise.
16020 (vsx_set_v4sf_p9): Likewise.
16021 (vsx_set_v4sf_p9_zero): Likewise.
16022 (vsx_insert_extract_v4sf_p9): Likewise.
16023 (vsx_insert_extract_v4sf_p9_2): Likewise.
16024 * config/rs6000/rs6000.md (sign extend splitter): Change
16025 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16026 (floatsi<mode>2_lfiwax_mem): Likewise.
16027 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16028 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16029 since a test for TARGET_P9_VECTOR was used.
16030 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16031 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16032 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16033 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16034 TARGET_P8_VECTOR test.
16035 (fix_trunc<mode>si2_stfiwx): Likewise.
16036 (fix_trunc<mode>si2_internal): Likewise.
16037 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16038 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16040 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16041 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16042 TARGET_P8_VECTOR test.
16043 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16044 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16045 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16047 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16048 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16049 since a test for TARGET_P9_VECTOR was used.
16050 (splitter for loading small constants): Likewise.
16052 2017-07-26 Andrew Pinski <apinski@cavium.com>
16054 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16057 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16060 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16061 (fp_valid_at): Likewise.
16063 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16065 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16066 (qdf24xx_addrcost_table): Likewise.
16067 (cortexa57_tunings): Update to use generic_branch_cost.
16068 (cortexa72_tunings): Likewise.
16069 (cortexa73_tunings): Likewise.
16070 (qdf24xx_tunings): Likewise.
16072 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16074 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16075 (thunderx2t99_branch_cost): Likewise.
16076 (cortexa35_tunings): Update to use generic_branch_cost.
16077 (cortexa53_tunings): Likewise.
16078 (cortexa57_tunings): Likewise.
16079 (cortexa72_tunings): Likewise.
16080 (cortexa73_tunings): Likewise.
16081 (thunderx2t99_tunings): Likewise.
16083 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16085 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16086 (sparc_option_override): Honour MASK_FSMULD.
16087 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16088 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16089 * config/sparc/sparc.opt (mfsmuld): New option.
16090 * doc/invoke.texi (mfsmuld): Document option.
16092 2017-07-26 Marek Polacek <polacek@redhat.com>
16094 PR middle-end/70992
16095 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16097 2017-07-26 Richard Biener <rguenther@suse.de>
16099 * gimple-match-head.c (do_valueize): Return OP if valueize
16101 (get_def): New helper to get at the def stmt of a SSA name
16102 if valueize allows.
16103 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16104 do_valueize to get at the def stmt.
16105 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16107 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16109 PR middle-end/46932
16110 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16112 2017-07-26 Martin Liska <mliska@suse.cz>
16115 * function.c (expand_function_start): Make expansion of
16116 nonlocal_goto_save_area after parm_birth_insn.
16118 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16120 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16121 from all CPU target flags enable members.
16123 2017-07-26 Richard Biener <rguenther@suse.de>
16125 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16126 (decision_tree::gen): Make 'type' const.
16127 (write_predicate): Likewise.
16129 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16131 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16132 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16133 (rs6000_option_override_internal): Likewise.
16134 (rs6000_expand_vector_set): Likewise.
16135 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16136 (TARGET_UPPER_REGS_SF): Likewise.
16137 (TARGET_UPPER_REGS_DI): Likewise.
16138 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16139 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16140 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16141 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16142 (Splitters for DI constants in Altivec registers): Likewise.
16143 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16144 (vsx_set_v4sf_p9): Likewise.
16145 (vsx_set_v4sf_p9_zero): Likewise.
16146 (vsx_insert_extract_v4sf_p9): Likewise.
16147 (vsx_insert_extract_v4sf_p9_2): Likewise.
16149 2017-07-25 Carl Love <cel@us.ibm.com>
16151 * doc/extend.texi: Update the built-in documentation file for the
16152 existing built-in functions
16153 vector signed char vec_cnttz (vector signed char);
16154 vector unsigned char vec_cnttz (vector unsigned char);
16155 vector signed short vec_cnttz (vector signed short);
16156 vector unsigned short vec_cnttz (vector unsigned short);
16157 vector signed int vec_cnttz (vector signed int);
16158 vector unsigned int vec_cnttz (vector unsigned int);
16159 vector signed long long vec_cnttz (vector signed long long);
16160 vector unsigned long long vec_cnttz (vector unsigned long long);
16162 2017-07-25 Andrew Pinski <apinski@cavium.com>
16164 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16165 accesses where the use is for the first operand of a BIT_INSERT.
16167 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16170 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16171 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16173 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16175 * config/i386/gstabs.h: Delete.
16176 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16178 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16180 * config/i386/i386.c (ix86_decompose_address): Do not check for
16181 register RTX when looking at index_reg or base_reg.
16182 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16184 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16186 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16187 to update EH info here.
16189 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16191 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16193 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16195 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16197 2017-07-25 Torsten Duwe <duwe@suse.de>
16199 * common.opt: Introduce -fpatchable-function-entry
16200 command line option, and its variables function_entry_patch_area_size
16201 and function_entry_patch_area_start.
16202 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16203 including a two-value parser.
16204 * target.def (print_patchable_function_entry): New target hook.
16205 * targhooks.h (default_print_patchable_function_entry): New function.
16206 * targhooks.c (default_print_patchable_function_entry): Likewise.
16207 * toplev.c (process_options): Switch off IPA-RA if
16208 patchable function entries are being generated.
16209 * varasm.c (assemble_start_function): Look at the
16210 patchable-function-entry command line switch and current
16211 function attributes and maybe generate NOP instructions by
16212 calling the print_patchable_function_entry hook.
16213 * doc/extend.texi: Document patchable_function_entry attribute.
16214 * doc/invoke.texi: Document -fpatchable_function_entry
16215 command line option.
16216 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16218 * doc/tm.texi: Re-generate.
16220 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16223 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16224 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16226 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16228 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16230 2017-07-25 Richard Biener <rguenther@suse.de>
16232 PR tree-optimization/81455
16233 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16234 not walk in cycles when looking for guards.
16236 2017-07-25 Richard Biener <rguenther@suse.de>
16238 PR tree-optimization/81529
16239 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16240 when optimizing backedge uses.
16242 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16244 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16246 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16247 to dl_section_ref. On AIX, append an expression to subtract
16248 the size of the section length to dl_section_ref.
16250 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16252 * configure.ac: If any of the config.* scripts fail, exit 1.
16253 * configure: Regenerate.
16255 2017-07-25 Richard Biener <rguenther@suse.de>
16257 PR middle-end/81546
16258 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16259 of immediate uses, be more verbose on errors.
16261 2017-07-25 Richard Biener <rguenther@suse.de>
16263 PR tree-optimization/81510
16264 * tree-vect-loop.c (vect_is_simple_reduction): When the
16265 reduction stmt is not inside the loop bail out.
16267 2017-07-25 Richard Biener <rguenther@suse.de>
16269 PR tree-optimization/81303
16270 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16271 profitability check against LOOP_VINFO_NITERSM1.
16273 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16275 * domwalk.c (cmp_bb_postorder): Simplify.
16276 (sort_bbs_postorder): New function. Use it...
16277 (dom_walker::walk): ...here to optimize common cases.
16279 2017-07-25 Martin Liska <mliska@suse.cz>
16282 * ipa-visibility.c (function_and_variable_visibility): Make the
16283 redirection just on target that supports aliasing.
16284 Fix GNU coding style.
16286 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16289 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16291 * config/arm/rtems.h: Likewise.
16292 * config/bfin/rtems.h: Likewise.
16293 * config/i386/rtemself.h: Likewise.
16294 * config/lm32/rtems.h: Likewise.
16295 * config/m32c/rtems.h: Likewise.
16296 * config/m68k/rtemself.h: Likewise.
16297 * config/microblaze/rtems.h: Likewise.
16298 * config/mips/rtems.h: Likewise.
16299 * config/moxie/rtems.h: Likewise.
16300 * config/nios2/rtems.h: Likewise.
16301 * config/powerpcspe/rtems.h: Likewise.
16302 * config/rs6000/rtems.h: Likewise.
16303 * config/rtems.h: Likewise.
16304 * config/sh/rtems.h: Likewise.
16305 * config/sh/rtemself.h: Likewise.
16306 * config/sparc/rtemself.h: Likewise.
16308 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16311 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16312 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16313 * tree-ssa-structalias.c (alias_get_name): Same.
16315 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16318 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16319 instructions if no du chain is found.
16321 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16323 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16325 2017-07-25 Richard Biener <rguenther@suse.de>
16327 PR middle-end/81505
16328 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16331 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16333 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16334 upper-regs options.
16335 (ISA_2_7_MASKS_SERVER): Likewise.
16336 (ISA_3_0_MASKS_IEEE): Likewise.
16337 (OTHER_P8_VECTOR_MASKS): Likewise.
16338 (OTHER_VSX_VECTOR_MASKS): Likewise.
16339 (POWERPC_MASKS): Likewise.
16340 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16341 duplicate list of options.
16342 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
16343 explicit -mupper-regs options.
16344 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
16345 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
16346 alias for -mupper-regs-df.
16347 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16348 (rs6000_init_hard_regno_mode_ok): Likewise.
16349 (rs6000_option_override_internal): Likewise.
16350 (rs6000_opt_masks): Likewise.
16351 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
16352 options in terms of whether -mvsx or -mpower8-vector was used.
16353 (TARGET_UPPER_REGS_DI): Likewise.
16354 (TARGET_UPPER_REGS_SF): Likewise.
16355 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
16356 -mupper-regs-* options.
16358 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16360 * passes.c (emergency_dump_function): Print some empty lines and a
16361 header before the RTL dump.
16363 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16365 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16367 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
16370 * config/aarch64/aarch64.c (aarch64_classify_symbol):
16371 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
16373 2017-07-24 Carl Love <cel@us.ibm.com>
16375 * config/rs6000/rs6000-c.c: Add support for built-in functions
16376 vector float vec_extract_fp32_from_shorth (vector unsigned short);
16377 vector float vec_extract_fp32_from_shortl (vector unsigned short);
16378 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
16379 vec_extract_fp_from_shortl): Add defines for the two builtins.
16380 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
16381 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
16383 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
16384 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
16385 * doc/extend.texi: Update the built-in documentation file for the
16386 new built-in function.
16388 2017-07-24 Jakub Jelinek <jakub@redhat.com>
16391 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
16393 * doc/generic.texi: Likewise.
16394 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16395 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16397 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
16399 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
16400 (aarch64_mls_elt_merge<mode>): Likewise.
16402 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
16404 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16405 having __cxa_atexit.
16407 2017-07-23 Michael Collison <michael.collison@arm.com>
16409 * config/arm/arm.c (arm_option_override): Deprecate
16410 use of -mstructure-size-boundary.
16411 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
16412 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
16414 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16417 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16418 Reduce cost estimate for direct moves.
16420 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
16423 * config/i386/i386.c (ix86_option_override_internal): Disable
16424 BMI, BMI2 and TBM instructions for -m16.
16426 2017-07-21 Carl Love <cel@us.ibm.com>
16428 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16429 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16430 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16431 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16432 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16433 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16434 VMULOSW): New enum "unspec" values.
16435 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16436 altivec_vmulosw): New patterns.
16437 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16438 VMULOSW): Add definitions.
16440 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
16442 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
16443 (qdf24xx): Likewise.
16444 * config/aarch64/aarch64-options-extensions.def (rdma); New.
16445 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
16446 (AARCH64_FL_V8_1): Renumber.
16447 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
16448 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
16449 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
16450 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
16451 rdma to feature modifiers list.
16453 2017-07-21 Yury Gribov <tetra2005@gmail.com>
16455 PR middle-end/56727
16456 * ipa-visibility (function_and_variable_visibility): Convert
16457 recursive PLT call to direct call if appropriate.
16459 2017-07-21 Andrew Pinski <apinski@cavium.com>
16461 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
16462 operand 1 to see if the types precision matches.
16463 * fold-const.c (operand_equal_p): Likewise.
16465 2017-07-21 Richard Biener <rguenther@suse.de>
16467 PR tree-optimization/81303
16468 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
16469 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
16470 (vect_peeling_hash_get_lowest_cost): Adjust.
16471 (vect_enhance_data_refs_alignment): Likewise. Use
16472 vect_get_peeling_costs_all_drs to compute the penalty for no
16473 peeling to match up costs.
16475 2017-07-21 Richard Biener <rguenther@suse.de>
16477 PR tree-optimization/81500
16478 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
16479 we didn't identify a reduction path.
16481 2017-07-21 Tom de Vries <tom@codesourcery.com>
16482 Cesar Philippidis <cesar@codesourcery.com>
16484 PR gcov-profile/81442
16485 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
16488 2017-07-21 Tom de Vries <tom@codesourcery.com>
16491 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
16493 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
16494 nvptx_override_options_after_change.
16496 2017-07-21 Ulrich Drepper <drepper@redhat.com>
16498 * dwarf2out.c (output_file_names): Avoid double testing for
16499 dwarf_version >= 5.
16501 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
16503 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
16505 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16507 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
16509 (try_crossjump_to_edge): Do not punt on partitioned functions.
16511 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16513 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
16514 Put all BBs reachable only via paths crossing cold region to cold
16516 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
16518 2016-07-21 Richard Biener <rguenther@suse.de>
16520 PR tree-optimization/81303
16521 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
16522 into account prologue and epilogue iterations when raising
16523 min_profitable_iters to sth at least covering one vector iteration.
16525 2017-07-21 Tamar Christina <tamar.christina@arm.com>
16527 * config/arm/arm.c (arm_test_cpu_arch_dat):
16530 2017-07-20 Nathan Sidwell <nathan@acm.org>
16532 Remove TYPE_METHODS.
16533 * tree.h (TYPE_METHODS): Delete.
16534 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
16535 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
16536 (dbxout_type_methods): Scan TYPE_FIELDS.
16537 (dbxout_type): Don't check TYPE_METHODS here.
16538 * function.c (use_register_for_decl): Always ignore register for
16539 class types when not optimizing.
16540 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
16541 * tree.c (free_lang_data_in_type): Stitch out member functions and
16542 templates from TYPE_FIELDS.
16543 (build_distinct_type_copy, verify_type_variant,
16544 verify_type): Member fns are on TYPE_FIELDS.
16545 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
16546 * tree-pretty-print.c (dump_generic_node): Likewise.
16548 2017-07-20 Jakub Jelinek <jakub@redhat.com>
16551 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
16552 V2TImode and V4TImode.
16553 (ix86_expand_vector_extract): Likewise.
16554 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
16555 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
16556 (ssescalarmode): Handle V4TImode and V2TImode.
16557 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
16558 (*vec_extractv2ti, *vec_extractv4ti): New insns.
16559 (VEXTRACTI128_MODE): New mode iterator.
16560 (splitter for *vec_extractv?ti first element): New.
16561 (VEC_INIT_MODE): New mode iterator.
16562 (vec_init<mode>): Consolidate 3 expanders into one using
16563 VEC_INIT_MODE mode iterator.
16565 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16567 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
16568 non_spilled_static_chain_regno_p.
16570 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16572 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
16574 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
16576 * bb-reorder.c (connect_traces): Allow copying of blocks within
16579 2017-07-20 Richard Biener <rguenther@suse.de>
16581 * gimple.h (gimple_phi_result): Add gphi * overload.
16582 (gimple_phi_result_ptr): Likewise.
16583 (gimple_phi_arg): Likewise. Adjust index assert to only
16584 allow actual argument accesses rather than all slots available
16586 (gimple_phi_arg_def): Add gphi * overload.
16587 * tree-phinodes.c (make_phi_node): Initialize only actual
16589 (resize_phi_node): Clear memory not covered by old node,
16590 do not initialize excess argument slots.
16591 (reserve_phi_args_for_new_edge): Initialize new argument slot
16594 2017-07-20 Bin Cheng <bin.cheng@arm.com>
16596 PR tree-optimization/81388
16598 2016-07-21 Bin Cheng <bin.cheng@arm.com>
16600 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
16601 by removing computation of may_be_zero.
16603 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16604 Tom de Vries <tom@codesourcery.com>
16606 PR middle-end/81030
16607 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
16608 when gimple level profile disagrees with what RTL expander did.
16610 2017-07-20 Richard Biener <rguenther@suse.de>
16612 PR tree-optimization/61171
16613 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
16614 (vect_analyze_stmt): Add slp instance parameter.
16615 (vectorizable_reduction): Likewise.
16616 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
16617 (vect_is_simple_reduction): Deal with chains not detected
16618 as SLP reduction chain, specifically not properly associated
16619 chains containing a mix of plus/minus.
16620 (get_reduction_op): Remove.
16621 (get_initial_defs_for_reduction): Simplify, pass in whether
16622 this is a reduction chain, pass in the SLP node for the PHIs.
16623 (vect_create_epilog_for_reduction): Get the SLP instance as
16625 (vectorizable_reduction): Get the SLP instance as arg.
16626 During analysis remember the SLP node with the PHIs in the
16627 instance. Simplify getting at the vectorized reduction PHIs.
16628 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
16629 through SLP instance.
16630 (vect_slp_analyze_operations): Likewise.
16631 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
16632 (vect_transform_stmt): Likewise.
16634 2017-07-20 Tom de Vries <tom@codesourcery.com>
16636 PR tree-optimization/81489
16637 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
16638 read of phi arg location to before loop that modifies phi.
16640 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16642 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
16645 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16647 PR middle-end/81331
16648 * except.c (execute): Fix ordering issue.
16650 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16652 PR rtl-optimization/81423
16653 * combine.c (make_compound_operation_int): Don't try to optimize
16654 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
16656 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16658 PR rtl-optimization/81423
16659 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
16660 with a constant that is -1 in the truncated to mode.
16662 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16664 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
16665 (determine_unlikely_bbs): ... here.
16666 * predict.h (propagate_unlikely_bbs_forward): Declare.
16667 * cfgexpand.c (pass_expand::execute): Use it.
16668 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
16670 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
16671 propagate_unlikely_bbs_forward.
16673 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16675 PR middle-end/81331
16676 * except.c (maybe_add_nop_after_section_switch): New function.
16679 2017-07-19 Tom de Vries <tom@codesourcery.com>
16681 * gimple.h (gimple_phi_set_arg): Make assert more strict.
16683 2017-07-19 Tom de Vries <tom@codesourcery.com>
16685 * gimple.h (gimple_phi_arg): Make assert more strict.
16687 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
16689 * config.gcc (powerpc*-*-*): Add mmintrin.h.
16690 * config/rs6000/mmintrin.h: New file.
16691 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
16693 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16695 PR tree-optimization/81346
16696 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
16698 2017-07-19 Tom de Vries <tom@codesourcery.com>
16700 * config/nvptx/nvptx.md (VECIM): Add V2DI.
16702 2017-07-19 Tom de Vries <tom@codesourcery.com>
16704 * config/nvptx/nvptx-modes.def: Add V2DImode.
16705 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
16706 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
16707 (nvptx_output_mov_insn): Handle lack of mov.b128.
16708 (nvptx_print_operand): Handle 'H' and 'L' codes.
16709 (nvptx_vector_mode_supported): Allow V2DImode.
16710 (nvptx_preferred_simd_mode): New function.
16711 (nvptx_data_alignment): New function.
16712 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
16713 nvptx_preferred_simd_mode.
16714 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
16716 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
16718 2017-07-19 Tom de Vries <tom@codesourcery.com>
16720 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
16721 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
16722 (nvptx_vector_mode_supported): New function. Allow V2SImode.
16723 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
16724 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
16725 (mov<VECIM>_insn): New define_insn.
16726 (define_expand "mov<VECIM>): New define_expand.
16728 2017-07-19 Tom de Vries <tom@codesourcery.com>
16730 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
16732 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16734 PR tree-optimization/81346
16735 * fold-const.h (fold_div_compare, range_check_type): Declare.
16736 * fold-const.c (range_check_type): New function.
16737 (build_range_check): Use range_check_type.
16738 (fold_div_compare): No longer static, rewritten into
16739 a match.pd helper function.
16740 (fold_comparison): Don't call fold_div_compare here.
16741 * match.pd (X / C1 op C2): New optimization using fold_div_compare
16742 as helper function.
16744 2017-07-19 Nathan Sidwell <nathan@acm.org>
16746 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
16747 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
16748 * tree.c (find_decls_types_r, verify_type): Use
16749 TYPE_{MIN,MAX}_VALUE_RAW.
16750 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
16751 (hash_tree): Likewise.
16752 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16754 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16757 2017-07-18 Tom de Vries <tom@codesourcery.com>
16759 PR middle-end/81464
16760 * omp-expand.c (expand_omp_for_static_chunk): Handle
16761 equal-argument loop exit phi.
16763 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
16766 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
16767 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
16768 operand 2 predicate.
16769 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
16770 operand 2 predicate.
16771 (ror,rol -> rorx splitters): Use const_int_operand as
16772 operand 2 predicate.
16774 2017-06-18 Richard Biener <rguenther@suse.de>
16776 PR tree-optimization/81410
16777 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
16778 the gap in the ! slp_perm SLP case after each group.
16780 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16782 PR middle-end/81463
16783 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
16786 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16788 PR middle-end/81462
16789 * predict.c (set_even_probabilities): Cleanup; do not affect
16790 probabilities that are already known.
16791 (combine_predictions_for_bb): Call even when count is set.
16793 2017-07-18 Nathan Sidwell <nathan@acm.org>
16795 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
16798 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16801 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
16802 optimization for loop niter analysis.
16804 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
16807 * config/avr/avr.c (avr_optimize_casesi): Don't use
16808 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
16810 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
16812 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
16813 body_cost_vec from _vect_peel_extended_info.
16814 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
16815 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
16818 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16820 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
16822 2017-07-18 Richard Biener <rguenther@suse.de>
16824 PR tree-optimization/80620
16825 PR tree-optimization/81403
16826 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
16827 info when re-using a VN table entry.
16829 2017-07-18 Richard Biener <rguenther@suse.de>
16831 PR tree-optimization/81418
16832 * tree-vect-loop.c (vectorizable_reduction): Properly compute
16833 vectype_in. Verify that with lane-reducing reduction operations
16834 we have a single def-use cycle.
16836 2017-07-17 Carl Love <cel@us.ibm.com>
16838 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
16840 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16841 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16842 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16843 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16844 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16845 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16846 VMULOSW): New enum "unspec" values.
16847 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16848 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16849 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16850 altivec_vmulosw): New patterns.
16851 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16852 VMULOSW): Add definitions.
16853 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
16855 * config/alpha/alpha.c: Include predict.h.
16857 2017-07-17 Yury Gribov <tetra2005@gmail.com>
16859 * tree-vrp.c (compare_assert_loc): Fix comparison function
16860 to return predictable results.
16862 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16864 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
16866 (subdi3): Likewise.
16867 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
16868 * doc/invoke.texi (mexpand-adddi): Update text.
16870 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16872 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
16873 that also clobbers the CC register. The old expand code is moved
16875 (*arc_clzsi2): ... here.
16876 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
16877 the CC register. The old expand code is moved to ...
16878 (arc_ctzsi2): ... here.
16880 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16882 * config/arc/arc.opt (mindexed-loads): Use initial value
16883 TARGET_INDEXED_LOADS_DEFAULT.
16884 (mauto-modify-reg): Use initial value
16885 TARGET_AUTO_MODIFY_REG_DEFAULT.
16886 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16887 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16888 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16889 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16891 2017-07-17 Martin Liska <mliska@suse.cz>
16894 * opts.c (finish_options): Do not allow -fgnu-tm
16895 w/ -fsanitize={kernel-,}address. Say sorry.
16897 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16900 * tree-loop-distribution.c (classify_partition): Only assert on
16901 numer of iterations.
16902 (merge_dep_scc_partitions): Delete prameter. Update function call.
16903 (distribute_loop): Remove code handling loop with unknown niters.
16904 (pass_loop_distribution::execute): Skip loop with unknown niters.
16906 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16909 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
16910 function sort_partitions_by_post_order.
16912 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16914 PR tree-optimization/81374
16915 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
16916 the max index of basic blocks, rather than number of basic blocks.
16918 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16920 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
16922 (arc_legitimate_pic_operand_p): Likewise.
16923 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
16925 (arc_needs_pcl_p): Likewise.
16926 (arc_legitimate_pc_offset_p): Likewise.
16927 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
16928 function is also used in constrains.md.
16929 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
16930 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
16931 PLUS. Only return true/false in known cases, otherwise assert.
16932 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
16933 is already called in arc_legitimate_constant_p.
16934 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
16936 (LEGITIMATE_PIC_OPERAND_P): Use
16937 arc_raw_symbolic_reference_mentioned_p function.
16938 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
16943 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16944 Andrew Burgess <andrew.burgess@embecosm.com>
16946 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
16947 (arc_return_address_register): New function.
16948 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
16949 (arc_handle_fndecl_attribute): Add naked attribute.
16950 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16951 (TARGET_WARN_FUNC_RETURN): Likewise.
16952 (arc_allocate_stack_slots_for_args): New function.
16953 (arc_warn_func_return): Likewise.
16954 (machine_function): Change type fn_type.
16955 (arc_compute_function_type): Consider new naked function type,
16956 change function return type.
16957 (arc_must_save_register): Adapt to handle new
16958 arc_compute_function_type's return type.
16959 (arc_expand_prologue): Likewise.
16960 (arc_expand_epilogue): Likewise.
16961 (arc_return_address_regs): Delete.
16962 (arc_return_address_register): New function.
16963 (arc_epilogue_uses): Use above function.
16964 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
16965 (arc_function_type): Change encoding, add naked type.
16966 (ARC_INTERRUPT_P): Change to handle the new encoding.
16967 (ARC_FAST_INTERRUPT_P): Likewise.
16968 (ARC_NORMAL_P): Define.
16969 (ARC_NAKED_P): Likewise.
16970 (arc_compute_function_type): Delete prototype.
16971 * config/arc/arc.md (in_ret_delay_slot): Use
16972 arc_return_address_register function.
16973 (simple_return): Likewise.
16974 (p_return_i): Likewise.
16976 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16978 PR tree-optimization/81428
16979 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
16980 can't be built for those types.
16982 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16984 Remove stuff dead since r239246.
16986 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
16987 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
16988 (avr_inform_devices): Remove dead stuff.
16990 2017-07-17 Tamar Christina <tamar.christina@arm.com>
16992 * config/arm/arm_neon.h: Fix softp typo.
16994 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16996 PR tree-optimization/81365
16997 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
16998 aggregate moves onto bb predecessor edges, make sure there are no
16999 loads that could alias the lhs in between the start of bb and the
17002 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17005 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17006 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17007 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17009 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17011 PR tree-optimization/81396
17012 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17013 (init_symbolic_number): Initialize it to 1.
17014 (perform_symbolic_merge): Add n_ops from both operands into the new
17016 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17017 without base_addr as useless if they need more than one operation.
17018 (bswap_replace): Handle !bswap case for NULL base_addr.
17020 2017-07-17 Tom de Vries <tom@codesourcery.com>
17023 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17026 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17028 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17029 conditional builtin define __FIX_LEON3FT_B2BST.
17031 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17033 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17034 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17037 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17039 PR rtl-optimization/81424
17040 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17041 to remove potential trapping from operands if -fnon-call-exceptions.
17043 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17045 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17046 profile_proability for scalling.
17047 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17049 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17051 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17053 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17055 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17056 fixpoint arithmetics.
17058 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17060 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17061 fixpoint arithmetics.
17063 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17065 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17066 fixpoint arithmetics.
17068 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17070 * profile-count.h (profile_probability::from_reg_br_prob_note,
17071 profile_probability::to_reg_br_prob_note): New functions.
17072 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17073 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17074 * predict.c (probability_reliable_p): Update.
17075 (edge_probability_reliable_p): Update.
17076 (br_prob_note_reliable_p): Update.
17077 (invert_br_probabilities): Update.
17078 (add_reg_br_prob_note): New function.
17079 (combine_predictions_for_insn): Update.
17080 * asan.c (asan_clear_shadow): Update.
17081 * cfgbuild.c (compute_outgoing_frequencies): Update.
17082 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17083 (update_br_prob_note): Update.
17084 (rtl_verify_edges): Update.
17085 (purge_dead_edges): Update.
17086 (fixup_reorder_chain): Update.
17087 * emit-rtl.c (try_split): Update.
17088 * ifcvt.c (cond_exec_process_insns): Update.
17089 (cond_exec_process_if_block): Update.
17090 (dead_or_predicable): Update.
17091 * internal-fn.c (expand_addsub_overflow): Update.
17092 (expand_neg_overflow): Update.
17093 (expand_mul_overflow): Update.
17094 * loop-doloop.c (doloop_modify): Update.
17095 * loop-unroll.c (compare_and_jump_seq): Update.
17096 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17097 * predict.h: Update.
17098 * reorg.c (mostly_true_jump): Update.
17100 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17101 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17102 * config/arc/arc.c: (emit_unlikely_jump): Update.
17103 * config/arm/arm.c: (emit_unlikely_jump): Update.
17104 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17105 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17106 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17107 (ix86_print_operand): Update.
17108 (ix86_split_fp_branch): Update.
17109 (predict_jump): Update.
17110 * config/ia64/ia64.c (ia64_print_operand): Update.
17111 * config/mmix/mmix.c (mmix_print_operand): Update.
17112 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17113 (rs6000_expand_split_stack_prologue): Update.
17114 * config/rs6000/rs6000.c: Update.
17115 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17116 (s390_expand_vec_movstr): Update.
17117 (s390_expand_cs_tdsi): Update.
17118 (s390_expand_split_stack_prologue): Update.
17119 * config/sh/sh.c (sh_print_operand): Update.
17120 (expand_cbranchsi4): Update.
17121 (expand_cbranchdi4): Update.
17122 * config/sparc/sparc.c (output_v9branch): Update.
17123 * config/spu/spu.c (get_branch_target): Update.
17124 (ea_load_store_inline): Update.
17125 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17126 * config/tilepro/tilepro.c: Update.
17128 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17130 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17131 (gimplify_save_expr): Likewise.
17133 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17135 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17137 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17139 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17141 * timevar.def (TV_IPA_FNSUMMARY): Define.
17143 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17145 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17146 to back store errata sensitive sequence from being generated.
17147 (sqrtdf2_fix): Likewise.
17149 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17151 * tree-ssa-threadupdate.c (compute_path_counts,
17152 update_joiner_offpath_counts): Use profile_probability.
17154 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17157 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17159 * config/arm/arm-c.c (arm_cpu_builtins): Define
17160 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17162 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17164 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17165 array entries to represent __ieee128 versions of the
17166 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17167 scalar_extract_sig, and scalar_insert_exp built-in functions.
17168 (altivec_resolve_overloaded_builtin): Add special case handling
17169 for the __builtin_scalar_insert_exp function, as represented by
17170 the P9V_BUILTIN_VEC_VSIEDP constant.
17171 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17172 exponent support for __ieee128 argument.
17173 (VSESQP): Add scalar extract signature support for __ieee128
17175 (VSTDCNQP): Add scalar test negative support for __ieee128
17177 (VSIEQP): Add scalar insert exponent support for __int128 argument
17178 with __ieee128 result.
17179 (VSIEQPF): Add scalar insert exponent support for __ieee128
17180 argument with __ieee128 result.
17181 (VSTDCQP): Add scalar test data class support for __ieee128
17183 (VSTDCNQP): Add overload support for scalar test negative with
17184 __ieee128 argument.
17185 (VSTDCQP): Add overload support for scalar test data class
17186 __ieee128 argument.
17187 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17188 UNSPEC_VSX_SXSIGDP.
17189 (UNSPEC_VSX_SIEXPQP): New constant.
17190 (xsxexpqp): New insn for VSX scalar extract exponent quad
17192 (xsxsigqp): New insn for VSX scalar extract significand quad
17194 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17195 precision with floating point argument.
17196 (xststdcqp): New expand for VSX scalar test data class quad
17198 (xststdcnegqp): New expand for VSX scalar test negative quad
17200 (xststdcqp): New insn to match expansions for VSX scalar test data
17201 class quad precision and VSX scalar test negative quad precision.
17202 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17203 special case operand checking to enforce that second operand of
17204 VSX scalar test data class with quad precision argument is a 7-bit
17206 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17207 prototypes and descriptions of __ieee128 versions of
17208 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17209 scalar_test_data_class, and scalar_test_neg built-in functions.
17211 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17213 PR tree-optimization/81162
17214 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17215 replace a negate with an add.
17217 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17219 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17221 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17223 * config/arm/arm-c.c (arm_cpu_builtins): Define
17224 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17226 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17228 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17229 (armv8-r): Set ARM Cortex-R52 as default CPU.
17230 * config/arm/arm-tables.opt: Regenerate.
17231 * config/arm/arm-tune.md: Regenerate.
17232 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17234 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17235 extension for -mcpu=cortex-r52.
17237 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17239 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17240 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17241 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17242 (fp-armv8): Define it as FP_ARMv8 only.
17243 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17244 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17246 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17247 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17248 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17249 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17250 than TARGET_FPU_ARMV8.
17251 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17252 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17253 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17255 * config/arm/neon.md (neon_vrint): Likewise.
17256 (neon_vcvt): Likewise.
17257 (neon_<fmaxmin_op><mode>): Likewise.
17258 (<fmaxmin><mode>3): Likewise.
17259 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17260 * config/arm/predicates.md (arm_cond_move_operator): Check against
17261 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17263 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17265 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17266 to top of function.
17268 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17270 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17271 loop in comment with memset.
17273 2017-07-14 Martin Liska <mliska@suse.cz>
17275 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17276 * dwarf2out.c (is_java): Remove the function.
17277 (output_pubname): Remove usage of the function.
17278 (lower_bound_default): Remove usage of DW_LANG_Java.
17279 (gen_compile_unit_die): Likewise.
17280 * gcc.c: Remove compiler defaults for .java and .zip files.
17281 * gimple-expr.c (remove_suffix): Change as there's no longer
17282 extension than 4-letter one.
17283 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17284 (gimplify_save_expr): Likewise.
17285 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17286 as it's possible even for other languages than Java.
17287 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17288 * lto-opts.c (lto_write_options): Remove reference to Java.
17289 * opts.c (strip_off_ending): Update file extension handling.
17290 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17291 * tree-eh.c (lower_resx): Likewise.
17292 * tree.c (free_lang_data_in_type): Remove dead code.
17293 (find_decls_types_r): Likewise.
17294 (build_common_builtin_nodes): Remove Java from a comment.
17295 (verify_type): Remove dead code.
17296 * varasm.c (assemble_external): Remove Java from a comment.
17298 2017-07-14 Martin Liska <mliska@suse.cz>
17300 * opts.c (finish_options): Add quotes.
17301 (common_handle_option): Likewise.
17303 2017-07-14 Martin Liska <mliska@suse.cz>
17305 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17306 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17307 Remove N_SO_PASCAL.
17308 * dwarf2out.c (lower_bound_default): Do not handle
17310 (gen_compile_unit_die): Likewise.
17311 * gcc.c: Remove default extension binding for GNU Pascal.
17312 * stmt.c: Remove Pascal language from a comment.
17313 * xcoffout.c: Likewise.
17315 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17318 * diagnostic-show-locus.c (fixit_cmp): New function.
17319 (layout::layout): Sort m_fixit_hints.
17320 (column_range::column_range): Assert that the values are valid.
17321 (struct char_span): New struct.
17322 (correction::overwrite): New method.
17323 (struct source_line): New struct.
17324 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17325 calls in terms of classes source_line and char_span, and
17326 correction::overwrite.
17327 (selftest::test_overlapped_fixit_printing_2): New function.
17328 (selftest::diagnostic_show_locus_c_tests): Call it.
17330 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17332 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17333 early if there is no lhs.
17335 2017-07-13 Martin Liska <mliska@suse.cz>
17337 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17338 (gen_reference_type_die): Likewise.
17339 * stor-layout.c: Remove Pascal-related comment.
17341 2017-07-13 Martin Liska <mliska@suse.cz>
17343 * opts.c (finish_options): Add quotes to error messages.
17344 (parse_sanitizer_options): Likewise.
17346 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
17348 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
17350 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
17352 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
17354 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17356 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
17358 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
17360 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17363 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17364 provides the hardware capability bits, define the macro
17365 __BUILTIN_CPU_SUPPORTS__.
17366 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17367 if GLIBC does not provide the hardware capability bits. Add a
17368 gcc_unreachable call if the built-in cpu function is neither
17369 __builtin_cpu_is nor __builtin_cpu_supports.
17370 (rs6000_get_function_versions_dispatcher): Change the warning
17371 that an old GLIBC is used which does not export the capability
17372 bits to be an error.
17373 * doc/extend.texi (target_clones attribute): Document the
17374 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
17375 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
17376 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
17377 the macros defined by GCC if the newer GLIBC is available.
17379 2017-07-12 Jeff Law <law@redhat.com>
17381 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
17382 remaining includes slightly.
17383 * config/riscv/riscv-builtins.c: Include profile-count.h.
17385 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17388 * config/avr/avr.c (avr_set_current_function): In diagnostic
17389 messages: Quote keywords and (parts of) identifiers.
17390 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
17393 2017-07-12 Carl Love <cel@us.ibm.com>
17395 * config/rs6000/rs6000-c.c: Add support for built-in functions
17396 vector bool char vec_revb (vector bool char);
17397 vector bool short vec_revb (vector short char);
17398 vector bool int vec_revb (vector bool int);
17399 vector bool long long vec_revb (vector bool long long);
17400 * doc/extend.texi: Update the built-in documentation file for the
17401 new built-in functions.
17403 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17405 * config/s390/s390.md: Remove movcc splitter.
17407 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17409 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
17410 load/store on condition.
17412 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17415 * config/avr/avr.c (avr_encode_section_info)
17416 [progmem && !TREE_READONLY]: Error if progmem object needs
17419 2017-07-11 Michael Collison <michael.collison@arm.com>
17421 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
17424 2017-07-11 Carl Love <cel@us.ibm.com>
17426 * config/rs6000/rs6000-c.c: Add support for builtins
17427 vector unsigned int vec_parity_lsbb (vector signed int);
17428 vector unsigned int vec_parity_lsbb (vector unsigned int);
17429 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
17430 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
17431 vector unsigned long long vec_parity_lsbb (vector signed long long);
17432 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
17433 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
17434 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
17435 * doc/extend.texi: Update the built-in documentation file for the
17436 new built-in functions.
17438 2017-07-11 David Malcolm <dmalcolm@redhat.com>
17440 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
17441 (layout::m_primary_loc): New field.
17442 (layout::layout): Initialize new field. Move location filtering
17443 logic from here to...
17444 (layout::maybe_add_location_range): ...this new method. Add
17445 support for filtering to just the lines already specified by other
17447 (layout::will_show_line_p): New method.
17448 (gcc_rich_location::add_location_if_nearby): New method.
17449 (selftest::test_add_location_if_nearby): New test function.
17450 (selftest::diagnostic_show_locus_c_tests): Call it.
17451 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
17454 2017-07-11 Tom de Vries <tom@codesourcery.com>
17456 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
17457 (bb_first_real_insn): New function.
17458 (nvptx_single): Add extra initialization of broadcasted condition
17461 2017-07-11 Nathan Sidwell <nathan@acm.org>
17463 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
17465 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
17467 * doc/extend.texi (AVR Function Attributes): Remove weblink to
17468 Binutils doc as TEXI will mess them up.
17469 * doc/invoke.texi (AVR Options): Same here.
17471 2017-07-11 Daniel Cederman <cederman@gaisler.com>
17473 * config/sparc/sparc.opt (mfix-ut700): New option.
17474 (mfix-gr712rc): Likewise.
17475 (sparc_fix_b2bst): New variable.
17476 * doc/invoke.texi (SPARC options): Document them.
17477 (ARM options): Fix warnings.
17478 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
17479 instructions to prevent sequences that can trigger the store-store
17480 errata for certain LEON3FT processors.
17481 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
17482 (sparc_option_override): Set sparc_fix_b2bst appropriately.
17483 * config/sparc/sparc.md (fix_b2bst): New attribute.
17484 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
17486 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
17489 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
17491 (*rsqrtsf2_sse): Ditto.
17493 (div<mode>3): Macroize insn from divdf3 and divsf3
17494 using MODEF mode iterator.
17496 2017-07-10 Martin Sebor <msebor@redhat.com>
17498 PR tree-optimization/80397
17499 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
17500 instead of testing for equality to INTEGER_TYPE.
17502 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
17504 * config.gcc: Remove uclibc from arc target spec.
17506 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
17508 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
17510 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17513 * lto-wrapper.c (remove_option): New function.
17514 (merge_and_complain): Merge PIC/PIE options more realistically.
17516 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17518 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
17522 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
17523 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
17524 * config.in: Regenerate.
17525 * configure: Regenerate.
17526 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
17527 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
17528 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
17529 (TARGET_GASISR_PROLOGUES): ...target mask.
17530 * common/config/avr/avr-common.c
17531 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
17532 Set -mgas-isr-prologues.
17533 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
17534 INSERT_PASS_BEFORE for it.
17535 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
17536 * config/avr/avr.c (avr_option_override)
17537 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
17538 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
17539 (avr_attribute_table) <no_gccisr>: Add new function attribute.
17540 (avr_set_current_function) <is_no_gccisr>: Init machine field.
17541 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
17543 (make_avr_pass_pre_proep): New function.
17544 (emit_push_sfr) <treg>: Add argument to function and use it
17545 instead of TMP_REG.
17546 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
17547 and set machine->gasisr.yes.
17548 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
17549 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
17550 __gcc_isr.n_pushed to .L__stack_usage.
17551 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
17552 (avr_asm_final_postscan_insn): ...this new static function.
17553 * config/avr/avr.h (machine_function)
17554 <is_no_gccisr, use_L__stack_usage>: New fields.
17555 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
17556 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
17557 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
17558 (gasisr, *gasisr): New expander and insn.
17559 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17560 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
17561 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
17563 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
17565 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
17568 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17570 Move jump-tables out of .text again.
17573 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
17574 (ASM_OUTPUT_ADDR_VEC): New function.
17575 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
17576 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
17577 INSN_ADDRESSes as asm comment.
17578 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
17579 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
17580 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
17581 * config/avr/avr.md (*tablejump): Adjust comment.
17582 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
17583 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
17585 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
17586 (avr_output_addr_vec): New proto.
17587 (avr_log_t) <insn_addresses>: New field.
17589 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
17592 * config/i386/i386.c (ix86_function_arg_advance): Set
17593 outgoing_args_on_stack to true if there are outgoing arguments
17595 (ix86_function_arg): Likewise.
17596 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
17597 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
17598 * config/i386/i386.h (machine_function): Add
17599 outgoing_args_on_stack.
17601 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
17603 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17604 supporting pthreds.
17605 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
17607 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
17609 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
17610 (REAL_H): Remove $(MACHMODE_H).
17611 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
17613 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
17614 $(MACHMODE_H) and double-int.h.
17615 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
17617 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
17620 2017-07-07 Andrew Pinski <apinski@cavium.com>
17622 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
17623 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
17625 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17627 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
17628 Add warning if GCC was not configured to link against a GLIBC that
17629 exports the hardware capability bits.
17630 (make_resolver_func): Make resolver function private and not a
17631 COMDAT function. Create the name with clone_function_name instead
17632 of make_unique_name.
17635 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
17636 correct operand in doing the split.
17638 2017-07-07 Carl Love <cel@us.ibm.com>
17640 * config/rs6000/rs6000-c: Add support for built-in function
17641 vector unsigned short vec_pack_to_short_fp32 (vector float,
17643 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
17644 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
17645 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
17646 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
17647 (convert_4f32_8i16): Add define_expand.
17648 * doc/extend.texi: Update the built-in documentation file for the
17649 new built-in function.
17651 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17653 * config/sparc/m8.md: New file.
17654 * config/sparc/sparc.md: Include m8.md.
17656 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17658 * config/sparc/sparc.opt: New option -mvis4b.
17659 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
17660 (sparc_option_override): Handle VIS4B.
17661 (enum sparc_builtins): Define
17662 SPARC_BUILTIN_DICTUNPACK{8,16,32},
17663 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
17664 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
17665 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
17666 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
17667 (check_constant_argument): New function.
17668 (sparc_vis_init_builtins): Define builtins
17669 __builtin_vis_dictunpack{8,16,32},
17670 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
17671 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
17672 __builtin_vis_fpcmpde{8,16,32}shl and
17673 __builtin_vis_fpcmpur{8,16,32}shl.
17674 (sparc_expand_builtin): Check that the constant operands to
17675 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
17676 constant and in range.
17677 * config/sparc/sparc-c.c (sparc_target_macros): Handle
17679 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
17680 (SPARC_IMM5_P): Likewise.
17681 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
17682 (enabled): Handle vis4b.
17683 (UNSPEC_DICTUNPACK): New unspec.
17684 (UNSPEC_FPCMPSHL): Likewise.
17685 (UNSPEC_FPUCMPSHL): Likewise.
17686 (UNSPEC_FPCMPDESHL): Likewise.
17687 (UNSPEC_FPCMPURSHL): Likewise.
17688 (cpu_feature): New CPU feature `vis4b'.
17689 (dictunpack{8,16,32}): New insns.
17690 (FPCSMODE): New mode iterator.
17691 (fpcscond): New code iterator.
17692 (fpcsucond): Likewise.
17693 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
17694 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
17695 (fpcmpde{8,16,32}{si,di}shl): Likewise.
17696 (fpcmpur{8,16,32}{si,di}shl): Likewise.
17697 * config/sparc/constraints.md: Define constraints `q' for unsigned
17698 2-bit integer constants and `t' for unsigned 5-bit integer
17700 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
17702 (imm5_operand_dictunpack16): Likewise.
17703 (imm5_operand_dictunpack32): Likewise.
17704 (imm2_operand): Likewise.
17705 * doc/invoke.texi (SPARC Options): Document -mvis4b.
17706 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
17707 ditunpack* and fpcmp*shl builtins.
17709 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17711 * config.gcc: Handle m8 in --with-{cpu,tune} options.
17712 * config.in: Add HAVE_AS_SPARC6 define.
17713 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
17715 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
17717 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
17718 (CPP_CPU_SPEC): Handle m8.
17719 (ASM_CPU_SPEC): Likewise.
17720 * config/sparc/sparc-opts.h (enum processor_type): Add
17722 * config/sparc/sparc.c (m8_costs): New struct.
17723 (sparc_option_override): Handle TARGET_CPU_m8.
17724 (sparc32_initialize_trampoline): Likewise.
17725 (sparc64_initialize_trampoline): Likewise.
17726 (sparc_issue_rate): Likewise.
17727 (sparc_register_move_cost): Likewise.
17728 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
17729 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
17730 (ASM_CPU64_DEFAULT_SPEC): Likewise.
17731 (CPP_CPU_SPEC): Handle M8.
17732 (ASM_CPU_SPEC): Likewise.
17733 (AS_M8_FLAG): Define.
17734 * config/sparc/sparc.md: Add m8 to the cpu attribute.
17735 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
17736 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
17738 * configure: Regenerate.
17739 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
17742 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17744 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
17746 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
17747 ("*movdi_insn_sp32"): Do not set v3pipe.
17748 ("*movsi_insn"): Likewise.
17749 ("*movdi_insn_sp64"): Likewise.
17750 ("*movsf_insn"): Likewise.
17751 ("*movdf_insn_sp32"): Likewise.
17752 ("*movdf_insn_sp64"): Likewise.
17753 ("*zero_extendsidi2_insn_sp64"): Likewise.
17754 ("*sign_extendsidi2_insn"): Likewise.
17755 ("*mov<VM32:mode>_insn"): Likewise.
17756 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17757 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17758 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17759 ("<vlop:code><VL:mode>3"): Likewise.
17760 ("*not_<vlop:code><VL:mode>3"): Likewise.
17761 ("*nand<VL:mode>_vis"): Likewise.
17762 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
17763 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
17764 ("one_cmpl<VL:mode>2"): Likewise.
17765 ("faligndata<VM64:mode>_vis"): Likewise.
17766 ("alignaddrsi_vis"): Likewise.
17767 ("alignaddrdi_vis"): Likweise.
17768 ("alignaddrlsi_vis"): Likewise.
17769 ("alignaddrldi_vis"): Likewise.
17770 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17771 ("bmaskdi_vis"): Likewise.
17772 ("bmasksi_vis"): Likewise.
17773 ("bshuffle<VM64:mode>_vis"): Likewise.
17774 ("cmask8<P:mode>_vis"): Likewise.
17775 ("cmask16<P:mode>_vis"): Likewise.
17776 ("cmask32<P:mode>_vis"): Likewise.
17777 ("pdistn<P:mode>_vis"): Likewise.
17778 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17780 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17782 * config/sparc/sparc.md ("subtype"): New insn attribute.
17783 ("*wrgsr_sp64"): Set insn subtype.
17784 ("*rdgsr_sp64"): Likewise.
17785 ("alignaddrsi_vis"): Likewise.
17786 ("alignaddrdi_vis"): Likewise.
17787 ("alignaddrlsi_vis"): Likewise.
17788 ("alignaddrldi_vis"): Likewise.
17789 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17790 ("fexpand_vis"): Likewise.
17791 ("fpmerge_vis"): Likewise.
17792 ("faligndata<VM64:mode>_vis"): Likewise.
17793 ("bshuffle<VM64:mode>_vis"): Likewise.
17794 ("cmask8<P:mode>_vis"): Likewise.
17795 ("cmask16<P:mode>_vis"): Likewise.
17796 ("cmask32<P:mode>_vis"): Likewise.
17797 ("fchksm16_vis"): Likewise.
17798 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
17799 ("fmean16_vis"): Likewise.
17800 ("fp<plusminus_insn>64_vis"): Likewise.
17801 ("<plusminus_insn>v8qi3"): Likewise.
17802 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17803 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
17804 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
17805 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
17806 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
17807 ("*movqi_insn"): Likewise.
17808 ("*movhi_insn"): Likewise.
17809 ("*movsi_insn"): Likewise.
17810 ("movsi_pic_gotdata_op"): Likewise.
17811 ("*movdi_insn_sp32"): Likewise.
17812 ("*movdi_insn_sp64"): Likewise.
17813 ("movdi_pic_gotdata_op"): Likewise.
17814 ("*movsf_insn"): Likewise.
17815 ("*movdf_insn_sp32"): Likewise.
17816 ("*movdf_insn_sp64"): Likewise.
17817 ("*zero_extendhisi2_insn"): Likewise.
17818 ("*zero_extendqihi2_insn"): Likewise.
17819 ("*zero_extendqisi2_insn"): Likewise.
17820 ("*zero_extendqidi2_insn"): Likewise.
17821 ("*zero_extendhidi2_insn"): Likewise.
17822 ("*zero_extendsidi2_insn_sp64"): Likewise.
17823 ("ldfsr"): Likewise.
17824 ("prefetch_64"): Likewise.
17825 ("prefetch_32"): Likewise.
17826 ("tie_ld32"): Likewise.
17827 ("tie_ld64"): Likewise.
17828 ("*tldo_ldub_sp32"): Likewise.
17829 ("*tldo_ldub1_sp32"): Likewise.
17830 ("*tldo_ldub2_sp32"): Likewise.
17831 ("*tldo_ldub_sp64"): Likewise.
17832 ("*tldo_ldub1_sp64"): Likewise.
17833 ("*tldo_ldub2_sp64"): Likewise.
17834 ("*tldo_ldub3_sp64"): Likewise.
17835 ("*tldo_lduh_sp32"): Likewise.
17836 ("*tldo_lduh1_sp32"): Likewise.
17837 ("*tldo_lduh_sp64"): Likewise.
17838 ("*tldo_lduh1_sp64"): Likewise.
17839 ("*tldo_lduh2_sp64"): Likewise.
17840 ("*tldo_lduw_sp32"): Likewise.
17841 ("*tldo_lduw_sp64"): Likewise.
17842 ("*tldo_lduw1_sp64"): Likewise.
17843 ("*tldo_ldx_sp64"): Likewise.
17844 ("*mov<VM32:mode>_insn"): Likewise.
17845 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17846 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17848 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17850 * config/sparc/sparc.md ("type"): New insn type viscmp.
17851 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
17853 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
17854 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
17855 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17856 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
17858 ("n7_vis_logical_11cycle"): Likewise.
17859 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
17860 * config/sparc/niagara2.md ("niag3_vis": Likewise.
17861 * config/sparc/niagara.md ("niag_vis"): Likewise.
17862 * config/sparc/ultra3.md ("us3_fga"): Likewise.
17863 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
17865 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17867 * config/sparc/sparc.md: New instruction type `bmask'.
17868 (bmaskdi_vis): Use the `bmask' type.
17869 (bmasksi_vis): Likewise.
17870 * config/sparc/ultra3.md (us3_array): Likewise.
17871 * config/sparc/niagara7.md (n7_array): Likewise.
17872 * config/sparc/niagara4.md (n4_array): Likewise.
17873 * config/sparc/niagara2.md (niag2_vis): Likewise.
17874 (niag3_vis): Likewise.
17875 * config/sparc/niagara.md (niag_vis): Likewise.
17877 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17879 * ipa-comdats.c: Remove optimize check from gate.
17880 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
17881 for functions not optimized.
17882 (ipa_fn_summary_read): Skip optimize check.
17883 (ipa_fn_summary_write): Likewise.
17884 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
17886 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
17888 (can_inline_edge_p): Check flag_pcc_struct_return for match.
17889 (check_callers): Give up on caller which is not optimized.
17890 (inline_small_functions): Likewise.
17891 (ipa_inline): Do not give up when not optimizing.
17892 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
17893 away unoptimizes cdtors.
17894 (whole_program_function_and_variable_visibility): Do
17895 ipa_discover_readonly_nonaddressable_vars in LTO mode.
17896 * ipa.c (process_references): Do not check optimize.
17897 (symbol_table::remove_unreachable_nodes): Update optimize check.
17898 (set_writeonly_bit): Update optimize check.
17899 (pass_ipa_cdtor_merge::gate): Do not check optimize.
17900 (pass_ipa_single_use::gate): Remove.
17902 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17904 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
17905 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
17906 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
17907 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
17908 permute_load, permute_store, adjust_extract, adjust_splat,
17909 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
17910 replace_swap_with_copy, dump_swap_insn_table,
17911 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
17912 recombine_lvx_pattern, recombine_stvx_pattern,
17913 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
17914 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
17915 to file rs6000-p8swap.c.
17916 * config/rs6000/rs6000-p8swap.c: New file.
17917 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
17918 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
17919 and rs6000*-*-* targets.
17921 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17923 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
17925 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17927 * lto-wrapper.c (merge_and_complain): Do not merge
17928 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
17929 fsigned_zeros, ftrapping_math, fwrapv.
17930 (append_compiler_options): Do not track these options.
17931 (append_linker_options): Likewie
17933 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17935 * cgraphunit.c (cgraph_node::finalize_function): When
17936 !flag_toplevel_reorde set no_reorder flag.
17937 (varpool_node::finalize_decl): Likewise.
17938 (symbol_table::compile): Drop no toplevel reorder path.
17940 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17942 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
17943 edges; zero probability is not better than uninitialized.
17945 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17947 * asan.h (asan_sanitize_allocas_p): Declare.
17948 * asan.c (asan_sanitize_allocas_p): New function.
17949 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
17950 (handle_builtin_alloca): Likewise.
17951 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
17952 if !asan_sanitize_allocas_p.
17953 * params.def (asan-instrument-allocas): Add new option.
17954 * params.h (ASAN_PROTECT_ALLOCAS): Define.
17955 * opts.c (common_handle_option): Disable allocas sanitization for
17958 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17960 * asan.c: Include gimple-fold.h.
17961 (get_last_alloca_addr): New function.
17962 (handle_builtin_stackrestore): Likewise.
17963 (handle_builtin_alloca): Likewise.
17964 (asan_emit_allocas_unpoison): Likewise.
17965 (get_mem_refs_of_builtin_call): Add new parameter, remove const
17966 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
17967 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
17968 (instrument_builtin_call): Pass gimple iterator to
17969 get_mem_refs_of_builtin_call.
17970 (last_alloca_addr): New global.
17971 * asan.h (asan_emit_allocas_unpoison): Declare.
17972 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
17973 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
17974 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
17975 if function calls alloca.
17976 * gimple-fold.c (replace_call_with_value): Remove static keyword.
17977 * gimple-fold.h (replace_call_with_value): Declare.
17978 * internal-fn.c: Include asan.h.
17979 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
17980 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
17982 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17984 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
17985 (C_SELFTEST_FLAGS): New.
17986 (CPP_SELFTEST_FLAGS): New.
17987 (SELFTEST_DEPS): New, from deps of s-selftest.
17988 (C_SELFTEST_DEPS): New, from deps of s-selftest.
17989 (CPP_SELFTEST_DEPS): New.
17990 (selftest): Add dependency on s-selftest-c++.
17991 (s-selftest): Rename to...
17992 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
17993 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
17994 than SELFTEST_FLAGS.
17995 (selftest-gdb): Rename to...
17996 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
17998 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
17999 (selftest-valgrind): Rename to...
18000 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18002 (selftest-valgrind): Reintroduce as an alias for
18003 selftest-c-valgrind.
18004 (s-selftest-c++): New.
18005 (selftest-c++-gdb): New.
18006 (selftest-c++-valgrind): New.
18008 2017-07-06 Olivier Hainque <hainque@adacore.com>
18010 * gcc.c (process_command): When deciding if undefined variables
18011 should be ignored when processing specs, accept "gcc -v" as well.
18013 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18015 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18016 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18018 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18020 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18021 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18022 * config/arm/arm-tables.opt: Regenerate.
18023 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18025 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18027 2017-07-06 Carl Love <cel@us.ibm.com>
18029 * ChangeLog: Clean up from mid air collision
18031 2017-07-06 Carl Love <cel@us.ibm.com>
18033 * config/rs6000/rs6000-c.c: Add support for built-in functions
18034 vector signed int vec_subc (vector signed int, vector signed int);
18035 vector signed __int128 vec_subc (vector signed __int128,
18036 vector signed __int128);
18037 vector unsigned __int128 vec_subc (vector unsigned __int128,
18038 vector unsigned __int128);
18039 vector signed int vec_sube (vector signed int, vector signed int,
18040 vector signed int);
18041 vector unsigned int vec_sube (vector unsigned int,
18042 vector unsigned int,
18043 vector unsigned int);
18044 vector signed __int128 vec_sube (vector signed __int128,
18045 vector signed __int128,
18046 vector signed__int128);
18047 vector unsigned __int128 vec_sube (vector unsigned __int128,
18048 vector unsigned __int128,
18049 vector unsigned __int128);
18050 vector signed int vec_subec (vector signed int, vector signed int,
18051 vector signed int);
18052 vector unsigned int vec_subec (vector unsigned int,
18053 vector unsigned int,
18054 vector unsigned int);
18055 vector signed __int128 vec_subec (vector signed __int128,
18056 vector signed __int128,
18057 vector signed__int128);
18058 vector unsigned __int128 vec_subec (vector unsigned __int128,
18059 vector unsigned __int128,
18060 vector unsigned __int128);
18061 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18062 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18063 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18064 BU_ALTIVEC_OVERLOAD_X definitions.
18065 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18066 * doc/extend.texi: Update the built-in documentation file for the new
18067 built-in functions.
18069 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18072 * diagnostic-show-locus.c (layout::layout): Use start and finish
18073 spelling location for the start and finish of each range.
18074 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18075 Add unused aspect param.
18076 * input.c (expand_location_1): Add "aspect" param, and use it
18077 to access the correct part of the location.
18078 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18080 (expand_location_to_spelling_point): Likewise.
18081 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18082 param, and pass it to expand_location_1.
18084 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18086 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18087 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18088 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18089 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18090 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18091 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18092 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18093 _mm_maskz_getmant_ss): New intrinsics.
18094 (__builtin_ia32_getexpss128_mask): Changed to ...
18095 __builtin_ia32_getexpss128_round ... this.
18096 (__builtin_ia32_getexpsd128_mask): Changed to ...
18097 __builtin_ia32_getexpsd128_round ... this.
18098 * config/i386/i386-builtin-types.def
18099 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18100 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18101 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18102 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18103 __builtin_ia32_getmantss_mask_round): New builtins.
18104 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18105 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18106 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18107 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18108 * config/i386/sse.md
18109 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18110 avx512f_sgetexp<mode><mask_scalar_name>
18111 <round_saeonly_scalar_name> ... this.
18112 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18113 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18114 vgetexp<ssescalarmodesuffix>
18115 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18116 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18117 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18118 avx512f_vgetmant<mode><mask_scalar_name>
18119 <round_saeonly_scalar_name> ... this.
18120 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18121 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18122 vgetmant<ssescalarmodesuffix>
18123 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18124 %0<mask_scalar_operand4>, %1, %2
18125 <round_saeonly_scalar_mask_op4>, %3} ... this.
18126 * config/i386/subst.md (mask_scalar_operand4,
18127 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18128 round_saeonly_scalar_nimm_predicate): New subst attributes.
18130 2017-07-06 Julia Koval <julia.koval@intel.com>
18132 * config/i386/i386.c (ix86_erase_embedded_rounding):
18133 Remove code for old rounding pattern.
18135 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18137 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18139 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18141 * doc/sourcebuild.texi (Test Directives, Variants of
18142 dg-require-support): Add documentation for dg-require-stack-check.
18144 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18146 * config/i386/subst.md (mask_scalar, round_scalar,
18147 round_saeonly_scalar): New meta-templates.
18148 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18149 round_scalar_mask_operand3, round_scalar_mask_op3,
18150 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18151 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18152 round_saeonly_scalar_constraint,
18153 round_saeonly_scalar_prefix): New subst attribute.
18154 * config/i386/sse.md
18155 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18156 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18157 <round_scalar_name> ... this.
18158 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18159 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18160 <round_scalar_name> ... this.
18161 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18162 <sse>_vm<code><mode>3<mask_scalar_name>
18163 <round_saeonly_scalar_name> ... this.
18164 (v<plusminus_mnemonic><ssescalarmodesuffix>
18165 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18166 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18167 v<plusminus_mnemonic><ssescalarmodesuffix>
18168 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18169 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18170 (v<multdiv_mnemonic><ssescalarmodesuffix>
18171 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18172 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18173 v<multdiv_mnemonic><ssescalarmodesuffix>
18174 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18175 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18176 (v<maxmin_float><ssescalarmodesuffix>
18177 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18178 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18179 v<maxmin_float><ssescalarmodesuffix>
18180 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18181 %0<mask_scalar_operand3>, %1, %<iptr>2
18182 <round_saeonly_scalar_mask_op3>} ... this.
18184 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18186 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18187 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18189 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18190 Alan Hayward <alan.hayward@arm.com>
18191 David Sherwood <david.sherwood@arm.com>
18193 * combine.c (simplify_if_then_else): Remove "enum" before
18195 * compare-elim.c (can_eliminate_compare): Likewise.
18196 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18198 (aarch64_lookup_simd_builtin_type): Likewise.
18199 (aarch64_simd_builtin_type): Likewise.
18200 (aarch64_init_simd_builtin_types): Likewise.
18201 (aarch64_simd_expand_args): Likewise.
18202 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18204 (aarch64_reverse_mask): Likewise.
18205 (aarch64_simd_emit_reg_reg_move): Likewise.
18206 (aarch64_gen_adjusted_ldpstp): Likewise.
18207 (aarch64_ccmp_mode_to_code): Likewise.
18208 (aarch64_operands_ok_for_ldpstp): Likewise.
18209 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18210 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18212 (aarch64_min_divisions_for_recip_mul): Likewise.
18213 (aarch64_reassociation_width): Likewise.
18214 (aarch64_get_condition_code_1): Likewise.
18215 (aarch64_simd_emit_reg_reg_move): Likewise.
18216 (aarch64_simd_attr_length_rglist): Likewise.
18217 (aarch64_reverse_mask): Likewise.
18218 (aarch64_operands_ok_for_ldpstp): Likewise.
18219 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18220 (aarch64_gen_adjusted_ldpstp): Likewise.
18221 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18223 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18224 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18225 (arm_lookup_simd_builtin_type): Likewise.
18226 (arm_simd_builtin_type): Likewise.
18227 (arm_init_simd_builtin_types): Likewise.
18228 (arm_expand_builtin_args): Likewise.
18229 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18230 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18231 (ft32_setup_incoming_varargs): Likewise.
18232 (ft32_function_arg): Likewise.
18233 (ft32_function_arg_advance): Likewise.
18234 (ft32_pass_by_reference): Likewise.
18235 (ft32_arg_partial_bytes): Likewise.
18236 (ft32_valid_pointer_mode): Likewise.
18237 (ft32_addr_space_pointer_mode): Likewise.
18238 (ft32_addr_space_legitimate_address_p): Likewise.
18239 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18241 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18242 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18243 (iamcu_alignment): Likewise.
18244 (canonicalize_vector_int_perm): Likewise.
18245 (ix86_noce_conversion_profitable_p): Likewise.
18246 (ix86_mpx_bound_mode): Likewise.
18247 (ix86_operands_ok_for_move_multiple): Likewise.
18248 * config/microblaze/microblaze-protos.h
18249 (microblaze_expand_conditional_branch_reg): Likewise.
18250 * config/microblaze/microblaze.c
18251 (microblaze_expand_conditional_branch_reg): Likewise.
18252 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18254 (rs6000_reassociation_width): Likewise.
18255 (rs6000_invalid_binary_op): Likewise.
18256 (fusion_p9_p): Likewise.
18257 (emit_fusion_p9_load): Likewise.
18258 (emit_fusion_p9_store): Likewise.
18259 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18261 (riscv_hard_regno_mode_ok_p): Likewise.
18262 (riscv_address_insns): Likewise.
18263 (riscv_split_symbol): Likewise.
18264 (riscv_legitimize_move): Likewise.
18265 (riscv_function_value): Likewise.
18266 (riscv_hard_regno_nregs): Likewise.
18267 (riscv_expand_builtin): Likewise.
18268 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18269 (riscv_build_integer): Likewise.
18270 (riscv_split_integer): Likewise.
18271 (riscv_legitimate_constant_p): Likewise.
18272 (riscv_cannot_force_const_mem): Likewise.
18273 (riscv_regno_mode_ok_for_base_p): Likewise.
18274 (riscv_valid_base_register_p): Likewise.
18275 (riscv_valid_offset_p): Likewise.
18276 (riscv_valid_lo_sum_p): Likewise.
18277 (riscv_classify_address): Likewise.
18278 (riscv_legitimate_address_p): Likewise.
18279 (riscv_address_insns): Likewise.
18280 (riscv_load_store_insns): Likewise.
18281 (riscv_force_binary): Likewise.
18282 (riscv_split_symbol): Likewise.
18283 (riscv_force_address): Likewise.
18284 (riscv_legitimize_address): Likewise.
18285 (riscv_move_integer): Likewise.
18286 (riscv_legitimize_const_move): Likewise.
18287 (riscv_legitimize_move): Likewise.
18288 (riscv_address_cost): Likewise.
18289 (riscv_subword): Likewise.
18290 (riscv_output_move): Likewise.
18291 (riscv_canonicalize_int_order_test): Likewise.
18292 (riscv_emit_int_order_test): Likewise.
18293 (riscv_function_arg_boundary): Likewise.
18294 (riscv_pass_mode_in_fpr_p): Likewise.
18295 (riscv_pass_fpr_single): Likewise.
18296 (riscv_pass_fpr_pair): Likewise.
18297 (riscv_get_arg_info): Likewise.
18298 (riscv_function_arg): Likewise.
18299 (riscv_function_arg_advance): Likewise.
18300 (riscv_arg_partial_bytes): Likewise.
18301 (riscv_function_value): Likewise.
18302 (riscv_pass_by_reference): Likewise.
18303 (riscv_setup_incoming_varargs): Likewise.
18304 (riscv_print_operand): Likewise.
18305 (riscv_elf_select_rtx_section): Likewise.
18306 (riscv_save_restore_reg): Likewise.
18307 (riscv_for_each_saved_reg): Likewise.
18308 (riscv_register_move_cost): Likewise.
18309 (riscv_hard_regno_mode_ok_p): Likewise.
18310 (riscv_hard_regno_nregs): Likewise.
18311 (riscv_class_max_nregs): Likewise.
18312 (riscv_memory_move_cost): Likewise.
18313 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18314 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18315 (rl78_addr_space_address_mode): Likewise.
18316 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18318 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18319 (rs6000_reassociation_width): Likewise.
18320 (rs6000_invalid_binary_op): Likewise.
18321 (fusion_p9_p): Likewise.
18322 (emit_fusion_p9_load): Likewise.
18323 (emit_fusion_p9_store): Likewise.
18324 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18325 (ok_for_simple_move_operands): Likewise.
18326 (ok_for_simple_move_strict_operands): Likewise.
18327 (ok_for_simple_arith_logic_operands): Likewise.
18328 (visium_legitimize_reload_address): Likewise.
18329 (visium_select_cc_mode): Likewise.
18330 (output_cbranch): Likewise.
18331 (visium_split_double_move): Likewise.
18332 (visium_expand_copysign): Likewise.
18333 (visium_expand_int_cstore): Likewise.
18334 (visium_expand_fp_cstore): Likewise.
18335 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18336 (visium_function_arg): Likewise.
18337 (visium_function_arg_advance): Likewise.
18338 (visium_libcall_value): Likewise.
18339 (visium_setup_incoming_varargs): Likewise.
18340 (visium_legitimate_constant_p): Likewise.
18341 (visium_legitimate_address_p): Likewise.
18342 (visium_legitimize_address): Likewise.
18343 (visium_secondary_reload): Likewise.
18344 (visium_register_move_cost): Likewise.
18345 (visium_memory_move_cost): Likewise.
18346 (prepare_move_operands): Likewise.
18347 (ok_for_simple_move_operands): Likewise.
18348 (ok_for_simple_move_strict_operands): Likewise.
18349 (ok_for_simple_arith_logic_operands): Likewise.
18350 (visium_function_value_1): Likewise.
18351 (rtx_ok_for_offset_p): Likewise.
18352 (visium_legitimize_reload_address): Likewise.
18353 (visium_split_double_move): Likewise.
18354 (visium_expand_copysign): Likewise.
18355 (visium_expand_int_cstore): Likewise.
18356 (visium_expand_fp_cstore): Likewise.
18357 (visium_split_cstore): Likewise.
18358 (visium_select_cc_mode): Likewise.
18359 (visium_split_cbranch): Likewise.
18360 (output_cbranch): Likewise.
18361 (visium_print_operand_address): Likewise.
18362 * expmed.c (flip_storage_order): Likewise.
18363 * expmed.h (emit_cstore): Likewise.
18364 (flip_storage_order): Likewise.
18365 * genrecog.c (validate_pattern): Likewise.
18366 * hsa-gen.c (gen_hsa_addr): Likewise.
18367 * internal-fn.c (expand_arith_overflow): Likewise.
18368 * ira-color.c (allocno_copy_cost_saving): Likewise.
18369 * lra-assigns.c (find_hard_regno_for_1): Likewise.
18370 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
18371 (process_invariant_for_inheritance): Likewise.
18372 * lra-eliminations.c (move_plus_up): Likewise.
18373 * omp-low.c (lower_oacc_reductions): Likewise.
18374 * simplify-rtx.c (simplify_subreg): Likewise.
18375 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
18376 (TARGET_CHKP_BOUND_MODE): Likewise..
18377 * targhooks.c (default_chkp_bound_mode): Likewise.
18378 (default_setup_incoming_vararg_bounds): Likewise.
18379 * targhooks.h (default_chkp_bound_mode): Likewise.
18380 (default_setup_incoming_vararg_bounds): Likewise.
18381 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
18382 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
18383 (have_whole_vector_shift): Likewise.
18384 * tree-vect-stmts.c (vectorizable_load): Likewise.
18385 * doc/tm.texi: Regenerate.
18387 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18389 Graceful degrade if Binutils PR21472 is not available.
18392 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
18393 .rodata in flash test fails.
18394 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
18395 * confgure: Regenerate.
18396 * config.in: Regenerate.
18397 * config/avr/avr.c (avr_asm_named_section)
18398 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
18399 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
18400 (avr_asm_init_sections): Same.
18402 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18404 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
18405 (fma<VH:mode>4_intrinsic): Likewise.
18406 (*fmsub<VCVTF:mode>4): Likewise.
18407 (*fmsub<VH:mode>4_intrinsic): Likewise.
18409 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18412 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
18413 Don't depend on "optimize > 0".
18414 (out_movhi_r_mr, out_movqi_mr_r): Same.
18415 (out_movhi_mr_r, out_movqi_r_mr): Same.
18416 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
18417 io_address_operand on "optimize > 0".
18419 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18421 * tree-loop-distribution.c: Add general explanantion on the pass.
18422 (generate_loops_for_partition): Mark distributed loop.
18423 (pg_add_dependence_edges): New parameter. Handle alias data
18424 dependence specially and record it in the parameter if asked.
18425 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
18426 (init_partition_graph_vertices, add_partition_graph_edge): New.
18427 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
18428 (free_partition_graph_vdata, build_partition_graph): New.
18429 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
18430 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
18431 (data_ref_segment_size, latch_dominated_by_data_ref): New.
18432 (compute_alias_check_pairs, version_loop_by_alias_check): New.
18433 (version_for_distribution_p, finalize_partitions): New.
18434 (distribute_loop): Handle alias data dependence specially. Factor
18435 out loop fusion code as functions and call these functions.
18437 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18439 * tree-loop-distribution.c (classify_partition): New parameter and
18440 better handle reduction statement.
18441 (rdg_build_partitions): Revise comment.
18442 (distribute_loop): Compute statements in all partitions and pass it
18443 to classify_partition.
18445 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18447 * tree-loop-distribution.c (enum partition_type): New.
18448 (struct partition): New field type.
18449 (partition_merge_into): Add parameter. Update partition type.
18450 (data_dep_in_cycle_p, update_type_for_merge): New functions.
18451 (build_rdg_partition_for_vertex): Compute partition type.
18452 (rdg_build_partitions): Dump partition type.
18453 (distribute_loop): Update calls to partition_merge_into.
18455 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18457 * tree-loop-distribution.c (struct ddr_hasher): New.
18458 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
18460 (classify_partition): Call get_data_dependence.
18461 (pg_add_dependence_edges): Ditto.
18462 (distribute_loop): Release data dependence hash table.
18464 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18466 * tree-loop-distribution.c (ref_base_address): Delete.
18467 (similar_memory_accesses): Rename ...
18468 (share_memory_accesses): ... to this. Check if partitions access
18469 the same memory reference.
18470 (distribute_loop): Call share_memory_accesses.
18472 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18474 * tree-loop-distribution.c (struct partition): New field recording
18475 its data reference.
18476 (partition_alloc, partition_free): Init and release data refs.
18477 (partition_merge_into): Merge data refs.
18478 (build_rdg_partition_for_vertex): Collect data refs for partition.
18479 (pg_add_dependence_edges): Change parameters from vector to bitmap.
18481 (distribute_loop): Remve data refs from vertice data of partition
18484 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18486 * tree-loop-distribution.c (params.h): Include header file.
18487 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
18488 (datarefs_vec): New global var.
18489 (create_rdg_vertices): Use datarefs_vec directly.
18490 (free_rdg): Don't free data references.
18491 (build_rdg): Update use. Don't free data references.
18492 (distribute_loop): Compute global variable for data references.
18493 Bail out if there are too many data references.
18495 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18497 * tree-loop-distribution.c (loop_nest): New global var.
18498 (build_rdg): Use loop directly, rather than loop nest.
18499 (pg_add_dependence_edges): Remove loop nest parameter. Use global
18501 (distribute_loop): Compute global variable loop nest. Update use.
18503 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18505 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
18506 (partition_merge_into): New parameter. Dump reason for fusion.
18507 (distribute_loop): Update use of partition_merge_into.
18509 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18511 * tree-loop-distribution.c (bb_top_order_index): New.
18512 (bb_top_order_index_size, bb_top_order_cmp): New.
18513 (stmts_from_loop): Use topological order.
18514 (pass_loop_distribution::execute): Compute and release topological
18515 order for basic blocks.
18517 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18519 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
18522 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18524 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
18525 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
18526 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
18527 * internal-fn.def (LOOP_DIST_ALIAS): New.
18528 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
18529 (fold_loop_internal_call): ... this.
18530 (vect_loop_dist_alias_call): New function.
18531 (set_uid_loop_bbs): Call fold_loop_internal_call.
18532 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
18535 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18538 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
18539 Require dead FLAGS_REG at the beginning of a peephole.
18541 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18544 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
18545 arguments in the call to __builtin_ia32_sbb_u32.
18546 (_subborrow_u64): Swap _X and _Y arguments in the call to
18547 __builtin_ia32_sbb_u64.
18549 2017-07-04 Jakub Jelinek <jakub@redhat.com>
18552 * tree-vrp.c (compare_assert_loc): Turn into a function template
18553 with stable template parameter. Only test if a->e is NULL,
18554 !a->e == !b->e has been verified already. Use e == NULL or
18555 e != NULL instead of e or ! e tests. If stable is true, don't use
18556 iterative_hash_expr, on the other side allow a or b or both NULL
18557 and sort the NULLs last.
18558 (process_assert_insertions): Sort using compare_assert_loc<false>
18559 instead of compare_assert_loc, later sort using
18560 compare_assert_loc<true> before calling process_assert_insertions_for
18561 in a loop. Use break instead of continue once seen NULL pointer.
18563 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18565 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18566 Cortex-R7 and Cortex-R8 processors.
18568 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18570 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
18571 uninitialized while src is not.
18573 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
18575 * common/config/arm/arm-common.c: Adjust include path for
18577 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
18578 (arm-cpu.h): Create in build directory. Adjust dependency rules.
18579 (arm-cpu-data.h): Likewise.
18580 (arm-cpu-cdata.h): Likewise.
18581 * config/arm/arm-cpu.h: Delete.
18582 * config/arm/arm-cpu-cdata.h: Delete.
18583 * config/arm/arm-cpu-data.h: Delete.
18585 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
18587 * config/arm/arm-cpus.in (cortex-a55): New.
18588 (cortex-a75): Likewise.
18589 (cortex-a75.cortex-a55): Likewise.
18590 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
18592 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
18593 * config/arm/arm-cpu-cdata.h: Regenerate.
18594 * config/arm/arm-cpu-data.h: Regenerate.
18595 * config/arm/arm-cpu.h: Regenerate.
18596 * config/arm/arm-tables.opt: Regenerate.
18597 * config/arm/arm-tune.md: Regenerate.
18599 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18601 * haifa-sched.c (sched_create_recovery_edges): Update profile.
18603 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18605 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
18608 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
18610 PR tree-optimization/81292
18611 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
18612 full_string_p, also call adjust_related_strinfos if the adjustment
18613 is simple, otherwise invalidate related strinfos.
18615 2017-07-04 Martin Liska <mliska@suse.cz>
18618 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
18619 newly created variable as DECL_IGNORED_P.
18621 2017-07-04 Martin Liska <mliska@suse.cz>
18624 * ipa-inline.c (inline_small_functions):
18625 Use xstrdup_for_dump.
18627 2017-07-04 Tom de Vries <tom@codesourcery.com>
18629 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
18631 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18634 * config/darwin.c (darwin_function_switched_text_sections):
18637 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
18639 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
18641 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18643 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
18645 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18647 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
18648 min_profitable_iters, and th as inclusive lower bounds.
18649 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
18650 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
18651 for min_profitable_iters and min_profitable_estimate.
18652 (vect_transform_loop): Treat th as an inclusive lower bound.
18653 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
18655 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18658 * config/darwin.c (darwin_function_switched_text_sections):
18659 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
18660 in two pieces, and suppress the use of buf.
18662 2017-07-03 Nathan Sidwell <nathan@acm.org>
18664 * hash-table.h (hash_table_mod1): Fix indentation.
18666 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18668 PR middle-end/81290
18669 * predict.c (force_edge_cold): Be more careful about propagation
18671 * profile-count.h (profile_probability::guessed,
18672 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
18674 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
18676 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
18678 * doc/invoke.texi (rcpc architecture extension): Document it.
18680 2017-07-03 Richard Biener <rguenther@suse.de>
18682 PR tree-optimization/60510
18683 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
18684 the scalar reduction PHI and use it.
18685 (vectorizable_reduction): Properly guard the single_defuse_cycle
18686 path for non-SLP reduction chains where we cannot use it.
18687 Rework reduc_def/index and vector type deduction. Rework
18688 vector operand gathering during reduction op code-gen.
18689 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
18690 chains dissolve the chain and leave it to non-SLP reduction
18693 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18695 * tree-data-ref.h (dr_alignment): Declare.
18696 * tree-data-ref.c (dr_alignment): New function.
18697 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
18698 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
18700 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
18702 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18704 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
18705 and base_misalignment fields.
18706 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
18707 * tree-data-ref.c: Include builtins.h.
18708 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
18709 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
18710 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
18711 * tree-vect-data-refs.c: Include tree-cfg.h.
18712 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
18713 fields instead of calculating an alignment here.
18714 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
18715 innermost_loop_behavior fields.
18717 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18719 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
18721 (DR_STEP_ALIGNMENT): New macro.
18722 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
18723 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
18724 (create_data_ref): Print it.
18725 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
18726 to tell whether the step preserves vector (mis)alignment.
18727 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18728 Move the check for an integer step and generalise to all INTEGER_CST.
18729 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
18730 Print the outer step alignment.
18732 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18734 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
18735 with offset_alignment.
18736 (DR_ALIGNED_TO): Delete.
18737 (DR_OFFSET_ALIGNMENT): New macro.
18738 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
18739 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
18740 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
18741 (create_data_ref): Likewise.
18742 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18743 (vect_analyze_data_refs): Likewise.
18744 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
18745 creating dummy innermost behavior.
18747 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18749 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
18750 with a "innermost_loop_behavior *" and refeence tree.
18751 * tree-data-ref.c (dr_analyze_innermost): Likewise.
18752 (create_data_ref): Update call accordingly.
18753 * tree-predcom.c (find_looparound_phi): Likewise.
18755 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18757 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
18758 fields with dr_wrt_vec_loop.
18759 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
18760 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
18761 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
18762 (vect_dr_behavior): New function.
18763 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18764 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
18765 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
18766 track whether the step preserves the misalignment.
18767 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18768 Use vect_dr_behavior.
18769 (vect_setup_realignment): Update call accordingly.
18770 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
18771 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
18772 call to vect_create_addr_base_for_vector_ref.
18773 (vect_create_cond_for_align_checks): Likewise.
18774 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
18775 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
18776 (vect_recog_mask_conversion_pattern): Likewise.
18777 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
18778 (new_stmt_vec_info): Remove redundant zeroing.
18780 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18782 * common/config/arm/arm-common.c (arm_be8_option): New function.
18783 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
18784 (ISA_ARMv6): Add isa_bit_be8.
18785 * config/arm/arm.h (arm_be8_option): Add prototype.
18786 (BE8_SPEC_FUNCTION): New define.
18787 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
18788 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
18789 (mlittle-endian): Similarly.
18790 (mbe8, mbe32): New options.
18791 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
18792 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
18794 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18796 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
18798 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18800 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
18801 (cleanup_tree_cfg_bb): Use it.
18802 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
18804 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
18806 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18809 * loop-doloop.c (add_test): Update profile.
18811 2017-07-03 Martin Liska <mliska@suse.cz>
18814 * sanopt.c (rewrite_usage_of_param): New function.
18815 (sanitize_rewrite_addressable_params): Likewise.
18816 (pass_sanopt::execute): Call rewrite_usage_of_param.
18818 2017-07-03 Richard Biener <rguenther@suse.de>
18820 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
18821 back to using VIEW_CONVERT_EXPR.
18823 2017-07-03 Martin Liska <mliska@suse.cz>
18826 * doc/extend.texi: Document when a resolver function is
18827 generated for target_clones.
18829 2017-07-03 Martin Liska <mliska@suse.cz>
18831 * asan.c (asan_emit_stack_protection): Unpoison just red zones
18832 and shadow memory of auto variables which are subject of
18833 use-after-scope sanitization.
18834 (asan_expand_mark_ifn): Add do set only when is_poison.
18836 2016-07-03 Richard Biener <rguenther@suse.de>
18838 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
18840 (vect_force_simple_reduction): Record reduction def -> phi mapping.
18841 (vectorizable_reduction): Perform reduction PHI creation when
18842 visiting a reduction PHI and adjust and simplify code generation
18843 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
18844 (vect_transform_loop): Visit reduction PHIs.
18845 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
18846 defs into the SLP tree.
18847 (vect_build_slp_tree): Reduction defs terminate the recursion.
18848 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
18850 (vect_get_vec_defs_for_stmt_copy): Export.
18851 (vect_get_vec_defs): Likewise.
18852 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
18854 (vect_get_vec_defs_for_stmt_copy): Declare.
18855 (vect_get_vec_defs): Likewise.
18857 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18859 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
18860 parameter with a "loop" parameter and use it instead of the
18861 loop containing DR_STMT. Don't check simple_iv when doing
18862 BB analysis. Describe the two analysis modes in the comment.
18864 2017-07-03 Tom de Vries <tom@codesourcery.com>
18866 PR tree-optimization/69468
18867 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
18868 (find_same_succ_bb): Handle ignore_edge_flags.
18870 2017-07-03 Tom de Vries <tom@codesourcery.com>
18872 PR tree-optimization/81192
18873 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
18875 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
18877 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
18879 2017-07-03 Tom de Vries <tom@codesourcery.com>
18881 PR tree-optimization/81192
18882 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
18883 BB_SAME_SUCC (bb) == NULL.
18885 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18887 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
18890 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18892 * dumpfile.c: Include profile-count.h
18893 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
18895 (insert_cond_bb): Update profile.
18896 * tree-cfg.h (insert_cond_bb): Update prototype.
18897 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
18898 * tree-dump.c: Do not include tree-cfg.
18900 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18902 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
18904 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18906 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
18909 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18911 * tree-complex.c (expand_complex_div_wide): update profile.
18913 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18914 Alan Hayward <alan.hayward@arm.com>
18915 David Sherwood <david.sherwood@arm.com>
18917 * Makefile.in (MACHMODE_H): Remove insn-modes.h
18918 (CORETYPES_H): New define.
18919 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
18920 (insn-modes-inline.h, s-modes-inline-h): New rules.
18921 (generated_files): Add insn-modes-inline.h.
18922 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
18923 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
18924 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
18925 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
18926 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
18927 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
18928 (build/gencodes.o, build/genconditions.o): Likewise.
18929 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
18930 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
18931 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
18932 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
18933 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
18934 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
18935 * coretypes.h: Include everything up to real.h for generators.
18936 Include insn-modes.h first. Include wide-int-print.h after
18937 wide-int.h. Include insn-modes-inline.h and then machmode.h.
18938 * machmode.h: Don't include insn-modes.h here.
18939 * function-tests.c: Remove includes of signop.h, machmode.h,
18940 double-int.h and wide-int.h.
18942 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
18944 * optc-save-gen.awk: Likewise.
18945 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
18946 * godump.c: Remove include of wide-int-print.h.
18947 * pretty-print.h: Likewise.
18948 * wide-int-print.cc: Likewise.
18949 * wide-int.cc: Likewise.
18950 * hash-map-tests.c: Remove include of signop.h.
18951 * hash-set-tests.c: Likewise.
18952 * rtl-tests.c: Likewise.
18953 * mkconfig.sh: Remove include of machmode.h.
18954 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
18956 (emit_insn_modes_inline_h): ...this new function. Emit the code
18957 into an insn-modes-inline.h header file, adding appropriate
18958 include guards and end comments.
18959 (emit_insn_modes_c_header): Remove include of machmode.h.
18960 (emit_min_insn_modes_c_header): Include coretypes.h rather than
18962 (main): Handle -i flag and call emit_insn_modes_inline_h when
18965 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18967 * tree-ssa-strlen.c (strinfo): Rename the length field to
18968 nonzero_chars. Add a full_string_p field.
18969 (compare_nonzero_chars, zero_length_string_p): New functions.
18970 (get_addr_stridx): Add an offset_out parameter.
18971 Use compare_nonzero_chars.
18972 (get_stridx): Update accordingly. Use compare_nonzero_chars.
18973 (new_strinfo): Update after above changes to strinfo.
18974 (set_endptr_and_length): Set full_string_p.
18975 (get_string_length): Update after above changes to strinfo.
18976 (unshare_strinfo): Update call to new_strinfo.
18977 (maybe_invalidate): Likewise.
18978 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
18979 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
18980 as a uhwi instead of an shwi. Update after above changes to
18981 strinfo and new_strinfo.
18982 (zero_length_string): Assert that chainsi contains full strings.
18983 Use zero_length_string_p. Update call to new_strinfo.
18984 (adjust_related_strinfos): Update after above changes to strinfo.
18985 Copy full_string_p from origsi.
18986 (adjust_last_stmt): Use zero_length_string_p.
18987 (handle_builtin_strlen): Update after above changes to strinfo and
18988 new_strinfo. Install the lhs as the string length if the previous
18989 entry didn't describe a full string.
18990 (handle_builtin_strchr): Update after above changes to strinfo
18992 (handle_builtin_strcpy): Likewise.
18993 (handle_builtin_strcat): Likewise.
18994 (handle_builtin_malloc): Likewise.
18995 (handle_pointer_plus): Likewise.
18996 (handle_builtin_memcpy): Likewise. Track nonzero characters
18997 that aren't necessarily followed by a nul terminator.
18998 (handle_char_store): Likewise.
19000 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19002 PR tree-optimization/80769
19003 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19004 for malloc and calloc. Document the new invariant that all related
19005 strinfos have delayed lengths or none do.
19006 (verify_related_strinfos): Move earlier in file.
19007 (set_endptr_and_length): New function, split out from...
19008 (get_string_length): ...here. Also set the lengths of related
19010 (zero_length_string): Assert that chainsi has known (rather than
19012 (adjust_related_strinfos): Likewise.
19014 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19016 PR tree-optimization/81136
19017 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19018 assert that two references with the same misalignment have the same
19019 compile-time misalignment if those compile-time misalignments
19022 2017-07-01 Andi Kleen <ak@linux.intel.com>
19024 * print-tree.c (print_node): Print all attributes.
19026 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19028 * cfg.c (scale_bbs_frequencies): New function.
19029 * cfg.h (scale_bbs_frequencies): Declare it.
19030 * cfgloopanal.c (single_likely_exit): Cleanup.
19031 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19033 (scale_loop_profile): Likewise.
19034 (loop_version): Likewise.
19035 (create_empty_loop_on_edge): Update.
19036 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19037 scale_loop_frequencies, scale_loop_profile, loopify,
19038 loop_version): Update prototypes.
19039 * modulo-sched.c (sms_schedule): Update.
19040 * predict.c (unlikely_executed_edge_p): Also check probability.
19041 (probably_never_executed_edge_p): Fix typo.
19042 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19043 * tree-parloops.c (gen_parallel_loop): Update.
19044 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19045 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19046 * tree-ssa-loop-split.c (split_loop): Update.
19047 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19048 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19049 (vect_loop_versioning): Update.
19050 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19052 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19054 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19056 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19058 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19059 to keep profile consistent.
19061 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19063 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19064 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19065 * profile-count.h (max_safe_multiplier): Make unsigned.
19066 (profile_count::guessed_zero): New.
19068 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19070 * bb-reorder.c (fix_up_crossing_landing_pad,
19071 fix_crossing_conditional_branches): Use make_single_succ_edge
19072 to keep profile consistent.
19074 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19076 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19079 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19082 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19083 the right scopes, make sure cond_jump isn't preserved between multiple
19084 iterations. Search for fallthru edge whenever there are 3+ edges and
19085 use find_fallthru_edge for it.
19087 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19089 Patch by Alexander Monakov <amonakov@ispras.ru>
19090 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19091 probabilities consistently.
19093 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19095 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19096 emit_cmp_and_jump_insns.
19098 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19101 * tree-inline.c (expand_call_inline): Combine profile statuses.
19103 2017-06-30 Andrew Pinski <apinski@cavium.com>
19105 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19106 fold_stmt returned true.
19108 2017-06-30 Nathan Sidwell <nathan@acm.org>
19110 * ggc.h (empty_string): Delete.
19111 * cfgexpand.c (expand_asm_stmt): Use plain "".
19112 * optabs.c (expand_asm_memory_barrier): Likewise.
19113 * stringpool.c (empty_string): Delete.
19114 (digit_vector, digit_string): Delete.
19115 (ggc_alloc_string): Use plain "", don't optimize single digit
19116 strings. Use ggc_alloc_atomic.
19118 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19120 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19121 comparison set and one other set, use the cost of the non-comparison
19124 2017-06-30 Nathan Sidwell <nathan@acm.org>
19126 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19129 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19131 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19132 loops. Remove now unneeded calls to gimple_switch_set_label() that
19133 just set removed labels to NULL_TREE.
19135 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19137 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19138 (set_range_info): ...here. Only call set_range_info_raw if domain
19140 (set_nonzero_bits): Call set_range_info_raw.
19141 * tree-ssanames.h (set_range_info_raw): New.
19143 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19146 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19147 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19148 of nonimmediate_operand and <store_mask_constraint> instead of m
19149 for the input operand. For V8FI iterator, always split if input
19150 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19151 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19152 <store_mask_predicate> instead of register_operand and
19153 <store_mask_constraint> instead of v for the input operand. Make
19154 sure both operands aren't MEMs for if not <mask_applied>.
19156 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19158 * lto-wrapper.c (copy_file) Close both file descriptors before
19161 2017-06-30 Martin Liska <mliska@suse.cz>
19164 * multiple_target.c (create_dispatcher_calls): Make ifunc
19165 also for function that don't have calls or are not referenced.
19167 2017-06-30 Richard Biener <rguenther@suse.de>
19169 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19170 analyze the first scalar stmt. Move vector type computation
19171 for the BB case here from ...
19172 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19173 live operation processing in the SLP case properly.
19175 2017-06-30 Richard Biener <rguenther@suse.de>
19177 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19179 2017-06-30 Martin Liska <mliska@suse.cz>
19182 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19183 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19185 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19187 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19190 2017-06-30 Richard Biener <rguenther@suse.de>
19192 PR tree-optimization/81249
19193 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19194 condition reduction result to original scalar type.
19196 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19198 * profile-count.h (enum profile_quality): Fix typos and whitespace
19201 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19203 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19204 type for branch probabilities.
19206 2017-06-29 Julian Brown <julian@codesourcery.com>
19207 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19209 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19210 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19211 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19212 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19214 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19216 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19217 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19218 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19219 CC usage from generic code to here.
19220 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19221 CC usage into the target macros.
19223 2017-06-29 Maya Rashish <coypu@sdf.org>
19225 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19228 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19230 * arm/arm-builtins.c: Include profile-count.h
19231 * except.c (sjlj_emit_function_enter): Use
19232 profile_probability::unlikely.
19234 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19236 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19237 and tocrel_offset be pointer args rather than implicitly using
19239 (legitimate_constant_pool_address_p, rs6000_emit_move,
19240 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19241 tocrel_offset and use in toc_relative_expr_p call.
19242 (print_operand, print_operand_address): Use static tocrel_base_oac
19243 and tocrel_offset_oac.
19244 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19247 2017-06-29 Maya Rashish <coypu@sdf.org>
19249 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19251 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19253 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19254 objects, take into account only the alignment of 'op0' and 'mode1' if
19257 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19259 * ccmp.c (ccmp_tree_comparison_p): New function.
19260 (ccmp_candidate_p): Update to use above function.
19261 (get_compare_parts): New function.
19262 (expand_ccmp_next): Update to use new functions.
19263 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19265 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19266 take mode as argument.
19267 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19268 * expr.c (expand_expr_real_1): Pass mode as argument.
19270 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19272 * combine.c (combine_instructions): Print insns to dump_file, together
19275 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19277 * asan.c (asan_emit_stack_protection): Update.
19278 (create_cond_insert_point): Update.
19279 * auto-profile.c (afdo_propagate_circuit): Update.
19280 * basic-block.h (struct edge_def): Turn probability to
19281 profile_probability.
19282 (EDGE_FREQUENCY): Update.
19283 * bb-reorder.c (find_traces_1_round): Update.
19284 (better_edge_p): Update.
19285 (sanitize_hot_paths): Update.
19286 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19287 (make_single_succ_edge): Update.
19288 (check_bb_profile): Update.
19289 (dump_edge_info): Update.
19290 (update_bb_profile_for_threading): Update.
19291 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19292 probabilitycount to 0.
19293 * cfgbuild.c (compute_outgoing_frequencies): Update.
19294 * cfgcleanup.c (try_forward_edges): Update.
19295 (outgoing_edges_match): Update.
19296 (try_crossjump_to_edge): Update.
19297 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19298 (expand_gimple_tailcall): Update.
19299 (construct_init_block): Use make_single_succ_edge.
19300 (construct_exit_block): Use make_single_succ_edge.
19301 * cfghooks.c (verify_flow_info): Update.
19302 (redirect_edge_succ_nodup): Update.
19303 (split_edge): Update.
19304 (account_profile_record): Update.
19305 * cfgloopanal.c (single_likely_exit): Update.
19306 * cfgloopmanip.c (scale_loop_profile): Update.
19307 (set_zero_probability): Remove.
19308 (duplicate_loop_to_header_edge): Update.
19309 * cfgloopmanip.h (loop_version): Update prototype.
19310 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19311 (force_nonfallthru_and_redirect): Update.
19312 (update_br_prob_note): Update.
19313 (rtl_verify_edges): Update.
19314 (purge_dead_edges): Update.
19315 (rtl_lv_add_condition_to_bb): Update.
19316 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19317 * cgraphunit.c (init_lowered_empty_function): Update.
19318 (cgraph_node::expand_thunk): Update.
19319 * cilk-common.c: Include profile-count.h
19320 * dojump.c (inv): Remove.
19321 (jumpifnot): Update.
19322 (jumpifnot_1): Update.
19323 (do_jump_1): Update.
19325 (do_jump_by_parts_greater_rtx): Update.
19326 (do_compare_rtx_and_jump): Update.
19327 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19328 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19329 * dwarf2cfi.c: Include profile-count.h
19330 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19331 (sjlj_emit_dispatch_table): Likewise.
19332 * explow.c: Include profile-count.h
19333 * expmed.c (emit_store_flag_force): Update.
19334 (do_cmp_and_jump): Update.
19335 * expr.c (compare_by_pieces_d::generate): Update.
19336 (compare_by_pieces_d::finish_mode): Update.
19337 (emit_block_move_via_loop): Update.
19338 (store_expr_with_bounds): Update.
19339 (store_constructor): Update.
19340 (expand_expr_real_2): Update.
19341 (expand_expr_real_1): Update.
19342 * expr.h (try_casesi, try_tablejump): Update prototypes.
19343 * gimple-pretty-print.c (dump_probability): Update.
19344 (dump_profile): New.
19345 (dump_gimple_label): Update.
19346 (dump_gimple_bb_header): Update.
19347 * graph.c (draw_cfg_node_succ_edges): Update.
19348 * hsa-gen.c (convert_switch_statements): Update.
19349 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
19350 (find_if_case_1): Update.
19351 (find_if_case_2): Update.
19352 * internal-fn.c (expand_arith_overflow_result_store): Update.
19353 (expand_addsub_overflow): Update.
19354 (expand_neg_overflow): Update.
19355 (expand_mul_overflow): Update.
19356 (expand_vector_ubsan_overflow): Update.
19357 * ipa-cp.c (good_cloning_opportunity_p): Update.
19358 * ipa-split.c (split_function): Use make_single_succ_edge.
19359 * ipa-utils.c (ipa_merge_profiles): Update.
19360 * loop-doloop.c (add_test): Update.
19361 (doloop_modify): Update.
19362 * loop-unroll.c (compare_and_jump_seq): Update.
19363 (unroll_loop_runtime_iterations): Update.
19364 * lra-constraints.c (lra_inheritance): Update.
19365 * lto-streamer-in.c (input_cfg): Update.
19366 * lto-streamer-out.c (output_cfg): Update.
19367 * mcf.c (adjust_cfg_counts): Update.
19368 * modulo-sched.c (sms_schedule): Update.
19369 * omp-expand.c (expand_omp_for_init_counts): Update.
19370 (extract_omp_for_update_vars): Update.
19371 (expand_omp_ordered_sink): Update.
19372 (expand_omp_for_ordered_loops): Update.
19373 (expand_omp_for_generic): Update.
19374 (expand_omp_for_static_nochunk): Update.
19375 (expand_omp_for_static_chunk): Update.
19376 (expand_cilk_for): Update.
19377 (expand_omp_simd): Update.
19378 (expand_omp_taskloop_for_outer): Update.
19379 (expand_omp_taskloop_for_inner): Update.
19380 * omp-simd-clone.c (simd_clone_adjust): Update.
19381 * optabs.c (expand_doubleword_shift): Update.
19382 (expand_abs): Update.
19383 (emit_cmp_and_jump_insn_1): Update.
19384 (expand_compare_and_swap_loop): Update.
19385 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
19386 * predict.c (predictable_edge_p): Update.
19387 (edge_probability_reliable_p): Update.
19388 (set_even_probabilities): Update.
19389 (combine_predictions_for_insn): Update.
19390 (combine_predictions_for_bb): Update.
19391 (propagate_freq): Update.
19392 (estimate_bb_frequencies): Update.
19393 (force_edge_cold): Update.
19394 * profile-count.c (profile_count::dump): Add missing space into dump.
19395 (profile_count::debug): Add newline.
19396 (profile_count::differs_from_p): Explicitly convert to unsigned.
19397 (profile_count::stream_in): Update.
19398 (profile_probability::dump): New member function.
19399 (profile_probability::debug): New member function.
19400 (profile_probability::differs_from_p): New member function.
19401 (profile_probability::differs_lot_from_p): New member function.
19402 (profile_probability::stream_in): New member function.
19403 (profile_probability::stream_out): New member function.
19404 * profile-count.h (profile_count_quality): Rename to ...
19405 (profile_quality): ... this one.
19406 (profile_probability): New.
19407 (profile_count): Update.
19408 * profile.c (compute_branch_probabilities): Update.
19409 * recog.c (peep2_attempt): Update.
19410 * sched-ebb.c (schedule_ebbs): Update.
19411 * sched-rgn.c (find_single_block_region): Update.
19412 (compute_dom_prob_ps): Update.
19413 (schedule_region): Update.
19414 * sel-sched-ir.c (compute_succs_info): Update.
19415 * stmt.c (struct case_node): Update.
19416 (do_jump_if_equal): Update.
19417 (get_outgoing_edge_probs): Update.
19418 (conditional_probability): Update.
19419 (emit_case_dispatch_table): Update.
19420 (expand_case): Update.
19421 (expand_sjlj_dispatch_table): Update.
19422 (emit_case_nodes): Update.
19423 * targhooks.c: Update.
19424 * tracer.c (better_p): Update.
19425 (find_best_successor): Update.
19426 * trans-mem.c (expand_transaction): Update.
19427 * tree-call-cdce.c: Update.
19428 * tree-cfg.c (gimple_split_edge): Upate.
19429 (move_sese_region_to_fn): Upate.
19430 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
19431 * tree-eh.c (lower_resx): Upate.
19432 (cleanup_empty_eh_move_lp): Upate.
19433 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19434 * tree-inline.c (copy_edges_for_bb): Update.
19435 (copy_cfg_body): Update.
19436 * tree-parloops.c (gen_parallel_loop): Update.
19437 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
19438 (gimple_gen_time_profiler): Update.
19439 * tree-ssa-dce.c (remove_dead_stmt): Update.
19440 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19441 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
19442 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
19443 (unloop_loops): Update.
19444 (try_peel_loop): Update.
19445 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19446 * tree-ssa-loop-split.c (connect_loops): Update.
19447 (split_loop): Update.
19448 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19449 (hoist_guard): Update.
19450 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
19451 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
19452 (value_replacement): Update.
19453 * tree-ssa-reassoc.c (branch_fixup): Update.
19454 * tree-ssa-tail-merge.c (replace_block_by): Update.
19455 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
19456 (create_edge_and_update_destination_phis): Update.
19457 (compute_path_counts): Update.
19458 (recompute_probabilities): Update.
19459 (update_joiner_offpath_counts): Update.
19460 (freqs_to_counts_path): Update.
19461 (duplicate_thread_path): Update.
19462 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
19463 (struct switch_conv_info): Update.
19464 (gen_inbound_check): Update.
19465 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19466 (vect_do_peeling): Update.
19467 (vect_loop_versioning): Update.
19468 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19469 (optimize_mask_stores): Update.
19470 * ubsan.c (ubsan_expand_null_ifn): Update.
19471 * value-prof.c (gimple_divmod_fixed_value): Update.
19472 (gimple_divmod_fixed_value_transform): Update.
19473 (gimple_mod_pow2): Update.
19474 (gimple_mod_pow2_value_transform): Update.
19475 (gimple_mod_subtract): Update.
19476 (gimple_mod_subtract_transform): Update.
19477 (gimple_ic): Update.
19478 (gimple_stringop_fixed_value): Update.
19479 (gimple_stringops_transform): Update.
19480 * value-prof.h: Update.
19482 2017-06-29 Carl Love <cel@us.ibm.com>
19484 * config/rs6000/rs6000-c.c: Add support for built-in functions
19485 vector signed int vec_signed (vector float);
19486 vector signed long long vec_signed (vector double);
19487 vector signed int vec_signed2 (vector double, vector double);
19488 vector signed int vec_signede (vector double);
19489 vector signed int vec_signedo (vector double);
19490 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
19491 instruction generator.
19492 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19493 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
19494 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
19496 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
19497 vunsignede_v2df): Add define_expands.
19498 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
19499 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
19500 VEC_UNSIGNEDO): Add definitions.
19501 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19502 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
19503 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
19504 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
19505 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
19506 * config/rs6000/altivec.h (vec_signed, vec_signed2,
19507 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
19508 vec_unsignede, vec_unsignedo): Add builtin defines.
19509 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
19511 * doc/extend.texi: Update the built-in documentation file for the
19512 new built-in functions.
19514 2017-06-29 Richard Biener <rguenther@suse.de>
19516 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
19517 reduction chains to LOOP_VINFO_REDUCTIONS.
19518 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
19519 SLP reductions after processing reduction chains.
19521 2017-06-29 Nathan Sidwell <nathan@acm.org>
19523 * builtins.c (fold_builtin_FUNCTION): Use
19524 lang_hooks.decl_printable_name.
19526 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
19528 PR middle-end/81194
19529 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
19530 with only one label.
19531 * stmt.c (expand_case): Assert NCASES is greater than one.
19533 2017-06-29 Richard Biener <rguenther@suse.de>
19535 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
19537 (group_case_labels): Likewise.
19538 (find_taken_edge): Push sanity checking on val to workers...
19539 (find_taken_edge_cond_expr): ... here
19540 (find_taken_edge_switch_expr): ... and here, handle cases
19541 with just a default label.
19542 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
19543 (group_case_labels): Likewise.
19544 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
19545 group_case_labels does anything cleanup the CFG again.
19547 2017-06-29 Bin Cheng <bin.cheng@arm.com>
19549 PR tree-optimization/81196
19550 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
19551 exit condition comparing two IVs.
19553 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
19555 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
19556 profile to the dummy entry at the end of the list of architectures.
19557 * config/arm/arm-cpu-cdata.h: Regenerated.
19559 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19560 Michael Collison <michael.collison@arm.com>
19563 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
19565 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
19566 (*aarch64_reg_<mode>3_minus_mask): New pattern.
19567 (*aarch64_<optab>_reg_di3_mask2): New pattern.
19568 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
19569 of shift when the shift amount is masked with constant equal to
19570 the size of the mode.
19571 * config/aarch64/predicates.md (subreg_lowpart_operator): New
19574 2017-06-29 Martin Liska <mliska@suse.cz>
19576 * config/i386/i386.opt: Change range from [1,5] to [0,5].
19578 2017-06-29 Yury Gribov <tetra2005@gmail.com>
19581 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
19584 (edge_growth_cache_entry::edge_growth_cache_entry): New
19586 (reset_edge_growth_cache): Update to use constructor.
19588 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19590 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
19591 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19592 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19594 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
19596 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
19597 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
19599 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
19601 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
19602 (*-linux-uclibc*): Add t-uclibc tmake_file.
19603 * config/t-musl: New.
19604 * config/t-uclibc: New.
19606 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
19608 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
19610 (gen_comm_data): Emit architectural setting of arch_prof.
19611 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
19613 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
19614 (armv8-m.base, armv8-m.main): Likewise.
19615 * arm-protos.h (arm_build_target): Add profile field.
19616 (arch_option): Likewise.
19617 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
19619 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
19620 arm_active_target.profile.
19622 2017-06-28 Richard Biener <rguenther@suse.de>
19624 PR middle-end/81227
19625 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
19626 TYPE_OVERFLOW_WRAPS.
19627 * match.pd (negate_expr_p): Likewise.
19628 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
19629 fold_build2, not fold_binary.
19631 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19633 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
19634 Convert memory address to Pmode.
19635 (aarch64_print_operand): Assert MEM operands are always Pmode.
19637 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19640 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
19641 Remove redundant if.
19642 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
19643 * config/arm/aarch-common-protos.h
19644 (aarch_forward_to_shift_is_not_shifted_re): Remove.
19645 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
19647 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
19650 * multiple_target.c (create_dispatcher_calls): Set the default
19651 clone to be static, not public.
19653 2017-06-28 Richard Biener <rguenther@suse.de>
19655 * tree-vect-loop.c (vectorizable_reduction): Move special
19656 cond reduction IV var creation ...
19657 (vect_create_epilog_for_reduction): ... here. Remove induction_index
19658 parameter. Use STMT_VINFO_VECTYPE.
19659 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
19662 2017-06-28 Martin Liska <mliska@suse.cz>
19665 * ipa-visibility.c (non_local_p): Handle visibility.
19667 2017-06-28 Martin Liska <mliska@suse.cz>
19670 * common.opt: Add IntegerRange to various options.
19671 * opt-functions.awk (integer_range_info): New function.
19672 * optc-gen.awk: Add integer_range_info to cl_options struct.
19673 * opts-common.c (decode_cmdline_option): Handle
19674 CL_ERR_INT_RANGE_ARG.
19675 (cmdline_handle_error): Likewise.
19676 * opts.c (print_filtered_help): Show valid interval in
19677 when --help is provided.
19678 * opts.h (struct cl_option): Add range_min and range_max fields.
19679 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
19681 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
19683 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
19684 (x * C EQ/NE y * C): New transformation.
19686 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
19688 * genmultilib (combination_space): Accept '+' in option names.
19690 2017-06-28 Martin Liska <mliska@suse.cz>
19693 * asan.c (instrument_derefs): Bail out inner references
19694 that are hard register variables.
19696 2017-06-28 Jakub Jelinek <jakub@redhat.com>
19699 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
19700 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
19702 2017-06-28 Richard Biener <rguenther@suse.de>
19704 * tree-vectorizer.h (vect_get_vec_defs): Remove.
19705 (vect_get_slp_defs): Adjust.
19706 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
19708 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
19710 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
19711 get_initial_defs_for_reduction instead of vect_get_vec_defs.
19712 (vectorizable_reduction): Adjust.
19713 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
19715 (vect_get_slp_defs): Likewise.
19716 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
19717 (vectorizable_bswap): Adjust.
19718 (vectorizable_call): Likewise.
19719 (vectorizable_conversion): Likewise.
19720 (vectorizable_assignment): Likewise.
19721 (vectorizable_shift): Likewise.
19722 (vectorizable_operation): Likewise.
19723 (vectorizable_store): Likewise.
19724 (vectorizable_condition): Likewise.
19725 (vectorizable_comparison): Likewise.
19727 2017-06-28 Michael Collison <michael.collison@arm.com>
19730 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
19732 (arm_gen_movmemqi): Removed unused variable 'i'.
19733 Convert 'for' loop into 'while' loop.
19734 (arm_expand_prologue): Remove last unnecessary set of insn.
19735 (thumb_pop): Remove unused variable 'pushed_words'.
19736 (thumb_exit): Remove last unnecessary set of regs_to_pop.
19738 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19740 * config/s390/predicates.md: Use s390_rel_address_ok_p.
19741 * config/s390/s390-protos.h: Add prototype of
19742 s390_rel_address_ok_p.
19743 * config/s390/s390.c (s390_got_symbol): New function.
19744 (s390_rel_address_ok_p): New function.
19745 (legitimize_pic_address): Use s390_rel_address_ok_p.
19746 (s390_load_got): Use s390_got_symbol.
19747 (s390_option_override): Issue error if
19748 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
19749 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
19751 * config/s390/s390.opt: New option mpic-data-is-text-relative.
19753 2017-06-27 Andrew Pinski <apinski@cavium.com>
19755 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
19756 (X * copysign (1.0, X)): New pattern.
19757 (X * copysign (1.0, -X)): New pattern.
19758 (copysign (-1.0, CST)): New pattern.
19760 2017-06-27 Joseph Myers <joseph@codesourcery.com>
19762 * genmultilib (combination_space): Remove variable.
19763 Validate reuse rules against regular expression for any sequence
19764 of multilib options in any order.
19766 2017-06-27 Michael Collison <michael.collison@arm.com>
19768 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
19769 call aarch64_split_simd_combine.
19770 * (aarch64_combine_internal<mode>): Delete pattern.
19771 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
19772 Allow register and subreg operands.
19774 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19776 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
19777 specific need, just fallback on defaults.
19778 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
19780 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19781 Olivier Hainque <hainque@adacore.com>
19783 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
19785 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
19786 targets. Pick a default if no particular attempt applied.
19787 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
19790 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19792 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
19793 (x86_64-wrs-vxworks7): Likewise.
19795 2017-06-27 Marek Polacek <polacek@redhat.com>
19798 * ubsan.c (instrument_null): Check get_base_address's result for null.
19800 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19802 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
19804 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19806 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
19807 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
19808 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
19809 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
19810 New function types.
19811 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
19812 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
19813 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
19814 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
19815 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
19816 BUILT_IN_FEUPDATEENV): New builtins.
19817 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
19818 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
19819 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
19820 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
19822 (builtin_structptr_types): Adjust size.
19823 * tree.c (builtin_structptr_types): Add four entries.
19825 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19826 Olivier Hainque <hainque@adacore.com>
19828 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
19829 (TLS_SYM): New local macro, forcing reference to __tls__ on
19830 link command lines for VxWorks 7 RTPs, triggering initialization
19832 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
19833 OS features TLS support, true for RTPs on VxWorks 7.
19834 * config/vxworks.c (vxworks_override_options): Setup emutls
19837 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19839 * predict.c (test_prediction_value_range): Use -1U instead of -1
19840 to avoid narrowing conversion warning.
19841 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
19842 to avoid narrowing conversion warning.
19843 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
19845 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
19847 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19849 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
19850 64bit configurations.
19851 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
19852 (SIZE_TYPE): Likewise.
19853 * config/vxworks.c (vxworks_emutls_var_fields): Use
19854 long_unsigned_type_node instead of unsigned_type_node as the offset
19855 field type, which is "pointer" mode in emutls.c.
19857 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19860 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
19862 PR middle-end/81207
19863 * gimple-fold.c (replace_call_with_call_and_fold): Handle
19864 gimple_vuse copying separately from gimple_vdef copying.
19866 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19868 * value-prof.c (free_hist): Remove call to memset and the enclosing if
19871 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19872 Olivier Hainque <hainque@adacore.com>
19874 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
19875 for all vxworks7 targets.
19876 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
19877 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
19878 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
19879 variations for VX6/VX7 and 32/64bits later on in ...
19880 (VXWORKS_LIB_SPEC): Leverage new macros.
19881 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
19882 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
19884 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19886 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
19887 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
19889 2017-06-26 Carl Love <cel@us.ibm.com>
19891 * config/rs6000/rs6000-c.c: Add support for built-in functions
19892 vector bool char vec_reve (vector bool char);
19893 vector signed char vec_reve (vector signed char);
19894 vector unsigned char vec_reve (vector unsigned char);
19895 vector bool int vec_reve (vector bool int);
19896 vector signed int vec_reve (vector signed int);
19897 vector unsigned int vec_reve (vector unsigned int);
19898 vector bool long long vec_reve (vector bool long long);
19899 vector signed long long vec_reve (vector signed long long);
19900 vector unsigned long long vec_reve (vector unsigned long long);
19901 vector bool short vec_reve (vector bool short);
19902 vector signed short vec_reve (vector signed short);
19903 vector double vec_reve (vector double);
19904 vector float vec_reve (vector float);
19905 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
19906 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
19907 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
19908 (altivec_vreve): New pattern.
19909 * config/rs6000/altivec.h (vec_reve): New define.
19910 * doc/extend.texi (vec_rev): Update the built-in documentation file
19911 for the new built-in functions.
19913 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19915 PR tree-optimization/71815
19916 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
19918 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
19920 (slsr_process_phi): Likewise.
19921 (replace_uncond_cands_and_profitable_phis): Don't replace a
19922 multiply candidate with a stride of 1 (copy or cast).
19923 (phi_incr_cost): Call uses_consumed_by_stmt rather than
19925 (lowest_cost_path): Likewise.
19926 (total_savings): Likewise.
19928 2017-06-26 Richard Biener <rguenther@suse.de>
19931 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
19932 Use def_builtin_pure for all gather builtins.
19934 2017-06-26 Richard Biener <rguenther@suse.de>
19936 PR tree-optimization/81203
19937 * tree-tailcall.c (find_tail_calls): Do not move stmts into
19938 non-dominating BBs.
19940 2017-06-26 Marek Polacek <polacek@redhat.com>
19943 * doc/invoke.texi: Document -Wmultistatement-macros.
19945 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
19947 * doc/sourcebuild.texi (ARM-specific attributes): Document new
19948 arm_neon_ok_no_float_abi effective target.
19950 2017-06-26 Richard Biener <rguenther@suse.de>
19952 PR tree-optimization/80928
19953 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
19954 (copy_bbs): Set BB_DUPLICATED flag early.
19955 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
19957 (execute_on_shrinking_pred): Likewise.
19958 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
19959 BB_DUPLICATED blocks.
19960 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
19961 iterate over all PHIs considering removal of *gsi.
19963 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
19965 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
19968 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19970 * config/rs6000/rs6000-string.c: (expand_block_clear,
19971 do_load_for_compare, select_block_compare_mode,
19972 compute_current_alignment, expand_block_compare,
19973 expand_strncmp_align_check, expand_strn_compare,
19974 expand_block_move, rs6000_output_load_multiple)
19975 Move functions related to string/block move/compare
19976 to a separate file.
19977 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
19978 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
19979 for this function which is now used in two files.
19980 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
19981 * config.gcc: Add rs6000-string.o to extra_objs for
19982 targets powerpc*-*-* and rs6000*-*-*.
19984 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
19987 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
19988 32-bit, since indexed is not valid for DImode.
19989 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
19990 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
19991 (define_peephole2 for Altivec d-form load): Add 32-bit support.
19992 (define_peephole2 for Altivec d-form store): Likewise.
19995 * multiple_target.c (create_dispatcher_calls): Only create the
19996 dispatcher call if the function is the default clone of a
19997 versioned function.
19999 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20001 PR middle-end/80902
20002 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20003 a call, force the call to not be a tail call.
20005 2017-06-23 Jeff Law <law@redhat.com>
20007 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20010 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20012 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20013 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20014 handling for early expansion of vector shifts (sl,sr,sra,rl).
20015 (builtin_function_type): Add vector shift right instructions
20016 to the unsigned argument list.
20018 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20020 rtl-optimizatoin/79286
20021 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20022 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20023 trap. PIC register plus a const unspec without offset can never trap.
20025 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20027 * tree.h (builtin_structptr_type): New type.
20028 (builtin_structptr_types): Declare new array.
20029 * tree.c (builtin_structptr_types): New array.
20030 (free_lang_data, build_common_tree_nodes): Use it.
20032 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20035 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20038 2017-06-22 Matt Turner <mattst88@gmail.com>
20040 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20041 Lake models to skylake case. Assume skylake for unknown
20042 models with clflushopt.
20044 2017-06-22 Jeff Law <law@redhat.com>
20046 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20047 frame sizes that do not satisfy aarch64_uimm12_shift.
20049 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20051 * profile-count.h (apply_probability,
20052 apply_scale, probability_in): Fix checks for zero.
20054 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20056 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20057 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20059 2016-06-22 Richard Biener <rguenther@suse.de>
20061 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20062 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20063 REDUC_MAX_EXPR support.
20064 (vectorizable_reduction): Likewise.
20065 (vect_create_epilog_for_reduction): Likewise.
20067 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20069 * match.pd (A / (1 << B) -> A >> B): New.
20070 * generic-match-head.c: Include optabs-tree.h.
20071 * gimple-match-head.c: Likewise.
20072 * optabs-tree.h (target_supports_op_p): New.
20073 * optabs-tree.c (target_supports_op_p): New.
20075 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20077 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20078 $gcc_cv_ld --help output.
20079 (gcc_cv_ld_demangle): Likewise.
20080 (gcc_cv_ld_eh_frame_hdr): Likewise.
20081 (gcc_cv_ld_pie): Likewise.
20082 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20083 (gcc_cv_ld_buildid): Likewise.
20084 (gcc_cv_ld_sysroot): Likewise.
20085 (ld_bndplt_support): Likewise.
20086 (ld_pushpopstate_support): Likewise.
20087 * configure: Regenerate.
20088 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20090 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20093 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20094 operands indexes to avoid gap between operands and match_dups.
20096 2017-06-21 Andrew Pinski <apinski@cavium.com>
20098 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20099 Increment Arith_shift and Arith_shift_reg by 1.
20100 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20102 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20103 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20104 (aarch64_strip_extend): Add new argument and test for it.
20105 (aarch64_cheap_mult_shift_p): New function.
20106 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20107 add a cost if it is true.
20108 Update calls to aarch64_strip_extend.
20109 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20111 2017-06-21 Andrew Pinski <apinski@cavium.com>
20113 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20115 (thunderxt88): Likewise.
20116 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20117 (thunderx_prefetch_tune): New variable.
20118 (thunderx2t99_prefetch_tune): Update for the correct values.
20119 (thunderxt88_tunings): New variable.
20120 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20121 generic_prefetch_tune.
20122 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20124 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20126 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20127 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20128 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20129 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20130 (aarch64_atomic_cas<mode>, GPI): Likewise.
20132 2017-06-21 Martin Liska <mliska@suse.cz>
20134 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20135 statements on cold and hot labels.
20136 * predict.c (tree_estimate_probability_bb): Remove the
20137 prediction from this place.
20139 2017-06-21 Martin Liska <mliska@suse.cz>
20141 PR tree-optimization/79489
20142 * gimplify.c (maybe_add_early_return_predict_stmt): New
20144 (gimplify_return_expr): Call the function.
20145 * predict.c (tree_estimate_probability_bb): Remove handling
20147 * predict.def: Update comment about early return predictor.
20148 * gimple-predict.h (is_gimple_predict): New function.
20149 * predict.def: Change default value of early return to 66.
20150 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20152 * passes.def: Put pass_strip_predict_hints to the beginning of
20155 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20157 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20158 FUNCTION_DECL declarations.
20159 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20161 (dwaf2out_decl): Likewise.
20162 * godump.c (go_early_global_decl): Skip call to the real debug hook
20163 for FUNCTION_DECL declarations.
20164 * passes.c (rest_of_decl_compilation): Skip call to the
20165 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20166 -fdump-go-spec is passed.
20168 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20170 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20171 Reorder for compactness.
20172 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20173 (def_builtin_pure, def_builtin_pure2): New functions.
20174 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20176 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20178 * match.pd (nop_convert): New predicate.
20179 ((A +- CST1) +- CST2): Allow some NOP conversions.
20181 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20184 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20185 with ctors/dtors if GOVD_SHARED is set.
20187 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20189 * config/aarch64/aarch64.md (movti_aarch64):
20190 Emit mov rather than orr.
20191 (movtf_aarch64): Likewise.
20192 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20193 Emit mov rather than orr.
20195 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20197 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20198 Swap alternatives, make integer dup more expensive.
20200 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20202 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20203 Return true for non-tls symbols.
20205 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20207 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20208 (cortex-a75): Likewise.
20209 (cortex-a75.cortex-a55): Likewise.
20210 * config/aarch64/aarch64-tune.md: Regenerate.
20211 * doc/invoke.texi (-mtune): Document new values for -mtune.
20213 2017-06-21 Tom de Vries <tom@codesourcery.com>
20215 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20216 stack_size feature.
20217 (Effective-Target Keywords, Other attributes): Suggest using
20218 dg-add-options stack_size feature to get stack limit in stack_size
20219 effective target documentation.
20221 2017-06-21 Julian Brown <julian@codesourcery.com>
20222 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20224 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20225 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20226 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20228 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20229 attribute type list for neon_multiply.
20230 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20231 attribute type list for neon_multiply.
20232 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20233 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20234 attribute type list for neon_multiply.
20235 * config/arm/types.md (crypto_pmull): Add.
20236 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20237 attribute type list.
20239 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20241 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20244 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20246 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20247 to make sure not to dereference a NULL cost_classes_ptr pointer.
20249 2017-06-20 Carl Love <cel@us.ibm.com>
20251 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20252 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20253 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20254 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20255 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20256 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20257 VMULOSW): New enum "unspec" values.
20258 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20259 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20260 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20261 altivec_vmulosw): New patterns.
20262 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20263 VMULOSW): Add definitions.
20265 2017-06-20 Julia Koval <julia.koval@intel.com>
20267 * config/i386/i386.c: Fix rounding expand for new pattern.
20268 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20270 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20272 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20273 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20275 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20277 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20280 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20282 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20283 architecture, then by implementer ID.
20284 * config/aarch64/aarch64-tune.md: Regenerate.
20286 2017-06-20 Richard Biener <rguenther@suse.de>
20288 PR middle-end/81097
20289 * fold-const.c (split_tree): Fold to type before negating.
20291 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20293 * diagnostic-show-locus.c
20294 (selftest::test_fixit_deletion_affecting_newline): New function.
20295 (selftest::diagnostic_show_locus_c_tests): Call it.
20297 2017-06-20 Andreas Schwab <schwab@suse.de>
20300 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20303 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20305 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20306 __ARM_FEATURE_COPROC according to support.
20308 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20310 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20311 Rewritten to avoid overflow for > 32-bit pointers.
20314 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20315 by removing enum keyword.
20316 (ubsan_type_descriptor): Likewise. Formatting fix.
20319 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20320 splitter): Require TARGET_SSE2 in the condition.
20322 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20325 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20326 for doing vector set of SFmode on ISA 3.0.
20327 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20328 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20330 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20331 SFmode value into a V4SF variable that was extracted from another
20332 V4SF variable without converting the element to double precision
20333 and back to single precision vector format.
20334 (vsx_insert_extract_v4sf_p9_2): Likewise.
20336 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20338 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20339 in UWHI to avoid undefined overflow.
20342 * ubsan.h (enum ubsan_encode_value_phase): New.
20343 (ubsan_encode_value): Change second argument to
20344 enum ubsan_encode_value_phase with default value of
20345 UBSAN_ENCODE_VALUE_GENERIC.
20346 * ubsan.c (ubsan_encode_value): Change second argument to
20347 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
20348 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
20349 create_tmp_var_raw instead of create_tmp_var and use a
20351 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
20352 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
20353 ubsan_encode_value callers.
20356 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
20357 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
20358 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
20360 2017-06-19 Richard Biener <rguenther@suse.de>
20362 PR middle-end/81118
20363 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20364 estimates if we changed anything.
20366 2017-06-19 Richard Biener <rguenther@suse.de>
20368 PR tree-optimization/80887
20369 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
20370 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
20371 simplified lookups, then reset mprts_hook.
20372 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
20374 (try_to_simplify): Likewise.
20376 2017-06-19 Martin Liska <mliska@suse.cz>
20379 * gimplify.c (gimplify_switch_expr):
20380 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
20382 2017-06-19 Martin Liska <mliska@suse.cz>
20384 * doc/install.texi: Document that PGO runs in 4 stages.
20386 2017-06-19 Martin Liska <mliska@suse.cz>
20389 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
20390 to dispatcher function name.
20391 * multiple_target.c (replace_function_decl): New function.
20392 (create_dispatcher_calls): Redirect both edges and references.
20394 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
20396 * profile-count.c (profile_count::dump): Dump quality.
20397 (profile_count::differs_from_p): Update for unsigned val.
20398 * profile-count.h (profile_count_quality): New enum.
20399 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
20401 2017-06-19 Richard Biener <rguenther@suse.de>
20403 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
20404 struct function as arg.
20405 (estimate_numbers_of_iterations): Export overload with loop arg.
20406 (free_numbers_of_iterations_estimates_loop): Use an overload of
20407 free_numbers_of_iterations_estimates instead.
20408 * tree-cfg.c (remove_bb): Adjust.
20409 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
20410 * tree-parloops.c (gen_parallel_loop): Likewise.
20411 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
20413 (tree_unroll_loops_completely): Likewise.
20414 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
20415 Use an overload instead and export.
20416 (estimated_loop_iterations): Adjust.
20417 (max_loop_iterations): Likewise.
20418 (likely_max_loop_iterations): Likewise.
20419 (estimate_numbers_of_iterations): Take struct function as arg
20421 (loop_exits_before_overflow): Adjust.
20422 (free_numbers_of_iterations_estimates_loop): Use an overload.
20423 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
20424 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
20426 2017-06-19 Richard Biener <rguenther@suse.de>
20429 * ipa-prop.c (find_constructor_constant_at_offset): Handle
20430 RANGE_EXPR conservatively.
20432 2017-06-16 Carl Love <cel@us.ibm.com>
20434 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20435 definitions for vec_float, vec_float2, vec_floato,
20436 vec_floate built-ins.
20437 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
20438 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
20440 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
20441 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
20442 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
20443 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
20444 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
20445 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
20446 vec_floato): Add builtin defines.
20447 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
20448 Update the built-in documentation file for the new built-in
20451 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20453 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
20454 (mthumb): Mark as the negative of -marm.
20456 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20458 * doc/invoke.texi (ARM Options, -mcpu): Document supported
20460 (ARM Options, -mtune): Document that this accepts the same
20461 extension options as -mcpu.
20462 (ARM Options, -mfpu): Document addition of -mfpu=auto.
20464 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20466 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
20467 permitted extensions.
20469 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20471 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
20472 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
20473 (armv8-m.main): Add option +nodsp.
20474 * config/arm/arm-cpu-cdata.h: Regenerated.
20476 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20478 * config/arm/t-fuchsia: New file.
20479 * config.gcc (arm*-*-fuchsia*): Use it.
20481 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20483 * config/arm/t-symbian: Rewrite for new option infrastructure.
20485 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20487 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
20488 (MULTILIB_REQUIRED): Likewise.
20490 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20492 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
20493 (MULTILIB_RESUE): Likewise.
20494 (MULTILIB_MATCHES): Likewise.
20495 (MULTLIB_REQUIRED): Likewise.
20497 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20499 * config/arm/t-rtems: Rewrite for new option framework.
20501 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20503 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
20504 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
20505 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
20506 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
20507 * config/arm/t-multilib: ... here.
20508 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
20509 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
20510 armv7-a and armv8*-a when A-profile libraries have not been built.
20511 * config/arm/t-rmprofile: Rewrite.
20513 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20515 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
20516 with a backslash. Remove the backslash after substituting unescaped
20518 * doc/fragments.texi (MULTILIB_REUSE): Document it.
20520 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20522 * config.gcc: (arm*-*-*): When building a-profile libraries, force
20523 the driver to pass through the default setting of -mfloat-abi.
20524 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
20526 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
20527 (all_feat_combs): New rule.
20528 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
20530 * config/arm/t-aprofile: Rewrite.
20532 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20534 * config/arm/arm.h (FPUTYPE_AUTO): Define.
20535 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
20536 fpu is not specified by the user/command-line.
20537 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
20538 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
20539 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
20540 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
20541 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
20542 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
20544 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20546 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
20547 * config/arm/t-arm-elf: Rewritten.
20549 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20551 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
20552 have some floating-point instructions.
20553 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
20554 (TARGET_MAYBE_HARD_FLOAT): New macro.
20555 * config/arm/arm-builtins.c (arm_init_builtins): Use
20556 TARGET_MAYBE_HARD_FLOAT.
20557 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
20559 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20561 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
20562 (configargs.h): Include it.
20563 (arm_print_hint_for_fpu_option): New function.
20564 (arm_parse_fpu_option): New function.
20565 (candidate_extension): New class.
20566 (arm_canon_for_multilib): New function.
20567 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
20568 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
20569 (ARCH_CANONICAL_SPECS): New macro.
20570 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
20572 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20574 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
20575 are set after handling multilib fragments. Set target_cpu_default2
20578 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20580 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
20582 (arm*-*-*): Set target_cpu_default2 to a quoted string.
20583 * config/arm/parsecpu.awk (check_cpu): Validate any extension
20585 (check_arch): Likewise.
20586 * config/arm/arm.c (arm_configure_build_target): Handle
20587 TARGET_CPU_DEFAULT being a string constant. Scan any feature
20588 options in the default.
20590 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20592 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
20593 when an option is an alias of another.
20594 * config/arm/parsecpu.awk (optalias): New parser token.
20595 (gen_comm_data): Mark non-alias options as such. Emit entries
20596 for extension aliases.
20597 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
20598 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
20599 (armv6kz, armv6zk, armv6t2): Likewise.
20600 (armv7): Make vfpv3-d16 an alias.
20601 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
20603 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
20604 Sort in canonical order.
20605 (armv8-a): Sort in canonical order.
20606 (armv8.1-a, armv8.2-a): Likewise.
20607 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
20609 (cortex-a9): Sort in canonical order.
20610 * config/arm/arm.c (selftests.h): Include it.
20611 (arm_test_cpu_arch_data): New function.
20612 (arm_run_self_tests): New function.
20613 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
20614 (targetm): Move declaration to the end of the file.
20615 * arm-cpu-cdata.h: Regenerated.
20617 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20619 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
20620 call to target_mode_check describing the type of option passed.
20621 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
20622 (arm_target_thumb_only): Use arm_parse_arch_option_name or
20623 arm_parse_cpu_option_name to match parameters against list of
20625 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
20626 arm_arch_core_flags data structure.
20627 * config/arm/arm-cpu_cdata.h: Regenerated.
20629 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20631 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
20633 (arm_print_hint_for_cpu_option): Likewise.
20634 (arm_print_hint_for_arch_option): Likewise.
20635 (arm_parse_cpu_option_name): Likewise.
20636 (arm_parse_arch_option_name): Likewise.
20637 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
20638 of entries in the all_fpus list.
20639 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
20640 (arm_parse_cpu_option_name): Declare.
20641 (arm_parse_arch_option_name): Declare.
20642 (arm_parse_option_features): Declare.
20643 (arm_intialize_isa): Declare.
20644 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
20646 (gen_comm_data): ... here. Make definitions non-static.
20647 * config/arm/arm-cpu-data.h: Regenerated.
20648 * config/arm/arm-cpu-cdata.h: Regenerated.
20650 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20652 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
20653 (cpu_arch_extension): New structure.
20654 (cpu_arch_option, arch_option, cpu_option): New structures.
20655 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
20656 architecture types.
20657 (gen_data): Generate new format data tables.
20658 * config/arm/arm.c (cpu_tune): New structure.
20659 (cpu_option, processors): Delete.
20660 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
20661 (arm_print_hint_for_cpu_option): ... this and ...
20662 (arm_print_hint_for_arch_option): ... this.
20663 (arm_parse_arch_cpu_name): Delete. Replace with ...
20664 (arm_parse_cpu_option_name): ... this and ...
20665 (arm_parse_arch_option_name): ... this.
20666 (arm_unrecognized_feature): Change type of target parameter to
20668 (arm_parse_arch_cpu_features): Delete. Replace with ...
20669 (arm_parse_option_features): ... this.
20670 (arm_configure_build_target): Rework to use new configuration data
20672 (arm_print_tune_info): Rework for new configuration data tables.
20673 * config/arm/arm-cpu-data.h: Regenerated.
20674 * config/arm/arm-cpu.h: Regenerated.
20676 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20678 * Makefile.in (OBJS): Move sbitmap.o from here ...
20679 (OBJS-libcommon): ... to here.
20681 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20683 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
20684 (ISA_ALL_CRYPTO): New macro.
20685 (ISA_ALL_SIMD): New macro
20686 (ISA_ALL_FP): New macro.
20687 * config/arm/arm.c (fpu_bitlist): Update initializer.
20688 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
20690 (arm9e): Add fpu. Add option for nofp
20691 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
20692 (arm926ej-s, arm1026ej-s): Likewise.
20693 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
20694 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
20695 neon-fp16, neon-vfpv4, nofp and nosimd.
20696 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
20697 (cortex-a8): Add fpu. Add option for nofp.
20698 (cortex-a9): Add fpu. Add options for nosimd and nofp.
20699 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
20700 (cortex-r4f): Add fpu.
20701 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
20702 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
20704 (cortex-r8): Likewise.
20705 (cortex-m4): Add fpu. Add option for nofp.
20706 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
20707 (cortex-a17.cortex-a7): Likewise.
20708 (cortex-a32): Add fpu. Add options for crypto and nofp.
20709 (cortex-a35, cortex-a53): Likewise.
20710 (cortex-a57): Add fpu. Add option for crypto.
20711 (cortex-a72, cortex-a73): Likewise.
20712 (exynos-m1): Likewise.
20713 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
20714 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
20715 (cortex-m33): Add fpu. Add option for nofp.
20716 * config/arm/arm-cpu-cdata.h: Regenerated
20717 * config/arm/arm-cpu-data.h: Regenerated.
20719 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20721 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
20722 (armv5te, armv5tej): Likewise.
20723 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
20724 (armv7): Add options fp and vfpv3-d16.
20725 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
20726 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
20728 (armv7ve): Likewise.
20729 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
20730 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
20731 (armv8-a): Add nocrypto option.
20732 (armv8.1-a, armv8.2-a): Likewise.
20733 (armv8-m.main): add options fp, fp.dp and nofp.
20735 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20737 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
20739 (armv8-a+crc): Delete.
20740 (armv8.1-a): Add options simd, crypto and nofp.
20741 (armv8.2-a): Add options fp16, simd, crypto and nofp.
20742 (armv8.2-a+fp16): Delete.
20743 (armv8-m.main): Add option dsp.
20744 (armv8-m.main+dsp): Delete.
20745 (cortex-a8): Add fpu. Add nofp option.
20746 (cortex-a9): Add fpu. Add nofp and nosimd options.
20747 * config/arm/parsecpu.awk (gen_data): Generate option tables and
20748 link to main cpu and architecture data structures.
20749 (gen_comm_data): Only put isa attributes from the main architecture
20751 (option): New statement for architecture and CPU entries.
20752 * arm.c (struct cpu_option): New structure.
20753 (struct processors): Add entry for options.
20754 (arm_unrecognized_feature): New function.
20755 (arm_parse_arch_cpu_name): Ignore any characters after the first
20757 (arm_parse_arch_cpu_feature): New function.
20758 (arm_configure_build_target): Separate out any CPU and architecture
20759 features and parse separately. Don't error out if -mfpu=auto is
20760 used with only an architecture string.
20761 (arm_print_asm_arch_directives): New function.
20762 (arm_file_start): Call it.
20763 * config/arm/arm-cpu-cdata.h: Regenerated.
20764 * config/arm/arm-cpu-data.h: Likewise.
20765 * config/arm/arm-tables.opt: Likewise.
20767 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20769 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
20770 assembler when it is not -mfpu=auto.
20772 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20774 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
20775 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
20776 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
20777 (ASM_CPU_SPEC): Rewrite.
20778 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
20779 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
20780 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
20781 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
20782 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
20783 copied string is NUL-terminated. Also strip any characters prefixed
20785 (arm_rewrite_selected_arch): New function.
20786 (arm_rewrite_march): New function.
20788 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20790 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
20791 (x_arm_cpu_string, x_arm_tune_string): Likewise.
20792 (march, mcpu, mtune): Convert to string-based options.
20793 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
20794 (arm_parse_arch_cpu_name): New function.
20795 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
20796 identify selected architecture or CPU.
20797 (arm_option_save): New function.
20798 (TARGET_OPTION_SAVE): Redefine.
20799 (arm_option_restore): Restore string options.
20800 (arm_option_print): Print string options.
20802 2017-06-16 Martin Sebor <msebor@redhat.com>
20804 PR tree-optimization/80933
20805 PR tree-optimization/80934
20806 * builtins.c (fold_builtin_3): Do not handle bcmp here.
20807 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
20808 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
20809 (gimple_fold_builtin): Call them.
20811 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20813 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
20814 as unlikely; update profile.
20816 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20818 * predict.c (force_edge_cold): Handle declaring edges impossible
20821 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20823 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
20825 (try_unroll_loop_completely): Fix reporting.
20827 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20829 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
20831 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
20834 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
20835 if given a non-constant argument for an intrinsic which requires a
20838 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20840 * profile.c (compare_freqs): New function.
20841 (branch_prob): Sort edge list.
20842 (find_spanning_tree): Assume that the list is priority sorted.
20844 2017-06-16 Richard Biener <rguenther@suse.de>
20846 PR tree-optimization/81090
20847 * passes.def (pass_record_bounds): Remove.
20848 * tree-pass.h (make_pass_record_bounds): Likewise.
20849 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
20850 make_pass_record_bounds): Likewise.
20851 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
20852 not free niter estimates at the beginning but at the end.
20853 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
20855 2017-06-16 Richard Biener <rguenther@suse.de>
20857 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
20858 initializer to workaround ICE in host GCC 4.8.
20860 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20862 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
20864 (clone_inlined_nodes): Update.
20866 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20868 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
20869 prefetch settings, and enable prefetching by default at -O3.
20871 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20873 * config/aarch64/aarch64.c (aarch64_override_options_internal):
20874 Set flag_prefetch_loop_arrays according to tuning data.
20876 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20878 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
20879 New tune structure.
20880 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
20881 [Unrelated to main purpose of the patch] Place the pointer field last
20882 to enable type checking errors when tune structure are wrongly merged.
20883 * config/aarch64/aarch64.c (generic_prefetch_tune,)
20884 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
20885 (thunderx2t99_prefetch_tune): New tune constants.
20886 (tune_params *_tunings): Update all tunings (no functional change).
20887 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
20888 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
20889 from tunings structures.
20891 2017-06-16 Jakub Jelinek <jakub@redhat.com>
20894 * ubsan.c (instrument_null): Add T argument, use it instead
20895 of computing it based on IS_LHS.
20896 (instrument_object_size): Likewise.
20897 (pass_ubsan::execute): Adjust instrument_null and
20898 instrument_object_size callers to pass gimple_get_lhs or
20899 gimple_assign_rhs1 result to it. Use instrument_null instead of
20900 calling get_base_address and instrument_mem_ref. Handle
20901 aggregate call arguments for object-size sanitization.
20903 2017-06-16 Yury Gribov <tetra2005@gmail.com>
20905 PR tree-optimization/81089
20906 * tree-vrp.c (is_masked_range_test): Validate operands of
20909 2017-06-15 Martin Sebor <msebor@redhat.com>
20912 * dumpfile.c (dump_register): Avoid calling memset to initialize
20913 a class with a default ctor.
20914 * gcc.c (struct compiler): Remove const qualification.
20915 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
20916 * hash-table.h: Ditto.
20917 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
20919 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
20920 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
20922 * params.h (struct param_info): Make struct members non-const.
20923 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
20924 with default initialization.
20925 * vec.h (vec_copy_construct, vec_default_construct): New helper
20927 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
20928 with vec_copy_construct.
20929 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
20930 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
20931 * doc/invoke.texi (-Wclass-memaccess): Document.
20933 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20935 * emit-rtl.h (is_leaf): Update comment about local
20936 register allocator.
20938 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20941 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
20942 for a variable to have a section before checking if the section has a
20944 Set section to.persistent if persistent attribute is set.
20945 Warn if .persistent attribute is used on an automatic variable.
20947 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
20949 PR rtl-optimization/80474
20950 * reorg.c (update_block): Do not ignore instructions in a delay slot.
20952 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
20954 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
20957 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
20959 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
20960 (casesi): Emit bounds checking as RTL.
20961 (casesi_internal_mips16_<mode>): Remove bounds checking.
20963 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
20965 * config/xtensa/xtensa.c (xtensa_option_override): Append
20966 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
20967 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
20968 xtensa_doloop_hooks): Define unconditionally.
20969 (xtensa_reorg_loops): Only call reorg_loops in the presence of
20971 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
20972 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
20973 for it in xtensa_option_override.
20974 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
20975 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
20977 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
20979 * doc/cppopts.texi: Document '-' special value to -MF.
20981 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
20983 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
20984 (cortex_a53_fconst): Likewise.
20985 (cortex_a53_fpmul): Likewise.
20986 (cortex_a53_f_load_64): Likewise.
20987 (cortex_a53_f_load_many): Likewise.
20988 (cortex_a53_advsimd_alu): Likewise.
20989 (cortex_a53_advsimd_alu_q): Likewise.
20990 (cortex_a53_advsimd_mul): Likewise.
20991 (cortex_a53_advsimd_mul_q): Likewise.
20992 (fpmac bypass): Add new bypass for fpmac-fpmac case.
20993 Add missing fmul, r2f_cvt and fconst cases.
20995 2017-06-14 Richard Biener <rguenther@suse.de>
20997 PR middle-end/81088
20998 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21000 (fold_binary_loc): When associating do not treat pre-existing
21001 TREE_OVERFLOW on literal constants as a reason to allow
21002 TREE_OVERFLOW on associated literal constants.
21004 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21006 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21007 (MASK_FEATURES): New macro.
21008 * config/sparc/sparc.c (sparc_option_override): Remove the special
21009 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21011 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21013 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21014 a division of 0 if non-call exceptions are enabled.
21016 2017-06-14 Andrew Pinski <apinski@cavium.com>
21017 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21020 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21021 Improve vector initialization code gen for only variable case.
21023 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21025 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21027 2017-06-14 Richard Biener <rguenther@suse.de>
21029 PR tree-optimization/81083
21030 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21033 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21035 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21036 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21037 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21038 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21039 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21040 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21042 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21044 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21045 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21047 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21049 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21051 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21053 * config/rs6000/t-rtems: Don't handle SPE.
21055 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21057 * config/rs6000/t-linux: Don't handle SPE.
21059 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21061 * config/rs6000/eabispe.h: Delete file.
21063 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21065 * config/rs6000/t-spe: Delete file.
21067 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21069 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21070 (rs6000_legitimate_offset_address_p): Return false for anything in
21071 V2SImode or V2SFmode.
21073 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21075 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21076 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21077 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21079 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21080 (rs6000_legitimate_offset_address_p): Ditto.
21081 (rs6000_emit_move): Ditto.
21082 (rs6000_init_builtins): Remove V4HI_type_node.
21084 2017-06-13 Martin Liska <mliska@suse.cz>
21087 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21088 (gate_asan): Likewise.
21089 * asan.h (asan_no_sanitize_address_p): Remove the function.
21090 (sanitize_flags_p): New function.
21091 * builtins.def: Fix coding style.
21092 * common.opt: Use renamed enum value.
21093 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21094 * doc/extend.texi: Document no_sanitize attribute.
21095 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21096 to SANITIZE_UNDEFINED_NONDEFAULT.
21097 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21098 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21099 Use sanitize_flags_p.
21100 * gimplify.c (gimplify_function_tree): Likewise.
21101 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21102 * opts.c (parse_no_sanitize_attribute): New function.
21103 (common_handle_option): Use renamed enum value.
21104 * opts.h (parse_no_sanitize_attribute): Declare.
21105 * tree.c (sanitize_flags_p): New function.
21106 * tree.h: Declared here.
21107 * tsan.c: Use sanitize_flags_p.
21108 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21109 (instrument_mem_ref): Likewise.
21110 (instrument_bool_enum_load): Likewise.
21111 (do_ubsan_in_current_function): Remove the function.
21112 (pass_ubsan::execute): Use sanitize_flags_p.
21113 * ubsan.h: Remove do_ubsan_in_current_function
21114 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21115 (dump_function_to_file): Use it here.
21117 2017-06-13 Martin Jambor <mjambor@suse.cz>
21119 PR tree-optimization/80803
21120 PR tree-optimization/81063
21121 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21122 (propagate_subaccesses_across_link): Enqueue subtree whenever
21123 necessary instead of relying on the caller.
21125 2017-06-13 Martin Jambor <mjambor@suse.cz>
21127 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21128 that have a first_link.
21129 (sort_and_splice_var_accesses): Do not check first_link before
21131 (subtree_mark_written_and_enqueue): Likewise.
21132 (propagate_all_subaccesses): Likewise and do not stop at first
21133 parent with a first_link.
21135 2017-06-13 Martin Jambor <mjambor@suse.cz>
21137 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21140 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21142 * match.pd: New pattern.
21144 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21146 * tree-vrp.c (is_masked_range_test): New function.
21147 (register_edge_assert_for): Determine ranges for
21150 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21152 PR tree-optimization/67328
21153 * fold-const.c (maskable_range_p): New function.
21154 (build_range_check): Generate bittests if possible.
21156 2017-06-13 Martin Liska <mliska@suse.cz>
21158 * gimple-pretty-print.c (dump_probability): Add new argument.
21159 (dump_edge_probability): Dump both probability and count.
21160 (dump_gimple_label): Likewise.
21161 (dump_gimple_bb_header): Likewise.
21163 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21166 * config/avr/avr-devices.c: Fix indentation.
21167 * config/avr/gen-avr-mmcu-specs.c: Dito.
21169 2017-06-13 Richard Biener <rguenther@suse.de>
21171 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21172 instead get vector type from stmt_info.
21173 (vectorizable_reduction): Adjust. Remove dead code.
21175 2017-06-13 Richard Biener <rguenther@suse.de>
21177 PR middle-end/81065
21178 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21179 case of C * (x * C2 + C3).
21180 (fold_addr_of_array_ref_difference): Properly fold index difference.
21182 2017-06-12 David S. Miller <davem@davemloft.net>
21185 * config/sparc/sparc.md (return expander): Emit frame blockage if
21186 function uses alloca.
21188 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21190 * combine.c (make_field_assignment): Check len rather than the mode
21191 precision when calling force_to_mode.
21193 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21195 Support multilibs and devices that see flash in RAM address range.
21198 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21199 (avr_mcu_t) <flash_pm_offset>: New field.
21200 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21201 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21202 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21203 (AVR_TINY_PM_OFFSET): Remove macro.
21204 * config/avr/avr.opt (-mshort-calls): New option.
21205 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21206 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21207 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21208 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21209 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21210 instead of avr_arch->have_jmp_call.
21211 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21212 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21213 avr_arch->flash_pm_offset to define.
21214 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21215 new field flash_pm_offset. Add entry for avrxmega3.
21216 (avr_texinfo): Add entry for avrxmega3.
21217 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21218 attiny212, attiny214,
21219 attiny412, attiny414, attiny416, attiny417,
21220 attiny814, attiny816, attiny817,
21221 attiny1614, attiny1616, attiny1617,
21222 attiny3214, attiny3216, attiny3217.
21223 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21224 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21225 (avr_print_operand_address) [AVR_TINY]: Same.
21226 (avr_asm_init_sections) <readonly_data_section>: Only patch
21227 callback if avr_arch->flash_pm_offset = 0.
21228 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21229 for rodata if avr_arch->flash_pm_offset != 0.
21230 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21231 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21232 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21233 (m_options): Append opt_rcall.
21234 (m_dirnames): Append dir_rcall.
21235 * config/avr/t-multilib: Regenerate.
21237 * configure.ac [target=avr]: Check whether avrxmega3 default
21238 linker description file works as needed.
21239 * configure: Regenerate.
21240 * doc/avr-mmcu.texi: Regenerate.
21241 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21242 <__AVR_ARCH__>: Document avrxmega3 and 103.
21243 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21244 <__AVR_SHORT_CALLS__>: Document it.
21245 <__AVR_PM_BASE_ADDRESS__>: Document it.
21246 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21247 (AVR Variable Attributes) <progmem>: Document this is
21248 not needed for avrxmega3.
21249 (AVR Named Address Spaces) <__flash>: Dito.
21251 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21253 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21255 2017-06-12 Doug Rupp <rupp@adacore.com>
21257 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21258 Append vxworks-stdint.h to the tm_file list.
21259 * config/vxworks-stdint.h: New file.
21261 2017-06-12 Martin Liska <mliska@suse.cz>
21263 PR tree-optimization/81041
21264 * tree-profile.c (gimple_gen_ic_func_profiler):
21265 Create an extra BB in profile-generate
21266 (gimple_gen_time_profiler): Likewise.
21268 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21270 PR tree-optimization/81003
21271 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21272 (update_range_test): Use it instead of force_gimple_operand_gsi.
21274 2017-06-12 Richard Biener <rguenther@suse.de>
21276 PR tree-optimization/81053
21277 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21278 with backedge value not defined in loop. Simplify def stmt
21281 2017-06-11 Tom de Vries <tom@codesourcery.com>
21284 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21286 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21287 nvptx_cannot_force_const_mem.
21289 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21291 * opts.c (finish_options): Move test for flag_split_stack after
21292 it has been initialized.
21294 2017-06-11 Jason Merrill <jason@redhat.com>
21296 * tree.h (id_equal): New.
21297 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21298 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21299 instead of strcmp of IDENTIFIER_POINTER.
21301 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21303 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21304 (mark_all_inlined_calls_cdtor): Fix formating.
21305 (inline_transform): Rescale profile before inlining.
21307 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21309 * cgraph.h (cgraph_edge::clone): Update prototype.
21310 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21311 (cgraph_node::create_clone): Update.
21312 (cgraph_node::create_version_clone): Update.
21313 * tree-inline.c (copy_bb): Update.
21314 (expand_call_inline): Update.
21316 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21318 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21319 factored out from ...
21320 (rs6000_emit_prologue): ... here.
21322 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21324 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21325 factored out from ...
21326 (rs6000_emit_prologue): ... here.
21328 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21330 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21332 (handle_missing_profiles): Fix computation of tp_first_run.
21333 (counts_to_freqs): Do not touch freqs when count is 0.
21335 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21337 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21340 2017-06-10 Tom de Vries <tom@codesourcery.com>
21342 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21343 attributes): Document signal effective target.
21345 2017-06-10 Tom de Vries <tom@codesourcery.com>
21347 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21348 Document effective target stack_size.
21350 2017-06-09 David Malcolm <dmalcolm@redhat.com>
21352 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
21353 to the edit_context if they can be auto-applied.
21355 2017-06-9 Ian Lance Taylor <iant@golang.org>
21357 * opts.c (finish_options): If -fsplit-stack, disable implicit
21358 -forder-blocks-and-partition.
21359 * doc/invoke.texi (Optimize Options): Document that when using
21360 -fsplit-stack -forder-blocks-and-partition is not implicitly
21363 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21365 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21366 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21367 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21368 * builtins.def (abort, trap, unreachable): Declare cold.
21369 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
21370 * tree-core.h (ECF_COLD): New.
21371 * tree.c (set_call_expr_flags): Handle ECF_COLD.
21372 (build_common_builtin_nodes): Mark unreachable and abort as cold.
21374 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21376 * predict.c (unlikely_executed_stmt_p): Cleanup.
21378 2017-06-09 Richard Biener <rguenther@suse.de>
21380 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
21381 model if the ref is always written to.
21383 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21385 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
21387 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21389 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
21392 2017-06-09 Tom de Vries <tom@codesourcery.com>
21395 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
21396 "target cannot support label values" when encountering LABEL_REF.
21398 2017-06-09 Martin Liska <mliska@suse.cz>
21400 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
21401 (gimple_gen_ic_func_profiler): Emit direct comparison
21402 of __gcov_indirect_call_callee with NULL.
21403 (gimple_gen_time_profiler): Change probability from
21404 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
21406 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21408 * profile.c (edge_gcov_counts): Turn to pointer.
21409 (compute_branch_probabilities, compute_branch_probabilities): Update.
21410 (branch_prob): Do not clear edge_gcov_count.
21411 * profile.h (edge_gcov_counts): Turn to pointer.
21412 (edge_gcov_count): Update.
21414 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21416 * gimple.h (gimple_check_failed): Mark cold.
21418 2017-06-09 Richard Biener <rguenther@suse.de>
21420 PR tree-optimization/66623
21421 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
21422 refactor check_reduction into two parts, properly computing
21423 whether we have to check reduction validity for outer loop
21426 2017-06-09 Richard Biener <rguenther@suse.de>
21428 PR tree-optimization/79483
21429 * graphite-scop-detection.c (order): New global.
21430 (get_order): Compute bb to order mapping that satisfies code
21431 generation constraints.
21432 (cmp_pbbs): New helper.
21433 (build_scops): Start domwalk at entry block, sort generated
21436 2017-06-09 Richard Biener <rguenther@suse.de>
21438 PR middle-end/81007
21439 * ipa-polymorphic-call.c
21440 (ipa_polymorphic_call_context::restrict_to_inner_class):
21441 Skip FIELD_DECLs with error_mark_node type.
21442 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
21445 2017-06-09 Martin Liska <mliska@suse.cz>
21447 * predict.c (struct branch_predictor): New struct.
21448 (test_prediction_value_range): New test.
21449 (predict_c_tests): New function.
21450 * selftest-run-tests.c (selftest::run_tests): Run the function.
21451 * selftest.h: Declare new tests.
21453 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
21456 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
21457 gen_add3_insn did not fail.
21458 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
21459 r0, construct that number in a temporary reg and add that reg to r0.
21460 If asked to put the result in r0 as well, fail.
21462 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
21464 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21465 for early expansion of vec_eqv.
21467 2017-06-08 Jakub Jelinek <jakub@redhat.com>
21469 PR middle-end/81005
21470 * ubsan.c (instrument_null): Avoid pointless code temporary.
21471 (pass_ubsan::execute): Instrument aggregate arguments of calls.
21473 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
21477 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
21480 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
21483 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21485 * predict.c (unlikely_executed_edge_p): Move ahead.
21486 (probably_never_executed_edge_p): Use it.
21488 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21490 PR middle-end/79988
21491 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
21492 gimple_call_builtin_p call.
21494 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21496 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
21497 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
21498 rtl_check_failed_type2, rtl_check_failed_code1,
21499 rtl_check_failed_code2, rtl_check_failed_code_mode,
21500 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
21501 rtvec_check_failed_bounds, rtl_check_failed_flag,
21502 _fatal_insn_not_found, _fatal_insn): Likewise.
21503 * tree.h (tree_contains_struct_check_failed,
21504 tree_check_failed, tree_not_check_failed,
21505 tree_class_check_failed, tree_range_check_failed,
21506 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
21507 tree_vec_elt_check_failed, phi_node_elt_check_failed,
21508 tree_operand_check_failed, omp_clause_check_failed,
21509 omp_clause_operand_check_failed, omp_clause_range_check_failed):
21512 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21514 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
21515 flag_branch_probabilities.
21516 * ipa-inline.c (edge_badness): Likewise.
21517 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
21518 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
21519 * predict.c (maybe_hot_frequency_p): Likewise.
21520 (probably_never_executed): Likewise.
21521 * sched-ebb.c (schedule_ebbs): Likewise.
21522 * sched-rgn.c (find_single_block_region): Likewise.
21523 * tracer.c (tail_duplicate): Likewise.
21525 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21527 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
21528 longer requires x_flag_profile_use.
21530 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21532 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
21533 instead of flag_reorder_blocks_and_partition.
21534 * dbxout.c (dbxout_function_end): Likewise.
21535 * dwarf2out.c (gen_subprogram_die): Likewise.
21536 * haifa-sched.c (sched_create_recovery_edges): Likewise.
21537 * hw-doloop.c (reorg_loops): Likewise.
21538 * varasm.c (assemble_start_function,
21539 assemble_end_function): Likewise.
21540 (decide_function_section): Do not check for
21541 flag_reorder_blocks_and_partition.
21543 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21545 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21547 (chkp_get_hard_register_fake_addr_expr): Ditto.
21548 (chkp_build_addr_expr): Add check for hard reg case.
21549 (chkp_parse_array_and_component_ref): Ditto.
21550 (chkp_find_bounds_1): Ditto.
21551 (chkp_process_stmt): Don't generate bounds store for
21554 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21556 * predict.c (maybe_hot_bb_p): Do not check profile status.
21557 (maybe_hot_edge_p): Likewise.
21558 (probably_never_executed): Check for zero counts even if profile
21560 (unlikely_executed_edge_p): New function.
21561 (unlikely_executed_stmt_p): New function.
21562 (unlikely_executed_bb_p): New function.
21563 (set_even_probabilities): Use unlikely predicates.
21564 (combine_predictions_for_bb): Likewise.
21565 (predict_paths_for_bb): Likewise.
21566 (predict_paths_leading_to_edge): Likewise.
21567 (determine_unlikely_bbs): New function.
21568 (estimate_bb_frequencies): Use it.
21569 (compute_function_frequency): Use zero counts even if profile is
21571 * profile-count.h: Fix typo.
21573 2017-08-08 Julia Koval <julia.koval@intel.com>
21575 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
21576 _mm512_mask_cvtsepi16_storeu_epi8,
21577 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
21578 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
21579 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
21580 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
21581 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
21582 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
21583 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
21584 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
21585 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
21586 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
21587 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
21588 __builtin_ia32_pmovuswb256mem_mask,
21589 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
21590 __builtin_ia32_pmovwb512mem_mask): New builtins.
21592 2017-08-08 Julia Koval <julia.koval@intel.com>
21594 PR target/73350,80862
21595 * config/i386/subst.md (round): Fix round pattern.
21596 * config/i386/i386.c (ix86_erase_embedded_rounding):
21597 Fix erasing rounding for the fixed pattern.
21599 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21601 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
21603 2017-06-08 Martin Liska <mliska@suse.cz>
21605 PR gcov-profile/80911
21606 * gcov.c (block_info::block_info): New constructor.
21608 2017-06-07 Carl Love <cel@us.ibm.com>
21610 * config/rs6000/rs6000-c: The return type of the following
21611 built-in functions was implemented as int not long long. Fix sign
21612 of return value for the unsigned version of vec_mulo and vec_mule.
21613 vector unsigned long long vec_bperm (vector unsigned long long,
21614 vector unsigned char)
21615 vector signed long long vec_mule (vector signed int,
21617 vector unsigned long long vec_mule (vector unsigned int,
21618 vector unsigned int)
21619 vector signed long long vec_mulo (vector signed int,
21621 vector unsigned long long vec_mulo (vector unsigned int,
21622 vector unsigned int)
21623 * doc/extend.texi: Fix the documentation for the built-in
21626 2017-06-07 Carl Love <cel@us.ibm.com>
21629 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
21632 2017-06-07 Carl Love <cel@us.ibm.com>
21634 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
21635 support, Generate doublehv for signed int/float for BE case only.
21637 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
21639 * doc/invoke.texi (mcx16): Rewrite.
21641 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21643 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
21644 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
21645 *mov<mode>_softfloat, and an anonymous splitter): Use
21646 nonimmediate_operand instead of rs6000_nonimmediate_operand.
21648 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21650 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
21652 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21653 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
21654 (rs6000_debug_reg_global): Adjust.
21655 (rs6000_init_hard_regno_mode_ok): Adjust.
21656 (rs6000_dbx_register_number): Adjust.
21657 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
21658 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21659 Remove SPE_ACC and SPEFSCR.
21660 (REG_ALLOC_ORDER): Ditto.
21661 (FRAME_POINTER_REGNUM): Change to 111.
21662 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
21663 (REG_CLASS_NAMES): Ditto.
21664 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
21665 (REGISTER_NAMES): Ditto.
21666 (ADDITIONAL_REG_NAMES): Ditto.
21667 (rs6000_reg_names): Ditto.
21668 * config/rs6000/rs6000.md: Renumber some register number
21671 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21673 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
21675 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21676 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
21678 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
21679 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21680 Delete the SPE high registers.
21681 (REG_ALLOC_ORDER): Ditto.
21682 (enum reg_class): Remove SPE_HIGH_REGS.
21683 (REG_CLASS_NAMES): Ditto.
21684 (REG_CLASS_CONTENTS): Delete the SPE high registers.
21685 (REGISTER_NAMES): Ditto.
21686 (rs6000_reg_names): Ditto.
21687 * doc/tm.texi.in: Remove SPE as example.
21688 * doc/tm.texi: Regenerate.
21690 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21692 * config/rs6000/8540.md (ppc8540_brinc): Delete.
21693 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
21694 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
21695 * config/rs6000/rs6000.md (type): Remove "brinc".
21697 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21699 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
21700 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
21701 * config/rs6000/linuxspe.h: Delete file.
21702 * config/rs6000/rs6000.md: Don't include spe.md.
21703 * config/rs6000/spe.h: Delete file.
21704 * config/rs6000/spe.md: Delete file.
21705 * config/rs6000/t-rs6000: Remove spe.md.
21707 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21709 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
21710 (reg_or_none500mem_operand): Delete.
21711 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
21712 instead of reg_or_none500mem_operand.
21714 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21716 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21717 handling of SPE flags.
21718 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
21720 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21722 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
21724 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
21725 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
21726 paired_divv2sf3): Similar.
21727 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
21728 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
21729 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
21731 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
21732 Rename the paired_* instruction patterns.
21733 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
21735 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
21736 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
21737 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
21738 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
21739 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
21740 PAIRED_VECTOR_MODE.
21741 (struct machine_function): Delete field spe_insn_chain_scanned_p.
21742 (spe_func_has_64bit_regs_p): Delete.
21743 (spe_expand_predicate_builtin): Delete.
21744 (spe_expand_evsel_builtin): Delete.
21745 (TARGET_DWARF_REGISTER_SPAN): Do not define.
21746 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
21747 (invalid_e500_subreg): Delete.
21748 (rs6000_legitimize_address): Always force_reg op2 as well, for
21749 paired single memory accesses.
21750 (rs6000_member_type_forces_blk): Delete.
21751 (rs6000_spe_function_arg): Delete.
21752 (rs6000_expand_unop_builtin): Delete SPE handling.
21753 (rs6000_expand_binop_builtin): Ditto.
21754 (spe_expand_stv_builtin): Delete.
21755 (bdesc_2arg_spe): Delete.
21756 (spe_expand_builtin): Delete.
21757 (spe_expand_predicate_builtin): Delete.
21758 (spe_expand_evsel_builtin): Delete.
21759 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
21760 (spe_init_builtins): Delete.
21761 (spe_func_has_64bit_regs_p): Delete.
21762 (savres_routine_name): Delete "info" parameter. Adjust callers.
21763 (rs6000_emit_stack_reset): Ditto.
21764 (rs6000_dwarf_register_span): Delete.
21765 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
21766 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
21767 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
21769 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
21771 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
21772 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
21773 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
21774 mulv2sf3, divv2sf3): Delete expanders.
21776 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21778 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
21780 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21782 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
21783 * config/rs6000/rs6000.c: Ditto.
21785 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21787 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
21788 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
21789 comparison_operator.
21791 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21793 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
21794 * config/rs6000/rs6000.opt: Ditto.
21795 * config/rs6000/t-rtems: Ditto.
21797 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21799 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
21800 TARGET_E500_SINGLE by 0, simplify.
21801 * config/rs6000/rs6000.c: Ditto.
21802 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
21803 (spe_build_register_parallel): Delete.
21804 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
21805 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
21806 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
21807 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
21808 (E500_CONVERT): Delete.
21809 * config/rs6000/spe.md: Remove many patterns and all define_constants.
21811 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21813 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
21814 * config/rs6000/dfp.md: Ditto.
21815 (negdd2, *negdd2_fpr): Merge.
21816 (absdd2, *absdd2_fpr): Merge.
21817 (negtd2, *negtd2_fpr): Merge.
21818 (abstd2, *abstd2_fpr): Merge.
21819 * config/rs6000/e500.h: Delete file.
21820 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
21821 TARGET_FPRS by 1 and simplify.
21822 * config/rs6000/rs6000-c.c: Ditto.
21823 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
21824 TARGET_DF_SPE by 0.
21825 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
21827 * config/rs6000/rs6000.md: Ditto.
21828 (floatdidf2, *floatdidf2_fpr): Merge.
21829 (move_from_CR_gt_bit): Delete.
21830 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
21831 (E500_CR_IOR_COMPARE): Delete.
21832 (All patterns that require !TARGET_FPRS): Delete.
21833 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
21835 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21837 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
21839 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21841 * graphds.c (add_edge): Intitialize edge's attached data.
21842 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
21843 pointer parameter. Call pointed function on each edge during
21844 graph traversing. Skip traversing the edge when the function
21846 (graphds_dfs, graphds_scc): Ditto.
21847 (for_each_edge): New parameter. Pass the new parameter to callback
21849 * graphds.h (skip_edge_callback): New function pointer type.
21850 (graphds_dfs, graphds_scc): New function pointer parameter.
21851 (graphds_edge_callback, for_each_edge): New parameter.
21853 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21855 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
21856 out code checking if runtime alias check is possible to below ...
21857 Call the new function.
21858 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
21859 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
21861 2017-06-07 Marek Polacek <polacek@redhat.com>
21864 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
21865 TYPE_OVERFLOW_WRAPS check.
21867 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21869 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
21870 if versioning is required.
21871 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
21872 peeling with the check for versioning.
21874 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21876 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
21877 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
21878 Set true to new parameter if new ssa variable is defined.
21879 (vect_gen_vector_loop_niters): Refactor. Set range information
21880 for the new vector loop bound variable.
21881 (vect_do_peeling): Ditto.
21883 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21885 * tree-affine.c (ssa.h): Include header file.
21886 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
21887 has wrapping overflow behavior.
21889 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21891 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
21893 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21895 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
21896 (tree_to_aff_combination): ... here.
21898 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21900 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
21901 reg_pressure model function.
21902 (ivopts_global_cost_for_size): Delete.
21903 (determine_set_costs, iv_ca_recount_cost): Call new model function
21904 ivopts_estimate_reg_pressure.
21906 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21908 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
21909 expensive than udiv. Remove floating point cases from mod.
21911 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21913 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
21914 Increase idiv cost.
21916 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21918 * config/aarch64/aarch64.md
21919 (copysignsf3): Fix mask generation.
21921 2017-06-07 Jakub Jelinek <jakub@redhat.com>
21923 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
21925 (class dump_manager): Add register_dumps method.
21926 * dumpfile.c: Include langhooks.h.
21927 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
21928 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
21929 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
21930 (dump_manager::dump_register): Start with 512 entries instead of 32.
21931 (dump_manager::register_dumps): New method.
21932 * toplev.c (general_init): Instead of invoking register_dumps
21933 langhook, invoke register_dumps method on the dump manager.
21934 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
21937 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
21939 * doc/md.texi: Clarify the restrictions on a define_insn condition.
21940 Say that # requires an associated define_split to exist, and that
21941 the define_split must be suitable for use after register allocation.
21943 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21945 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
21946 (compute_outgoing_frequencies): Also initialize zero counts.
21947 (find_many_sub_basic_blocks): Do not produce uninitialized profile
21948 around loops; preserve more of profile when nothing changes.
21950 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
21952 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
21954 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
21955 * config/arm/arm-cpu-cdata.h: Regenerate.
21956 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
21957 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
21958 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
21959 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
21960 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
21962 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
21964 * config/arm/t-rmprofile: Likewise.
21965 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
21967 2017-06-06 David S. Miller <davem@davemloft.net>
21970 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
21971 blockage if function uses alloca.
21973 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21975 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
21976 New "uid" fields to hold pretty-print IDs of group and ref.
21977 Memory references are now identified as <group_id>:<ref_id>
21978 instead of using [random] addresses.
21979 (dump_mem_details): Simplify, no functional change.
21980 (dump_mem_ref): Simplify and make output more concise.
21981 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
21982 (find_or_create_group): Initialize group uid.
21983 (record_ref): Initialize ref uid. Improve debug output.
21984 (prune_group_by_reuse, should_issue_prefetch_p,)
21985 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
21986 (mark_nontemporal_store, determine_loop_nest_reuse):
21987 Improve debug output.
21989 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21991 * dbgcnt.def (prefetch): New debug counter.
21992 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
21993 (schedule_prefetches): Stop issueing prefetches if debug counter
21996 2017-06-06 Tom de Vries <tom@codesourcery.com>
21998 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
21999 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22001 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22003 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22004 Use aarch64_reg_or_zero predicate for operand 4.
22005 (aarch64_compare_and_swap<mode> define_insn_and_split):
22006 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22007 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22009 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22011 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22012 (arm_compute_save_core_reg_mask): This.
22013 (thumb1_compute_save_reg_mask): Rename into ...
22014 (thumb1_compute_save_core_reg_mask): This.
22015 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22016 (arm_compute_frame_layout): Likewise.
22018 2017-06-06 Richard Biener <rguenther@suse.de>
22020 PR tree-optimization/80974
22021 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22022 keep or clear leaders SSA info.
22024 2017-06-06 Tom de Vries <tom@codesourcery.com>
22026 * config/nvptx/nvptx.c (split_mode_p): New function.
22027 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22029 2017-06-06 Tom de Vries <tom@codesourcery.com>
22031 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22033 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22036 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22039 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22041 * shrink-wrap.c (handle_simple_exit): Update profile.
22042 (try_shrink_wrapping): Upate profile.
22044 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22046 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22047 (tree_guess_outgoing_edge_probabilities): New.
22048 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22049 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22051 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22053 * ipa-split.c (split_function): Initialize return bb profile.
22055 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22057 * profile.c (compute_branch_probabilities): Also initialize
22058 EXIT_BLOCK profile.
22060 2017-06-06 Richard Biener <rguenther@suse.de>
22062 PR tree-optimization/80928
22063 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22064 (vect_analyze_loop_operations): Properly guard analysis for
22066 (vect_transform_loop): Likewise.
22067 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22068 (vect_model_induction_cost): Do not cost for pure SLP.
22069 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22070 of induction in inner loop vectorization.
22071 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22072 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22073 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22075 (vect_analyze_slp_cost_1): Cost induction.
22076 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22077 (vect_get_slp_vect_defs): Likewise.
22078 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22079 (vect_transform_stmt): Handle SLP reductions.
22080 * tree-vectorizer.h (vectorizable_induction): Adjust.
22082 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22084 * config/rs6000/rs6000.c (make_resolver_func): Update
22085 init_lowered_empty_function call.
22087 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22089 * doc/invoke.texi: Document the -fprofile-abs-path option.
22090 * common.opt (fprofile-abs-path): New option.
22091 * gcov-io.h (gcov_write_filename): Declare.
22092 * gcov-io.c (gcov_write_filename): New function.
22093 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22094 * profile.c (output_location): Likewise.
22096 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22098 * shring-wrap.c: Revert accidental commit.
22100 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22102 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22104 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22106 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22108 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22109 profile in callgraph edge.
22110 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22111 (apply_scale): Likewise.
22112 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22113 Also scale profile when inlining function with zero profile.
22114 (initialize_cfun): Update exit block profile even when it is zero.
22115 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22116 when profile is read.
22118 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22120 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22121 (CLONE_*): New constants to define the processors we can generate
22122 code for with the target_clone attribute.
22123 (rs6000_clone_map): New array to identify which clone processors
22124 the current program is running on.
22125 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22126 target_clone attribute.
22127 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22128 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22129 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22130 (cpu_expand_builtin): Add support for target_clone attribute.
22131 (rs6000_valid_attribute_p): Allow "default" attribute.
22132 (get_decl_name): New debug function to simplify printing the
22133 current function name in debugging statements.
22134 (rs6000_clone_priority): New functions to support the target_clone
22135 attribute, and be able to generate code to switch between ISA 2.05
22136 through ISA 3.0 (power6 through power9).
22137 (rs6000_compare_version_priority): Likewise.
22138 (rs6000_get_function_versions_dispatcher): Likewise.
22139 (make_resolver_func): Likewise.
22140 (add_condition_to_bb): Likewise.
22141 (dispatch_function_versions): Likewise.
22142 (rs6000_generate_version_dispatcher_body): Likewise.
22143 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22144 (fusion_gpr_load_p): Fix a spacing issue.
22145 * doc/extend.texi (Common Function Attributes): Document that the
22146 PowerPC supports the target_clone attribute.
22148 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22150 * config/arm/arm.h: explain F symbol found in description of ARM
22151 register allocation in its legend.
22153 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22155 * config/mips/frame-header-opt.c: Include profile-count.h.
22156 * config/riscv/riscv.c: Include profile-count.h
22158 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22160 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22162 (sm_set_flag_if_changed): Add bbs field.
22163 (execute_sm_if_changed_flag_set): Pass BBS.
22164 (execute_sm): Update.
22166 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22168 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22171 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22173 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22174 (peephole2): New peephole2 to emit the above.
22175 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22177 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22179 * config/aarch64/aarch64.c (define_peephole2 above
22180 *sub_<shift>_<mode>): New peephole.
22182 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22184 * config/i386/i386.c (make_resolver_func): Update.
22185 * Makefile.in: Add profile-count.h and profile-count.o
22186 * auto-profile.c (afdo_indirect_call): Update to new API.
22187 (afdo_set_bb_count): Update.
22188 (afdo_propagate_edge): Update.
22189 (afdo_propagate_circuit): Update.
22190 (afdo_calculate_branch_prob): Update.
22191 (afdo_annotate_cfg): Update.
22192 * basic-block.h: Include profile-count.h
22193 (struct edge_def): Turn count to profile_count.
22194 (struct basic_block_def): Likewie.
22195 (REG_BR_PROB_BASE): Move to profile-count.h
22196 (RDIV): Move to profile-count.h
22197 * bb-reorder.c (max_entry_count): Turn to profile_count.
22198 (find_traces): Update.
22199 (rotate_loop):Update.
22200 (connect_traces):Update.
22201 (sanitize_hot_paths):Update.
22202 * bt-load.c (migrate_btr_defs): Update.
22203 * cfg.c (RDIV): Remove.
22204 (init_flow): Use alloc_block.
22205 (alloc_block): Uninitialize count.
22206 (unchecked_make_edge): Uninitialize count.
22207 (check_bb_profile): Update.
22208 (dump_edge_info): Update.
22209 (dump_bb_info): Update.
22210 (update_bb_profile_for_threading): Update.
22211 (scale_bbs_frequencies_int): Update.
22212 (scale_bbs_frequencies_gcov_type): Update.
22213 (scale_bbs_frequencies_profile_count): New.
22214 * cfg.h (update_bb_profile_for_threading): Update.
22215 (scale_bbs_frequencies_profile_count): Declare.
22216 * cfgbuild.c (compute_outgoing_frequencies): Update.
22217 (find_many_sub_basic_blocks): Update.
22218 * cfgcleanup.c (try_forward_edges): Update.
22219 (try_crossjump_to_edge): Update.
22220 * cfgexpand.c (expand_gimple_tailcall): Update.
22221 (construct_exit_block): Update.
22222 * cfghooks.c (verify_flow_info): Update.
22223 (dump_bb_for_graph): Update.
22224 (split_edge): Update.
22225 (make_forwarder_block): Update.
22226 (duplicate_block): Update.
22227 (account_profile_record): Update.
22228 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22229 (get_estimated_loop_iterations): Update.
22230 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22231 (single_likely_exit): Update.
22232 * cfgloopmanip.c (scale_loop_profile): Update.
22234 (set_zero_probability): Update.
22235 (lv_adjust_loop_entry_edge): Update.
22236 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22237 (purge_dead_edges): Update.
22238 (rtl_account_profile_record): Update.
22239 * cgraph.c (cgraph_node::create): Uninitialize count.
22240 (symbol_table::create_edge): Uninitialize count.
22241 (cgraph_update_edges_for_call_stmt_node): Update.
22242 (cgraph_edge::dump_edge_flags): Update.
22243 (cgraph_node::dump): Update.
22244 (cgraph_edge::maybe_hot_p): Update.
22245 * cgraph.h: Include profile-count.h
22246 (create_clone), create_edge, create_indirect_edge): Update.
22247 (cgraph_node): Turn count to profile_count.
22248 (cgraph_edge0: Likewise.
22249 (make_speculative, clone): Update.
22250 (create_edge): Update.
22251 (init_lowered_empty_function): Update.
22252 * cgraphclones.c (cgraph_edge::clone): Update.
22253 (duplicate_thunk_for_node): Update.
22254 (cgraph_node::create_clone): Update.
22255 * cgraphunit.c (cgraph_node::analyze): Update.
22256 (cgraph_node::expand_thunk): Update.
22257 * final.c (dump_basic_block_info): Update.
22258 * gimple-streamer-in.c (input_bb): Update.
22259 * gimple-streamer-out.c (output_bb): Update.
22260 * graphite.c (print_global_statistics): Update.
22261 (print_graphite_scop_statistics): Update.
22262 * hsa-brig.c: Include basic-block.h.
22263 * hsa-dump.c: Include basic-block.h.
22264 * hsa-gen.c (T sum_slice): Update.
22265 (convert_switch_statements):Update.
22266 * hsa-regalloc.c: Include basic-block.h.
22267 * ipa-chkp.c (chkp_produce_thunks): Update.
22268 * ipa-cp.c (struct caller_statistics): Update.
22269 (init_caller_stats): Update.
22270 (gather_caller_stats): Update.
22271 (ipcp_cloning_candidate_p): Update.
22272 (good_cloning_opportunity_p): Update.
22273 (get_info_about_necessary_edges): Update.
22274 (dump_profile_updates): Update.
22275 (update_profiling_info): Update.
22276 (update_specialized_profile): Update.
22277 (perhaps_add_new_callers): Update.
22278 (decide_about_value): Update.
22279 (ipa_cp_c_finalize): Update.
22280 * ipa-devirt.c (struct odr_type_warn_count): Update.
22281 (struct decl_warn_count): Update.
22282 (struct final_warning_record): Update.
22283 (possible_polymorphic_call_targets): Update.
22284 (ipa_devirt): Update.
22285 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22286 * ipa-icf.c (sem_function::merge): Update.
22287 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22288 * ipa-inline.c (compute_uninlined_call_time): Update.
22289 (compute_inlined_call_time): Update.
22290 (want_inline_small_function_p): Update.
22291 (want_inline_self_recursive_call_p): Update.
22292 (edge_badness): Update.
22293 (lookup_recursive_calls): Update.
22294 (recursive_inlining): Update.
22295 (inline_small_functions): Update.
22296 (dump_overall_stats): Update.
22297 (dump_inline_stats): Update.
22298 * ipa-profile.c (ipa_profile_generate_summary): Update.
22299 (ipa_propagate_frequency): Update.
22300 (ipa_profile): Update.
22301 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22302 * ipa-utils.c (ipa_merge_profiles): Update.
22303 * loop-doloop.c (doloop_modify): Update.
22304 * loop-unroll.c (report_unroll): Update.
22305 (unroll_loop_runtime_iterations): Update.
22306 * lto-cgraph.c (lto_output_edge): Update.
22307 (lto_output_node): Update.
22308 (input_node): Update.
22309 (input_edge): Update.
22310 (merge_profile_summaries): Update.
22311 * lto-streamer-in.c (input_cfg): Update.
22312 * lto-streamer-out.c (output_cfg): Update.
22313 * mcf.c (create_fixup_graph): Update.
22314 (adjust_cfg_counts): Update.
22315 (sum_edge_counts): Update.
22316 * modulo-sched.c (sms_schedule): Update.
22317 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22318 * predict.c (maybe_hot_count_p): Update.
22319 (probably_never_executed): Update.
22320 (dump_prediction): Update.
22321 (combine_predictions_for_bb): Update.
22322 (propagate_freq): Update.
22323 (handle_missing_profiles): Update.
22324 (counts_to_freqs): Update.
22325 (rebuild_frequencies): Update.
22326 (force_edge_cold): Update.
22327 * predict.h: Include profile-count.h
22328 (maybe_hot_count_p, counts_to_freqs): UPdate.
22329 * print-rtl-function.c: Do not include cfg.h
22330 * print-rtl.c: Include basic-block.h
22331 * profile-count.c: New file.
22332 * profile-count.h: New file.
22333 * profile.c (is_edge_inconsistent): Update.
22334 (correct_negative_edge_counts): Update.
22335 (is_inconsistent): Update.
22336 (set_bb_counts): Update.
22337 (read_profile_edge_counts): Update.
22338 (compute_frequency_overlap): Update.
22339 (compute_branch_probabilities): Update; Initialize and deinitialize
22341 (branch_prob): Update.
22342 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
22343 (edge_gcov_count): New.
22344 (bb_gcov_count): New.
22345 * shrink-wrap.c (try_shrink_wrapping): Update.
22346 * tracer.c (better_p): Update.
22347 * trans-mem.c (expand_transaction): Update.
22348 (ipa_tm_insert_irr_call): Update.
22349 (ipa_tm_insert_gettmclone_call): Update.
22350 * tree-call-cdce.c: Update.
22351 * tree-cfg.c (gimple_duplicate_sese_region): Update.
22352 (gimple_duplicate_sese_tail): Update.
22353 (gimple_account_profile_record): Update.
22354 (execute_fixup_cfg): Update.
22355 * tree-inline.c (copy_bb): Update.
22356 (copy_edges_for_bb): Update.
22357 (initialize_cfun): Update.
22358 (freqs_to_counts): Update.
22359 (copy_cfg_body): Update.
22360 (expand_call_inline): Update.
22361 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22362 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
22363 (try_unroll_loop_completely): Update.
22364 (try_peel_loop): Update.
22365 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22366 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22367 * tree-ssa-loop-split.c (connect_loops): Update.
22368 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22369 * tree-ssa-reassoc.c (branch_fixup): Update.
22370 * tree-ssa-tail-merge.c (replace_block_by): Update.
22371 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
22372 (compute_path_counts): Update.
22373 (update_profile): Update.
22374 (recompute_probabilities): Update.
22375 (update_joiner_offpath_counts): Update.
22376 (estimated_freqs_path): Update.
22377 (freqs_to_counts_path): Update.
22378 (clear_counts_path): Update.
22379 (ssa_fix_duplicate_block_edges): Update.
22380 (duplicate_thread_path): Update.
22381 * tree-switch-conversion.c (case_bit_test_cmp): Update.
22382 (struct switch_conv_info): Update.
22383 * tree-tailcall.c (decrease_profile): Update.
22384 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22385 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22386 * value-prof.c (check_counter): Update.
22387 (gimple_divmod_fixed_value): Update.
22388 (gimple_mod_pow2): Update.
22389 (gimple_mod_subtract): Update.
22390 (gimple_ic_transform): Update.
22391 (gimple_stringop_fixed_value): Update.
22392 * value-prof.h (gimple_ic): Update.
22394 2017-06-02 Carl Love <cel@us.ibm.com>
22396 * config/rs6000/rs6000-c: Add support for built-in functions
22397 vector double vec_doublee (vector signed int);
22398 vector double vec_doublee (vector unsigned int);
22399 vector double vec_doublee (vector float);
22400 vector double vec_doubleh (vector signed int);
22401 vector double vec_doubleh (vector unsigned int);
22402 vector double vec_doubleh (vector float);
22403 vector double vec_doublel (vector signed int);
22404 vector double vec_doublel (vector unsigned int);
22405 vector double vec_doublel (vector float);
22406 vector double vec_doubleo (vector signed int);
22407 vector double vec_doubleo (vector unsigned int);
22408 vector double vec_doubleo (vector float);.
22409 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
22410 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
22412 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
22413 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
22414 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
22416 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
22417 vec_doublel, vec_doubleh.
22418 * doc/extend.texi: Update the built-in documentation file for the
22419 new built-in functions.
22421 2017-06-02 David Malcolm <dmalcolm@redhat.com>
22424 * ipa-inline-analysis.c (free_growth_caches): Set
22425 edge_removal_hook_holder to NULL after removing it.
22427 2017-06-02 Sudakshina Das <sudi.das@arm.com>
22429 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
22430 comparision with zero.
22432 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22433 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22434 for early expansion of vec_min and vec_max builtins.
22435 (builtin_function_type): Add min/max unsigned variants to those
22436 identified as having unsigned arguments.
22438 2017-06-02 Olivier Hainque <hainque@adacore.com>
22440 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
22442 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22444 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
22445 Use VALL_F16 iterator rather than VALL.
22447 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22449 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
22450 Emit CBNZ inside loop when doing a strong exchange and comparing
22451 against zero. Generate the CC flags after the loop.
22453 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
22455 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
22456 (dl_section_ref): New.
22457 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
22458 On AIX, append an expression to subtract the size of the
22459 section length to dl_section_ref.
22461 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22463 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22464 for early expansion of vector absolute builtins.
22466 2017-06-02 Richard Biener <rguenther@suse.de>
22468 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
22469 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
22471 2017-06-02 Richard Biener <rguenther@suse.de>
22473 PR tree-optimization/80948
22474 * tree-tailcall.c (find_tail_calls): Track stmts to move in
22475 stmt order as well.
22477 2017-06-02 Richard Biener <rguenther@suse.de>
22479 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
22481 * tree-vect-stmts.c (process_use): Do not mark backedge defs
22482 for inductions as relevant.
22484 2017-06-02 Richard Biener <rguenther@suse.de>
22486 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
22487 (vectorizable_induction): ... this. Remove dead code.
22489 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
22491 * builtins. (expand_builtin_alloca): Remove second parameter and
22492 infer its value from the first parameter instead.
22493 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
22495 2017-06-02 Jakub Jelinek <jakub@redhat.com>
22497 PR rtl-optimization/80903
22498 * loop-doloop.c (add_test): Unshare sequence.
22500 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
22502 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
22504 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
22506 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
22508 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
22509 xlogue_layout::get_instance, logue_layout::xlogue_layout,
22510 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
22511 (xlogue_layout::get_stub_rtx): Make static.
22512 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
22513 (xlogue_layout::compute_stub_managed_regs): Rename to...
22514 (xlogue_layout::count_stub_managed_regs): ...this.
22515 (xlogue_layout::is_stub_managed_reg): New function.
22516 (xlogue_layout::m_stub_names): Rename to...
22517 (xlogue_layout::s_stub_names): ...this, make static.
22518 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
22519 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
22520 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
22521 xlogue_layout::s_stub_names): Instantiate statics.
22522 (stub_managed_regs): Remove.
22523 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
22524 (disable_call_ms2sysv_xlogues): Rename to...
22525 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
22526 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
22528 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
22529 change after reload_completed.
22530 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
22532 (ix86_expand_prologue): Likewise.
22533 (ix86_expand_epilogue): Likewise.
22534 (ix86_expand_split_stack_prologue): Likewise.
22535 (ix86_compute_frame_layout): Remove frame parameter ...
22536 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
22537 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
22539 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
22540 (ix86_frame): Move from here ...
22541 * config/i386/i386.h (ix86_frame): ... to here.
22542 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
22543 complete ix86_frame data structure instead. Remove some_ld_name.
22545 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22547 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
22548 symbols that hold a DECL_VALUE_EXPR.
22550 2017-06-01 Martin Jambor <mjambor@suse.cz>
22552 PR tree-optimization/80898
22553 * tree-sra.c (process_subtree_disqualification): Removed.
22554 (disqualify_candidate): Do not acll
22555 process_subtree_disqualification.
22556 (subtree_mark_written_and_enqueue): New function.
22557 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
22558 RHS has been disqualified and re-queue LHS if necessary. Apart
22559 from that, ignore disqualified RHS.
22561 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22563 * config/s390/s390.c (s390_emit_epilogue): Disable early return
22564 address fetch for z10 or later.
22566 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22568 * config/arc/arc.md (tst_movb): Add guard when splitting.
22570 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22572 * config/arc/arc.c (arc_can_eliminate): Test against
22573 arc_frame_pointer_needed.
22575 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22577 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
22578 to prevent store reordering.
22579 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
22580 (type): Add block type.
22581 (stack_tie): Define special instruction to be used in
22584 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22586 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
22587 constraint. It is not valid for the pattern.
22588 (noncommutative_binary_comparison): Likewise.
22590 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22592 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
22595 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22597 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
22598 be used by the reg-alloc.
22600 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22602 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
22603 reg-alloc when having mul64 or mul32x16 instructions.
22604 (mulsidi3): Likewise.
22605 (umulsidi3): Likewise.
22606 (mulsi32x16): New pattern.
22607 (mulsi64): Likewise.
22608 (mulsidi64): Likewise.
22609 (umulsidi64): Likewise.
22610 (MUL32x16_REG): Define.
22611 (mul64_600): Use MUL32x16_REG.
22612 (mac64_600): Likewise.
22613 (umul64_600): Likewise.
22614 (umac64_600): Likewise.
22616 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22618 * config/arc/arc.md (mulsi3_700): Make it commutative.
22620 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
22622 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
22624 (*sign_extendsidi2_insn): Likewise for movstosw.
22626 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22628 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
22629 the type of the input discriminant value. Convert the
22630 discriminant value of signedness vary.
22632 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
22634 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
22635 Add to -Wall section.
22637 2017-06-01 Richard Biener <rguenther@suse.de>
22639 PR middle-end/66313
22640 * fold-const.c (fold_plusminus_mult_expr): If the factored
22641 factor may be zero use a wrapping type for the inner operation.
22642 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
22643 and handle moved defs.
22644 (process_assignment): Properly guard the unary op case. Return a
22645 tri-state indicating that moving the stmt before the call may allow
22646 to continue. Pass through to_move.
22647 (find_tail_calls): Handle moving unrelated defs before
22650 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
22653 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
22654 splitter result in the canonical way.
22656 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
22658 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
22659 also for 32bit target. Update insn attributes.
22660 (zero-extendsidi2 splitter): Allow all registers for operand 1.
22662 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
22664 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
22665 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
22666 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
22667 (_mm_maskz_min_ss): New intrinsics.
22669 2017-05-31 Martin Liska <mliska@suse.cz>
22671 * tree-vect-loop.c (vect_create_epilog_for_reduction):
22672 Change comment style to one we normally use.
22673 (vectorizable_reduction): Likewise.
22674 (vectorizable_induction): Likewise.
22675 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
22676 (vectorizable_call): Likewise.
22677 (vectorizable_simd_clone_call): Likewise.
22678 (vectorizable_conversion): Likewise.
22679 (vectorizable_assignment): Likewise.
22680 (vectorizable_shift): Likewise.
22681 (vectorizable_operation): Likewise.
22682 (vectorizable_store): Likewise.
22683 (vectorizable_load): Likewise.
22684 * tree-vectorizer.h: Likewise.
22686 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
22688 * passes.c (emergency_dump_function): New.
22689 * tree-pass.h (emergency_dump_function): Declare.
22690 * plugin.c (plugins_internal_error_function): Remove.
22691 * plugin.h (plugins_internal_error_function): Remove declaration.
22692 * toplev.c (internal_error_function): New static function. Use it...
22693 (general_init): ...here.
22695 2017-05-31 Graham Markall <graham.markall@embecosm.com>
22697 * config/arc/arc.c (arc_print_operand): Handle constant operands.
22698 (arc_rtx_costs): Add costs for new patterns.
22699 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
22700 * config/arc/predicates.md: Add _1_2_3_operand predicate.
22702 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22704 * tree-ssa-strlen.c (get_next_strinfo): New function.
22705 (get_stridx_plus_constant): Use it.
22706 (zero_length_string): Likewise.
22707 (adjust_related_strinfos): Likewise.
22708 (adjust_last_stmt): Likewise.
22710 2017-05-31 Richard Biener <rguenther@suse.de>
22713 * config/i386/i386.c (ix86_expand_builtin): Remove assert
22714 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
22716 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22718 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
22719 loop_vinfo argument and use of dependence distance vectors.
22720 Check instead whether the two references differ only in their
22721 initial value and assume that they have the same alignment if the
22722 difference is a multiple of the vector alignment.
22723 (vect_analyze_data_refs_alignment): Update call accordingly.
22725 2017-05-31 Martin Liska <mliska@suse.cz>
22728 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
22730 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22732 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
22733 (create_intersect_range_checks): Move from ...
22734 * tree-data-ref.c (create_intersect_range_checks_index)
22735 (create_intersect_range_checks): ... to here.
22736 (create_runtime_alias_checks): New function factored from ...
22737 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
22738 here. Call above function.
22739 * tree-data-ref.h (create_runtime_alias_checks): New function.
22741 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22743 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
22744 segment length for dr_b and compute it in wide_int.
22746 2017-05-31 Richard Biener <rguenther@suse.de>
22748 PR tree-optimization/80906
22749 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
22750 and pass through iv_map.
22751 (copy_bb_and_scalar_dependences): Adjust.
22752 (translate_pending_phi_nodes): Likewise.
22753 (copy_loop_close_phi_args): Handle code-generating IVs instead
22756 2017-05-30 David Malcolm <dmalcolm@redhat.com>
22758 * diagnostic-color.c (color_dict): Add "type-diff".
22759 (parse_gcc_colors): Update comment.
22760 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
22761 -fdiagnostics-show-template-tree and -fno-elide-type.
22762 (GCC_COLORS): Add type-diff to example.
22764 (-fdiagnostics-show-template-tree): New.
22765 (-fno-elide-type): New.
22766 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
22767 the pp_format_decoder callback. Call any m_format_postprocessor's
22769 (pretty_printer::pretty_printer): Initialize
22770 m_format_postprocessor.
22771 (pretty_printer::~pretty_printer): Delete any
22772 m_format_postprocessor.
22773 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
22774 (class format_postprocessor): New class.
22775 (struct pretty_printer::format_decoder): Document the new parameters.
22776 (struct pretty_printer::m_format_postprocessor): New field.
22777 * tree-diagnostic.c (default_tree_printer): Update for new
22778 bool and const char ** params.
22779 * tree-diagnostic.h (default_tree_printer): Likewise.
22781 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
22783 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
22784 (lwa_operand): Delete rs6000_gen_cell_microcode test.
22785 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22786 rs6000_gen_cell_microcode code.
22787 (rs6000_final_prescan_insn): Delete.
22788 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
22789 "warn-cell-microcode" entries.
22790 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
22791 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
22792 throughout. Change cc_reg_not_micro_cr0_operand to
22793 cc_reg_not_cr0_operand throughout.
22794 (*extendhi<mode>2_noload): Delete.
22795 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
22796 (mwarn-cell-microcode): Delete.
22797 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
22798 -mgen-cell-microcode and -mwarn-cell-microcode.
22800 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
22803 * config/i386/constraints.md (Yd): New constraint.
22805 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
22806 and (?Yd, r) alternatives. Update insn attributes.
22807 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
22808 and (?*Yd, r) alternatives. Update insn attributes.
22809 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
22811 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
22813 * gimplify.c (gimplify_modify_expr): Don't create a
22814 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
22817 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
22819 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
22821 2017-05-30 Richard Biener <rguenther@suse.de>
22823 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
22824 and reduc_def fields.
22825 (STMT_VINFO_REDUC_TYPE): New define.
22826 (STMT_VINFO_REDUC_DEF): Likewise.
22827 (vect_force_simple_reduction): Adjust prototype.
22828 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
22829 (vect_is_simple_reduction): Remove check_reduction argument.
22830 (vect_force_simple_reduction): Adjust and set
22831 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22832 (vectorizable_reduction): Do not re-do reduction analysis
22833 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22834 * tree-parloops.c (gather_scalar_reductions): Adjust.
22836 2017-05-30 Richard Biener <rguenther@suse.de>
22838 PR middle-end/80901
22839 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
22842 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
22844 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
22845 Introduce unknown_misalignment parameter and remove vf.
22846 (vect_peeling_hash_get_lowest_cost):
22847 Pass unknown_misalignment parameter.
22848 (vect_enhance_data_refs_alignment):
22849 Fix unsupportable data ref treatment.
22851 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22853 * tree-vect-data-refs.c (vect_get_data_access_cost):
22854 Workaround for SLP handling.
22855 (vect_enhance_data_refs_alignment):
22856 Compute costs for doing no peeling at all, compare to the best
22857 peeling costs so far and avoid peeling if cheaper.
22859 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22861 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
22862 Return peeling info and set costs to zero for unlimited cost
22864 (vect_enhance_data_refs_alignment): Also inspect all datarefs
22865 with unknown misalignment. Compute and costs for unknown
22866 misalignment, compare them to the costs for known misalignment
22867 and choose the cheapest for peeling.
22869 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22871 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
22872 (vect_get_peeling_costs_all_drs): Create function.
22873 (vect_peeling_hash_get_lowest_cost):
22874 Use vect_get_peeling_costs_all_drs.
22875 (vect_peeling_supportable): Create function.
22876 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
22878 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22880 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
22881 DR_HAS_NEGATIVE_STEP.
22882 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
22883 (vect_enhance_data_refs_alignment): Use.
22884 (vect_duplicate_ssa_name_ptr_info): Use.
22885 * tree-vectorizer.h (dr_misalignment): Use.
22886 (known_alignment_for_access_p): Use.
22888 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22891 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
22893 (has_section_name): New function.
22895 2017-05-30 Martin Liska <mliska@suse.cz>
22898 * auto-profile.c (get_function_decl_from_block): Fix
22901 2017-05-30 Richard Biener <rguenther@suse.de>
22903 PR middle-end/80876
22904 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
22906 2017-05-30 Martin Liska <mliska@suse.cz>
22908 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
22909 * dumpfile.h (struct dump_file_info): Remove ctors.
22911 2017-05-30 Martin Liska <mliska@suse.cz>
22913 * predict.def: Fix GNU coding style.
22915 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22917 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22918 Mark 'to' argument with ATTRIBUTE_UNUSED.
22920 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22922 * config/xtensa/xtensa.c (xtensa_emit_call): Use
22923 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
22924 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
22927 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
22929 * doc/install.texi (Options specification): Restore entry of
22930 --enable-sjlj-exceptions.
22932 2017-05-27 Michael Eager <eager@eagercon.com>
22935 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
22937 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
22939 * config/microblaze/microblaze.h
22940 (FIXED_REGISTERS): Update in macro.
22941 (CALL_USED_REGISTERS): Update in macro.
22943 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
22945 * doc/install.texi: Add links to macOS binary distributions.
22947 2017-05-27 Jakub Jelinek <jakub@redhat.com>
22951 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22953 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22955 2017-05-26 Martin Liska <mliska@suse.cz>
22957 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
22959 2017-05-26 Martin Liska <mliska@suse.cz>
22961 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
22962 always leading ';; '.
22963 (dump_bb_info): Likewise.
22964 (brief_dump_cfg): Likewise.
22965 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
22966 * dumpfile.c: Remove usage of TDF_VERBOSE.
22967 * dumpfile.h (enum dump_kind): Likewise.
22968 (dump_gimple_bb_header): Do not use TDF_COMMENT.
22969 * print-tree.c (debug_verbose): Remove.
22970 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
22971 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
22972 * tree-diagnostic.c (default_tree_printer): Replace
22973 TDF_DIAGNOSTIC with TDF_SLIM.
22975 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22977 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
22978 in parameter loop, rather than loop_vinfo.
22979 (create_intersect_range_checks): Ditto.
22980 (vect_create_cond_for_alias_checks): Update call to above functions.
22982 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22984 PR tree-optimization/80815
22985 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
22986 for merging runtime alias checks. Handle negative DR_STEPs.
22988 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22990 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
22992 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
22993 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
22994 out code pruning runtime alias checks.
22995 * tree-data-ref.c (prune_runtime_alias_test_list): New function
22996 factored out from above.
22997 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22999 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23001 (prune_runtime_alias_test_list): New decalaration.
23003 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23005 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23006 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23007 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23008 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23009 (operator==, comp_dr_with_seg_len_pair): Ditto.
23010 (vect_prune_runtime_alias_test_list): Ditto.
23012 2017-05-26 Martin Liska <mliska@suse.cz>
23015 * params.def: Bound partial-inlining-entry-probability param.
23017 2017-05-26 Marek Polacek <polacek@redhat.com>
23020 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23023 2017-05-26 Richard Biener <rguenther@suse.de>
23025 PR tree-optimization/80842
23026 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23029 2017-05-26 Richard Biener <rguenther@suse.de>
23031 PR tree-optimization/80844
23032 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23034 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23036 * doc/md.texi (Machine Constraints): Update x86 family
23037 machine constraints section to match 'config/i386/constraints.md'.
23039 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23041 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23043 2017-05-25 Nathan Sidwell <nathan@acm.org>
23045 * doc/invoke.texi (--enable-languages): Update documentation.
23047 2017-05-25 Martin Liska <mliska@suse.cz>
23049 * dumpfile.c: Add TDF_FOLDING.
23050 * dumpfile.h (enum dump_kind): Likewise.
23051 * genmatch.c (dt_simplify::gen_1): Use it.
23053 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23055 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23057 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23059 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23060 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23062 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23064 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23065 * match.pd (X == C): Rewrite it here.
23066 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23067 with_certain_nonzero_bits2): New predicates.
23068 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23070 2017-05-24 Nathan Sidwell <nathan@acm.org>
23072 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23075 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23078 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23080 * config/powerpcspe: New port. Files are copied from the rs6000
23081 port, with "rs6000" in filenames replaced by "powerpcspe".
23083 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23085 PR rtl-optimization/80754
23086 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23088 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23090 * config/sparc/sparc.md (length): Return the correct value for -mflat
23091 sibcalls to match output_sibcall.
23093 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23097 * config/rs6000/rs6000.c (struct machine_function): Add new field
23099 (rs6000_get_separate_components): Init that field, use it.
23100 (rs6000_components_for_bb): Use the field.
23102 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23104 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23106 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23108 PR middle-end/80823
23109 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23111 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23114 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23115 address against address_operand predicate.
23116 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23118 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23120 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23121 parameters passed indirectly.
23123 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23125 * config/i386/i386.md (*movdi_internal): Remove SSE4
23126 alternative 18 (?r, *v). Update insn attributes.
23127 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23128 Update insn attributes.
23129 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23130 Update insn attributes.
23131 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23132 alternative 1 (r, v). Remove isa attribute.
23133 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23134 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23135 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23137 2017-05-23 Tom de Vries <tom@codesourcery.com>
23139 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23142 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23144 * cgraphunit.c (symbol_table::process_new_functions): Update.
23145 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23146 (inline_generate_summary): Rename to ...
23147 (ipa_fn_summary_generate): ... this one.
23148 (inline_read_summary): Rename to ...
23149 (ipa_fn_summary_read): ... this one.
23150 (inline_write_summary): Rename to ...
23151 (ipa_fn_summary_write): ... this one.
23152 (inline_free_summary): Rename to ...
23153 (ipa_free_fn_summary): ... this one.
23154 (pass_data_local_fn_summary, pass_local_fn_summary,
23155 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23156 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23157 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23158 make_pass_ipa_fn_summary): New.
23159 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23160 inline_write_summary, inline_free_summary): Remove.
23161 (ipa_free_fn_summary) : New.
23162 * ipa-inline.c (ipa_inline): Update.
23163 (pass_ipa_inline): Do not generate summaries.
23164 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23166 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23167 and add pass_ipa_fn_summary.
23168 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23170 (make_pass_inline_parameters): Remove.
23172 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23174 * omp-low.c (struct omp_context): Remove "default_kind" member.
23177 * omp-offload.c (execute_oacc_device_lower): Remove the
23178 parallelism dimensions function attributes for unparallelized
23179 OpenACC kernels constructs.
23181 2017-05-23 Martin Liska <mliska@suse.cz>
23183 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23185 (cgraph_edge::make_speculative): Likewise.
23186 (cgraph_edge::resolve_speculation): Likewise.
23187 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23188 (cgraph_node::dump): Likewise.
23189 * cgraph.h: Likewise.
23190 * cgraphunit.c (analyze_functions): Likewise.
23191 (symbol_table::compile): Likewise.
23192 * ipa-cp.c (print_all_lattices): Likewise.
23193 (determine_versionability): Likewise.
23194 (initialize_node_lattices): Likewise.
23195 (ipcp_verify_propagated_values): Likewise.
23196 (estimate_local_effects): Likewise.
23197 (update_profiling_info): Likewise.
23198 (create_specialized_node): Likewise.
23199 (perhaps_add_new_callers): Likewise.
23200 (decide_about_value): Likewise.
23201 (decide_whether_version_node): Likewise.
23202 (identify_dead_nodes): Likewise.
23203 (ipcp_store_bits_results): Likewise.
23204 * ipa-devirt.c (dump_targets): Likewise.
23205 (ipa_devirt): Likewise.
23206 * ipa-icf.c (sem_item::dump): Likewise.
23207 (sem_function::equals): Likewise.
23208 (sem_variable::equals): Likewise.
23209 (sem_item_optimizer::read_section): Likewise.
23210 (sem_item_optimizer::execute): Likewise.
23211 (congruence_class::dump): Likewise.
23212 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23213 (dump_inline_summary): Likewise.
23214 (estimate_node_size_and_time): Likewise.
23215 (inline_analyze_function): Likewise.
23216 * ipa-inline-transform.c (inline_call): Likewise.
23217 * ipa-inline.c (report_inline_failed_reason): Likewise.
23218 (want_early_inline_function_p): Likewise.
23219 (edge_badness): Likewise.
23220 (update_edge_key): Likewise.
23221 (inline_small_functions): Likewise.
23222 * ipa-profile.c (ipa_profile): Likewise.
23223 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23224 (ipa_make_edge_direct_to_target): Likewise.
23225 (remove_described_reference): Likewise.
23226 (ipa_impossible_devirt_target): Likewise.
23227 (propagate_controlled_uses): Likewise.
23228 (ipa_print_node_params): Likewise.
23229 (ipcp_transform_function): Likewise.
23230 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23231 (propagate_pure_const): Likewise.
23232 * ipa-reference.c (generate_summary): Likewise.
23233 (read_write_all_from_decl): Likewise.
23234 (propagate): Likewise.
23235 (ipa_reference_read_optimization_summary): Likewise.
23236 * ipa-utils.c (ipa_merge_profiles): Likewise.
23237 * ipa.c (walk_polymorphic_call_targets): Likewise.
23238 (symbol_table::remove_unreachable_nodes): Likewise.
23239 (ipa_single_use): Likewise.
23240 * passes.c (execute_todo): Likewise.
23241 * predict.c (drop_profile): Likewise.
23242 * symtab.c (symtab_node::get_dump_name): New function.
23243 (symtab_node::dump_name): Likewise.
23244 (symtab_node::dump_asm_name): Likewise.
23245 (symtab_node::dump_references): Likewise.
23246 (symtab_node::dump_referring): Likewise.
23247 (symtab_node::dump_base): Likewise.
23248 (symtab_node::debug_symtab): Likewise.
23249 * tree-sra.c (convert_callers_for_node): Likewise.
23250 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23251 * value-prof.c (init_node_map): Likewise.
23253 2017-05-23 Martin Liska <mliska@suse.cz>
23255 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23256 and symtab_node::debug_symtab to symbol_table::debug.
23257 * cgraphunit.c (analyze_functions): Use the renamed function.
23258 (symbol_table::compile): Likewise.
23259 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23260 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23261 * passes.c (execute_todo): Likewise.
23262 * symtab.c (symbol_table::dump): New function.
23263 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23265 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23267 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23268 that nonconst implies exec.
23270 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23272 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23273 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23274 (inline_edge_summary_vec): Turn into ...
23275 (ipa_call_summaries): ... this one.
23276 (redirect_to_unreachable, edge_set_predicate,
23277 evaluate_properties_for_edge, inline_summary_alloc,
23278 reset_ipa_call_summary, reset_inline_summary,
23279 inline_summary_t::duplicate): Update.
23280 (inline_edge_duplication_hook): Turn to ...
23281 (ipa_call_summary_t::duplicate): ... this one.
23282 (inline_edge_removal_hook): Turn to ...
23283 (ipa_call_summary_t::remove): ... this one.
23284 (dump_inline_edge_summary): Turn to ...
23285 (dump_ipa_call_summary): ... this one.
23286 (estimate_function_body_sizes): Update.
23287 (inline_update_callee_summaries): Update.
23288 (remap_edge_change_prob): Update.
23289 (remap_edge_summaries): Update.
23290 (inline_merge_summary): Update.
23291 (do_estimate_edge_time): Update.
23292 (inline_generate_summary): Update.
23293 (inline_read_section): Update.
23294 (inline_read_summary): Update.
23295 (inline_free_summary): Update.
23296 * ipa-inline.c (can_inline_edge_p): Update.
23297 (compute_inlined_call_time): Update.
23298 (want_inline_small_function_p): Update.
23299 (edge_badness): Update.
23300 (early_inliner): Update.
23301 * ipa-inline.h (inline_edge_summary): Turn to ...
23302 (ipa_call_summary): ... this one.
23303 (ipa_call_summary_t): New class.
23304 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23305 (ipa_call_summaries): New.
23306 (inline_edge_summary): Remove.
23307 (estimate_edge_growth): Update.
23308 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23309 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23310 * ipa-split.c (execute_split_functions): Update.
23311 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23313 2017-05-23 Tom de Vries <tom@codesourcery.com>
23315 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23316 attributes): Document rdrand effective target.
23318 2017-05-23 Tom de Vries <tom@codesourcery.com>
23320 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23321 attributes): Sort alphabetically.
23323 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23325 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23327 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23330 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23331 V2DF/V2DI splat into two separate patterns, one that handles
23332 registers, and the other that only handles memory. Drop support
23333 for splatting from a GPR on ISA 2.07 and then splitting the
23334 splat into direct move and splat.
23335 (vsx_splat_<mode>_reg): Likewise.
23336 (vsx_splat_<mode>_mem): Likewise.
23338 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23340 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23342 2017-05-22 Jakub Jelinek <jakub@redhat.com>
23344 PR middle-end/80809
23345 * omp-low.c (finish_taskreg_remap): New function.
23346 (finish_taskreg_scan): If unit size of ctx->record_type
23347 is non-constant, unshare the size expression and replace
23348 decls in it with possible outer var refs.
23350 PR middle-end/80809
23351 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
23352 GOVD_SHARED rather than GOVD_PRIVATE with it.
23353 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
23354 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
23356 PR middle-end/80853
23357 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
23358 as last argument to build_outer_var_ref for pointer bases of array
23359 section reductions.
23361 2017-05-19 Martin Sebor <msebor@redhat.com>
23363 * print-tree.c (print_node): Print DECL_READ_P flag.
23365 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23367 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23368 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23369 * cgraph.c: Likewise.
23370 * cgraphunit.c: Likewise.
23371 * gengtype.c: Likewise.
23372 * ipa-cp.c: Likewise.
23373 * ipa-devirt.c: Likewise.
23374 * ipa-icf.c: Likewise.
23375 * ipa-predicate.c: Likewise.
23376 * ipa-profile.c: Likewise.
23377 * ipa-prop.c: Likewise.
23378 * ipa-split.c: Likewise.
23380 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
23381 edge_predicate_pool, dump_inline_hints,
23382 inline_summary::account_size_time, redirect_to_unreachable,
23383 edge_set_predicate, set_hint_predicate,
23384 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
23385 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
23386 inline_summary_t::remove, remap_hint_predicate_after_duplication,
23387 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
23388 ipa_call_summary_t::remove, initialize_growth_caches,
23389 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
23390 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
23391 mark_modified, unmodified_parm_1, unmodified_parm,
23392 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
23393 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
23394 compute_bb_predicates, will_be_nonconstant_expr_predicate,
23395 will_be_nonconstant_predicate, record_modified_bb_info,
23396 get_minimal_bb, record_modified, param_change_prob,
23397 phi_result_unknown_predicate, predicate_for_phi_result,
23398 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
23399 estimate_function_body_sizes, compute_inline_parameters,
23400 compute_inline_parameters_for_curren, pass_data_inline_parameters,
23401 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
23402 inline_update_callee_summaries, remap_edge_change_prob,
23403 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
23404 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
23405 inline_analyze_function, inline_summary_t::insert,
23406 inline_generate_summary, read_ipa_call_summary, inline_read_section,
23407 inline_read_summary, write_ipa_call_summary, inline_write_summary,
23408 inline_free_summary): Move to ipa-fnsummary.h
23409 (predicate_t): Remove.
23410 * ipa-fnsummary.c: New file.
23411 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
23412 (enum inline_hints_vals, inline_hints, agg_position_info,
23413 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
23414 inline_summaries, ipa_call_summary, ipa_call_summary_t,
23415 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
23416 dump_inline_summary, dump_inline_hints, inline_generate_summary,
23417 inline_read_summary, inline_write_summary, inline_free_summary,
23418 inline_analyze_function, initialize_inline_failed,
23419 inline_merge_summary, inline_update_overall_summary,
23420 compute_inline_parameters): Move to ipa-fnsummary.h
23421 * ipa-fnsummary.h: New file.
23422 * ipa-inline-transform.h: Include ipa-inline.h.
23423 * ipa-inline.c: LIkewise.
23425 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23427 * ipa-inline.c (edge_badness): Use inlined_time instead of
23428 inline_summaries->get.
23430 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23432 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
23434 2017-05-22 Nathan Sidwell <nathan@acm.org>
23436 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
23437 (fdump-lang): Document 'raw' option.
23438 * dumpfile.h (TDI_tu): Delete.
23439 * dumpfile.c (dump_files): Remove translation-unit.
23440 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
23442 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
23444 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
23445 command option from $(AWK) call.
23446 * config/avr/genmultilib.awk: Simplify and rewrite so that it
23447 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
23448 [FORMAT]: Remove handling of variable.
23449 * config/avr/t-multilib: Regenerate.
23451 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23453 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
23455 (dump_inline_summary): Do not print self_time.
23456 (estimate_function_body_sizes): Do not set self_time.
23457 (compute_inline_parameters): Likewise.
23458 (inline_read_section, inline_write_summary): Do not stream self_time.
23459 * ipa-inline.h (inline_summary): Drop self_time.
23461 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23463 * ipa-inline-analysis.c (account_size_time): Rename to ...
23464 (inline_summary::account_size_time): ... this one.
23465 (reset_ipa_call_summary): Turn to ...
23466 (ipa_call_summary::reset): ... this one.
23467 (reset_inline_summary): Turn to ...
23468 (inline_summary::reset): ... this one.
23469 (inline_summary_t::remove): Update.
23470 (inline_summary_t::duplicate): Update.
23471 (ipa_call_summary_t::remove): Update.
23472 (dump_inline_summary): Update.
23473 (estimate_function_body_sizes): Update.
23474 (compute_inline_parameters): Update.
23475 (estimate_node_size_and_time): Update.
23476 (inline_merge_summary): Update.
23477 (inline_update_overall_summary): Update.
23478 (inline_read_section): Update.
23479 (inline_write_summary): Update.
23480 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
23481 add account_size_time and reset member functions.
23482 (ipa_call_summary): Add reset function.
23483 * ipa-predicate.h (predicate::operator &): Constify.
23485 2017-05-22 Richard Biener <rguenther@suse.de>
23487 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
23489 2017-05-19 Jason Merrill <jason@redhat.com>
23491 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
23493 2017-05-19 Marek Polacek <polacek@redhat.com>
23496 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
23497 TYPE_OVERFLOW_WRAPS checks.
23499 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
23501 * tree-core.h (enum omp_clause_default_kind): Add
23502 "OMP_CLAUSE_DEFAULT_PRESENT".
23503 * tree-pretty-print.c (dump_omp_clause): Handle it.
23504 * gimplify.c (enum gimplify_omp_var_data): Add
23505 "GOVD_MAP_FORCE_PRESENT".
23506 (gimplify_adjust_omp_clauses_1): Map it to
23507 "GOMP_MAP_FORCE_PRESENT".
23508 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
23510 * gimplify.c (oacc_default_clause): Clarify.
23512 2017-05-19 Nathan Sidwell <nathan@acm.org>
23514 LANG_HOOK_REGISTER_DUMPS
23515 * toplev.c (general_init): Call register dump lang hook.
23516 * doc/invoke.texi: Document -fdump-lang option family.
23517 * dumpfile.c (dump_files): Remove class dump here.
23518 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
23519 * dumpfile.h (tree_dump_index): Remove TDI_class.
23520 * langhooks-def.h (lhd_register_dumps): Declare.
23521 (LANG_HOOKS_REGISTER_DUMPS): Define.
23522 (LANG_HOOKS_INITIALIZER): Add it.
23523 * langhooks.c (lhd_register_dumps): Define.
23524 * langhooks.h (struct lang_hooks): Add register_dumps.
23526 2017-05-19 Nathan Sidwell <nathan@acm.org>
23528 * context.h (context::set_passes): New.
23529 * context.c (context::context): Do not create pass manager.
23530 * toplev.c (general_init): Create pass manager here.
23532 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
23534 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
23535 use this splitter if two add or or instructions would also work for
23536 the constant we want to generate.
23538 2017-05-19 Richard Biener <rguenther@suse.de>
23541 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
23542 predicate evaluation.
23544 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23546 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
23548 * ipa-inline.c (want_inline_small_function_p): Do not cast to
23551 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23553 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23554 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23555 (inline_edge_summary_vec): Turn into ...
23556 (ipa_call_summaries): ... this one.
23557 (redirect_to_unreachable, edge_set_predicate,
23558 evaluate_properties_for_edge, inline_summary_alloc,
23559 reset_ipa_call_summary, reset_inline_summary,
23560 inline_summary_t::duplicate): Update.
23561 (inline_edge_duplication_hook): Turn to ...
23562 (ipa_call_summary_t::duplicate): ... this one.
23563 (inline_edge_removal_hook): Turn to ...
23564 (ipa_call_summary_t::remove): ... this one.
23565 (dump_inline_edge_summary): Turn to ...
23566 (dump_ipa_call_summary): ... this one.
23567 (estimate_function_body_sizes): Update.
23568 (inline_update_callee_summaries): Update.
23569 (remap_edge_change_prob): Update.
23570 (remap_edge_summaries): Update.
23571 (inline_merge_summary): Update.
23572 (do_estimate_edge_time): Update.
23573 (inline_generate_summary): Update.
23574 (inline_read_section): Update.
23575 (inline_read_summary): Update.
23576 (inline_free_summary): Update.
23577 * ipa-inline.c (can_inline_edge_p): Update.
23578 (compute_inlined_call_time): Update.
23579 (want_inline_small_function_p): Update.
23580 (edge_badness): Update.
23581 (early_inliner): Update.
23582 * ipa-inline.h (inline_edge_summary): Turn to ...
23583 (ipa_call_summary): ... this one.
23584 (ipa_call_summary_t): New class.
23585 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23586 (ipa_call_summaries): New.
23587 (inline_edge_summary): Remove.
23588 (estimate_edge_growth): Update.
23589 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23590 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23591 * ipa-split.c (execute_split_functions): Update.
23592 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23594 2017-05-19 Richard Biener <rguenther@suse.de>
23596 PR middle-end/80764
23597 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
23599 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
23601 * config/rs6000/rs6000.c (struct machine_function): Add field
23602 fpr_is_wrapped_separately.
23603 (rs6000_get_separate_components): Use 64 components. Handle the
23604 new FPR components.
23605 (rs6000_components_for_bb): Handle the FPR components.
23606 (rs6000_emit_prologue_components): Handle the FPR components.
23607 (rs6000_emit_epilogue_components): Handle the FPR components.
23608 (rs6000_set_handled_components): Handle the FPR components.
23609 (rs6000_emit_prologue): Don't output prologue code for those FPRs
23610 that are already separately shrink-wrapped.
23611 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
23612 that are already separately shrink-wrapped.
23614 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23617 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
23620 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
23621 (define_peephole2 for Altivec d-form load): Add peepholes to catch
23622 cases where the register allocator uses a move and an offsettable
23623 memory operation to/from a FPR register on ISA 2.06/2.07.
23624 (define_peephole2 for Altivec d-form store): Likewise.
23626 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
23629 * config/i386/mmx.md (*mov<mode>_internal): Enable
23630 alternatives 11, 12, 13 and 14 also for 32bit targets.
23631 Remove alternatives 15, 16, 17 and 18.
23632 * config/i386/sse.md (vec_concatv2di): Change
23633 alternative (!x, *y) to (x, ?!*Yn).
23635 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
23637 * dumpfile.h (enum dump_kind): Remove stray comma.
23639 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23641 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
23642 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
23643 predicate::num_conditions
23644 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
23645 (CHANGED): turn into predicate::changed.
23646 (agg_position_info): Move to ipa-predicate.h
23647 (add_condition, predicate::add_clause, predicate::operator &=,
23648 predicate::or_with, predicate::evaluate, predicate::probability,
23649 dump_condition, dump_clause, predicate::dump,
23650 predicate::remap_after_duplication, predicate::remap_after_inlining,
23651 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
23652 (evaluate_conditions_for_known_args): Update.
23653 (set_cond_stmt_execution_predicate): Update.
23654 * ipa-inline.h: Include ipa-predicate.h
23655 (condition, inline_param_summary, conditions, agg_position_info,
23656 predicate): Move to ipa-predicate.h
23657 * ipa-predicate.c: New file.
23658 * ipa-predicate.h: New file.
23660 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
23662 * final.c (leaf_function_p): Check we are not in a sequence.
23664 2017-05-18 Martin Liska <mliska@suse.cz>
23666 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
23667 * dumpfile.c (dump_register): Use new enum dump_kind.
23668 (get_dump_file_name): Likewise.
23669 (dump_enable_all): Likewise.
23670 (dump_switch_p_1): Likewise.
23671 (enable_rtl_dump_file): Remove usage of TDF_RTL.
23672 * dumpfile.h (enum dump_kind): New enum type.
23673 (struct dump_file_info): Create constructor and
23674 format fields and comments.
23675 * passes.c (pass_manager::register_one_dump_file):
23677 * statistics.c (statistics_early_init): Likewise.
23678 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
23679 TDF_TREE with TDF_SLIM.
23680 (gather_memory_references_ref): Likewise.
23682 2017-05-18 Martin Liska <mliska@suse.cz>
23684 * vec.h (struct vnull): Use it.
23686 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23688 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
23689 (true_predicate, false_predicate, true_predicate_p,
23690 false_predicate_p): Remove.
23691 (single_cond_predicate, not_inlined_predicate): Turn to member function
23693 (add_condition): Update.
23694 (add_clause): Turn to...
23695 (predicate::add_clause): ... this one; update; allow passing NULL
23697 (and_predicates): Turn to ...
23698 (predicate::operator &=): ... this one.
23699 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
23700 (or_predicates): Turn to ...
23701 (predicate::or_with): ... this one.
23702 (evaluate_predicate): Turn to ...
23703 (predicate::evaluate): ... this one.
23704 (predicate_probability): Turn to ...
23705 (predicate::probability): ... this one.
23706 (dump_condition): Update.
23707 (dump_predicate): Turn to ...
23708 (predicate::dump): ... this one.
23709 (account_size_time): Update.
23710 (edge_set_predicate): Update.
23711 (set_hint_predicate): UPdate.
23712 (evaluate_conditions_for_known_args): Update.
23713 (evaluate_properties_for_edge): Update.
23714 (remap_predicate_after_duplication): Turn to...
23715 (predicate::remap_after_duplication): ... this one.
23716 (remap_hint_predicate_after_duplication): Update.
23717 (inline_summary_t::duplicate): UPdate.
23718 (dump_inline_edge_summary): Update.
23719 (dump_inline_summary): Update.
23720 (set_cond_stmt_execution_predicate): Update.
23721 (set_switch_stmt_execution_predicate): Update.
23722 (compute_bb_predicates): Update.
23723 (will_be_nonconstant_expr_predicate): Update.
23724 (will_be_nonconstant_predicate): Update.
23725 (phi_result_unknown_predicate): Update.
23726 (predicate_for_phi_result): Update.
23727 (array_index_predicate): Update.
23728 (estimate_function_body_sizes): Update.
23729 (estimate_node_size_and_time): Update.
23730 (estimate_ipcp_clone_size_and_time): Update.
23731 (remap_predicate): Rename to ...
23732 (predicate::remap_after_inlining): ... this one.
23733 (remap_hint_predicate): Update.
23734 (inline_merge_summary): Update.
23735 (inline_update_overall_summary): Update.
23736 (estimate_size_after_inlining): Update.
23737 (read_predicate): Rename to ...
23738 (predicate::stream_in): ... this one.
23739 (read_inline_edge_summary): Update.
23740 (write_predicate): Rename to ...
23741 (predicate::stream_out): ... this one.
23742 (write_inline_edge_summary): Update.
23743 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
23744 (clause_t): Turn to uint32_t
23745 (predicate): Turn to class; implement constructor and operators
23747 (size_time_entry): Update.
23748 (inline_summary): Update.
23749 (inline_edge_summary): Update.
23751 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
23753 * fold-const.c (fold_binary_loc): Move transformation...
23754 * match.pd (C - X CMP X): ... here.
23756 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
23758 * config/sparc/sparc.c (sparc_option_override): Set function
23759 alignment for -mcpu=niagara7 to 64 to match the I$ line.
23760 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
23762 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
23764 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
23766 2017-05-18 Marek Polacek <polacek@redhat.com>
23769 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
23770 (pass_ubsan::execute): Call gimple_assign_single_p instead of
23771 gimple_assign_load_p.
23773 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
23775 PR middle-end/80692
23776 * real.c (do_compare): Give decimal_do_compare preference over
23777 comparing just the signs.
23779 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
23781 * doc/md.texi (Canonicalization of Instructions): Describe the
23782 canonical form of instructions that inherently set a condition
23785 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
23787 PR middle-end/80775
23788 * tree-cfg.c: Move deletion of unreachable case statements to after
23789 the merging of consecutive case labels.
23791 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
23793 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
23794 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
23795 restoring of callee-saved registers.
23797 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
23799 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
23800 * config/visium/visium.c (single_set_and_flags): Likewise.
23801 * config/visium/visium.md (Substitutions): Likewise.
23803 2017-05-17 Martin Liska <mliska@suse.cz>
23805 * cfg.c: Introduce dump_flags_t type and
23806 use it instead of int type.
23808 * cfghooks.c: Likewise.
23809 * cfghooks.h (struct cfg_hooks): Likewise.
23810 * cfgrtl.c: Likewise.
23811 * cfgrtl.h: Likewise.
23812 * cgraph.c (cgraph_node::get_body): Likewise.
23813 * coretypes.h: Likewise.
23814 * domwalk.c: Likewise.
23815 * domwalk.h: Likewise.
23816 * dumpfile.c (struct dump_option_value_info): Likewise.
23817 (dump_enable_all): Likewise.
23818 (dump_switch_p_1): Likewise.
23819 (opt_info_switch_p): Likewise.
23820 * dumpfile.h (enum tree_dump_index): Likewise.
23821 (struct dump_file_info): Likewise.
23822 * genemit.c: Likewise.
23823 * generic-match-head.c: Likewise.
23824 * gengtype.c (open_base_files): Likewise.
23825 * gimple-pretty-print.c: Likewise.
23826 * gimple-pretty-print.h: Likewise.
23827 * graph.c (print_graph_cfg): Likewise.
23828 * graphite-scop-detection.c (dot_all_sese): Likewise.
23829 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
23830 * loop-unroll.c (report_unroll): Likewise.
23831 * passes.c (pass_manager::register_one_dump_file): Likewise.
23832 * print-tree.c: Likewise.
23833 * statistics.c: Likewise.
23834 * tree-cfg.c: Likewise.
23835 * tree-cfg.h: Likewise.
23836 * tree-dfa.c: Likewise.
23837 * tree-dfa.h: Likewise.
23838 * tree-dump.c (dump_function): Likewise.
23839 * tree-dump.h (struct dump_info): Likewise.
23840 * tree-pretty-print.c: Likewise.
23841 * tree-pretty-print.h: Likewise.
23842 * tree-ssa-live.c: Likewise.
23843 * tree-ssa-live.h: Likewise.
23844 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
23845 * tree-vect-loop.c: Likewise.
23846 * tree-vect-slp.c: Likewise.
23848 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
23849 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23851 PR tree-optimization/80457
23852 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
23853 of all arguments to a statement as scalar_to_vec operations.
23854 (vectorizable_call): Adjust call to vect_model_simple_cost for
23856 (vectorizable_conversion): Likewise.
23857 (vectorizable_assignment): Likewise.
23858 (vectorizable_shift): Likewise.
23859 (vectorizable_operation): Likewise.
23860 (vectorizable_comparison): Likewise.
23861 (vect_is_simple_cond): Record the def types for operands.
23862 (vectorizable_condition): Likewise, call vect_model_simple_cost.
23863 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
23864 for statement argument count.
23866 2017-05-16 Carl Love <cel@us.ibm.com>
23868 * config/rs6000/rs6000-c: Add support for built-in functions
23869 vector unsigned long long vec_bperm (vector unsigned long long,
23870 vector unsigned char)
23871 vector signed long long vec_mule (vector signed int,
23873 vector unsigned long long vec_mule (vector unsigned int,
23874 vector unsigned int)
23875 vector signed long long vec_mulo (vector signed int,
23877 vector unsigned long long vec_mulo (vector unsigned int,
23878 vector unsigned int)
23879 vector signed char vec_sldw (vector signed char,
23880 vector signed char,
23882 vector unsigned char vec_sldw (vector unsigned char,
23883 vector unsigned char,
23885 vector signed short vec_sldw (vector signed short,
23886 vector signed short,
23888 vector unsigned short vec_sldw (vector unsigned short,
23889 vector unsigned short,
23891 vector signed int vec_sldw (vector signed int,
23894 vector unsigned int vec_sldw (vector unsigned int,
23895 vector unsigned int,
23897 vector signed long long vec_sldw (vector signed long long,
23898 vector signed long long,
23900 vector unsigned long long vec_sldw (vector unsigned long long,
23901 vector unsigned long long,
23903 * config/rs6000/rs6000-c: Add support for built-in functions
23904 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
23905 * config/rs6000/altivec.h: Add defintion for vec_sldw.
23906 * doc/extend.texi: Update the built-in documentation for the
23907 new built-in functions.
23909 2017-05-16 Marek Polacek <polacek@redhat.com>
23913 * tree.c (save_expr): Don't fold the expression.
23915 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
23917 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
23918 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
23919 and (?*y,m). Update insn attributes.
23921 2017-05-16 Martin Liska <mliska@suse.cz>
23923 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
23924 flags argument of print_gimple_stmt, print_gimple_expr,
23925 print_generic_stmt and print_generic_expr.
23926 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
23927 * coretypes.h: Likewise.
23928 * except.c (dump_eh_tree): Likewise.
23929 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
23930 * gimple-pretty-print.h: Likewise.
23931 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
23932 (backprop::push_to_worklist): Likewise.
23933 (backprop::pop_from_worklist): Likewise.
23934 (backprop::process_use): Likewise.
23935 (backprop::intersect_uses): Likewise.
23936 (note_replacement): Likewise.
23937 * gimple-ssa-store-merging.c
23938 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
23939 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
23940 (pass_store_merging::execute): Likewise.
23941 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
23942 (ssa_base_cand_dump_callback): Likewise.
23943 (dump_incr_vec): Likewise.
23944 (replace_refs): Likewise.
23945 (replace_mult_candidate): Likewise.
23946 (create_add_on_incoming_edge): Likewise.
23947 (create_phi_basis): Likewise.
23948 (insert_initializers): Likewise.
23949 (all_phi_incrs_profitable): Likewise.
23950 (introduce_cast_before_cand): Likewise.
23951 (replace_one_candidate): Likewise.
23952 * gimplify.c (gimplify_expr): Likewise.
23953 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
23954 (set_rename): Likewise.
23955 (rename_uses): Likewise.
23956 (copy_loop_phi_nodes): Likewise.
23957 (add_close_phis_to_merge_points): Likewise.
23958 (copy_loop_close_phi_args): Likewise.
23959 (copy_cond_phi_args): Likewise.
23960 (graphite_copy_stmts_from_block): Likewise.
23961 (translate_pending_phi_nodes): Likewise.
23962 * graphite-poly.c (print_pdr): Likewise.
23963 (dump_gbb_cases): Likewise.
23964 (dump_gbb_conditions): Likewise.
23965 (print_scop_params): Likewise.
23966 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
23967 (build_cross_bb_scalars_use): Likewise.
23968 (gather_bbs::before_dom_children): Likewise.
23969 * hsa-dump.c (dump_hsa_immed): Likewise.
23970 * ipa-cp.c (print_ipcp_constant_value): Likewise.
23971 (get_replacement_map): Likewise.
23972 * ipa-inline-analysis.c (dump_condition): Likewise.
23973 (estimate_function_body_sizes): Likewise.
23974 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
23975 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
23976 * ipa-prop.c (ipa_dump_param): Likewise.
23977 (ipa_print_node_jump_functions_for_edge): Likewise.
23978 (ipa_modify_call_arguments): Likewise.
23979 (ipa_modify_expr): Likewise.
23980 (ipa_dump_param_adjustments): Likewise.
23981 (ipa_dump_agg_replacement_values): Likewise.
23982 (ipcp_modif_dom_walker::before_dom_children): Likewise.
23983 * ipa-pure-const.c (check_stmt): Likewise.
23984 (pass_nothrow::execute): Likewise.
23985 * ipa-split.c (execute_split_functions): Likewise.
23986 * omp-offload.c (dump_oacc_loop_part): Likewise.
23987 (dump_oacc_loop): Likewise.
23988 * trans-mem.c (tm_log_emit): Likewise.
23989 (tm_memopt_accumulate_memops): Likewise.
23990 (dump_tm_memopt_set): Likewise.
23991 (dump_tm_memopt_transform): Likewise.
23992 * tree-cfg.c (gimple_verify_flow_info): Likewise.
23993 (print_loop): Likewise.
23994 * tree-chkp-opt.c (chkp_print_addr): Likewise.
23995 (chkp_gather_checks_info): Likewise.
23996 (chkp_get_check_result): Likewise.
23997 (chkp_remove_check_if_pass): Likewise.
23998 (chkp_use_outer_bounds_if_possible): Likewise.
23999 (chkp_reduce_bounds_lifetime): Likewise.
24000 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24001 (chkp_mark_completed_bounds): Likewise.
24002 (chkp_register_incomplete_bounds): Likewise.
24003 (chkp_mark_invalid_bounds): Likewise.
24004 (chkp_maybe_copy_and_register_bounds): Likewise.
24005 (chkp_build_returned_bound): Likewise.
24006 (chkp_get_bound_for_parm): Likewise.
24007 (chkp_build_bndldx): Likewise.
24008 (chkp_get_bounds_by_definition): Likewise.
24009 (chkp_generate_extern_var_bounds): Likewise.
24010 (chkp_get_bounds_for_decl_addr): Likewise.
24011 * tree-chrec.c (chrec_apply): Likewise.
24012 * tree-data-ref.c (dump_data_reference): Likewise.
24013 (dump_subscript): Likewise.
24014 (dump_data_dependence_relation): Likewise.
24015 (analyze_overlapping_iterations): Likewise.
24016 * tree-inline.c (expand_call_inline): Likewise.
24017 (tree_function_versioning): Likewise.
24018 * tree-into-ssa.c (dump_defs_stack): Likewise.
24019 (dump_currdefs): Likewise.
24020 (dump_names_replaced_by): Likewise.
24021 (dump_update_ssa): Likewise.
24022 (update_ssa): Likewise.
24023 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24024 * tree-parloops.c (build_new_reduction): Likewise.
24025 (try_create_reduction_list): Likewise.
24026 (ref_conflicts_with_region): Likewise.
24027 (oacc_entry_exit_ok_1): Likewise.
24028 (oacc_entry_exit_single_gang): Likewise.
24029 * tree-pretty-print.h: Likewise.
24030 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24031 (get_scalar_evolution): Likewise.
24032 (add_to_evolution): Likewise.
24033 (get_loop_exit_condition): Likewise.
24034 (analyze_evolution_in_loop): Likewise.
24035 (analyze_initial_condition): Likewise.
24036 (analyze_scalar_evolution): Likewise.
24037 (instantiate_scev): Likewise.
24038 (number_of_latch_executions): Likewise.
24039 (gather_chrec_stats): Likewise.
24040 (final_value_replacement_loop): Likewise.
24041 (scev_const_prop): Likewise.
24042 * tree-sra.c (dump_access): Likewise.
24043 (disqualify_candidate): Likewise.
24044 (create_access): Likewise.
24045 (reject): Likewise.
24046 (maybe_add_sra_candidate): Likewise.
24047 (create_access_replacement): Likewise.
24048 (analyze_access_subtree): Likewise.
24049 (analyze_all_variable_accesses): Likewise.
24050 (sra_modify_assign): Likewise.
24051 (initialize_constant_pool_replacements): Likewise.
24052 (find_param_candidates): Likewise.
24053 (decide_one_param_reduction): Likewise.
24054 (replace_removed_params_ssa_names): Likewise.
24055 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24056 * tree-ssa-copy.c (dump_copy_of): Likewise.
24057 (copy_prop_visit_cond_stmt): Likewise.
24058 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24059 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24060 (record_equivalences_from_stmt): Likewise.
24061 * tree-ssa-dse.c (compute_trims): Likewise.
24062 (delete_dead_call): Likewise.
24063 (delete_dead_assignment): Likewise.
24064 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24065 (forward_propagate_into_cond): Likewise.
24066 (pass_forwprop::execute): Likewise.
24067 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24068 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24070 (move_computations_worker): Likewise.
24071 (execute_sm): Likewise.
24072 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24073 (remove_exits_and_undefined_stmts): Likewise.
24074 (remove_redundant_iv_tests): Likewise.
24075 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24076 (adjust_iv_update_pos): Likewise.
24077 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24078 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24079 (value_replacement): Likewise.
24080 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24081 * tree-ssa-pre.c (print_pre_expr): Likewise.
24082 (get_representative_for): Likewise.
24083 (create_expression_by_pieces): Likewise.
24084 (insert_into_preds_of_block): Likewise.
24085 (eliminate_insert): Likewise.
24086 (eliminate_dom_walker::before_dom_children): Likewise.
24087 (eliminate): Likewise.
24088 (remove_dead_inserted_code): Likewise.
24089 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24090 * tree-ssa-reassoc.c (get_rank): Likewise.
24091 (eliminate_duplicate_pair): Likewise.
24092 (eliminate_plus_minus_pair): Likewise.
24093 (eliminate_not_pairs): Likewise.
24094 (undistribute_ops_list): Likewise.
24095 (eliminate_redundant_comparison): Likewise.
24096 (update_range_test): Likewise.
24097 (optimize_range_tests_var_bound): Likewise.
24098 (optimize_vec_cond_expr): Likewise.
24099 (rewrite_expr_tree): Likewise.
24100 (rewrite_expr_tree_parallel): Likewise.
24101 (linearize_expr): Likewise.
24102 (break_up_subtract): Likewise.
24103 (linearize_expr_tree): Likewise.
24104 (attempt_builtin_powi): Likewise.
24105 (attempt_builtin_copysign): Likewise.
24106 (transform_stmt_to_copy): Likewise.
24107 (transform_stmt_to_multiply): Likewise.
24108 (dump_ops_vector): Likewise.
24109 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24110 (print_scc): Likewise.
24111 (set_ssa_val_to): Likewise.
24112 (visit_reference_op_store): Likewise.
24113 (visit_use): Likewise.
24114 (sccvn_dom_walker::before_dom_children): Likewise.
24115 (run_scc_vn): Likewise.
24116 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24118 (expr_hash_elt::print): Likewise.
24119 (const_and_copies::pop_to_marker): Likewise.
24120 (const_and_copies::record_const_or_copy_raw): Likewise.
24121 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24122 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24123 (dump_predicates): Likewise.
24124 (find_uninit_use): Likewise.
24125 (warn_uninitialized_phi): Likewise.
24126 (pass_late_warn_uninitialized::execute): Likewise.
24127 * tree-ssa.c (verify_vssa): Likewise.
24128 (verify_ssa): Likewise.
24129 (maybe_optimize_var): Likewise.
24130 * tree-vrp.c (dump_value_range): Likewise.
24131 (dump_all_value_ranges): Likewise.
24132 (dump_asserts_for): Likewise.
24133 (register_edge_assert_for_2): Likewise.
24134 (vrp_visit_cond_stmt): Likewise.
24135 (vrp_visit_switch_stmt): Likewise.
24136 (vrp_visit_stmt): Likewise.
24137 (vrp_visit_phi_node): Likewise.
24138 (simplify_cond_using_ranges_1): Likewise.
24139 (fold_predicate_in): Likewise.
24140 (evrp_dom_walker::before_dom_children): Likewise.
24141 (evrp_dom_walker::push_value_range): Likewise.
24142 (evrp_dom_walker::pop_value_range): Likewise.
24143 (execute_early_vrp): Likewise.
24145 2017-05-16 Richard Biener <rguenther@suse.de>
24147 * dwarf2out.c (loc_list_from_tree_1): Do not create
24148 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24150 2017-05-16 Richard Biener <rguenther@suse.de>
24152 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24154 (note_variable_value_in_expr): If we resolved the decl ref
24155 do not push to the stack.
24157 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24159 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24160 operations in fast-math mode.
24161 (vaddq_f16): Likewise.
24162 (vmul_f16): Likewise.
24163 (vmulq_f16): Likewise.
24164 (vsub_f16): Likewise.
24165 (vsubq_f16): Likewise.
24166 * config/arm/neon.md (add<mode>3): New.
24168 (fma:<VH:mode>3): New. Also remove outdated comment.
24171 2017-05-16 Martin Liska <mliska@suse.cz>
24175 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24176 (odr_types_equivalent_p): Likewise.
24178 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24180 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24182 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24185 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24186 non-interunit SSE move alternatives with '?'.
24187 (zero-extendsidi peephole2): New peephole to skip intermediate
24188 general register in SSE zero-extend sequence.
24190 2017-05-15 Jeff Law <law@redhat.com>
24192 * reorg.c (relax_delay_slots): Create a new variable to hold
24193 the temporary target rather than clobbering TARGET_LABEL.
24195 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24196 missing argument to extract_bit_field call.
24197 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24199 2017-05-15 Martin Liska <mliska@suse.cz>
24202 * gcc.c (process_command): Do not allow empty argument of -o option.
24204 2017-05-15 Renlin Li <renlin.li@arm.com>
24206 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24207 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24208 * config/aarch64/constraints.md (Usf): Add long call check.
24209 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24210 (call_value): Likewise.
24211 (sibcall): Likewise.
24212 (sibcall_value): Likewise.
24214 (call_value_insn): New.
24215 (sibcall_insn): Update rtx pattern.
24216 (sibcall_value_insn): Likewise.
24217 (call_internal): Remove.
24218 (call_value_internal): Likewise.
24219 (sibcall_internal): Likewise.
24220 (sibcall_value_internal): Likewise.
24221 (call_reg): Likewise.
24222 (call_symbol): Likewise.
24223 (call_value_reg): Likewise.
24224 (call_value_symbol): Likewise.
24226 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24229 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24231 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24233 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24234 compatible with CCGOCmode and with CCZmode.
24236 2017-05-14 Martin Sebor <msebor@redhat.com>
24238 PR middle-end/77671
24239 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24240 (gimple_fold_builtin_snprintf): Same.
24241 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24242 (gimple_fold_builtin_snprintf): Same.
24243 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24244 of character types.
24245 (is_call_safe): New function.
24246 (try_substitute_return_value): Call it.
24247 (try_simplify_call): New function.
24248 (pass_sprintf_length::handle_gimple_call): Call it.
24250 2017-05-14 Martin Sebor <msebor@redhat.com>
24252 PR middle-end/80669
24253 * builtins.c (expand_builtin_stpncpy): Simplify.
24255 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24257 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24258 * config/i386/i386.h
24259 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24260 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24261 (struct machine_function): Add new members call_ms2sysv,
24262 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24263 (struct machine_frame_state): New fields sp_realigned and
24264 sp_realigned_offset.
24265 * config/i386/i386.c
24266 (enum xlogue_stub): New enum.
24267 (enum xlogue_stub_sets): New enum.
24268 (class xlogue_layout): New class.
24269 (struct ix86_frame): New fields stack_realign_allocate_offset,
24270 stack_realign_offset and outlined_save_offset. Modify comments to
24271 detail stack layout when using out-of-line stubs.
24272 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24273 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24274 -mcall-ms2sysv-xlogues.
24275 (stub_managed_regs): New static variable.
24276 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24277 registers managed by out-of-line stub.
24278 (disable_call_ms2sysv_xlogues): New function.
24279 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24280 m->call_ms2sysv when appropriate and compute frame layout for
24282 (sp_valid_at, fp_valid_at): New inline functions.
24283 (choose_basereg): New function.
24284 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24286 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24287 Use align parameter of choose_baseaddr to generated aligned SSE movs
24289 (pro_epilogue_adjust_stack): Modify to track
24290 machine_frame_state::sp_realigned.
24291 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24292 (ix86_nsaved_sseregs): Likewise.
24293 (ix86_emit_save_regs): Likewise.
24294 (ix86_emit_save_regs_using_mov): Likewise.
24295 (ix86_emit_save_sse_regs_using_mov): Likewise.
24296 (get_scratch_register_on_entry): Likewise.
24297 (gen_frame_set): New function.
24298 (gen_frame_load): Likewise.
24299 (gen_frame_store): Likewise.
24300 (emit_outlined_ms2sysv_save): Likewise.
24301 (emit_outlined_ms2sysv_restore): Likewise.
24302 (ix86_expand_prologue): Modify stack re-alignment code and call
24303 emit_outlined_ms2sysv_save when appropriate.
24304 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24305 parameter rtx_insn *insn, which allows the function to be used to only
24306 generate the notes.
24307 (ix86_expand_epilogue): Modify validity checks of frame and stack
24308 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24309 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24310 * config/i386/predicates.md
24311 (save_multiple): New predicate.
24312 (restore_multiple): Likewise.
24313 * config/i386/sse.md
24314 (save_multiple<mode>): New pattern.
24315 (save_multiple_realign<mode>): Likewise.
24316 (restore_multiple<mode>): Likewise.
24317 (restore_multiple_and_return<mode>): Likewise.
24318 (restore_multiple_leave_return<mode>): Likewise.
24319 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24321 2017-05-14 Julia Koval <julia.koval@intel.com>
24323 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24324 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24325 (__builtin_ia32_xsetbv): New builtins.
24326 * config/i386/i386.c (ix86_expand_special_args_builtin):
24328 (ix86_expand_builtin): Special expand for new intrinsics.
24329 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24330 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24331 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24333 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24335 * cfganal.c (inverted_post_order_compute): Change argument type
24337 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24338 * df-core.c (rest_of_handle_df_initialize): Adjust.
24339 (rest_of_handle_df_finish): Likewise.
24340 (df_analyze_1): Likewise.
24341 (df_analyze): Likewise.
24342 (loop_inverted_post_order_compute): Change argument to be a vec *.
24343 (df_analyze_loop): Adjust.
24344 (df_get_n_blocks): Likewise.
24345 (df_get_postorder): Likewise.
24346 * df.h (struct df_d): Change field to be a vec.
24347 * lcm.c (compute_laterin): Adjust.
24348 (compute_available): Likewise.
24349 * lra-lives.c (lra_create_live_ranges_1): Likewise.
24350 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
24351 * tree-ssa-pre.c (compute_antic): Likewise.
24353 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24355 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
24356 (depth_first_search::depth_first_search): Change structure init
24357 function to this constructor.
24358 (depth_first_search::add_bb): Rename function to this member.
24359 (depth_first_search::execute): Likewise.
24360 (flow_dfs_compute_reverse_finish): Adjust.
24362 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24364 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24365 (longest_simple_path): Likewise.
24366 * shrink-wrap.c (spread_components): Likewise.
24367 (disqualify_problematic_components): Likewise.
24368 (emit_common_heads_for_components): Likewise.
24369 (emit_common_tails_for_components): Likewise.
24370 (insert_prologue_epilogue_for_components): Likewise.
24372 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24374 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
24377 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24379 * df-core.c (df_set_blocks): Start using auto_bitmap.
24380 (df_compact_blocks): Likewise.
24381 * df-problems.c (df_rd_confluence_n): Likewise.
24382 * df-scan.c (df_insn_rescan_all): Likewise.
24383 (df_process_deferred_rescans): Likewise.
24384 (df_update_entry_block_defs): Likewise.
24385 (df_update_exit_block_uses): Likewise.
24386 (df_entry_block_bitmap_verify): Likewise.
24387 (df_exit_block_bitmap_verify): Likewise.
24388 (df_scan_verify): Likewise.
24389 * lra-constraints.c (lra_constraints): Likewise.
24390 (undo_optional_reloads): Likewise.
24391 (lra_undo_inheritance): Likewise.
24392 * lra-remat.c (calculate_gen_cands): Likewise.
24393 (do_remat): Likewise.
24394 * lra-spills.c (assign_spill_hard_regs): Likewise.
24395 (spill_pseudos): Likewise.
24396 * tree-ssa-pre.c (bitmap_set_and): Likewise.
24397 (bitmap_set_subtract_values): Likewise.
24399 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24401 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
24402 management with auto_bitmap.
24403 (fix_inter_tick): Likewise.
24404 (fix_recovery_deps): Likewise.
24405 * ira.c (add_store_equivs): Likewise.
24406 (find_moveable_pseudos): Likewise.
24407 (split_live_ranges_for_shrink_wrap): Likewise.
24408 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
24409 (rtx_reuse_manager::seen_def_p): Likewise.
24410 (rtx_reuse_manager::set_seen_def): Likewise.
24411 * print-rtl.h (class rtx_reuse_manager): Likewise.
24413 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24415 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
24417 (migrate_btr_def): Likewise.
24418 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
24419 * df-core.c (loop_post_order_compute): Likewise.
24420 (loop_inverted_post_order_compute): Likewise.
24421 * hsa-common.h: Likewise.
24422 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
24423 * init-regs.c (initialize_uninitialized_regs): Likewise.
24424 * ipa-inline.c (resolve_noninline_speculation): Likewise.
24425 (inline_small_functions): Likewise.
24426 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
24427 * ira.c (combine_and_move_insns): Likewise.
24428 (build_insn_chain): Likewise.
24429 * loop-invariant.c (find_invariants): Likewise.
24430 * lower-subreg.c (propagate_pseudo_copies): Likewise.
24431 * predict.c (tree_predict_by_opcode): Likewise.
24432 (predict_paths_leading_to): Likewise.
24433 (predict_paths_leading_to_edge): Likewise.
24434 (estimate_loops_at_level): Likewise.
24435 (estimate_loops): Likewise.
24436 * shrink-wrap.c (try_shrink_wrapping): Likewise.
24437 (spread_components): Likewise.
24438 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
24439 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
24440 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
24441 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
24442 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
24443 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
24444 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
24445 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
24446 (mark_threaded_blocks): Likewise.
24447 (thread_through_all_blocks): Likewise.
24448 * tree-ssa.c (verify_ssa): Likewise.
24449 (execute_update_addresses_taken): Likewise.
24450 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
24452 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24454 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
24456 (post_order_compute): Likewise.
24457 (inverted_post_order_compute): Likewise.
24458 (pre_and_rev_post_order_compute_fn): Likewise.
24460 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24462 * genrecog.c (int_set::int_set): Explicitly construct our
24463 auto_vec base class.
24464 * vec.h (auto_vec::auto_vec): New constructor.
24466 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24468 * bitmap.h (class auto_bitmap): New constructor taking
24469 bitmap_obstack * argument.
24471 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24473 * bitmap.h (class auto_bitmap): Change type of m_bits to
24474 bitmap_head, and adjust ctor / dtor and member operators.
24476 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
24478 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
24479 when returned register mode doesn't match original mode.
24481 2017-05-12 Jeff Law <law@redhat.com>
24482 Jakub Jelinek <jakub@redhat.com>
24484 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
24485 we look for cc setter after the compare-elim changes.
24486 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
24487 within the vector to match what compare-elim now expects.
24488 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
24489 (xorsi3_flags, one_cmplsi2_flags): Likewise.
24491 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
24492 after the compare-elim changes.
24493 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
24494 the vector to match what compare-elim now expects.
24495 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
24496 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
24497 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
24498 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
24499 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
24501 * config/visium/visium.c (single_set_and_flags): Fix where
24502 we look for cc setter after the compare-elim changes.
24503 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
24504 with the vector to match what compare-elim now expects.
24505 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
24506 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
24507 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
24508 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
24509 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
24510 (neg<mode>2_insn_set_overflow): Likewise.
24512 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
24514 PR middle-end/79794
24515 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
24516 maybe_expand_insn call, set ops[0].target. If still set after call,
24517 set alt_rtl. Add extra arg to recursive calls.
24518 (extract_bit_field): Add alt_rtl argument. Pass to
24520 * expmed.h (extract_bit_field): Fix prototype.
24521 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
24522 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
24523 to extract_bit_field_calls.
24524 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
24525 Pass alt_rtl to extract_bit_field calls.
24526 * calls.c (store_unaligned_arguments_into_psuedos)
24527 load_register_parameters): Pass extra NULL to extract_bit_field calls.
24528 * optabs.c (maybe_legitimize_operand): Clear op->target when call
24530 * optabs.h (struct expand_operand): Add target bitfield.
24532 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24534 * compare-elim.c (try_eliminate_compare): Canonicalize
24535 operation with embedded compare to
24536 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
24537 (set (reg) (operation)].
24539 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
24541 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24544 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
24545 cost of adding a carry flag for ADC instruction.
24546 [case MINUS]: Ignore the cost of subtracting a carry flag
24547 for SBB instruction.
24549 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
24551 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
24553 * config/rs6000/bmiintrin.h: New file.
24554 * config/rs6000/bmi2intrin.h: New file.
24555 * config/rs6000/x86intrin.h: New file.
24557 2017-05-12 Jeff Law <law@redhat.com>
24559 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
24562 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
24564 PR middle-end/80707
24565 * tree-cfg.c: Remove cfg edges of unreachable case statements.
24567 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24569 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24570 early expansion of vector divide builtins.
24571 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
24572 builtins identified as having unsigned arguments.
24574 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24576 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
24577 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
24578 expansion of vector logical operations (and, andc, or, xor,
24581 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24583 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
24584 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
24586 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24588 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24589 early GIMPLE expansion of vector multiplies.
24591 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24593 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
24594 TARGET_HAVE_MOVT conditional.
24595 (movt splitter): Likewise.
24597 2017-05-12 Richard Biener <rguenther@suse.de>
24599 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
24600 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24601 Fold all stmts not inplace.
24603 2017-05-12 Richard Biener <rguenther@suse.de>
24605 PR tree-optimization/80713
24606 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
24607 inserted_exprs bit for not removed stmts.
24609 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
24611 PR middle-end/69921
24612 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
24613 parallelized" attribute for parallelized OpenACC kernels.
24614 * omp-offload.c (execute_oacc_device_lower): Use it.
24616 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
24617 Set "oacc kernels" attribute.
24618 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
24619 parameter. Adjust all users.
24620 (oacc_fn_attrib_kernels_p): Remove function.
24621 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
24622 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
24623 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
24624 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
24625 assert "oacc kernels" attribute is set.
24627 2017-05-11 Carl Love <cel@us.ibm.com>
24629 * config/rs6000/rs6000-c: Add support for built-in functions
24630 vector unsigned char vec_popcnt (vector signed char)
24631 vector unsigned char vec_popcnt (vector unsigned char)
24632 vector unsigned short vec_popcnt (vector signed short)
24633 vector unsigned short vec_popcnt (vector unsigned short)
24634 vector unsigned int vec_popcnt (vector signed int)
24635 vector unsigned int vec_popcnt (vector unsigned int)
24636 vector unsigned long long vec_popcnt (vector signed long long)
24637 vector unsigned long long vec_popcnt (vector unsigned long long)
24638 vector signed long long vec_slo (vector signed long long,
24639 vector signed char)
24640 vector signed long long vec_slo (vector signed long long,
24641 vector unsigned char)
24642 vector unsigned long long vec_slo (vector unsigned long long,
24643 vector signed char)
24644 vector unsigned long long vec_slo (vector unsigned long long,
24645 vector unsigned char)
24646 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
24647 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
24648 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
24649 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
24650 * doc/extend.texi: Update the built-in documentation file for the
24651 new built-in functions.
24653 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
24655 * attribs.h (sorted_attr_string): Move machine independent
24656 functions for target clone support from the i386 port to common
24657 code. Rename ix86_function_versions to common_function_versions.
24658 Rename make_name to make_unique_name.
24659 (common_function_versions): Likewise.
24660 (make_unique_name): Likewise.
24661 (make_dispatcher_decl): Likewise.
24662 (is_function_default_version): Likewise.
24663 * attribs.c (attr_strcmp): Likewise.
24664 (sorted_attr_string): Likewise.
24665 (common_function_versions): Likewise.
24666 (make_unique_name): Likewise.
24667 (make_dispatcher_decl): Likewise.
24668 (is_function_default_version): Likewise.
24669 * config/i386/i386.c (attr_strcmp): Likewise.
24670 (sorted_attr_string): Likewise.
24671 (ix86_function_versions): Likewise.
24672 (make_name): Likewise.
24673 (make_dispatcher_decl): Likewise.
24674 (is_function_default_version): Likewise.
24675 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
24677 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24680 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
24681 Account for direct move costs for vec_construct of integer
24684 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
24687 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
24688 (UNSPEC_STX_ATOMIC): Ditto.
24689 (loaddi_via_sse): New insn.
24690 (storedi_via_sse): Ditto.
24691 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
24692 Update corresponding peephole2 patterns.
24693 (atomic_storedi_fpu): Ditto.
24695 2017-05-11 Julia Koval <julia.koval@intel.com>
24697 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
24698 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
24700 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
24701 (__builtin_ia32_rsqrt14ss_mask): New builtins.
24702 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
24704 2017-05-11 Nathan Sidwell <nathan@acm.org>
24706 * graphite-poly.c: Include dumpfile.h.
24708 * dumpfle.h (dump_function): Declare here ...
24709 * tree-dump.h (dump_function): ... not here.
24710 * dumpfile.c: #include tree-cfg.h.
24711 (dump_function): Move here from ...
24712 * tree-dump.c (dump_function): ... here.
24713 * gimplify.c: #include splay-tree.h, not tree-dump.h.
24714 * graphite-poly.c: Don't include tree-dump.h.
24715 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
24716 * print-tree.c: Likewise.
24717 * stor-layout.c: Likewise.
24718 * tree-nested.c: Likewise.
24720 * dumpfile.c (dump_start): Use TDF_FLAGS.
24721 (dump_enable_all): Fix TDF_KIND check thinko.
24723 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
24725 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24726 array entries to represent two legal parameterizations of the
24727 overloaded __builtin_cmpb function, as represented by the
24728 P6_OV_BUILTIN_CMPB constant.
24729 (altivec_resolve_overloaded_builtin): Add special case handling
24730 for the __builtin_cmpb function, as represented by the
24731 P6_OV_BUILTIN_CMPB constant.
24732 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
24733 (BU_P6_64BIT_2): New macro.
24734 (BU_P6_OVERLOAD_2): New macro
24735 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
24736 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
24737 (CMPB): Add overload support to represent both 32-bit and 64-bit
24738 compare-bytes function.
24739 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
24740 support for TARGET_CMPB.
24741 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
24742 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24743 documentation of the __builtin_cmpb overloaded built-in function.
24745 2017-05-11 Richard Biener <rguenther@suse.de>
24747 PR tree-optimization/80705
24748 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
24749 bases are not vectorizable.
24751 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24753 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
24754 when counting register pressure.
24756 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24758 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
24759 (iv_ca_more_deps): Renamed to ...
24760 (iv_ca_compare_deps): ... this.
24761 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
24763 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24765 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
24767 (determine_group_iv_costs): ... here.
24768 (find_inv_vars_cb): Record inv var if it's not recorded before.
24770 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24772 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
24773 (get_shiftadd_cost): Ditto.
24775 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24777 * tree-ssa-address.c: Include header file.
24778 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
24780 (add_to_parts): Refactor.
24781 (addr_to_parts): New parameter. Update use of move_hint_to_base.
24782 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
24785 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24787 PR tree-optimization/53090
24788 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
24790 (extract_cond_operands): Detect condition with IV on both sides
24791 and return COMP_IV_EXPR_2.
24792 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
24793 (rewrite_use_compare): Simplify by removing call to function
24794 extract_cond_operands.
24796 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24798 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
24799 (extract_cond_operands): Detect condition comparing against non-
24800 invariant bound and return appropriate enum value.
24801 (find_interesting_uses_cond): Update use of extract_cond_operands.
24802 Handle its return value accordingly.
24803 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
24805 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24807 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
24808 nonlinear iv_use computation in loop invariant sensitive way.
24810 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24812 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
24813 (find_iv_candidates): Call relate_compare_use_with_all_cands.
24815 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24817 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
24818 (dump_cand): Support iv_cand.inv_exprs.
24819 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
24821 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
24824 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24826 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
24828 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
24829 as local function. Include necessary header files.
24830 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
24832 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24834 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
24836 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24838 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
24839 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
24840 RSHIFT_EXPR and BIT_NOT_EXPR.
24842 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24844 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
24845 (adjust_setup_cost): New parameter supporting round up adjustment.
24846 (struct address_cost_data): Delete.
24847 (force_expr_to_var_cost): Don't bound cost with spill_cost.
24848 (split_address_cost, ptr_difference_cost): Delete.
24849 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
24850 (struct ainc_cost_data): New struct.
24851 (get_address_cost_ainc): New function.
24852 (get_address_cost, get_computation_cost): Reimplement.
24853 (determine_group_iv_cost_address): Record inv_expr for all uses of
24855 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
24856 (iv_ca_has_deps): Reimplemented to ...
24857 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
24859 (iv_ca_extend): Call iv_ca_more_deps.
24861 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24863 * tree-ssa-address.c (struct mem_address): Move to header file.
24864 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
24865 * tree-ssa-address.h (struct mem_address): Move from C file.
24866 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
24868 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24870 * tree-affine.h (aff_combination_type): New interface.
24871 (aff_combination_zero_p): Remove static.
24872 (aff_combination_const_p): New interface.
24873 (aff_combination_singleton_var_p): New interfaces.
24875 2017-05-11 Richard Biener <rguenther@suse.de>
24877 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24878 Skip unreachable blocks and destinations.
24879 (eliminate): Move stmt removal and fixup ...
24880 (fini_eliminate): ... here. Skip inserted exprs.
24881 (pass_pre::execute): Move fini_pre after fini_eliminate.
24882 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
24883 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
24884 PRE to get rid of dead code that has invalid SSA form and
24885 split critical edges again.
24887 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24889 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24891 2017-05-11 Richard Biener <rguenther@suse.de>
24893 * passes.c (execute_function_todo): Verify loops if they are
24894 said to be up-to-date.
24895 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
24896 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
24898 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
24901 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
24902 handle calling assemble_external ourself.
24905 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
24906 modes with zero size. Enhance comment.
24908 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24910 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
24911 built-ins for vec_xl and vec_xst with short and char pointer
24914 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24916 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
24917 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
24918 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
24919 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
24920 (_mm_maskz_min_round_ss): New intrinsics.
24921 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
24922 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
24923 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
24924 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
24925 (__builtin_ia32_minss_mask_round): New builtins.
24926 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24927 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24928 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
24930 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
24931 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
24933 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
24936 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24938 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
24939 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
24940 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
24941 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
24942 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
24943 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
24944 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
24945 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24946 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24947 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
24948 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
24949 (__builtin_ia32_mulss_mask_round): New builtins.
24950 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24951 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24952 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
24954 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
24955 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24957 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24960 2017-05-10 Julia Koval <julia.koval@intel.com>
24962 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
24963 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
24964 (_mm256_setr_m128i): New intrinsics.
24966 2017-05-10 Julia Koval <julia.koval@intel.com>
24968 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
24969 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
24970 (_mm_maskz_rcp14_ss): New intrinsics.
24971 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
24972 (__builtin_ia32_rcp14ss_mask): New builtins.
24973 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
24975 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
24977 PR tree-optimization/51513
24978 * tree-cfg.c (gimple_seq_unreachable_p): New function.
24979 (assert_unreachable_fallthru_edge_p): Use it.
24980 (group_case_labels_stmt): Likewise.
24981 * tree-cfg.h: Prototype it.
24982 * stmt.c: Include cfghooks.h and tree-cfg.h.
24983 (emit_case_dispatch_table) <gap_label>: New local variable.
24984 Use it to fill dispatch table gaps.
24985 Test for default_label before updating probabilities.
24986 (expand_case) <default_label>: Remove unneeded initialization.
24987 Test for unreachable default case statement and remove its edge.
24988 Set default_label accordingly.
24989 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
24991 2017-05-10 Carl Love <cel@us.ibm.com>
24993 * config/rs6000/rs6000-c: Add support for built-in functions
24994 vector signed char vec_neg (vector signed char)
24995 vector signed short int vec_neg (vector short int)
24996 vector signed int vec_neg (vector signed int)
24997 vector signed long long vec_neg (vector signed long long)
24998 vector float vec_neg (vector float)
24999 vector double vec_neg (vector double)
25000 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25002 * config/rs6000/altivec.h: Add define for vec_neg
25003 * doc/extend.texi: Update the built-in documentation for the
25004 new built-in functions.
25006 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25008 PR tree-optimization/77644
25009 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25011 2017-05-10 Nathan Sidwell <nathan@acm.org>
25013 * dumpfile.h (TDI_lang_all): New.
25014 (TDF_KIND): New. Renumber others
25015 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25017 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25019 (get_dump_file_name): Adjust suffix generation.
25020 (dump_enable_all): Use TDF_KIND.
25021 * doc/invoke.texi (-fdump-lang-all): Document.
25023 * dumpfile.h: Tabify.
25025 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25028 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25029 Move member access before delete.
25031 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25033 * tree-inline.c (expand_call_inline): Split block at stmt
25036 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25039 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25040 are now unused after splitting mov{sf,sd}_hardfloat.
25041 (f32_lr2): Likewise.
25042 (f32_lm): Likewise.
25043 (f32_lm2): Likewise.
25044 (f32_li): Likewise.
25045 (f32_li2): Likewise.
25046 (f32_lv): Likewise.
25047 (f32_sr): Likewise.
25048 (f32_sr2): Likewise.
25049 (f32_sm): Likewise.
25050 (f32_sm2): Likewise.
25051 (f32_si): Likewise.
25052 (f32_si2): Likewise.
25053 (f32_sv): Likewise.
25054 (f32_dm): Likewise.
25055 (f32_vsx): Likewise.
25056 (f32_av): Likewise.
25057 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25058 For movsf, order stores so the VSX stores occur before the GPR
25059 store which encourages the register allocator to use a traditional
25060 FPR instead of a GPR. For movsd, order the stores so that the GPR
25061 store comes before the VSX stores to allow the power6 to work.
25062 This is due to the power6 not having a 32-bit integer store
25063 instruction from a FPR.
25064 (movsf_hardfloat): Likewise.
25065 (movsd_hardfloat): Likewise.
25067 2017-05-09 Martin Sebor <msebor@redhat.com>
25069 PR translation/80280
25070 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25073 PR translation/80280
25074 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25075 data member added in r247778.
25076 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25078 2017-05-09 Nathan Sidwell <nathan@acm.org>
25080 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25082 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25085 2017-05-09 Marek Polacek <polacek@redhat.com>
25087 * doc/invoke.texi: Fix typo.
25089 2017-05-09 Richard Biener <rguenther@suse.de>
25091 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25092 (vrp_val_is_min): Likewise.
25093 (set_value_range_to_value): Likewise.
25094 (set_value_range_to_nonnegative): Likewise.
25095 (gimple_assign_nonzero_p): Likewise.
25096 (gimple_stmt_nonzero_p): Likewise.
25097 (vrp_int_const_binop): Likewise. Remove unreachable case.
25098 (adjust_range_with_scev): Adjust comments.
25099 (compare_range_with_value): Likewise.
25100 (extract_range_from_phi_node): Likewise.
25101 (test_for_singularity): Likewise.
25103 2017-05-09 Richard Biener <rguenther@suse.de>
25105 * tree-vrp.c (get_single_symbol): Add assert that we don't
25106 get overflowed constants as invariant part.
25107 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25108 checks. Use wi::cmp instead of recursing for integer constants.
25109 (compare_values): Just ignore whether we assumed undefined
25110 overflow instead of failing the compare.
25111 (extract_range_for_var_from_comparison_expr): Add comment before the
25112 TREE_NO_WARNING sets.
25113 (test_for_singularity): Likewise.
25114 (extract_range_from_comparison): Do not disable optimization
25115 when we assumed undefined overflow.
25116 (extract_range_basic): Remove init of unused var.
25118 2017-05-09 Richard Biener <rguenther@suse.de>
25120 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25121 (extract_range_from_multiplicative_op_1): Adjust.
25122 (extract_range_from_binary_expr_1): Use int_const_binop.
25124 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25127 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25128 rs6000_store_data_bypass_p in seven define_bypass directives and
25129 in several comments.
25130 * config/rs6000/rs6000-protos.h: Add prototype for
25131 rs6000_store_data_bypass_p function.
25132 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25133 function implements slightly different (rs6000-specific) semantics
25134 than store_data_bypass_p, returning false rather than aborting
25135 with assertion error when arguments do not satisfy the
25136 requirements of store data bypass.
25137 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25138 rs6000_store_data_bypass_p.
25140 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25142 * config/xtensa/xtensa-protos.h
25143 (xtensa_initial_elimination_offset): New declaration.
25144 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25145 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25146 macro definition, add case for FRAME_POINTER_REGNUM when
25147 FRAME_GROWS_DOWNWARD.
25148 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25149 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25150 xtensa_initial_elimination_offset.
25152 2017-05-08 Nathan Sidwell <nathan@acm.org>
25154 * doc/invoke.texi: Alphabetize -fdump options.
25156 2017-05-08 Martin Sebor <msebor@redhat.com>
25158 PR translation/80280
25159 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25161 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25163 * target.def (compute_frame_layout): New optional target hook.
25164 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25165 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25166 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25168 * reload1.c (verify_initial_elim_offsets): Likewise.
25169 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25170 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25171 (arm_get_frame_offsets): Split up into this ...
25172 (arm_compute_frame_layout): ... and this function.
25174 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25176 * config/aarch64/constraints.md (Usa): New constraint.
25177 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25179 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25181 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25182 with_multilib_list after it has been checked.
25184 2017-05-08 Richard Biener <rguenther@suse.de>
25186 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25187 (bitmap_set_subtract_values): Likewise.
25189 2017-05-08 Richard Biener <rguenther@suse.de>
25191 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25192 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25194 (gimple_stmt_nonzero_warnv_p): Rename to ...
25195 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25197 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25198 (extract_range_basic): Adjust, do not disable propagation on
25199 strict overflow sensitive simplification.
25200 (vrp_visit_cond_stmt): Likewise.
25202 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25204 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25205 body size unconditionally.
25207 2017-05-07 Jeff Law <law@redhat.com>
25210 2017-05-06 Jeff Law <law@redhat.com>
25211 PR tree-optimization/78496
25212 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25215 PR tree-optimization/78496
25216 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25217 (simplify_stmt_using_ranges): Call it.
25218 (vrp_dom_walker::before_dom_children): Extract equivalences
25219 from an ASSERT_EXPR with an equality comparison against a
25222 2017-05-06 Jeff Law <law@redhat.com>
25224 PR tree-optimization/78496
25225 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25228 PR tree-optimization/78496
25229 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25230 (simplify_stmt_using_ranges): Call it.
25231 (vrp_dom_walker::before_dom_children): Extract equivalences
25232 from an ASSERT_EXPR with an equality comparison against a
25235 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25237 * lra-constraints.c (lra_copy_reg_equiv): New function.
25238 (split_reg): Use it to copy equivalence information from the
25239 original register to the spill register.
25241 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25243 PR rtl-optimization/75964
25244 * simplify-rtx.c (simplify_const_relational_operation): Remove
25245 invalid handling of comparisons of integer ABS.
25247 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25249 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25250 initialize to zero.
25251 (init_regs): Remove declaration.
25252 (function_arg_advance_32): Initialize error_p as boolean variable.
25254 2017-05-05 Nathan Sidwell <nathan@acm.org>
25256 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25257 lines. Use for (;;).
25259 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25261 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25262 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25263 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25264 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25265 VF=2 that require versioning.
25267 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25269 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25272 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25274 * diagnostic.h (diagnostic_override_option_index): Convert from
25275 macro to inline function.
25277 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25279 * diagnostic.c (last_module_changed_p): New function.
25280 (set_last_module): New function.
25281 (diagnostic_report_current_module): Convert macro usage to
25282 the above functions.
25283 * diagnostic.h (diagnostic_context::last_module): Strengthen
25284 from const line_map * to const line_map_ordinary *.
25285 (diagnostic_last_module_changed): Delete macro.
25286 (diagnostic_set_last_module): Delete macro.
25288 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25290 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25291 with diagnostic_report_diagnostic.
25292 (diagnostic_n_impl_richloc): Likewise.
25293 * diagnostic.h (report_diagnostic): Delete macro.
25294 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25295 with diagnostic_report_diagnostic.
25296 * substring-locations.c (format_warning_va): Likewise.
25298 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25300 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25301 save/restor of format_spec. Move option-printing code to...
25302 (print_option_information): ...this new function, and
25303 reimplement by simply printing to the pretty_printer,
25304 rather than appending to the format string.
25306 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25308 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25309 handling logic into...
25310 (update_effective_level_from_pragmas): ...this new function.
25312 2017-05-04 Andrew Waterman <andrew@sifive.com>
25314 * config/riscv/riscv.opt (mstrict-align): New option.
25315 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25316 (SLOW_UNALIGNED_ACCESS): Define.
25317 (riscv_slow_unaligned_access): Declare.
25318 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25320 (riscv_slow_unaligned_access): New variable.
25321 (rocket_tune_info): Set slow_unaligned_access to true.
25322 (optimize_size_tune_info): Set slow_unaligned_access to false.
25323 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25324 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25325 (riscv_option_override): Set riscv_slow_unaligned_access.
25326 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25328 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25330 * config/riscv/riscv.md: Unify indentation.
25332 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25337 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25339 (extendsi<mode>2): Add support for doing sign extension via
25340 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25341 don't have ISA 3.0 instructions.
25342 (extendsi<mode>2 splitter): Likewise.
25343 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
25344 generate the normal insns since SImode can now go in vector
25345 registers. Disallow the special UNSPECs needed for previous
25346 machines to hide SImode being used. Add new insns
25347 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
25348 (fix_trunc<mode>si2_stfiwx): Likewise.
25349 (fix_trunc<mode>si2_internal): Likewise.
25350 (fixuns_trunc<mode>si2): Likewise.
25351 (fixuns_trunc<mode>si2_stfiwx): Likewise.
25352 (fctiw<u>z_<mode>_smallint): Likewise.
25353 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
25354 of floating point to 32-bit integer from doing a direct move to
25355 the GPR registers to do a store.
25356 (fctiwz_<mode>): Break long line.
25358 2017-05-05 Bin Cheng <bin.cheng@arm.com>
25360 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
25361 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
25362 (addr_list, addr_offset_valid_p): New.
25363 (split_address_groups): Check offset validity with above function.
25364 (gt-tree-ssa-loop-ivopts.h): Include header file.
25366 2017-05-05 Nathan Sidwell <nathan@acm.org>
25368 * config.gcc (arm*-*-*): Add missing 'fi'.
25370 2017-05-05 Steve Ellcey <sellcey@cavium.com>
25372 * doc/invoke.texi (-fopt-info): Explicitly say order of options
25373 included in -fopt-info does not matter.
25374 * doc/optinfo.texi (-fopt-info): Fix description of default
25375 behavour. Explicitly say order of options included in -fopt-info
25378 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25380 * config.gcc: Allow combinations of aprofile and rmprofile values for
25381 --with-multilib-list.
25382 * config/arm/t-multilib: New file.
25383 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
25384 variables. Remove setting of ISA and floating-point ABI in
25385 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
25386 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
25387 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
25388 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
25390 * config/arm/t-rmprofile: Likewise except for the matches changes.
25391 * doc/install.texi (--with-multilib-list): Document the combination of
25392 aprofile and rmprofile values and warn about pitfalls in doing that.
25394 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25396 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
25397 (movdi_aarch64): Likewise.
25399 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25401 PR tree-optimization/80632
25402 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
25404 (build_arrays): Initialize it for virtual phis.
25405 (fix_phi_nodes): Use it for virtual phis.
25407 PR tree-optimization/80558
25408 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
25409 [x, y] op z into [x op, y op z] for op & or | if conditions
25412 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
25413 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25416 * config/arm/arm.md (use_literal_pool): Remove.
25417 (64-bit immediate split): No longer takes cost into consideration
25418 if arm_disable_literal_pool is enabled.
25419 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
25420 used when arm_disable_literal_pool is enabled.
25421 (arm_max_const_double_inline_cost): Remove use of
25422 arm_disable_literal_pool.
25423 (push_minipool_fix): Add assert.
25424 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
25425 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
25426 (no_literal_pool_sf_immediate): New.
25428 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25430 PR tree-optimization/80613
25431 * tree-ssa-dce.c (propagate_necessity): Remove cases for
25432 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
25434 2017-05-05 Richard Biener <rguenther@suse.de>
25436 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
25438 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
25440 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
25441 of this flag from insn conditions due to removal from r247495.
25443 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25445 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
25447 (arm_early_store_addr_dep_ptr): Likewise.
25448 * config/arm/aarch-common-protos.h
25449 (arm_early_load_addr_dep_ptr): Add prototype.
25450 (arm_early_store_addr_dep_ptr): Likewise.
25451 * config/arm/cortex-a53.md: Add new bypasses.
25453 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25455 * tree.c (next_type_uid): Change type to unsigned.
25456 (type_hash_canon): Decrement back next_type_uid if
25457 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
25458 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
25461 2017-05-04 Martin Sebor <msebor@redhat.com>
25463 * builtins.c: Fix a trivial typo in a comment.
25465 PR middle-end/79234
25466 * builtins.c (check_sizes): Adjust to handle reading past the end.
25467 Avoid printing excessive upper bound of ranges. Use %E to print
25468 tree nodes instead of converting them to %wu.
25469 (expand_builtin_memchr): New function.
25470 (compute_dest_size): Rename...
25471 (compute_objsize): ...to this.
25472 (expand_builtin_memcpy): Adjust.
25473 (expand_builtin_mempcpy): Adjust.
25474 (expand_builtin_strcat): Adjust.
25475 (expand_builtin_strcpy): Adjust.
25476 (check_strncat_sizes): Adjust.
25477 (expand_builtin_strncat): Adjust.
25478 (expand_builtin_strncpy): Adjust and simplify.
25479 (expand_builtin_memset): Adjust.
25480 (expand_builtin_bzero): Adjust.
25481 (expand_builtin_memcmp): Adjust.
25482 (expand_builtin): Handle memcmp.
25483 (maybe_emit_chk_warning): Check strncat just once.
25485 2017-05-04 Martin Sebor <msebor@redhat.com>
25487 PR preprocessor/79214
25488 PR middle-end/79222
25489 PR middle-end/79223
25490 * builtins.c (check_sizes): Add inlining context and issue
25491 warnings even when -Wno-system-headers is set.
25492 (check_strncat_sizes): Same.
25493 (expand_builtin_strncat): Same.
25494 (expand_builtin_memmove): New function.
25495 (expand_builtin_stpncpy): Same.
25496 (expand_builtin): Handle memmove and stpncpy.
25498 2017-05-04 Bin Cheng <bin.cheng@arm.com>
25500 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
25501 which is not used any more.
25503 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25505 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
25507 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25509 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
25510 (cortexa53_tunings): Likewise.
25511 (cortexa57_tunings): Likewise.
25512 (cortexa72_tunings): Likewise.
25513 (cortexa73_tunings): Likewise.
25515 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25517 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
25518 Set loop alignment to 8.
25520 2017-05-04 Martin Sebor <msebor@redhat.com>
25522 PR translation/80280
25523 * builtins.c (expand_builtin_object_size): Add missing quoting to
25524 %D and like directives.
25525 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
25526 (hsa_type_for_tree_type): Same.
25527 (verify_function_arguments): Same.
25528 * symtab.c (symbol_table::change_decl_assembler_name): Same.
25529 * varasm.c (get_section): Same.
25532 2017-05-04 Martin Sebor <msebor@redhat.com>
25534 PR translation/80280
25535 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
25537 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25539 * config/aarch64/aarch64.c (generic_addrcost_table):
25540 Change HI/TI mode setting.
25542 2017-05-04 Martin Jambor <mjambor@suse.cz>
25544 PR tree-optimization/80622
25545 * tree-sra.c (comes_initialized_p): New function.
25546 (build_accesses_from_assign): Only set write lazily when
25547 comes_initialized_p is false.
25548 (analyze_access_subtree): Use comes_initialized_p.
25549 (propagate_subaccesses_across_link): Assert !comes_initialized_p
25550 instead of testing for PARM_DECL.
25552 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25554 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
25555 constraint on operand 0 to allow more general addressing modes.
25556 Adjust output template.
25557 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
25559 * config/aarch64/aarch64-protos.h
25560 (aarch64_address_valid_for_prefetch_p): Declare prototype.
25561 * config/aarch64/constraints.md (Dp): New address constraint.
25562 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
25565 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
25567 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
25568 update use of estimate_ipcp_clone_size_and_time.
25569 (estimate_local_effects): Update use of
25570 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
25571 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
25572 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
25573 Return nonspecialized time.
25575 2017-05-04 Richard Biener <rguenther@suse.de>
25577 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
25578 for the last VUSE which def dominates the PHI. Directly call
25580 (get_continuation_for_phi_1): Remove.
25582 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
25584 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
25585 to explain the use of truncating division. Cap the number of
25586 iterations to the maximum given by nb_iterations_upper_bound,
25589 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25591 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
25592 * configure: Regenerate.
25593 * config.in: Regenerate.
25594 * config/i386/driver-mingw32.c: new file.
25595 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
25596 * config.host: Link driver-mingw32.o on MinGW host.
25597 * doc/install.texi: Document new --enable-mingw-wildcard configure
25600 2017-05-04 Marek Polacek <polacek@redhat.com>
25602 PR tree-optimization/80612
25603 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
25605 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25606 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
25608 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
25609 (movt splitter): Likewise.
25610 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
25611 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
25612 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
25613 block for Thumb-1 with MOVT.
25614 (thumb2_legitimate_address_p): Move code block ...
25615 (can_avoid_literal_pool_for_label_p): ... into this new function.
25616 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
25618 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
25619 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
25620 "M-profile targets with the MOVT instruction".
25622 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25624 * config/arm/arm-builtins.c (arm_init_builtins): Rename
25625 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
25626 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
25628 2017-05-04 Martin Liska <mliska@suse.cz>
25630 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
25631 variable cond_code.
25633 2017-05-04 Richard Biener <rguenther@suse.de>
25635 * tree.c (array_at_struct_end_p): Handle arrays at struct
25636 end with flexarrays more conservatively. Refactor and treat
25637 arrays of arrays or aggregates more strict. Fix
25638 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
25639 * tree.c (array_at_struct_end_p): Adjust prototype.
25640 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
25641 * gimple-fold.c (get_range_strlen): Likewise.
25642 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
25644 2017-05-04 Richard Biener <rguenther@suse.de>
25646 PR tree-optimization/31130
25647 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
25649 (supports_overflow_infinity): Likewise.
25650 (is_negative_overflow_infinity): Likewise.
25651 (is_positive_overflow_infinity): Likewise.
25652 (is_overflow_infinity): Likewise.
25653 (stmt_overflow_infinity): Likewise.
25654 (overflow_infinity_range_p): Likewise.
25655 (usable_range_p): Remove as always returning true.
25656 (make_overflow_infinity): Remove.
25657 (negative_overflow_infinity): Likewise.
25658 (positive_overflow_infinity): Likewise.
25659 (avoid_overflow_infinity): Likewise.
25660 (set_value_range): Adjust accordingly.
25661 (set_value_range_to_nonnegative): Likewise, remove now unused
25662 overflow_infinity arg.
25663 (vrp_operand_equal_p): Adjust.
25664 (update_value_range): Likewise.
25665 (range_int_cst_singleton_p): Likewise.
25666 (operand_less_p): Likewise.
25667 (compare_values_warnv): Likewise.
25668 (extract_range_for_var_from_comparison_expr): Likewise.
25669 (vrp_int_const_binop): Likewise.
25670 (zero_nonzero_bits_from_vr): Likewise.
25671 (extract_range_from_multiplicative_op_1): Likewise.
25672 (extract_range_from_binary_expr_1): Likewise.
25673 (extract_range_from_unary_expr): Likewise.
25674 (extract_range_from_comparison): Likewise.
25675 (extract_range_basic): Likewise.
25676 (adjust_range_with_scev): Likewise.
25677 (compare_ranges): Likewise.
25678 (compare_range_with_value): Likewise.
25679 (dump_value_range): Likewise.
25680 (test_for_singularity): Likewise, remove strict_overflow_p parameter
25682 (simplify_cond_using_ranges): Adjust.
25684 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25686 * brig-builtins.def: Added a builtin for class_f64.
25687 * builtin-types.def: Added a builtin type needed by class_f64.
25689 2017-05-03 Jason Merrill <jason@redhat.com>
25691 * timevar.def: Add TV_CONSTEXPR.
25693 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25695 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
25697 2017-05-03 Martin Jambor <mjambor@suse.cz>
25699 * ipa-prop.c (ipa_update_after_lto_read): Removed.
25700 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
25701 * ipa-cp.c (ipcp_propagate_stage): Do not call
25702 ipa_update_after_lto_read.
25703 * ipa-inline.c (ipa_inline): Likewise.
25705 2017-05-03 Martin Jambor <mjambor@suse.cz>
25707 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
25708 tag. Added a default constructor and a destructor.
25709 (ipa_edge_args_sum_t): New class;
25710 (ipa_edge_args_sum): Declare.
25711 (ipa_edge_args_vector): Remove declaration.
25712 (IPA_EDGE_REF): Use ipa_edge_args_sum.
25713 (ipa_free_edge_args_substructures): Remove declaration.
25714 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
25715 (ipa_edge_args_info_available_for_edge_p): Likewise.
25716 * ipa-prop.c (ipa_edge_args_vector): Removed.
25717 (edge_removal_hook_holder): Likewise.
25718 (edge_duplication_hook_holder): Likewise.
25719 (ipa_edge_args_sum): New variable.
25720 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
25721 ipa_edge_args_vector.
25722 (ipa_free_edge_args_substructures): Likewise.
25723 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
25724 ipa_edge_args_vector.
25725 (ipa_edge_removal_hook): Turned into method
25726 ipa_edge_args_sum_t::remove.
25727 (ipa_edge_duplication_hook): Turned into method
25728 ipa_edge_args_sum_t::duplicate.
25729 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
25730 registering edge hooks.
25731 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
25732 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
25733 ipa_edge_args_sum instead of ipa_edge_args_vector.
25734 * ipa-profile.c (ipa_profile): Likewise.
25736 2017-05-03 Martin Jambor <mjambor@suse.cz>
25738 * symbol-summary.h (function_summary): New method exists.
25739 (function_summary::symtab_removal): Deallocate through release.
25740 (call_summary): New class.
25741 (gt_ggc_mx): New overload.
25742 (gt_pch_nx): Likewise.
25743 (gt_pch_nx): Likewise.
25745 2017-05-03 Jeff Law <law@redhat.com>
25747 PR tree-optimization/78496
25748 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
25749 from simplify_cond_using_ranges. Split off code to walk
25750 backwards through casts into ...
25751 (simplify_cond_using_ranges_2): New function.
25752 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
25753 (execute_vrp): After identifying jump threads, call
25754 simplify_cond_using_ranges_2.
25756 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
25759 * ipa-inline.h (inline_summary): Add ctor.
25760 (create_ggc): Do not use ggc_cleared_alloc.
25762 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
25763 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25765 * gcc.c (handle_braces): Support escaping in switch matching
25767 * doc/invoke.texi (Spec Files): Document it.
25768 Remove superfluous @code markup in items.
25770 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25772 * diagnostic-show-locus.c (struct column_range): New struct.
25773 (get_affected_columns): New function.
25774 (get_printed_columns): New function.
25775 (struct correction): New struct.
25776 (correction::ensure_capacity): New function.
25777 (correction::ensure_terminated): New function.
25778 (struct line_corrections): New struct.
25779 (line_corrections::~line_corrections): New dtor.
25780 (line_corrections::add_hint): New function.
25781 (layout::print_trailing_fixits): Reimplement in terms of the new
25783 (selftest::test_overlapped_fixit_printing): New function.
25784 (selftest::diagnostic_show_locus_c_tests): Call it.
25786 2017-05-03 Nathan Sidwell <nathan@acm.org>
25788 Canonicalize canonical type hashing
25789 * tree.h (type_hash_canon_hash): Declare.
25790 * tree.c (type_hash_list, attribute_hash_list): Move into
25791 type_hash_canon_hash.
25792 (build_type_attribute_qual_variant): Break out hash code calc into
25793 type_hash_canon_hash.
25794 (type_hash_canon_hash): New. Generic type hash computation.
25795 (build_range_type_1, build_array_type_1, build_function_type,
25796 build_method_type_directly, build_offset_type, build_complex_type,
25797 make_vector_type): Call it.
25799 2017-05-03 Richard Biener <rguenther@suse.de>
25801 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
25802 When all DRs have unknown misaligned do not always peel
25803 when there is a store but apply the same costing model as if
25804 there were only loads.
25806 2017-05-03 Richard Biener <rguenther@suse.de>
25809 PR tree-optimization/80492
25810 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25811 compare_base_decls returning dont-know properly.
25813 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
25815 * config/arm/iterators.md (CCSI): New mode iterator.
25816 (arch): New mode attribute.
25817 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
25818 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
25819 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
25820 code iterator for success result mode.
25821 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
25822 the corresponding new insn generators.
25824 2017-05-03 Bin Cheng <bin.cheng@arm.com>
25827 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25828 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25830 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
25832 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
25833 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
25834 (DDR_A): Wrap DDR argument in brackets.
25835 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
25836 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
25837 (DDR_REVERSED_P): Likewise.
25839 2017-05-03 Jakub Jelinek <jakub@redhat.com>
25841 PR tree-optimization/79472
25842 * tree-switch-conversion.c (struct switch_conv_info): Add
25843 contiguous_range and default_case_nonstandard fields.
25844 (collect_switch_conv_info): Compute contiguous_range and
25845 default_case_nonstandard fields, don't clear final_bb if
25846 contiguous_range and only the default case doesn't have the required
25848 (check_all_empty_except_final): Set default_case_nonstandard instead
25849 of failing if contiguous_range and the default case doesn't have empty
25851 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
25852 and only the default case doesn't have the required constants. Skip
25854 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
25855 if default_case_nonstandard.
25856 (build_constructors): Build constant 1 just once. Assert that default
25857 values aren't inserted in between cases if contiguous_range. Skip
25859 (build_arrays): Skip virtual phis.
25860 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
25861 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
25862 Handle virtual phis.
25863 (gen_inbound_check): Handle default_case_nonstandard case.
25864 (process_switch): Adjust check_final_bb caller. Call
25865 gather_default_values with the first non-default case instead of
25866 default case if default_case_nonstandard.
25868 2017-05-02 Nathan Sidwell <nathan@acm.org>
25870 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
25871 check. Fix formatting.
25873 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
25875 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
25876 errors when comparing specialized and unspecialized times.
25878 2017-05-02 David Malcolm <dmalcolm@redhat.com>
25880 * diagnostic-show-locus.c
25881 (layout::should_print_annotation_line_p): Make private.
25882 (layout::print_annotation_line): Make private.
25883 (layout::annotation_line_showed_range_p): Make private.
25884 (layout::show_ruler): Make private.
25885 (layout::print_source_line): Make private. Pass in line and
25886 line_width, rather than calling location_get_source_line. Drop
25888 (layout::print_leading_fixits): New method.
25889 (layout::print_any_fixits): Rename to...
25890 (layout::print_trailing_fixits): ...this, and make private.
25891 Don't print newline fixits.
25892 (diagnostic_show_locus): Move logic for printing one row into...
25893 (layout::print_line): ...this new function. Move the
25894 location_get_source_line call and error-handling from
25895 print_source_line to here. Call print_leading_fixits, and rename
25896 print_any_fixits to print_trailing_fixits.
25897 (selftest::test_fixit_insert_containing_newline): Update now that
25898 newlines are partially supported.
25899 (selftest::test_fixit_insert_containing_newline_2): New test.
25900 (selftest::test_fixit_replace_containing_newline): Update comments.
25901 (selftest::diagnostic_show_locus_c_tests): Call the new test.
25902 * edit-context.c (class added_line): New class.
25903 (class edited_line): Describe newline handling in comment.
25904 (edited_line::actually_edited_p): New method.
25905 (edited_line::print_content): Delete redundant decl.
25906 (edited_line::m_predecessors): New field.
25907 (edited_file::print_content): Call edited_line::print_content.
25908 (edited_file::print_diff): Update to support newlines.
25909 (edited_file::print_diff_hunk): Likewise.
25910 (edited_file::print_run_of_changed_lines): New function.
25911 (edited_file::print_diff_line): Convert to...
25912 (print_diff_line): ...this.
25913 (edited_file::get_effective_line_count): New function.
25914 (edited_line::edited_line): Initialize new field m_predecessors.
25915 (edited_line::~edited_line): Clean up m_predecessors.
25916 (edited_line::apply_fixit): Handle newlines.
25917 (edited_line::get_effective_line_count): New function.
25918 (edited_line::print_content): New function.
25919 (edited_line::print_diff_lines): New function.
25920 (selftest::test_applying_fixits_insert_containing_newline): New
25922 (selftest::test_applying_fixits_replace_containing_newline): New
25924 (selftest::insert_line): New function.
25925 (selftest::test_applying_fixits_multiple_lines): Add example of
25927 (selftest::edit_context_c_tests): Call the new tests.
25929 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25931 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
25932 parameter cand. Update dump information.
25933 (get_computation_cost): Update uses.
25935 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25937 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
25938 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
25939 (get_computation_at, rewrite_use_address): Update use of
25940 get_computation_aff.
25942 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25944 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
25945 (get_computation): Delete.
25946 (get_computation_cost): Implement like get_computation_cost_at.
25947 Use get_computation_at.
25948 (get_computation_cost_at): Delete.
25949 (rewrite_use_nonlinear_expr): Use get_computation_at.
25950 (rewrite_use_compare, remove_unused_ivs): Ditto.
25952 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25954 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
25956 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25958 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
25959 (ivopts_global_cost_for_size): Rename parameter and update uses.
25960 (iv_ca_recount_cost): Update uses.
25961 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
25962 candidates seperately in n_invs and n_cands.
25963 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
25965 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25967 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
25968 (find_inv_vars_cb): New.
25969 (find_depends): Renamed to ...
25970 (find_inv_vars): ... this.
25971 (add_candidate_1, force_var_cost): Call find_inv_vars.
25972 (split_address_cost, determine_group_iv_cost_cond): Ditto.
25974 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25976 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
25977 inv_vars. Add inv_exprs.
25978 (struct iv_cand): Rename depends_on to inv_vars.
25979 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
25980 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
25981 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
25982 (dump_cand): Dump inv_vars.
25983 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
25984 (record_invariant, find_depends, add_candidate_1): Ditto.
25985 (set_group_iv_cost, force_var_cost): Ditto.
25986 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
25987 (get_computation_cost_at, get_computation_cost): Ditto.
25988 (determine_group_iv_cost_generic): Ditto.
25989 (determine_group_iv_cost_address): Ditto.
25990 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
25991 (determine_group_iv_costs): Ditto.
25992 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
25993 (iv_ca_set_remove_invariants): Renamed to ...
25994 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
25995 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
25996 (iv_ca_set_add_invariants): Renamed to ...
25997 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
25998 (iv_ca_set_cp): Use iv_ca_set_add_invs.
25999 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26000 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26001 (create_new_ivs): Remove useless dump.
26003 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26005 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26007 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26008 (iv_ca_set_no_cp, create_new_iv): Ditto.
26010 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26012 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26014 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26016 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26017 function tree_check2.
26019 2017-05-02 Martin Liska <mliska@suse.cz>
26021 * doc/gcov.texi: Add missing preposition.
26022 * gcov.c (function_info::function_info): Properly fill up
26023 all member variables.
26025 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26027 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26029 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26031 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26033 2017-05-02 Martin Liska <mliska@suse.cz>
26036 * lto-streamer-in.c (lto_read_tree_1): Remove
26037 LTO_STREAMER_DEBUG.
26038 * lto-streamer.c (struct tree_hash_entry): Likewise.
26039 (struct tree_entry_hasher): Likewise.
26040 (tree_entry_hasher::hash): Likewise.
26041 (tree_entry_hasher::equal): Likewise.
26042 (lto_streamer_init): Likewise.
26043 (lto_orig_address_map): Likewise.
26044 (lto_orig_address_get): Likewise.
26045 (lto_orig_address_remove): Likewise.
26046 * lto-streamer.h: Likewise.
26047 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26048 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26050 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26052 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26053 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26054 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26055 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26056 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26057 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26058 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26059 (mm_maskz_sub_ss): New intrinsics.
26060 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26061 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26062 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26063 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26064 (__builtin_ia32_subss_mask_round): New builtins.
26065 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26066 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26067 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26069 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26070 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26072 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26075 2017-05-02 Martin Jambor <mjambor@suse.cz>
26077 PR tree-optimization/78687
26078 * tree-sra.c (access): New field parent.
26079 (process_subtree_disqualification): New function.
26080 (disqualify_candidate): Call it.
26081 (build_accesses_from_assign): Reset write flag if creating an
26083 (build_access_subtree): Fill in parent field and also prpagate
26084 down grp_write flag.
26085 (create_artificial_child_access): New parameter set_grp_write, set
26086 grp_write to its value.
26087 (propagate_subaccesses_across_link): Also propagate grp_write flag
26089 (propagate_all_subaccesses): Push the closest parent back to work
26090 queue if add_access_to_work_queue returned true.
26092 2017-05-02 Richard Biener <rguenther@suse.de>
26094 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26095 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26096 -fstrict-overflow documentation.
26097 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26098 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26099 flag_strict_overflow.
26100 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26101 * lto-opts.c (lto_write_options): Do not stream it.
26102 * lto-wrapper.c (merge_and_complain): Do not handle it.
26103 * opts.c (default_options_table): Do not set -fstrict-overflow.
26104 (finish_options): Likewise do not clear it when sanitizing.
26105 * simplify-rtx.c (simplify_const_relational_operation): Do not
26106 test flag_strict_overflow.
26108 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26110 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26111 using enabled attribute.
26112 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26113 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26114 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26115 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26116 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26117 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26118 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26119 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26120 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26121 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26123 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26125 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26127 2017-05-02 Richard Biener <rguenther@suse.de>
26129 PR tree-optimization/80591
26131 2017-04-10 Richard Biener <rguenther@suse.de>
26133 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26136 2017-05-02 Richard Biener <rguenther@suse.de>
26138 PR tree-optimization/80549
26139 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26140 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26141 headers if they do not have a preheader.
26143 2017-05-02 Martin Liska <mliska@suse.cz>
26146 * common.opt: Fix typo.
26147 * doc/invoke.texi: Likewise.
26149 2017-05-01 Jan Beulich <jbeulich@suse.com>
26151 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26152 swapping, add (x,x,m,x,n) alternative.
26154 2017-05-01 Nathan Sidwell <nathan@acm.org>
26156 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26157 unnecessary unadjusted_alignment check.
26159 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26162 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26164 * gimplify.c (gimplify_cilk_detach): New function.
26165 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26166 * tree-core.h: Document EXPR_CILK_SPAWN.
26167 * tree.h (EXPR_CILK_SPAWN): Define.
26169 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26171 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26172 to use new fixit_hint representation, using the "replace" logic.
26173 (get_line_span_for_fixit_hint): Likewise.
26174 (layout::print_any_fixits): Likewise.
26175 (selftest::test_one_liner_many_fixits): Rename to...
26176 (selftest::test_one_liner_many_fixits_1): ...this, and update
26177 comment and expected output to reflect that the multiple fix-it
26178 hints are now consolidated into one insertion.
26179 (selftest::test_one_liner_many_fixits_2): New test.
26180 (selftest::test_diagnostic_show_locus_one_liner): Update for
26182 (selftest::test_fixit_consolidation): Update for fix-it API
26184 * diagnostic.c (print_parseable_fixits): Likewise.
26185 * edit-context.c (edited_line::m_line_events): Convert from
26186 auto_vec <line_event *> to auto_vec <line_event>.
26187 (class line_event): Convert from abstract base class to a concrete
26188 class, taking over the role of replace_event.
26189 (class insert_event): Delete.
26190 (class replace_event): Rename to class line_event. Convert to
26192 (edit_context::add_fixits): Reimplement.
26193 (edit_context::apply_insert): Delete.
26194 (edit_context::apply_replace): Rename to...
26195 (edit_context::apply_fixit): ...this. Convert to half-open range.
26196 (edited_file::apply_insert): Delete.
26197 (edited_file::apply_replace): Rename to...
26198 (edited_file::apply_fixit): ...this.
26199 (edited_line::~edited_line): Drop deletion of events.
26200 (edited_line::apply_insert): Delete.
26201 (edited_line::apply_replace): Rename to...
26202 (edited_line::apply_fixit): ...this. Convert to half-open range.
26203 Update for change to type of m_line_events.
26204 * edit-context.h (edit_context::apply_insert): Delete.
26205 (edit_context::apply_replace): Rename to...
26206 (edit_context::apply_fixit): ...this.
26208 2017-05-01 Martin Sebor <msebor@redhat.com>
26210 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26213 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26216 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26217 __get_cpuid_max returns 0.
26218 (__get_cpuid_count): Ditto.
26220 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26222 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26223 replacement expression is another instance of one of its arguments.
26225 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26228 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26229 check for stack push/pop autoinc.
26230 * config/i386/i386.c (ix86_agi_dependent): Return false
26231 if the only reason why modified_in_p returned true is that
26232 addr is SP based and set_insn is a push or pop.
26234 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26236 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26239 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26242 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26243 (account_size_time): Use two predicates - exec_pred and
26245 (evaluate_conditions_for_known_args): Compute both clause and
26247 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26248 (inline_summary_t::duplicate): Update.
26249 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26251 (compute_inline_parameters): Likewise.
26252 (estimate_edge_size_and_time): Update caluclation of time.
26253 (estimate_node_size_and_time): Compute both time and nonspecialized
26255 (estimate_ipcp_clone_size_and_time): Update.
26256 (inline_merge_summary): Update.
26257 (do_estimate_edge_time): Update.
26258 (do_estimate_edge_size): Update.
26259 (do_estimate_edge_hints): Update.
26260 (inline_read_section, inline_write_summary): Stream both new predicates.
26261 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26263 (compute_inlined_call_time): Cleanup.
26264 (big_speedup_p): Update.
26265 (edge_badness): Update.
26266 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26267 (size_time_entry): Replace predicate by exec_predicate and
26268 nonconst_predicate.
26269 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26270 (estimate_edge_time): Return also nonspec_time.
26271 (reset_edge_growth_cache): Update.
26273 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26275 PR rtl-optimization/80491
26276 * ifcvt.c (noce_process_if_block): When looking for x setter
26277 with missing else_bb, don't check only the insn right before
26278 cond_earliest, but look for the last insn that x is modified in
26279 within the same bb.
26281 PR rtl-optimization/80491
26282 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26284 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26286 PR tree-optimization/80487
26287 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26289 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26291 PR tree-optimization/79697
26292 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26293 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26294 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26296 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26297 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26299 2017-04-28 Martin Sebor <msebor@redhat.com>
26301 PR tree-optimization/80523
26302 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26303 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26305 (maybe_warn, format_directive, parse_directive): Use new functions.
26306 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26308 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26310 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26312 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26314 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26315 target_header_dir): Set correctly.
26316 * configure: Regenerated.
26317 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26318 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26319 instead of SYSTEM_HEADER_DIR.
26321 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26323 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26324 (estimate_local_effects): Likewise.
26325 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26326 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26327 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26328 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26329 do_estimate_edge_time, estimate_edge_time): Likewise.
26330 * ipa-inline-analysis.c (estimate_node_size_and_time,
26331 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26332 (estimate_time_after_inlining): Remove.
26334 2017-04-28 Martin Liska <mliska@suse.cz>
26336 * doc/gcov.texi: Enhance documentation of gcov.
26338 2017-04-28 Martin Liska <mliska@suse.cz>
26340 * doc/gcov.texi: Sort options in alphabetic order.
26341 * doc/gcov-dump.texi: Likewise.
26342 * doc/gcov-tool.texi: Likewise.
26343 * gcov.c (print_usage): Likewise.
26344 * gcov-dump.c (print_usage): Likewise.
26345 * gcov-tool.c (print_merge_usage_message): Likewise.
26346 (print_rewrite_usage_message): Likewise.
26347 (print_overlap_usage_message): Likewise.
26349 2017-04-28 Martin Liska <mliska@suse.cz>
26351 PR gcov-profile/53915
26352 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
26354 2017-04-28 Martin Liska <mliska@suse.cz>
26356 PR gcov-profile/79891
26357 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
26358 is marked by compiler as living on a line.
26359 (get_cycles_count): Remove usage of the union.
26360 (output_intermediate_file): Likewise.
26361 (find_source): Fix GNU coding style.
26362 (accumulate_line_counts): Remove old non-all block mode.
26363 (output_lines): Remove usage of the union.
26364 * profile.c (output_location): Include all BBs, even if
26365 belonging to a same line (and file) as a previous BB.
26367 2017-04-28 Martin Liska <mliska@suse.cz>
26369 * gcov.c (process_args): Handle new argument 'w'.
26370 (read_graph_file): Assign ID to BBs.
26371 (output_branch_count): Display BB # if verbose flag is set.
26372 (output_lines): Likewise for arcs.
26373 (print_usage): Add '--verbose' option help.
26374 * doc/gcov.texi: Document --verbose (-w) option.
26376 2017-04-28 Martin Liska <mliska@suse.cz>
26378 * gcov.c (struct block_location_info): New struct.
26379 (process_file): Fill up the new structure.
26380 (read_graph_file): Replace usage of encoding by the newly added
26382 (add_line_counts): Likewise.
26383 (accumulate_line_counts): Remove usage of the union.
26384 (function_info::function_info): New function.
26385 (function_info::~function_info): Likewise.
26386 (process_file): Call delete instead of release_function.
26387 (release_function): Release the function.
26388 (release_structures): Call delete instead of release_function.
26389 (solve_flow_graph): Replace usage of num_blocks.
26390 (find_exception_blocks): Likewise.
26391 (output_lines): Fix GNU coding style.
26393 2017-04-28 Martin Liska <mliska@suse.cz>
26396 * coverage.c (coverage_remove_note_file): New function.
26397 * coverage.h: Declare the function.
26398 * toplev.c (finalize): Clean if an error has been seen.
26400 2017-04-28 Martin Liska <mliska@suse.cz>
26402 PR gcov-profile/80031
26403 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
26404 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
26405 * gcov.c (read_graph_file): Read just number of blocks.
26406 * profile.c (branch_prob): Do not stream 0 flags per a basic
26409 2017-04-28 Martin Liska <mliska@suse.cz>
26411 * gcov-dump.c (tag_*): Add new argument to declarations.
26412 (dump_gcov_file): Likewise.
26413 (tag_blocks): Add and use new argument depth.
26414 (tag_arcs): Likewise.
26415 (tag_lines): Likewise.
26416 (tag_counters): Likewise.
26417 (tag_summary): Likewise.
26418 (dump_working_sets): Use depth to do a proper indentation.
26420 2017-04-28 Jakub Jelinek <jakub@redhat.com>
26423 * cgraph.h (symtab_node::debug_symtab): No longer inline.
26424 * symtab.c (symtab_node::debug_symtab): Move definition here.
26426 2017-04-28 Richard Biener <rguenther@suse.de>
26428 * lto-streamer.h (LTO_major_version): Bump to 7.
26430 2017-04-28 Richard Biener <rguenther@suse.de>
26432 * tree-vrp.c (assert_info): New struct.
26433 (add_assert_info): New helper.
26434 (register_edge_assert_for_2): Refactor to add asserts to a vector
26436 (register_edge_assert_for_1): Likewise.
26437 (register_edge_assert_for): Likewise.
26438 (finish_register_edge_assert_for): New helper actually registering
26439 asserts where live on edge.
26440 (find_conditional_asserts): Adjust.
26441 (find_switch_asserts): Likewise.
26442 (evrp_dom_walker::try_find_new_range): Generalize.
26443 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
26445 2017-04-27 Marek Polacek <polacek@redhat.com>
26448 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
26449 arg10 and arg11 to itype.
26451 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
26453 * doc/extend.texi (Object Size Checking): Improve grammar.
26455 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
26458 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
26459 that the logic for permitting reciprocal estimates matches that
26462 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26465 * tree.c (type_cache_hasher::equal): Only compare
26466 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
26467 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
26468 non-aggregate element types.
26469 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
26470 about the flag on ARRAY_TYPEs in the comment, formatting fix.
26472 2017-04-27 Richard Biener <rguenther@suse.de>
26474 PR middle-end/80533
26475 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
26476 stripping ARRAY_REFs from MEM_EXPR make sure we're not
26477 keeping a reference to a trailing array.
26479 2017-04-27 Richard Biener <rguenther@suse.de>
26481 PR middle-end/80539
26482 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
26483 being in loop-closed SSA form conservatively.
26484 (chrec_fold_multiply_poly_poly): Likewise.
26486 2017-04-27 Tamar Christina <tamar.christina@arm.com>
26488 PR middle-end/79665
26489 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
26490 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
26492 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26495 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
26496 (aarch64_function_arg_alignment): Return unsigned int again, but still
26497 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
26498 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26499 Don't emit -Wpsabi note.
26500 (aarch64_function_arg_boundary): Likewise.
26501 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26504 2017-04-26 Nathan Sidwell <nathan@acm.org>
26506 * tree.h (crc32_unsigned_n): Declare.
26507 (crc32_unsigned, crc32_unsigned): Make inline.
26508 * tree.c (crc32_unsigned_bits): Replace with ...
26509 (crc32_unsigned_n): ... this.
26510 (crc32_unsigned, crc32_byte): Remove.
26511 (crc32_string): Remove unnecessary braces.
26513 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26515 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
26516 * ipa-inline-analysis.c (MAX_TIME): Remove.
26517 (account_size_time): Use sreal for time.
26518 (dump_inline_summary): Update.
26519 (estimate_function_body_sizes): Update.
26520 (estimate_edge_size_and_time): Update.
26521 (estimate_calls_size_and_time): Update.
26522 (estimate_node_size_and_time): Update.
26523 (inline_merge_summary): Update.
26524 (inline_update_overall_summary): Update.
26525 (estimate_time_after_inlining): Update.
26526 (inline_read_section): Update.
26527 (inline_write_summary): Update.
26528 * ipa-inline.c (compute_uninlined_call_time): Update.
26529 (compute_inlined_call_time): Update.
26530 (recursive_inlining): Update.
26531 (inline_small_functions): Update.
26532 (dump_overall_stats): Update.
26533 * ipa-inline.h: Include sreal.h.
26534 (size_time_entry): Turn time to sreal.
26535 (inline_summary): Turn self_time nad time to sreal.
26537 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26539 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
26541 (sreal::stream_out, sreal::stream_in): New.
26542 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
26544 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26546 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
26549 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
26552 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
26554 (dimode_scalar_chain::compute_convert_gain): Ditto.
26555 (dimode_scalar_chain::make_vector_copies): Ditto.
26556 (dimode_scalar_chain::convert_reg): Ditto.
26557 (dimode_scalar_chain::convert_insn): Ditto.
26558 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
26559 (VI248_AVX512BW_1): New mode iterator.
26560 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
26561 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
26564 2017-04-25 Martin Sebor <msebor@redhat.com>
26566 PR tree-optimization/80497
26567 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
26568 constants are representable in HOST_WIDE_INT.
26569 (parse_directive): Ditto.
26571 2017-04-25 Martin Sebor <msebor@redhat.com>
26574 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
26575 (new_zero_array): Adjust signature.
26576 (dom_info::dom_init): Used unsigned rather that size_t.
26577 (dom_info::dom_info): Same.
26579 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26580 Jakub Jelinek <jakub@redhat.com>
26583 * config/arm/arm.c: Include gimple.h.
26584 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26585 returns negative, increment ncrn only if it returned positive.
26586 (arm_needs_doubleword_align): Return int instead of bool,
26587 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
26588 members, but if there is any such non-FIELD_DECL
26589 > PARM_BOUNDARY aligned decl, return -1 instead of false.
26590 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26591 returns negative, increment nregs only if it returned positive.
26592 (arm_setup_incoming_varargs): Likewise.
26593 (arm_function_arg_boundary): Emit -Wpsabi note if
26594 arm_needs_doubleword_align returns negative, return
26595 DOUBLEWORD_ALIGNMENT only if it returned positive.
26597 2017-04-25 Marek Polacek <polacek@redhat.com>
26600 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26601 first argument to type.
26603 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
26606 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
26607 type checks to test for compatibility instead of equality.
26609 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26610 Jakub Jelinek <jakub@redhat.com>
26613 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
26615 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
26616 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
26617 the alignment computation, but return their maximum in warn_alignment.
26618 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26619 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
26621 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
26622 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26625 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26627 * config/arc/simdext.md (dmpyh): Fix typo.
26629 2017-04-25 Richard Biener <rguenther@suse.de>
26631 PR tree-optimization/80492
26632 * alias.c (compare_base_decls): Handle registers with asm
26633 specification conservatively.
26634 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26635 compare_base_decls returning dont-know properly.
26637 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26639 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
26640 (legitimate_offset_address_p): New function.
26641 (arc_legitimate_address_p): Use above function.
26643 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26645 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
26647 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26649 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
26650 ACCH registers whenever they are available.
26652 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26654 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
26655 double regs fix when not used.
26657 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26659 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
26661 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
26662 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
26664 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26666 * config/arc/arc.c (arc_output_addsi): Check for h-register class
26667 when emitting short ADD instructions.
26669 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26671 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
26673 (cmpsi_cc_c_insn): Likewise.
26674 (cbranchsi4_scratch): Compute proper instruction length using
26675 compact_hreg_operand.
26676 * config/arc/predicates.md (compact_hreg_operand): New predicate.
26678 2017-04-25 Richard Biener <rguenther@suse.de>
26680 PR middle-end/80509
26681 * passes.c (pass_manager::pass_manager): Initialize
26682 m_name_to_pass_map.
26684 2017-04-25 Richard Biener <rguenther@suse.de>
26686 PR tree-optimization/79201
26687 * tree-ssa-sink.c (statement_sink_location): Handle calls.
26689 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26692 * config/s390/vector.md: Split MEM->GPR vector moves for
26693 non-s_operand addresses.
26695 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26698 * config/s390/predicates.md (reload_const_wide_int_operand): New
26700 * config/s390/s390.md ("movti"): Remove d/P alternative.
26701 ("movti_bigconst"): New pattern definition.
26703 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26706 * s390-protos.h (s390_expand_cs_hqi): Removed.
26707 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
26708 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
26709 modes as well as CCZ1mode and CCZmode.
26710 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
26711 signature of s390_emit_compare_and_swap.
26712 (s390_expand_cs_hqi): Likewise, make static.
26713 (s390_expand_cs_tdsi): Generate an explicit compare before trying
26714 compare-and-swap, in some cases.
26715 (s390_expand_cs): Wrapper function.
26716 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
26718 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
26719 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
26720 patterns for small and large integers. Forbid symref memory operands.
26721 Move expander to s390.c. Require cc register.
26722 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
26723 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
26724 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
26725 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
26726 symref memory operands. Remove CC mode and call s390_match_ccmode
26728 ("atomic_exchange<mode>"): Allow and implement all integer modes.
26730 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26732 * config/s390/s390.md (define_peephole2): New peephole to help
26733 combining the load-and-test pattern with volatile memory.
26735 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26737 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
26738 with CCZmode for TARGET_Z196.
26740 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26742 PR rtl-optimization/80501
26743 * combine.c (make_compound_operation_int): Set subreg_code to SET
26744 even for AND with mask of the sign bit of mode.
26746 PR rtl-optimization/80500
26747 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
26748 sum's initial value.
26750 2017-04-25 Julian Brown <julian@codesourcery.com>
26751 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26753 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
26755 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
26757 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
26759 2017-04-25 Julian Brown <julian@codesourcery.com>
26760 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26762 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
26763 (thunderx2t99_sha): New Reservation.
26765 2017-04-25 Julian Brown <julian@codesourcery.com>
26766 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26768 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
26769 type for 1-element load.
26771 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
26773 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
26775 2017-04-24 Martin Jambor <mjambor@suse.cz>
26777 PR tree-optimization/80293
26778 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
26779 char arrays not totally scalarizable if it is false.
26780 (analyze_all_variable_accesses): Pass correct value in the new
26781 parameter. Add a statistics counter.
26783 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
26785 PR middle-end/79931
26786 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
26788 2017-04-24 Richard Biener <rguenther@suse.de>
26790 PR tree-optimization/80494
26791 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
26792 out for complex types.
26794 2017-04-24 Richard Biener <rguenther@suse.de>
26796 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
26797 * tree-ssa-sccvn.c (print_scc): Print SCC size.
26798 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
26799 (DFS): Adjust and never fail.
26800 (sccvn_dom_walker::fail): Remove.
26801 (sccvn_dom_walker::before_dom_children): Adjust.
26802 (run_scc_vn): Likewise and never fail.
26803 * tree-ssa-pre.c (pass_pre::execute): Adjust.
26804 (pass_fre::execute): Likewise.
26806 2017-04-24 Richard Biener <rguenther@suse.de>
26808 PR tree-optimization/79725
26809 * tree-ssa-sink.c (statement_sink_location): Return whether
26810 failure reason was zero uses. Move that check later.
26811 (sink_code_in_bb): Deal with zero uses by removing the stmt
26814 2017-04-24 Richard Biener <rguenther@suse.de>
26817 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
26818 pointer-based references.
26820 2017-04-24 Richard Biener <rguenther@suse.de>
26823 * pass_manager.h (pass_manager::operator new): Remove.
26824 (pass_manager::operator delete): Likewise.
26825 * passes.c (pass_manager::operator new): Remove.
26826 (pass_manager::operator delete): Likewise.
26827 (pass_manager::pass_manager): Zero individual pass members.
26829 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
26832 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
26833 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
26834 Check "XEXP (src, 1)" operand here.
26835 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
26836 Check "XEXP (src, 1)" operand here.
26837 (dimode_scalar_chain::make_vector_copies): Detect count register
26838 of a shift instruction. Zero extend count register from QImode
26839 to DImode to satisfy vector shift pattern count operand predicate.
26840 Substitute vector shift count operand with a DImode copy.
26841 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
26844 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
26846 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
26847 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26848 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
26849 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26850 (UNSPEC_NOREX_MEM): Remove definition.
26852 2017-04-21 Richard Biener <rguenther@suse.de>
26854 PR tree-optimization/79547
26855 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26856 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
26857 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
26858 without any constraints.
26860 2017-04-21 Richard Biener <rguenther@suse.de>
26862 PR tree-optimization/78847
26863 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
26865 2017-04-21 Richard Biener <rguenther@suse.de>
26867 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
26868 (build_distinct_type_copy): Likewise.
26869 (build_variant_type_copy): Likewise.
26870 * tree.c (build_qualified_type): Pass down mem-stat info.
26871 (build_distinct_type_copy): Likewise.
26872 (build_variant_type_copy): Likewise.
26874 2017-04-21 Richard Biener <rguenther@suse.de>
26876 PR tree-optimization/80237
26877 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
26879 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
26880 for a simplified result.
26882 2016-04-21 Richard Biener <rguenther@suse.de>
26884 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
26885 sth as strict as a simple_iv but a chrec without symbols and an
26886 operand defined in the loop we are peeling (and not some subloop).
26887 (propagate_constants_for_unrolling): Propagate all constants.
26889 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26892 * config/i386/i386.c (print_reg): Remove assert for disalowed
26893 regno values, call output_operand_lossage instead.
26895 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26898 * config/i386/constraints.md (Yc): New register constraint.
26899 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
26900 Use Yc constraint for alternative 2 of operand 0. Remove
26901 preferred_for_speed attribute.
26903 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26905 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
26906 lastprivate clauses in SIMT case.
26908 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
26910 * doc/invoke.texi (-Wextra-semi): Document new warning option.
26912 2017-04-20 Richard Biener <rguenther@suse.de>
26914 PR tree-optimization/57796
26915 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
26916 as N scalar stores.
26917 (vect_model_load_cost): Cost gathers as N scalar loads.
26919 2017-04-20 Richard Biener <rguenther@suse.de>
26921 * ggc-page.c (ggc_allocated_p): Rename to ...
26922 (safe_lookup_page_table_entry): ... this and return the lookup
26924 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
26926 2017-04-20 Richard Biener <rguenther@suse.de>
26928 PR tree-optimization/80453
26929 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
26930 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
26931 from the conditions.
26932 (vn_phi_eq): Pass them down.
26933 (vn_phi_lookup): Record them.
26934 (vn_phi_insert): Likewise.
26936 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
26938 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
26939 uninitialized variable warning to avoid buffer overrun.
26941 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26944 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
26945 is suppressed for '{ 0 }' in C.
26947 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26949 * BASE-VER: Set to 8.0.0.
26951 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
26953 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
26954 priority .init_array and .fini_array section with SECTION_NOTYPE
26957 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26959 PR middle-end/80423
26960 * tree.h (build_array_type): Add typeless_storage default argument.
26961 * tree.c (type_cache_hasher::equal): Also compare
26962 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
26963 (build_array_type): Add typeless_storage argument, set
26964 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
26966 (build_nonshared_array_type): Adjust build_array_type_1 caller.
26967 (build_array_type): Likewise. Add typeless_storage argument.
26969 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26970 Jakub Jelinek <jakub@redhat.com>
26972 PR tree-optimization/80426
26973 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
26974 operation on symbolic operands, also compute the overflow for the
26975 invariant part when the operation degenerates into a negation.
26977 2017-04-19 Jakub Jelinek <jakub@redhat.com>
26980 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
26981 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
26984 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
26986 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
26989 * config/avr/avr.c (tree.h): Include it.
26990 (cgraph.h): Include it.
26991 (avr_encode_section_info): Don't warn for uninitialized progmem
26992 variable if it's just an alias.
26994 2017-04-19 Richard Biener <rguenther@suse.de>
26997 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
26998 when needed by AutoPGO.
27000 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27003 * doc/lto.texi: Remove an extra 'that'.
27005 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27007 PR rtl-optimization/80429
27008 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27009 are only used in debug insns.
27011 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27012 Vladimir Makarov <vmakarov@redhat.com>
27014 * config/sparc/predicates.md (input_operand): Add comment. Return
27015 true for any memory operand when LRA is in progress.
27016 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27018 2017-04-18 Jeff Law <law@redhat.com>
27021 * mips.md ({return,simple_return}_internal): Do not overwrite
27024 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27026 PR tree-optimization/80443
27027 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27028 instead of adding 1, subtract -1 and similarly instead of subtracting
27031 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27033 PR rtl-optimization/80357
27034 * haifa-sched.c (tmp_bitmap): New variable.
27035 (model_recompute): Handle duplicate use records.
27036 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27037 (free_global_sched_pressure_data): Free it.
27039 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27042 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27043 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27044 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27045 instead of SYSTEM_HEADER_DIR.
27047 2017-04-18 Jeff Law <law@redhat.com>
27049 PR middle-end/80422
27050 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27051 predecessors after walking up the insn chain.
27053 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27056 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27057 sizetype type into debug info.
27059 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27062 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27063 unneeded test for TARGET_UPPER_REGS_SF.
27064 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27066 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27069 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27070 instead of gsi_after_labels.
27072 2017-04-18 Jeff Law <law@redhat.com>
27074 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27078 2017-04-13 Jeff Law <law@redhat.com>
27079 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27080 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27082 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27085 * config/avr/avr.c (intl.h): Include it.
27086 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27088 2017-04-18 Martin Liska <mliska@suse.cz>
27090 PR gcov-profile/78783
27091 * gcov-tool.c (gcov_output_files): Validate that destination
27092 file is either removed by the tool or by a user.
27094 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27095 Guy Benyei <guybe@mellanox.com>
27097 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27098 block, and do not negate it, the stored id is already negative.
27100 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27102 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27104 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27107 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27108 masks of options that should be turned off if the VSX vector
27109 options are turned off.
27110 (OTHER_P8_VECTOR_MASKS): Likewise.
27111 (OTHER_VSX_VECTOR_MASKS): Likewise.
27112 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27113 rs6000_disable_incompatible_switches to validate no type switches
27115 (rs6000_incompatible_switch): New function to disallow turning on
27116 other vector options if -mno-vsx, -mno-power8-vector, or
27117 -mno-power9-vector are specified.
27119 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27121 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27123 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27125 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27126 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27127 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27128 (ARG_POINTER_CFA_OFFSET): Likewise.
27130 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27132 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27133 conditions to take advantage of various optimizations.
27135 2017-04-13 Jeff Law <law@redhat.com>
27137 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27138 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27139 (zero_extendsidi2_dext): Likewise.
27141 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27144 * fold-const.c (fold_ternary_loc): Revert
27145 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27148 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27150 PR rtl-optimization/80343
27151 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27152 new scratch pseudo.
27154 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27157 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27158 to ubsan_encode_value.
27160 2017-04-13 Jeff Law <law@redhat.com>
27162 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27163 appearing in DEBUG_INSNs.
27165 2017-04-13 Martin Liska <mliska@suse.cz>
27167 PR gcov-profile/80413
27168 * gcov-io.c (gcov_write_string): Copy to buffer just when
27169 allocated size is greater than zero.
27171 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27174 * dwarf2out.c (decls_for_scope): Ignore declarations of
27175 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27177 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27180 * ipa-visibility.c (non_local_p): Fix typos.
27181 (localize_node): When localizing symbol in same comdat group,
27182 dissolve the group only when we know external symbols are going
27184 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27186 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27188 PR tree-optimization/79390
27189 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27190 order does not result in usable sequence, retry with reversed operand
27196 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27197 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27198 op0 instead of fold_convert_loc (loc, type, arg0).
27200 2017-04-12 Jeff Law <law@redhat.com>
27202 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27203 has a delay slot in the generated code.
27205 * config/cris/cris.md (cris_preferred_reload_class): Return
27206 GENNONACR_REGS rather than GENERAL_REGS.
27208 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27211 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27212 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27213 signedness of the result type.
27215 2017-04-12 Richard Biener <rguenther@suse.de>
27216 Jeff Law <law@redhat.com>
27218 PR tree-optimization/80359
27219 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27220 trim stores to TARGET_MEM_REFs.
27222 2017-04-12 Richard Biener <rguenther@suse.de>
27224 PR tree-optimization/79390
27225 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27226 threading case even more.
27228 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27231 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27232 for quad_address_p for TImode, instead of just not indexed_address.
27234 2017-04-12 Richard Biener <rguenther@suse.de>
27235 Bernd Edlinger <bernd.edlinger@hotmail.de>
27237 PR middle-end/79671
27238 * alias.c (component_uses_parent_alias_set_from): Handle
27239 TYPE_TYPELESS_STORAGE.
27240 (get_alias_set): Likewise.
27241 * tree-core.h (tree_type_common): Add typeless_storage flag.
27242 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27243 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27244 for types containing members with TYPE_TYPELESS_STORAGE.
27245 (place_field): Likewise.
27246 (layout_type): Likewise for ARRAY_TYPE.
27247 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27248 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27249 TYPE_TYPELESS_STORAGE.
27250 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27252 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27255 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27256 first argument to type.
27258 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27262 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27263 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27264 (rs6000_expand_binop_builtin): Likewise.
27265 (rs6000_expand_ternop_builtin): Likewise; also add missing
27266 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27267 vshasigma built-ins.
27268 * doc/extend.texi: Document that vec_xxpermdi's third argument
27269 must be a constant.
27271 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27273 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27274 Use shift_const cost parameter when calculating gain of STV shifts.
27276 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27278 PR rtl-optimization/70478
27279 * lra-constraints.c (process_alt_operands): Check memory for
27280 disfavoring memory insn operand.
27282 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27284 PR middle-end/80100
27285 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27286 left shift in unsigned HOST_WIDE_INT type.
27288 PR rtl-optimization/80385
27289 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27290 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27293 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27294 if they have any depend clauses.
27296 2017-04-11 Martin Liska <mliska@suse.cz>
27299 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27300 * ipa-split.c (split_function): Create a local comdat symbol
27301 if caller is in a comdat group.
27303 2017-04-11 Martin Liska <mliska@suse.cz>
27306 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27309 2017-04-11 Martin Sebor <msebor@redhat.com>
27311 PR middle-end/80364
27312 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27313 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27315 (directive::set_width, directive::set_precision, format_character):
27317 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27320 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27323 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27324 conflict, set target->arch_name instead of target->cpu_name.
27326 2017-04-11 Richard Biener <rguenther@suse.de>
27328 PR tree-optimization/80374
27329 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27330 build_zero_cst, remove fold_convertible_p check again.
27332 2017-04-11 Martin Liska <mliska@suse.cz>
27335 * ubsan.c (instrument_object_size): Do not instrument register
27338 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27341 * config/i386/i386-builtin-types.def
27342 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
27343 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
27344 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
27345 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
27346 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
27347 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
27348 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
27349 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
27350 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
27351 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
27352 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
27353 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
27354 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
27355 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
27356 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
27357 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
27358 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
27359 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
27360 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
27361 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
27362 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
27363 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27364 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27365 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27366 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27367 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27368 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27369 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
27370 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
27371 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
27372 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
27373 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
27374 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
27375 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
27376 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
27377 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
27378 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
27379 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
27380 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
27381 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
27382 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
27383 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
27384 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
27385 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
27386 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
27387 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
27389 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
27390 flag to second_arg_count, handle 4 argument function type _COUNT
27391 aliases, handle second_arg_count on second argument rather than last.
27393 2017-04-10 Jeff Law <law@redhat.com>
27395 PR tree-optimization/80374
27396 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
27397 record anything if we can not convert integer_zero_node to the
27400 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27403 * config/rs6000/rs6000.c (rs6000_option_override_internal):
27404 Enhance special handling given to the TARGET_P9_MINMAX option in
27405 relation to certain other options.
27407 2017-04-10 Bin Cheng <bin.cheng@arm.com>
27409 PR tree-optimization/80153
27410 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
27411 remove POINTER_PLUS_EXPR's base part directly, rather than through
27414 2017-04-10 Richard Biener <rguenther@suse.de>
27415 Bin Cheng <bin.cheng@arm.com>
27417 PR tree-optimization/80153
27418 * tree-affine.c (aff_combination_to_tree): Get base pointer from
27419 the first element of pointer type aff_tree. Build result expr in
27421 (add_elt_to_tree): Convert to type unconditionally. Remove other
27422 fold_convert calls.
27423 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
27424 (rewrite_use_nonlinear_expr): Check invariant using iv information.
27426 2017-04-10 Richard Biener <rguenther@suse.de>
27428 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27431 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
27433 PR rtl-optimization/70478
27434 * lra-constraints.c (curr_small_class_check): New.
27435 (update_and_check_small_class_inputs): New.
27436 (process_alt_operands): Update curr_small_class_check. Disfavor
27437 alternative insn memory operands. Check available regs for small
27440 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27443 * config/mips/mips.opt (-mvirt): Update description.
27444 * doc/invoke.texi (-mvirt): Likewise.
27446 2017-04-10 Richard Biener <rguenther@suse.de>
27448 PR middle-end/80362
27449 * fold-const.c (fold_binary_loc): Look at unstripped ops when
27450 looking for NEGATE_EXPR in -A / -B to A / B folding.
27452 2017-04-10 Martin Liska <mliska@suse.cz>
27454 PR gcov-profile/80224
27455 * gcov.c (print_usage): Fix usage string.
27456 (get_gcov_intermediate_filename): Remove.
27457 (output_gcov_file): Use both for normal and intermediate format.
27458 (generate_results): Do not initialize special file for
27459 intermediate format.
27461 2017-04-10 Richard Biener <rguenther@suse.de>
27463 PR tree-optimization/80304
27464 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
27467 2017-04-10 Nathan Sidwell <nathan@acm.org>
27470 * config/rs6000/rs6000.c (rs6000_vector_type): New.
27471 (rs6000_init_builtins): Use it.
27473 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27475 * config/arm/arm.md (<mrc>): Add mode to SET source.
27476 (<mrrc>): Likewise.
27478 2017-04-10 Richard Biener <rguenther@suse.de>
27480 PR middle-end/80344
27481 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
27483 2017-04-10 Jakub Jelinek <jakub@redhat.com>
27486 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
27487 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
27488 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
27489 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
27490 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
27491 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
27492 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
27493 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
27494 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
27495 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
27496 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
27497 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
27498 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
27499 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
27500 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
27501 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
27502 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
27503 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
27504 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
27505 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
27506 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
27507 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
27508 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
27510 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
27512 PR rtl-optimization/70478
27513 * lra-constraints.c: Reverse the last patch.
27515 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
27517 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
27518 Add comment for WCHAR_T.
27520 2017-04-08 Martin Liska <mliska@suse.cz>
27523 2017-04-07 Martin Liska <mliska@suse.cz>
27526 * ipa-split.c (split_function): Add function part to a same comdat
27529 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27532 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
27534 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
27536 * rs6000/rs6000.c (vec_load_pendulum): Rename...
27537 (vec_pairing): ...to this.
27538 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
27539 (rs6000_sched_init): Adjust for name change.
27540 (struct rs6000_sched_context): Likewise.
27541 (rs6000_init_sched_context): Likewise.
27542 (rs6000_set_sched_context): Likewise.
27544 2017-04-07 Jakub Jelinek <jakub@redhat.com>
27550 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
27552 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
27553 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
27554 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
27556 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
27558 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
27560 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27562 PR rtl-optimization/70703
27563 * ira-color.c (update_conflict_hard_regno_costs): Use
27564 int64_t instead of HOST_WIDE_INT.
27566 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27568 PR rtl-optimization/70478
27569 * lra-constraints.c (process_alt_operands): Disfavor alternative
27570 insn memory operands.
27572 2017-04-07 Jeff Law <law@redhat.com>
27574 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
27575 CALL and NOTE_INSN_CALL_ARG_LOCATION.
27577 2017-04-07 Martin Liska <mliska@suse.cz>
27580 * config/aarch64/aarch64.c (aarch64_process_target_attr):
27581 Show error message instead of an ICE.
27583 2017-04-07 Martin Liska <mliska@suse.cz>
27586 * ipa-split.c (split_function): Add function part to a same comdat
27589 2017-04-07 Richard Biener <rguenther@suse.de>
27591 PR middle-end/80341
27592 * tree.c (get_unwidened): Also handle ! for_type case for
27594 * convert.c (do_narrow): Split out from ...
27595 (convert_to_integer_1): ... here. Do not pass final truncation
27596 type to get_unwidened for TRUNC_DIV_EXPR.
27598 2017-04-07 Richard Biener <rguenther@suse.de>
27600 * tree-affine.c (wide_int_ext_for_comb): Take type rather
27602 (aff_combination_const): Adjust.
27603 (aff_combination_scale): Likewise.
27604 (aff_combination_add_elt): Likewise.
27605 (aff_combination_add_cst): Likewise.
27606 (aff_combination_convert): Likewise.
27607 (add_elt_to_tree): Likewise. Remove unused argument.
27608 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
27610 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
27612 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
27614 * config/arm/arm.c (arm_default_short_enums): Use
27615 ARM_DEFAULT_SHORT_ENUMS.
27616 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
27618 2017-04-06 Jakub Jelinek <jakub@redhat.com>
27621 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
27622 members with redundant out-of-class redeclaration.
27624 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27627 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
27628 * config/i386/i386.md (*zero_extendsidi2):
27629 Add (?*x,*x) and (?*v,*v) alternatives.
27631 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27634 * config/i386/i386.c (ix86_expand_builtin)
27635 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
27636 mode from insn data. Convert operands to insn operand mode.
27637 Copy operands that don't satisfy insn predicate to a register.
27639 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
27641 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
27644 2017-04-06 Richard Biener <rguenther@suse.de>
27646 PR tree-optimization/80334
27647 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
27648 preserve alignment of accesses.
27650 2017-04-06 Richard Biener <rguenther@suse.de>
27652 PR tree-optimization/80262
27653 * tree-sra.c (build_ref_for_offset): Preserve address-space
27655 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
27656 Drop useless address-space information on MEM_REF offsets.
27658 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
27660 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
27662 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27664 PR rtl-optimization/70703
27665 * ira-color.c (update_conflict_hard_regno_costs): Use
27666 HOST_WIDE_INT instead of long.
27668 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
27671 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
27672 not defined for x86_64 target. Add -mmmx target option when __SSE2__
27674 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
27675 for x86_64 target. Handle -m3dnowa option.
27677 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27679 PR rtl-optimization/70703
27680 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
27681 (update_conflict_hard_regno_costs): Use long instead of unsigned
27682 arithmetic for cost calculation.
27684 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27685 Bernd Edlinger <bernd.edlinger@hotmail.de>
27688 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
27691 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
27694 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
27695 ptr_mode with Pmode throughout.
27696 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
27697 into probe_stack_range and use DImode.
27699 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
27702 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
27703 call_eh_return is true.
27705 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27707 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
27708 Initialize last_match_fntype_index.
27710 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27713 * tree-nvr.c: Include internal-fn.h.
27714 (pass_return_slot::execute): Ignore internal calls without
27717 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27718 Richard Biener <rguenther@suse.de>
27721 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
27722 captures used multiple times, except for the last use.
27723 * generic-match-head.c: Include gimplify.h.
27725 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27727 PR tree-optimization/79390
27728 * target.h (struct noce_if_info): Declare.
27729 * targhooks.h (default_noce_conversion_profitable_p): Declare.
27730 * target.def (noce_conversion_profitable_p): New target hook.
27731 * ifcvt.h (struct noce_if_info): New type, moved from ...
27732 * ifcvt.c (struct noce_if_info): ... here.
27733 (noce_conversion_profitable_p): Renamed to ...
27734 (default_noce_conversion_profitable_p): ... this. No longer
27736 (noce_try_store_flag_constants, noce_try_addcc,
27737 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
27738 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
27739 instead of noce_conversion_profitable_p.
27740 * config/i386/i386.c: Include ifcvt.h.
27741 (ix86_option_override_internal): Don't override
27742 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
27743 (ix86_noce_conversion_profitable_p): New function.
27744 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
27745 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
27746 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
27747 * doc/tm.texi: Regenerated.
27749 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27751 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
27754 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
27757 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
27758 instructions for small multiply cores.
27760 2017-04-04 Jeff Law <law@redhat.com>
27762 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
27764 (mips_expand_vec_perm_const): Initialize elements in orig_perm
27765 that are not set by the loop over the elements.
27767 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27770 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
27771 int mode, convert_modes it to mode as unsigned, otherwise use
27772 lowpart_subreg to mode rather than SImode.
27773 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
27774 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
27775 Use DImode instead of SImode for the shift count operand.
27776 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
27779 2017-04-04 Richard Biener <rguenther@suse.de>
27781 PR middle-end/80281
27782 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
27783 arithmetic done for the negate or the plus. Simplify.
27784 (A - (-B) -> A + B): Likewise.
27785 * fold-const.c (split_tree): Make sure to not negate pointers.
27787 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
27789 PR rtl-optimization/60818
27790 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
27791 a compare of comparisons with the thing compared if this results
27792 in a different machine mode.
27794 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
27796 * alias.c (base_alias_check): Fix typo in comment.
27797 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
27798 * cgraphunit.c (symbol_table::compile): Likewise.
27799 * collect2.c (maybe_run_lto_and_relink): Likewise.
27800 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
27801 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
27802 * config/avr/avr.c (avr_map_op_t): Likewise.
27803 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
27804 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
27805 * config/epiphany/epiphany.md (movcc): Likewise.
27806 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
27807 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
27809 * config/mips/mips.c (mips_save_restore_reg): Likewise.
27810 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
27811 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
27812 * config/sh/sh.c (sh_rtx_costs): Likewise.
27813 * fold-const.c (fold_truth_andor): Likewise.
27814 * genautomata.c (collapse_flag): Likewise.
27815 * gengtype.h (struct type::u::s): Likewise.
27816 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
27817 * input.c (FORMAT_AMOUNT): Likewise.
27818 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
27819 (known_aggs_to_agg_replacement_list): Likewise.
27820 * ipa-inline-analysis.c: Likewise.
27821 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
27822 * ipa-polymorphic-call.c
27823 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
27824 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
27825 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
27827 * modulo-sched.c (apply_reg_moves): Likewise.
27828 * omp-expand.c (build_omp_regions_1): Likewise.
27829 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
27830 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
27831 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
27832 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
27833 * value-prof.c: Likewise.
27834 * var-tracking.c (val_reset): Likewise.
27836 2017-04-03 Richard Biener <rguenther@suse.de>
27838 PR tree-optimization/80275
27839 * fold-const.c (split_address_to_core_and_offset): Handle
27842 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
27844 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
27845 descriptors is at least equal to that of functions.
27847 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27849 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
27851 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27854 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
27855 (mov<IMOD4:mode>): New expander.
27856 (*mov<IMOD4:mode>_internal): New insn and split pattern.
27858 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
27860 PR rtl-optimization/79405
27861 * fwprop.c (propagations_left): New variable.
27862 (forward_propagate_into): Decrement it.
27863 (fwprop_init): Initialize it.
27864 (fw_prop): If the variable has reached zero, stop propagating.
27865 (fwprop_addr): Ditto.
27867 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27870 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
27871 a FUNCTION_DECL, pass it as decl instead of origin to
27874 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
27876 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
27879 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
27882 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
27883 TARGET_VSX_SMALL_INTEGER.
27885 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27887 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27888 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
27890 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27892 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
27893 extraction from odd-numbered MSA register.
27895 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27897 PR middle-end/80173
27898 * expmed.c (store_bit_field_1): Don't attempt to create
27899 a word subreg out of hard registers wider than word if they
27900 have HARD_REGNO_NREGS of 1 for their mode.
27902 PR middle-end/80163
27903 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
27904 conversions to integer types wider than word and pointer.
27907 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
27908 (rtx_equal_for_cselib_p): Pass 0 to it.
27909 * cselib.c (cselib_hasher::equal): Likewise.
27910 (rtx_equal_for_cselib_1): Add depth argument. If depth
27911 is 128, don't look up VALUE locs and punt. Increment
27912 depth in recursive calls when walking VALUE locs.
27914 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
27916 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
27917 (make_gcov_file_name): Use the canonical path name for generating
27919 (read_line): Fix handling of files with ascii null bytes.
27921 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
27923 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
27924 to initialise a vector register instead
27925 of using a const_int.
27927 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27929 PR translation/80189
27930 * gimplify.c (omp_default_clause): Use %qs instead of %s in
27931 diagnostic messages.
27933 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
27936 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
27937 (dfp_diex_<mode>): Update mode of operand 1.
27938 * doc/extend.texi (dxex, dxexq): Document change to return type.
27939 (diex, diexq): Document change to argument type.
27941 2017-03-30 Martin Jambor <mjambor@suse.cz>
27944 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
27945 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
27946 it reflects the signature changes performed at the callee side.
27947 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
27948 to cgraph_build_function_type_skip_args.
27949 (build_function_decl_skip_args): Adjust call to the above function.
27951 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27954 * config/i386/sse.md
27955 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
27956 register as dest whenever it is a MEM not rtx_equal_p to the
27957 corresponding dup operand, and when forcing into reg move the
27958 reg into the memory afterwards.
27959 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
27960 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
27961 for the force_reg mode.
27962 (avx512vl_vextractf128<mode>): Use register as dest either
27963 always when a MEM, or when it is a MEM not rtx_equal_p to the
27964 corresponding dup operand, or even not when it is a CONST_VECTOR
27965 depending on the mode and lo vs. hi.
27966 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
27968 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
27969 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
27970 Likewise. Require that operands[2] is even.
27971 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
27972 Remove extraneous parens. Require that operands[2] is a multiple
27974 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
27975 operands[0] is a MEM if <mask_applied>, the predicates/constraints
27976 disallow memory then.
27978 2017-03-30 Richard Biener <rguenther@suse.de>
27980 PR tree-optimization/77498
27981 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
27982 to non-constants over backedges.
27984 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
27986 PR rtl-optimization/80233
27987 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
27988 as last_combined_insn. Do not test for BARRIER_P separately.
27990 2017-03-29 Andreas Schwab <schwab@suse.de>
27993 * calls.c (prepare_call_address): Convert funexp to Pmode before
27994 copying to temp reg.
27996 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27998 PR tree-optimization/80158
27999 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28000 Handle possible future case of more than one alternate
28002 (replace_rhs_if_not_dup): Likewise.
28003 (replace_one_candidate): Likewise.
28005 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28007 PR rtl-optimization/80193
28008 * ira.c (ira): Do not check allocation for LRA.
28010 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28012 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28013 (nvptx_output_simt_exit): Declare.
28014 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28015 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28016 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28017 (nvptx_declare_function_name): ...here. Emit declaration of local
28018 memory space buffer for omp_simt_enter insn.
28019 (nvptx_output_unisimt_switch): New.
28020 (nvptx_output_softstack_switch): New.
28021 (nvptx_output_simt_enter): New.
28022 (nvptx_output_simt_exit): New.
28023 * config/nvptx/nvptx.h (struct machine_function): New fields
28024 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28025 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28026 (UNSPECV_SIMT_EXIT): Ditto.
28027 (omp_simt_enter_insn): New insn.
28028 (omp_simt_enter): New expansion.
28029 (omp_simt_exit): New insn.
28030 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28032 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28033 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28034 (expand_GOMP_SIMT_EXIT): New.
28035 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28036 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28037 (GOMP_SIMT_EXIT): Ditto.
28038 * target-insns.def (omp_simt_enter): New insn.
28039 (omp_simt_exit): Ditto.
28040 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28042 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28043 (lower_rec_input_clauses): Likewise.
28044 (lower_lastprivate_clauses): Handle SIMT privatization.
28046 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28047 (ompdevlow_adjust_simt_enter): New.
28048 (find_simtpriv_var_op): New.
28049 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28050 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28052 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28053 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28054 (copy_decl_for_dup_finish): Ditto.
28056 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28058 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28061 * config/i386/i386.c (ix86_option_override_internal): Always
28062 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28064 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28066 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28068 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28070 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28071 mark new edge's irreducible flag accordign to it.
28072 (vect_do_peeling): Check loop preheader edge's irreducible flag
28073 and pass it to function slpeel_add_loop_guard.
28075 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28077 PR tree-optimization/80218
28078 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28079 Update block frequencies and counts.
28081 2017-03-28 Richard Biener <rguenther@suse.de>
28083 PR tree-optimization/78644
28084 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28085 of a simplification result we may not use it at all.
28087 2017-03-28 Richard Biener <rguenther@suse.de>
28090 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28091 without arguments, generate default definition of a SSA name.
28093 2017-03-28 Richard Biener <rguenther@suse.de>
28095 PR middle-end/80222
28096 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28097 TYPE_REF_CAN_ALIAS_ALL references.
28098 * fold-const.c (fold_indirect_ref_1): Likewise.
28100 2017-03-28 Martin Liska <mliska@suse.cz>
28103 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28104 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28106 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28107 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28109 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28110 (EXTRA_SPECS): Define.
28111 (SUBTARGET_EXTRA_SPECS): Likewise.
28112 (SUBTARGET_CPP_SPEC): Likewise.
28113 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28114 SUBTARGET_EXTRA_SPECS.
28115 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28117 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28119 * config/arc/simdext.md (vst64_insn): Update pattern.
28120 (vld32wh_insn): Likewise.
28121 (vld32wl_insn): Likewise.
28122 (vld64_insn): Likewise.
28123 (vld32_insn): Likewise.
28125 2017-03-28 Marek Polacek <polacek@redhat.com>
28128 * fold-const.c (fold_comparison): Use protected_set_expr_location
28129 instead of SET_EXPR_LOCATION.
28131 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28133 * tree.c (add_expr): Avoid name lookup warning.
28135 2017-03-27 Jeff Law <law@redhat.com>
28137 PR tree-optimization/80216
28138 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28139 function name. Limit recursion depth.
28140 (record_temporary_equivalences): Corresponding changes.
28142 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28144 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28147 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28150 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28152 * cfgcleanup.c (reg_note_cfa_p): New array.
28153 (insns_have_identical_cfa_notes): New function.
28154 (old_insns_match_p): Don't cross-jump in between /f
28155 and non-/f instructions. If both i1 and i2 are frame related,
28156 verify all CFA notes, their order and content.
28158 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28161 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28162 HImode and SImode with zero extend to DImode to one insn.
28163 (bswap<mode>2_extenddi): Likewise.
28164 (bswapsi2_extenddi): Likewise.
28165 (bswaphi2_extendsi): Likewise.
28166 (bswaphi2): Combine bswap HImode and SImode into one insn.
28167 Separate memory insns from swapping register.
28168 (bswapsi2): Likewise.
28169 (bswap<mode>2): Likewise.
28170 (bswaphi2_internal): Delete, no longer used.
28171 (bswapsi2_internal): Likewise.
28172 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28173 store, and gpr<-gpr swap insns.
28174 (bswap<mode>2_store): Likewise.
28175 (bswaphi2_reg): Register only splitter, combine with the splitter.
28176 (bswaphi2 splitter): Likewise.
28177 (bswapsi2_reg): Likewise.
28178 (bswapsi2 splitter): Likewise.
28179 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28180 the insns into load, store, and register/register insns.
28181 (bswapdi2_ldbrx): Likewise.
28182 (bswapdi2_load): Likewise.
28183 (bswapdi2_store): Likewise.
28184 (bswapdi2_reg): Likewise.
28186 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28188 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28189 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28191 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28194 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28196 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28197 special handling for target option conflicts between dform
28198 options (-mpower9-dform, -mpower9-dform-vector,
28199 -mpower9-dform-scalar) and -mno-direct-move.
28201 2017-03-27 Richard Biener <rguenther@suse.de>
28203 PR tree-optimization/80181
28204 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28206 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28208 * config/arc/predicates.md (move_double_src_operand): Replace the
28209 call to move_double_src_operand with a call to address_operand.
28211 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28213 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28214 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28215 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28217 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28219 * config/arc/predicates.md (long_immediate_loadstore_operand):
28220 Consider scaled addresses cases.
28222 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28224 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28225 restored when in interrupt.
28226 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28227 doesn't have delay slot.
28229 2017-03-27 Richard Biener <rguenther@suse.de>
28232 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28233 inlined thunk clones.
28235 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28238 * asan.c (instrument_derefs): Copy over last operand from
28239 original COMPONENT_REF to the new COMPONENT_REF with
28240 DECL_BIT_FIELD_REPRESENTATIVE.
28241 * ubsan.c (instrument_object_size): Likewise.
28243 2017-03-27 Richard Biener <rguenther@suse.de>
28245 PR tree-optimization/80170
28246 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28247 sure DR/SCEV didnt fold in constants we do not see when looking
28248 at the reference base alignment.
28250 2017-03-27 Richard Biener <rguenther@suse.de>
28252 PR middle-end/80171
28253 * gimple-fold.c (fold_ctor_reference): Properly guard against
28254 NULL return value from canonicalize_constructor_val.
28256 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28259 * config/i386/i386.c (ix86_expand_builtin)
28260 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28261 flags reg setting and flags reg using instructions.
28262 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28263 clobbering instructions to zero extend op2.
28265 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28267 * doc/install.texi (Configuration) <--with-aix-soname>:
28268 Update link to AIX ld.
28270 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28272 PR rtl-optimization/80160
28273 PR rtl-optimization/80159
28274 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28275 reg_alternate_class into account.
28277 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28280 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28281 to consider in curr_insn_transform.
28283 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28285 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28286 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28287 and emit_mode_inner.
28289 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28291 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28292 argument to the overloaded builtin variants. Use the new flag to
28293 deprecate certain builtin variants.
28294 * config/s390/s390-builtin-types.def: Add new builtin types.
28295 * config/s390/s390-builtins.h: Support new flags field for
28296 overloaded builtins.
28297 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28298 (s390_macro_to_expand): Enable vector float data type.
28299 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28300 builtins by incrementing the __VEC__ version number.
28301 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28303 (s390_resolve_overloaded_builtin): Emit error messages depending
28304 on the builtin flags.
28305 * config/s390/s390.c (s390_expand_builtin): Support additional
28306 flags argument. Change error message to match the messages
28307 emitted in s390-c.c.
28308 * config/s390/s390.md: New UNSPEC_* constants.
28309 (op_type): Add new instruction types.
28310 * config/s390/vecintrin.h: Add new builtins and test data class
28312 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28313 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28314 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28315 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28316 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28317 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28319 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28320 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28321 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28322 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28324 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28325 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28326 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28327 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28329 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28330 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28331 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28332 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28333 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28334 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28335 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28337 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28338 ("vec_scatter_element<V_HW_4:mode>_DI")
28339 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28340 ("vec_fpint<mode>", "vflls")
28341 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28342 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
28343 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
28344 ("*vec_cmphe<mode>_cc"): ... these.
28346 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
28347 mode constant instead of magic value.
28349 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28351 * config/s390/s390.c (s390_expand_vec_compare): Support other
28352 vector floating point modes than just V2DF.
28353 (s390_expand_vcond): Likewise.
28354 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
28355 (s390_cannot_change_mode_class): Prevent mode changes between TF
28356 and V1TF in vector registers.
28357 * config/s390/s390.md (DF, SF): New mode attributes.
28358 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
28359 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
28360 SFmode support for VRs.
28361 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
28363 (VFT, VF_HW): New mode iterators.
28364 (vw, sdx): New mode attributes.
28365 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28366 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28367 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28368 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28369 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
28370 also the new vector floating point modes. Renaming to ...
28372 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
28373 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
28374 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
28375 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
28376 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
28377 ("vec_unordered<mode>"): ... these.
28379 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
28380 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
28381 ("*vec_extendv2df"): New insn definitions.
28383 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28385 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
28386 ("mulditi3_2", "*muldi3_sign"): New patterns.
28387 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
28388 rename the pattern definition.
28390 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28392 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
28394 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
28396 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28398 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
28399 instruction if possible.
28400 * config/s390/vector.md (vec_halfnumelts): New mode
28402 ("*vec_vllezlf<mode>"): New pattern.
28404 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28406 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
28407 ("popcountv4si2", "popcountv2di2"): Rename to ...
28408 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
28409 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
28411 ("popcount<mode>2_vxe"): New pattern.
28413 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28415 * common/config/s390/s390-common.c (processor_flags_table): Add
28417 * config.gcc: Add arch12.
28418 * config/s390/driver-native.c (s390_host_detect_local_cpu):
28419 Default to arch12 for unknown CPU model numbers.
28420 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
28421 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
28422 PROCESSOR_max sanity check.
28423 * config/s390/s390-opts.h (enum processor_type): Add
28425 * config/s390/s390.c (processor_table): Add arch12.
28426 (s390_expand_builtin): Add check for B_VXE flag.
28427 (s390_issue_rate): Add PROCESSOR_ARCH12.
28428 (s390_get_sched_attrmask): Likewise.
28429 (s390_get_unit_mask): Likewise.
28430 (s390_sched_score): Enable z13 scheduling for arch12.
28431 (s390_sched_reorder): Likewise.
28432 (s390_sched_variable_issue): Likewise.
28433 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
28435 (s390_tune_attr): Use z13 scheduling also for arch12.
28436 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
28437 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
28438 (TARGET_VXE_P): New macros.
28439 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
28440 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
28441 * config/s390/s390.opt: Add arch12 as processor_type.
28443 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28445 * config/s390/s390.md
28446 ("fixuns_truncdddi2", "fixuns_trunctddi2")
28447 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
28448 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
28450 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
28451 Rename expanders to ...
28453 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
28454 ("fixuns_truncdddi2_emu"): ... these.
28456 ("fixuns_trunc<mode>si2_emu"): New expander.
28458 ("*fixuns_truncdfdi2_z13"): Rename to ...
28459 ("*fixuns_truncdfdi2_vx"): ... this.
28461 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28463 * config/s390/2964.md: Remove the single element vector compare
28464 instructions which are no longer used.
28465 * config/s390/s390.c (s390_select_ccmode): Remove handling of
28467 (s390_canonicalize_comparison): Remove handling of DFmode
28469 (s390_expand_vec_compare_scalar): Remove function.
28470 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
28471 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
28473 ("*cmp<mode>_ccs"): Add wfcdb instruction.
28475 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28477 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
28479 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
28480 will anyway by matched by mov<mode>_64dfp.
28482 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28484 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
28485 vlef/vstef. Add missing operand to vleif.
28487 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28489 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
28490 pair for all vector types with 64 bit elements.
28491 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
28492 * config/s390/vector.md (V_HW_64): ... here.
28493 (V_128_NOSINGLE): New mode iterator.
28494 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
28495 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
28496 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
28497 ("*vec_load_pairv2di"): Change to ...
28498 ("*vec_load_pair<mode>"): ... this one.
28500 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28502 * config/s390/constraints.md: Add comments.
28503 (jKK): Reject element sizes > 8 bytes.
28504 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
28506 * config/s390/s390.md: Add the s_operand checks formerly in
28507 s390_split_ok_p to various splitters where they are still
28509 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
28510 for 128 bit vectors. Plus two splitters.
28512 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28514 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
28517 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28520 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
28521 error if the boundary argument is not constant.
28523 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28525 PR rtl-optimization/80112
28526 * loop-doloop.c (doloop_condition_get): Don't check condition
28527 if cmp isn't SET with IF_THEN_ELSE src.
28529 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28531 PR tree-optimization/80158
28532 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
28533 replacing a candidate statement, also replace it for the
28534 candidate's alternate interpretation.
28535 (replace_rhs_if_not_dup): Likewise.
28536 (replace_one_candidate): Likewise.
28538 2017-03-24 Richard Biener <rguenther@suse.de>
28540 PR tree-optimization/80167
28541 * graphite-isl-ast-to-gimple.c
28542 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
28544 (translate_isl_ast_to_gimple::get_rename): Likewise.
28546 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28548 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
28549 handling of certain combinations of target options, including the
28550 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
28551 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
28553 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28556 * config/arm/arm.md (*load_multiple): Add reload_completed to
28557 matching condition.
28559 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28560 Richard Biener <rguenth@suse.de>
28562 PR tree-optimization/79908
28563 PR tree-optimization/80136
28564 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28565 been cast away, gimplify_and_add suffices.
28567 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
28569 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
28571 2017-03-23 Richard Biener <rguenther@suse.de>
28573 PR tree-optimization/80032
28574 * gimplify.c (gimple_push_cleanup): Forced unconditional
28575 cleanups still have to go to the conditional_cleanups
28578 2017-03-22 Jakub Jelinek <jakub@redhat.com>
28580 PR tree-optimization/80072
28581 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
28583 (next_operand_entry_id): Change type to unsigned int.
28584 (sort_by_operand_rank): Make sure to return the right return value
28585 even if unsigned fields are bigger than INT_MAX.
28586 (struct oecount): Change cnt and id type to unsigned int.
28587 (oecount_hasher::equal): Formatting fix.
28588 (oecount_cmp): Make sure to return the right return value
28589 even if unsigned fields are bigger than INT_MAX.
28590 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
28593 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
28594 TREE_READONLY on result if writing it more than once.
28597 * doc/invoke.texi (-fsanitize=thread): Document that with
28598 -fnon-call-exceptions atomics are not able to throw
28602 * tsan.c: Include tree-eh.h.
28603 (instrument_builtin_call): Call maybe_clean_eh_stmt or
28604 maybe_clean_or_replace_eh_stmt where needed.
28605 (instrument_memory_accesses): Add cfg_changed argument.
28606 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
28607 if it returned true.
28608 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
28610 PR rtl-optimization/63191
28611 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
28612 wrapper function, moved the whole old content into ...
28613 (ix86_delegitimize_address_1): ... this. New inline function.
28614 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
28615 true as last argument instead of ix86_delegitimize_address.
28617 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28619 * config/aarch64/aarch64.c (generic_branch_cost): Copy
28620 cortexa57_branch_cost.
28622 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28624 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
28626 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28629 * doc/md.texi (Constraints): Document wA constraint.
28630 * config/rs6000/constraints.md (wA): New.
28631 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
28632 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
28633 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
28634 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
28636 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
28639 * gimplify.c (is_oacc_declared): New function.
28640 (oacc_default_clause): Use it to set default flags for acc declared
28641 variables inside parallel regions.
28642 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
28643 declared variables.
28644 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
28645 declare attribute to any decl as necessary.
28647 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
28650 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
28651 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
28652 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
28653 (arm_arch_lpae): This.
28654 * config/arm/arm.c (arm_arch7ve): Rename into ...
28655 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
28656 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
28659 2017-03-22 Martin Liska <mliska@suse.cz>
28662 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
28663 error message instead of an ICE.
28665 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28667 * doc/extend.texi (6.11 Additional Floating Types): Revise.
28669 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
28671 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28673 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28676 2017-03-21 Martin Sebor <msebor@redhat.com>
28678 * doc/extend.texi: Use "cannot" instead of "can't."
28679 * doc/hostconfig.texi: Same.
28680 * doc/install.texi: Same.
28681 * doc/invoke.texi: Same.
28682 * doc/loop.texi: Same.
28683 * doc/md.texi: Same.
28684 * doc/objc.texi: Same.
28685 * doc/rtl.texi: Same.
28686 * doc/tm.texi: Same.
28687 * doc/tm.texi.in: Same.
28688 * doc/trouble.texi: Same.
28690 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
28693 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
28694 (collect_checksum_attributes): Set it.
28695 (die_checksum_ordered): Use it.
28697 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28699 PR tree-optimization/79908
28700 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
28701 change: For a VA_ARG whose LHS has been cast away, use
28702 force_gimple_operand to construct the side effects.
28704 2017-03-21 David Malcolm <dmalcolm@redhat.com>
28706 PR translation/80001
28707 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
28708 more amenable to translation.
28709 (oacc_loop_auto_partitions): Likewise.
28711 2017-03-21 Marek Polacek <polacek@redhat.com>
28712 Martin Sebor <msebor@redhat.com>
28714 PR tree-optimization/80109
28715 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
28716 on INTEGRAL_TYPE_P.
28718 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28719 Segher Boessenkool <segher@kernel.crashing.org>
28722 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
28723 check reg_used_between_p between insn and one of succ or succ2
28724 depending on if succ is artificial insn not inserted into insn
28727 2017-03-21 Martin Liska <mliska@suse.cz>
28729 PR gcov-profile/80081
28730 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
28731 * doc/gcc.texi: Include gcov-dump stuff.
28732 * doc/gcov-dump.texi: New file.
28734 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
28736 PR rtl-optimization/79150
28737 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
28738 conditional jump, if the jump is the last insn of the loop.
28740 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28741 Richard Biener <rguenth@suse.de>
28743 PR tree-optimization/79908
28744 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28745 been cast away, use force_gimple_operand to construct the side
28748 2017-03-21 Martin Liska <mliska@suse.cz>
28750 PR libfortran/79956
28751 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
28754 2017-03-21 Brad Spengler <spender@grsecurity.net>
28757 * plugin.c (htab_hash_plugin): New function.
28758 (add_new_plugin): Use it and adjust.
28759 (parse_plugin_arg_opt): Adjust.
28760 (init_one_plugin): Likewise.
28762 2017-03-21 Richard Biener <rguenther@suse.de>
28764 PR tree-optimization/80032
28765 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
28766 if set force the cleanup to happen unconditionally.
28767 (gimplify_target_expr): Push inserted clobbers with force_uncond
28768 to avoid them being removed by control-dependent DCE.
28770 2017-03-21 Richard Biener <rguenther@suse.de>
28772 PR tree-optimization/80122
28773 * tree-inline.c (copy_bb): Do not expans va-arg packs or
28774 va_arg_pack_len when the inlined call stmt requires pack
28776 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
28778 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28781 * tsan.c (instrument_builtin_call): If the memory model argument
28782 is not a constant, assume it is valid.
28785 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
28788 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28790 PR rtl-optimization/79910
28791 * combine.c (can_combine_p): Do not allow combining an I0 or I1
28792 if its dest is used by an insn before I2 (other than the combined
28793 insns themselves, which are properly handled already).
28795 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28798 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28800 * combine.c (record_used_regs): New static function.
28801 (try_combine): Handle situations where there is an additional
28802 instruction between I2 and I3 which needs to have a LOG_LINK
28806 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28808 * combine.c (try_combine): Delete redundant i1 test. Call
28809 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28811 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28814 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
28815 alternatives 13/14.
28817 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28819 PR tree-optimization/80054
28820 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
28821 the optimization if a PHI or any of its arguments is not dominated
28822 by the candidate's basis. Use gphi* rather than gimple* as
28824 (replace_profitable_candidates): Clean up a gimple* variable that
28825 should be a gphi* variable.
28827 2017-03-20 Martin Sebor <msebor@redhat.com>
28830 * doc/extend.texi (attribute constructor): Document present limitation.
28832 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
28835 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
28836 __POWER9_VECTOR__ #ifdef control, change template definition to
28837 use Power9-specific built-in function.
28838 (vec_any_eq): Likewise.
28839 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
28840 to control outcomes from this test.
28841 (vector_ae_<mode>p): For VEC_F modes, likewise.
28843 2017-03-20 Ian Lance Taylor <iant@google.com>
28845 * config/i386/i386.c (ix86_function_regparm): Save an extra
28846 register for -fsplit-stack with DECL_STATIC_CHAIN.
28848 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28851 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
28852 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
28854 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28856 * config/riscv/riscv.c (riscv_print_operand): Use "fence
28858 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
28861 2017-03-20 Marek Polacek <polacek@redhat.com>
28864 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
28866 2017-03-20 Richard Biener <rguenther@suse.de>
28868 PR tree-optimization/80113
28869 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
28870 allocate extra SSA name for PHI def.
28871 (add_close_phis_to_outer_loops): Likewise.
28872 (add_close_phis_to_merge_points): Likewise.
28873 (copy_loop_close_phi_args): Likewise.
28874 (copy_cond_phi_nodes): Likewise.
28876 2017-03-20 Martin Liska <mliska@suse.cz>
28878 PR middle-end/79753
28879 * tree-chkp.c (chkp_build_returned_bound): Do not build
28880 returned bounds for a LHS that's not a BOUNDED_P type.
28882 2017-03-20 Martin Liska <mliska@suse.cz>
28886 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
28887 COMPLEX_CST and VECTOR_CST.
28889 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28892 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
28893 target operand. A new splitter adds the clobber statement in case
28894 the target operand is dead anyway.
28896 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
28898 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
28899 to age-old versions of binutils and glibc.
28901 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
28903 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
28905 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28907 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
28909 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28911 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
28912 requirement for binutils 2.13.
28914 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28916 * combine.c (try_combine): Delete redundant i1 test. Call
28917 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28919 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
28921 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
28922 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
28924 <riscv64-*-elf>: Re-arrange section
28925 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
28926 <riscv32-*-linux>: Likewise.
28927 <riscv64-*-elf>: Likewise
28928 <riscv64-*-linux>: Likewise.
28930 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
28933 * aarch64.opt(verbose-cost-dump): Fix typo.
28935 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
28938 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
28939 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
28941 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28943 * reload.c (find_reloads): When reloading a nonoffsettable address,
28944 use RELOAD_OTHER for it and its address reloads.
28946 PR rtl-optimization/79910
28947 * combine.c (record_used_regs): New static function.
28948 (try_combine): Handle situations where there is an additional
28949 instruction between I2 and I3 which needs to have a LOG_LINK
28952 2017-03-17 Jeff Law <law@redhat.com>
28954 PR tree-optimization/71437
28955 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
28956 conditional in the hash table first.
28957 (vrp_dom_walker::before_dom_children): Extract condition from
28958 ASSERT_EXPR. Record condition, its inverion and any implied
28959 conditions as well.
28961 2017-03-17 Marek Polacek <polacek@redhat.com>
28962 Markus Trippelsdorf <markus@trippelsdorf.de>
28964 PR tree-optimization/80079
28965 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
28968 2017-03-17 Richard Biener <rguenther@suse.de>
28970 PR middle-end/80075
28971 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
28972 Properly verify the LHS before the RHS possibly claims to be
28974 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
28977 2017-03-17 Martin Jambor <mjambor@suse.cz>
28979 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
28980 (List of -O2 options): Likewise.
28981 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
28984 2017-03-17 Tom de Vries <tom@codesourcery.com>
28986 * gcov-dump.c (print_usage): Print bug_report_url.
28988 2017-03-17 Richard Biener <rguenther@suse.de>
28990 PR middle-end/80050
28991 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
28992 (parser::peek): Likewise.
28994 2017-03-17 Richard Biener <rguenther@suse.de>
28996 PR tree-optimization/80048
28997 * sese.c (free_sese_info): Properly release rename_map and
28998 copied_bb_map elements.
29000 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29002 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29003 Add linked-list forward and backlinks. Insert on
29004 construction, remove on destruction.
29005 (class pass_store_merging): Add m_stores_head field.
29006 (pass_store_merging::terminate_and_process_all_chains):
29007 Iterate over m_stores_head list.
29008 (pass_store_merging::terminate_all_aliasing_chains):
29010 (pass_store_merging::execute): Check for debug stmts first.
29011 Push new chains onto the m_stores_head stack.
29013 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29016 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29017 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29018 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29020 2017-03-16 Jeff Law <law@redhat.com>
29022 PR tree-optimization/71437
29023 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29024 member function. Implementation moved into after_dom_children
29025 member function and into the threader's thread_outgoing_edges
29027 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29028 some code into new thread_outgoing_edges.
29029 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29030 definition. Simplify marker handling (do it here). Assume we always
29031 have the available expression and the const/copies tables.
29032 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29034 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29035 * tree-vrp.c (equiv_stack): No longer file scoped.
29036 (vrp_dom_walker): New class.
29037 (vrp_dom_walker::before_dom_children): New member function.
29038 (vrp_dom_walker::after_dom_children): Likewise.
29039 (identify_jump_threads): Setup domwalker. Use it rather than
29040 walking edges in a random order by hand. Simplify setup/finalization.
29041 (finalize_jump_threads): Remove.
29042 (vrp_finalize): Do not call identify_jump_threads here.
29043 (execute_vrp): Do it here instead and call thread_through_all_blocks
29046 PR tree-optimization/71437
29047 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29049 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29051 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29052 (dom_opt_dom_walker::thread_across_edge): Remove
29053 handle_dominating_asserts argument. All callers changed.
29054 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29055 changes. Remove calls to lhs_of_dominating_assert. Other
29056 uses of handle_dominating_asserts turn into unconditional code
29057 (simplify_control_stmt_condition_1): Likewise.
29058 (simplify_control_stmt_condition): Likewise.
29059 (thread_through_normal_block, thread_across_edge): Likewise.
29060 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29061 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29062 object if it is not an SSA_NAME.
29063 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29064 before calling into the VRP specific simplifiers.
29065 (identify_jump_threads): Remove handle_dominating_asserts
29068 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29071 * tree-diagnostic.c (default_tree_printer): No longer static.
29072 * tree-diagnostic.h (default_tree_printer): New prototype.
29074 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29076 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29077 Change ins into fmov.
29079 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29081 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29082 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29083 Use h_con constraint for operand 1.
29084 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29085 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29087 2017-03-15 Jeff Law <law@redhat.com>
29089 PR tree-optimization/71437
29090 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29091 (record_temporary_equivalences): Use it.
29093 PR tree-optimization/71437
29094 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29095 tree-ssa-scopedtables.
29096 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29097 (record_conditions, record_cond, vuse_eq): Likewise.
29098 (record_edge_info): Adjust to API tweak of record_conditions.
29099 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29100 (record_temporary_equivalences, optimize_stmt): Likewise.
29101 (eliminate_redundant_computations): Likewise.
29102 (record_equivalences_from_stmt): Likewise.
29103 * tree-ssa-scopedtables.c: Include options.h and params.h.
29104 (vuse_eq): New function, moved from tree-ssa-dom.c
29105 (build_and_record_new_cond): Likewise.
29106 (record_conditions): Likewise. Accept vector of conditions rather
29107 than edge_equivalence structure for first argument.
29108 for the first argument.
29109 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29110 from tree-ssa-dom.c.
29111 (avail_exprs_stack::record_cond): Likewise.
29112 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29113 from tree-ssa-dom.c.
29114 (avail_exprs_stack): Add new member functions lookup_avail_expr
29116 (record_conditions): Declare.
29118 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29121 * lra-constraints.c (process_alt_operands): Increase reject for
29122 reloading an input/output operand.
29124 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29127 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29128 insns to convert from signed/unsigned char/short to IEEE 128-bit
29130 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29132 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29135 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29136 subreg of inner mode for values already in registers.
29138 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29140 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29141 iteration reg is used after the loop.
29143 2017-03-14 Martin Sebor <msebor@redhat.com>
29145 PR tree-optimization/79800
29146 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29147 precision in negative-positive range.
29148 (format_floating): Call non-const overload with adjusted precision.
29150 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29153 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29156 2017-03-14 Martin Sebor <msebor@redhat.com>
29158 PR middle-end/80020
29159 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29160 * builtins.def (aligned_alloc): Use it.
29163 * Makefile.in (GTFILES): Add calls.c.
29164 * calls.c: Include "gt-calls.h".
29166 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29168 PR rtl-optimization/79728
29169 * regs.h (struct target_regs): New field
29170 x_contains_allocatable_regs_of_mode.
29171 (contains_allocatable_regs_of_mode): New macro.
29172 * reginfo.c (init_reg_sets_1): Initialize it, and change
29173 contains_reg_of_mode so it includes global regs as well.
29174 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29175 rather than contains_regs_of_mode.
29177 2017-03-14 Martin Liska <mliska@suse.cz>
29179 * doc/invoke.texi: Document options that can't be combined with
29180 -fcheck-pointer-bounds.
29182 2017-03-14 Martin Liska <mliska@suse.cz>
29184 PR middle-end/79831
29185 * doc/invoke.texi (-Wchkp): Document the option.
29187 2017-03-14 Martin Liska <mliska@suse.cz>
29189 * Makefile.in: Install gcov-dump.
29191 2017-03-14 Martin Liska <mliska@suse.cz>
29193 * multiple_target.c (expand_target_clones): Bail out for
29194 an invalid attribute.
29196 2017-03-14 Richard Biener <rguenther@suse.de>
29198 * alias.c (struct alias_set_entry): Pack properly.
29199 * cfgloop.h (struct loop): Likewise.
29200 * cse.c (struct set): Likewise.
29201 * ipa-utils.c (struct searchc_env): Likewise.
29202 * loop-invariant.c (struct invariant): Likewise.
29203 * lra-remat.c (struct cand): Likewise.
29204 * recog.c (struct change_t): Likewise.
29205 * rtl.h (struct address_info): Likewise.
29206 * symbol-summary.h (function_summary): Likewise.
29207 * tree-loop-distribution.c (struct partition): Likewise.
29208 * tree-object-size.c (struct object_size_info): Likewise.
29209 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29210 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29211 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29212 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29213 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29214 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29215 (struct _stmt_vec_info): Likewise.
29217 2017-03-14 Martin Liska <mliska@suse.cz>
29220 * multiple_target.c (create_dispatcher_calls): Check that
29221 a target can create a function dispatcher.
29223 2017-03-14 Martin Liska <mliska@suse.cz>
29226 * multiple_target.c (expand_target_clones): Drop local.local
29227 flag for default implementation.
29229 2017-03-14 Richard Biener <rguenther@suse.de>
29231 PR tree-optimization/80030
29232 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29234 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29236 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29237 gcc_fallthrough() instead of __attribute__((fallthrough));
29239 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29241 * doc/gcc.texi: Remove "up" link to (DIR).
29242 * doc/gccint.texi: Ditto.
29244 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29246 * doc/install.texi (Specific) <avr>: Remove reference to
29249 2017-03-13 Jeff Law <law@redhat.com>
29251 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29252 attribute rather than comments.
29254 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29255 match_scratch operand is highest.
29257 2017-03-13 Martin Liska <mliska@suse.cz>
29259 PR middle-end/78339
29260 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29261 is a CHKP clone, use original declaration.
29263 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29265 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29266 (arc_conditional_register_usage): Use a different allocation order
29267 when optimizing for size.
29268 * common/config/arc/arc-common.c (arc_option_optimization_table):
29269 Section anchors default on when optimizing for size.
29271 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29273 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29275 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29277 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29278 * config/arc/arc.md (cpu_facility): Add cd variant.
29279 (*movqi_insn): Add code density variant.
29280 (*movhi_insn): Likewise.
29281 (*movqi_insn): Likewise.
29282 (*addsi3_mixed): Likewise.
29283 (subsi3_insn): Likewise.
29285 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29287 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29289 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29291 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29292 expressions with MINUS and UNARY ops.
29294 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29297 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29299 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29300 between vec_select and vector argument.
29301 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29302 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29303 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29304 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29305 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29306 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29308 2017-03-13 Richard Biener <rguenther@suse.de>
29311 * params.def (vect-max-peeling-for-alignment): Fix typo.
29313 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29315 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29316 issue that only occurred with binutils below 2.18.
29318 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29320 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29321 refer to binutils 2.11/2.12 minimum.
29323 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29325 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29326 ftp.kernel.org and simplify binutils requirement.
29328 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29330 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29332 (Optimize Options): Ditto. Also remove redundancy.
29334 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29336 PR translation/79848
29337 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29339 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29340 to G_ to avoid double translation.
29342 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29344 PR translation/79923
29345 * auto-profile.c (get_combined_location): Convert leading
29346 character of diagnostics to lower case and remove trailing period.
29347 (read_profile): Likewise for various diagnostics.
29348 * config/arm/arm.c (arm_option_override): Remove trailing period
29349 from various diagnostics.
29350 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
29351 (msp430_expand_delay_cycles): Likewise.
29353 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29356 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
29357 full command-line argument, rather than just "str".
29358 (aarch64_validate_march): Likewise.
29359 (aarch64_validate_mtune): Likewise.
29361 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
29363 PR rtl-optimization/78911
29364 * lra-assigns.c (must_not_spill_p): New function.
29365 (spill_for): Use it.
29367 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29369 PR tree-optimization/79981
29370 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
29371 ATOMIC_COMPARE_EXCHANGE ifn result.
29372 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
29373 IFN_ATOMIC_COMPARE_EXCHANGE.
29375 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29378 * opts.c (parse_sanitizer_options): Add missing question mark to
29379 "did you mean" message.
29381 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29383 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
29385 (VMULEUH_UNS): Likewise.
29386 (VMULOUB_UNS): Likewise.
29387 (VMULOUH_UNS): Likewise.
29388 * config/rs6000/rs6000.c (builtin_function_type): Remove
29389 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
29391 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29394 * read-rtl-function.c (function_reader::read_rtx_operand): Update
29395 x with result of extra_parsing_for_operand_code_0.
29396 (function_reader::extra_parsing_for_operand_code_0): Convert
29397 return type from void to rtx, returning x. When reading
29398 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
29399 larger size containing struct block_symbol.
29401 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
29403 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
29404 -mfloat128-hardware without -m64.
29406 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
29409 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
29410 entries to the case statement that marks unsigned arguments to
29411 overloaded functions.
29413 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29415 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
29416 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
29418 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
29421 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
29422 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
29424 2017-03-10 Martin Liska <mliska@suse.cz>
29428 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
29429 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
29432 2017-03-10 Olivier Hainque <hainque@adacore.com>
29434 * tree-switch-conversion (array_value_type): Start by resetting
29435 candidate type to it's main variant.
29437 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29439 PR rtl-optimization/79909
29440 * combine.c (try_combine): Use simplify_replace_rtx on individual
29441 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
29442 of the whole CALL_INSN_FUNCTION_USAGE.
29444 PR tree-optimization/79972
29445 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
29446 get_range_info on SSA_NAMEs. Formatting fixes.
29448 2017-03-10 Richard Biener <rguenther@suse.de>
29449 Jakub Jelinek <jakub@redhat.com>
29451 PR tree-optimization/77975
29452 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
29453 edge to be constant.
29454 (get_val_for): For constant x return it. Formatting fix.
29455 (loop_niter_by_eval): Avoid pointless looping if the next iteration
29456 would use the same bases as the current one.
29458 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29460 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
29461 instead of vec_select for V1TImode.
29462 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
29464 (VSX_LE_128): Add V1TI to this mode iterator.
29465 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
29466 (*vsx_le_perm_store_<mode>): Likewise.
29467 (pre-reload splitter for VSX stores): Likewise.
29468 (post-reload splitter for VSX stores): Likewise.
29469 (*vsx_xxpermdi2_le_<mode>): Likewise.
29470 (*vsx_lxvd2x2_le_<mode>): Likewise.
29471 (*vsx_stxvd2x2_le_<mode>): Likewise.
29473 2017-03-09 Michael Eager <eager@eagercon.com>
29475 Correct failures with --enable-checking=yes,rtl.
29477 * config/microblaze/microblaze.c (microblaze_expand_shift):
29478 Replace GET_CODE test with CONST_INT_P and INTVAL test with
29479 test for const0_rtx.
29480 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
29481 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
29483 2017-03-09 Richard Biener <rguenther@suse.de>
29485 PR tree-optimization/79977
29486 * graphite-scop-detection.c (scop_detection::merge_sese):
29487 Handle the case of extra exits to blocks dominating the entry.
29489 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
29491 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29494 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
29496 PR rtl-optimization/79949
29497 * lra-constraints.c (process_alt_operands): Check memory when
29498 trying to predict a cycle. Print about the overall increase.
29500 2017-03-09 Richard Biener <rguenther@suse.de>
29502 PR middle-end/79971
29503 * gimple-expr.c (useless_type_conversion_p): Preserve
29504 TYPE_SATURATING for fixed-point types.
29506 2017-03-09 Richard Biener <rguenther@suse.de>
29509 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
29510 alignment of BLKmode params.
29512 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29515 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
29516 (VALL_NO_V2Q): Likewise.
29518 * config/aarch64/aarch64-simd.md
29519 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
29521 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
29522 VALL_NO_V2Q mode iterator.
29523 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
29525 2017-03-09 Martin Liska <mliska@suse.cz>
29527 PR tree-optimization/79631
29528 * tree-chkp-opt.c (chkp_is_constant_addr): Call
29529 tree_int_cst_sign_bit just for INTEGER constants.
29531 2017-03-09 Martin Liska <mliska@suse.cz>
29535 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
29538 2017-03-09 Marek Polacek <polacek@redhat.com>
29541 * tree.c (inchash::add_expr): Handle TREE_VEC.
29543 2017-03-09 Martin Liska <mliska@suse.cz>
29546 (chkp_narrow_size_and_offset): New function.
29547 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
29548 (void chkp_parse_bit_field_ref): New function.
29549 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
29550 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
29552 2017-03-09 Martin Liska <mliska@suse.cz>
29555 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
29556 (chkp_find_bounds_1): Remove gcc_unreachable.
29558 2017-03-09 Jakub Jelinek <jakub@redhat.com>
29561 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
29562 BUILT_IN_SYNC*, determine the access type from the size suffix and
29563 always build a MEM_REF with that type. Handle forgotten
29564 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
29567 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
29568 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
29569 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
29570 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
29571 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
29572 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
29573 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
29574 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
29575 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
29576 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
29577 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
29578 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
29579 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
29580 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
29581 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
29582 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
29583 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
29584 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
29585 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
29586 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
29587 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
29588 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
29589 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
29590 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
29591 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
29592 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
29593 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
29594 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
29595 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
29596 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
29597 definitions outside of __OPTIMIZE__ guarded section.
29600 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
29601 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
29602 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
29603 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
29606 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29608 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
29609 ("vfenez<mode>"): Add missing constraints.
29611 2017-03-08 Martin Sebor <msebor@redhat.com>
29614 * config/nds32/nds32.c (nds32_option_override):
29615 Fix misspelled diagnostic.
29617 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29620 * gimplify.c (gimplify_omp_for): Replace index var in outer
29621 taskloop statement with an artificial variable and add
29622 OMP_CLAUSE_PRIVATE clause for it.
29624 2017-03-08 Richard Biener <rguenther@suse.de>
29626 PR tree-optimization/79955
29627 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
29628 for accesses that are completely outside of the variable.
29630 2017-03-08 Andrew Haley <aph@redhat.com>
29632 PR tree-optimization/79943
29633 * tree-ssa-loop-split.c (compute_new_first_bound): When
29634 calculating the new upper bound, (END-BEG) should be added, not
29637 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29639 * config/avr/avr.md (setmemhi): Make sure match_dup
29640 operand number comes before match_scratch.
29642 2017-03-08 Richard Biener <rguenther@suse.de>
29644 PR tree-optimization/79920
29645 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
29646 with ncopies == 1 to ...
29647 (vect_transform_slp_perm_load): ... here. Properly compute
29648 all element loads by iterating VF times over the group. Do
29649 not handle ncopies (computed in a broken way) in
29650 vect_create_mask_and_perm.
29652 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29655 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
29656 is a uniform vector, use uniform_vector_p return value instead of
29657 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
29659 2017-03-07 Marek Polacek <polacek@redhat.com>
29661 PR middle-end/79809
29662 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
29663 (alloca_call_type): Likewise.
29665 2017-03-07 Martin Liska <mliska@suse.cz>
29667 * gcov.c (process_args): Put comment to correct location.
29669 2017-03-07 Martin Liska <mliska@suse.cz>
29671 PR middle-end/68270
29672 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
29673 Use array_at_struct_end_p instead of DECL_CHAIN (field).
29674 (chkp_narrow_bounds_for_field): Likewise.
29675 (chkp_parse_array_and_component_ref): Pass one more argument to
29678 2017-03-07 Richard Biener <rguenther@suse.de>
29680 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
29683 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
29685 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
29686 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
29688 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29691 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
29692 to end of description.
29693 (PARAM_MAX_STORES_TO_MERGE): Likewise.
29695 2017-03-07 Jakub Jelinek <jakub@redhat.com>
29697 PR rtl-optimization/79901
29698 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
29700 (*avx512f_<code><mode>3<mask_name>): ... this.
29701 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
29702 iterator instead of VI8_AVX2_AVX512BW.
29704 PR rtl-optimization/79901
29705 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
29706 min/max expander, expand it using expand_vec_cond_expr.
29709 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
29712 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29715 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
29716 to void * for PCH reasons.
29717 * dwarf2out.c (output_loc_operands, output_die): Cast
29718 v.val_vec.array to unsigned char *.
29720 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
29723 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
29726 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
29728 PR rtl-optimization/79571
29729 * lra-constraints.c (process_alt_operands): Calculate static
29730 reject and subtract it from overall when only addresses will be
29733 2017-03-06 Julia Koval <julia.koval@intel.com>
29736 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
29737 incoming stack boundary to 128 for 64-bit targets.
29739 2017-03-06 Richard Biener <rguenther@suse.de>
29741 PR tree-optimization/79894
29742 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
29743 to NULL after folding it.
29745 2017-03-06 Richard Biener <rguenther@suse.de>
29747 PR tree-optimization/79824
29748 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
29749 check disabling peeling for gaps.
29751 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
29753 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
29754 attributes): Document gettimeofday.
29756 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29758 * config/s390/s390.c (s390_option_override_internal): Set
29759 PARAM_MIN_VECT_LOOP_BOUND
29761 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29763 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
29764 * config/s390/s390.md: Likewise.
29766 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29769 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
29770 (<avx2_avx512>_perm<mode>): Rename to ...
29771 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
29773 (<avx512>_perm<mode>_mask): Rename to ...
29774 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
29776 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
29777 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
29778 instead of VI8F_256_512.
29779 (avx512f_perm<mode>): New define_expand.
29780 (avx512f_perm<mode>_mask): Likewise.
29781 (avx512f_perm<mode>_1<mask_name>): New define_insn.
29782 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
29784 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29786 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
29787 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
29789 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
29791 2017-03-06 Martin Liska <mliska@suse.cz>
29794 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
29795 when having a SSA NAME w/o VAR_DECL assigned to it.
29797 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29799 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
29800 msa_dpsub_<su>_d): Fix MODE for vec_select.
29802 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29804 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
29806 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
29808 2017-03-06 Richard Biener <rguenther@suse.de>
29810 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
29811 * plugin.c (register_plugin_info): Likewise.
29812 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
29814 2017-03-05 Jakub Jelinek <jakub@redhat.com>
29816 * config/i386/sse.md (sse_storehps, sse_storelps,
29817 avx_<castmode><avxsizesuffix>_<castmode>,
29818 avx512f_<castmode><avxsizesuffix>_<castmode>,
29819 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
29820 in condition that at least one operand is not a MEM.
29822 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29824 PR middle-end/79805
29825 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
29826 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
29828 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
29829 gimple_call_nothrow_p flag based on whether original builtin can throw.
29830 If it can, emit following stmts on the fallthrough edge.
29831 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
29832 don't create new bb if inserting just debug stmts on the edge, try to
29833 insert them on the fallthru bb or just reset debug stmts.
29835 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
29838 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
29839 restore recog_data (including the operand rtxes inside it) around
29840 the call to get_insn_template.
29842 2017-03-03 Martin Sebor <msebor@redhat.com>
29844 PR tree-optimization/79699
29845 * context.c (context::~context): Free MPFR caches to avoid
29846 a memory leak on program exit.
29848 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29850 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
29851 Use wide_int::ulow () instead of .elt (0).
29853 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29855 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
29856 (*pushxf): Limit oF constraint to 32bit targets and add oC
29857 constraint for 64bit targets.
29858 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
29859 (*pushdf): Change rmF constraint to rmC.
29861 2017-03-03 Martin Liska <mliska@suse.cz>
29863 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
29864 Remove unused variable.
29866 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29869 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
29870 is a memory operand, increase num_memory.
29871 (ix86_expand_args_builtin): Likewise.
29873 2017-03-03 Jan Hubicka <jh@suse.cz>
29876 * ipa-devirt.c (maybe_record_node): Properly handle
29877 __cxa_pure_virtual visibility.
29879 2017-03-03 Martin Liska <mliska@suse.cz>
29881 PR tree-optimization/79803
29882 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
29884 (pass_loop_prefetch::execute): Disabled optimization if an
29885 assumption about L1 cache size is not met.
29887 2017-03-03 Martin Liska <mliska@suse.cz>
29889 PR rtl-optimization/79574
29890 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
29891 (hash_scan_set): Likewise.
29892 (dump_hash_table): Likewise.
29893 (hoist_code): Likewise.
29895 2017-03-03 Richard Biener <rguenther@suse.de>
29897 * fixed-value.c (fixed_from_string): Restore use of elt (1)
29898 in place of uhigh ().
29899 (fixed_convert_from_real): Likewise.
29901 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29904 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
29906 2017-03-03 Richard Biener <rguenther@suse.de>
29908 PR middle-end/79818
29909 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
29910 TYPE_OVERFLOW_UNDEFINED check.
29912 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29914 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
29916 (vector_ae_<mode>_p): Likewise.
29917 (vector_nez_<mode>_p): Likewise.
29918 (vector_ne_v2di_p): Likewise.
29919 (vector_ae_v2di_p): Likewise.
29920 (vector_ne_<mode>_p): Likewise.
29921 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
29923 (vsx_tsqrt<mode>2_fe): Likewise.
29925 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
29928 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
29930 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29932 PR rtl-optimization/79780
29933 * cprop.c (one_cprop_pass): When second and further conditional trap
29934 in a single basic block is turned into an unconditional trap, turn it
29935 into a deleted note to avoid RTL verification failures.
29937 2017-03-02 Richard Biener <rguenther@suse.de>
29939 * fold-const.c (const_binop): Use ulow () instead of elt (0).
29941 2017-03-02 Richard Biener <rguenther@suse.de>
29943 PR tree-optimization/79345
29945 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
29946 param and abort the walk, returning -1 if it is hit.
29947 (walk_aliased_vdefs): Take a limit param and pass it on.
29948 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
29949 defaulting to 0 and return a signed int.
29950 * tree-ssa-uninit.c (struct check_defs_data): New struct.
29951 (check_defs): New helper.
29952 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
29953 about uninitialized memory.
29954 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
29955 bogus uninitialized warning.
29956 (fixed_convert_from_real): Likewise.
29958 2017-03-02 Bin Cheng <bin.cheng@arm.com>
29960 PR tree-optimization/66768
29961 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
29962 iv_use if base object can't be determined.
29964 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29966 PR tree-optimization/79345
29967 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
29968 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
29969 (get_pattern_stats): Initialize it.
29970 * genemit.c (gen_expand): Verify match_scratch numbers come after
29971 match_operand/match_dup numbers.
29972 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
29973 match_scratch numbers.
29974 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
29976 * config/s390/s390.md (trunctdsd2): Likewise.
29978 2017-03-02 Richard Biener <rguenther@suse.de>
29980 * wide-int.h (wide_int_storage::operator=): Implement in terms
29983 2017-03-02 Richard Biener <rguenther@suse.de>
29985 PR tree-optimization/79777
29986 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
29987 the to insert expression to sth existing.
29989 2017-03-01 Martin Sebor <msebor@redhat.com>
29991 PR middle-end/79692
29992 * gimple-ssa-sprintf.c
29993 (directive::known_width_and_precision): New function.
29994 (format_integer): Use it.
29995 (get_mpfr_format_length): Consider the full range of precision
29996 when computing %g output with the # flag. Set the likely byte
29997 count to 3 rather than 1 when precision is indeterminate.
29998 (format_floating): Correct the lower bound of precision.
30000 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30002 * doc/invoke.texi: Document default code model for 64-bit Linux.
30004 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30007 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30008 udiv rather than div since input pattern is unsigned.
30010 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30012 * config/i386/i386.c (print_reg): Warn for values of
30013 unsupported size in integer register.
30015 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30018 * config/rs6000/predicates.md (current_file_function_operand): Do
30019 not allow self calls to be local if the function is replaceable.
30021 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30024 * config/rs6000/altivec.h (vec_ctz and others): Change the
30025 preprocessor macro that controls conditional compilation from
30026 _ARCH_PWR9 to __POWER9_VECTOR__.
30027 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30028 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30029 control (instead of _ARCH_PWR9 control) so that template
30030 definition uses power9-specific function.
30031 (vec_any_eq): Likewise.
30032 (vec_all_ne): Change macro definition to use a power9-specific
30033 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30034 _ARCH_PWR9 control).
30035 (vec_any_eq) Likewise.
30036 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30037 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30038 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30039 support for xvcmpnedp instruction.
30040 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30041 macro expansion so that Power9 implementation of vec_all_ne does
30042 not use the AltiVec predicate framework.
30043 (VCMPNEH_P): Likewise.
30044 (VCMPNEW_P): Likewise.
30045 (VCMPNED_P): Likewise.
30046 (VCMPNEFP_P): Likewise.
30047 (VCMPNEDP_P): Likewise.
30048 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30049 implementation of vec_any_eq to not use AltiVec predicate
30051 (VCMPAEH_P): Likewise.
30052 (VCMPAEW_P): Likewise.
30053 (VCMPAED_P): Likewise.
30054 (VCMPAEFP_P): Likewise.
30055 (VCMPAEDP_P): Likewise.
30056 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30057 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30058 not use the AltiVec predicate framework.
30059 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30060 of vec_any_eq to not use AltiVec predicate framework.
30061 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30062 support for predefined __POWER9_VECTOR__ macro to indicate that
30063 Power9 instruction selection is enabled.
30064 (altivec_overloaded_builtins): Remove extraneous
30065 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30066 function argument types RS6000_BTI_bool_V16QI and
30067 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30068 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30069 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30070 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30071 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30072 Power9 for implementations of vec_cmpne. Change the signature for
30073 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30074 (representing vec_all_ne) to remove the previously described first
30075 argument of type RS6000_BTI_INTSI, as this was an artifact of
30076 reliance on the AltiVec predicate framework, which is no longer
30077 used in the implementation of these functions. Add
30078 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30079 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30080 since, unlike the AltiVec predicate framework implementation, we
30081 do not share function descriptors between vec_alle and vec_anyeq.
30082 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30083 set of modes that receive special treatment even when
30084 TARGET_P9_VECTOR is true. The special treatment emits code that
30085 does not depend on Power9 instructions.
30086 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30087 define_expand to not rely on AltiVec predicate framework.
30088 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30090 (vector_ne_v2di_p): Change this define_expand to not rely on
30091 AltiVec predicate framework.
30092 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30094 (vector_ne_<mode>_p): Change this define_expand to not rely on
30095 AltiVec predicate framework.
30096 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30098 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30099 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30100 define_insn pattern.
30101 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30102 define_insn pattern because the xvcmpne<VSs>. instruction is not
30104 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30105 instruction is not supported.
30107 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30109 * config/nvptx/nvptx.c: Include intl.h.
30111 2017-03-01 Martin Jambor <mjambor@suse.cz>
30114 * ipa-prop.h (ipa_bits): Removed field known.
30115 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30116 to pointers. Adjusted their comments to warn about their sharing.
30117 (ipcp_transformation_summary): Change bits to a vector of pointers.
30118 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30119 (ipa_get_ipa_bits_for_value): Declare.
30120 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30121 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30122 (ipa_bits_hash_table): Likewise.
30123 (ipa_vr_ggc_hash_traits): Likewise.
30124 (ipa_vr_hash_table): Likewise.
30125 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30126 being pointers and vr_known being removed.
30127 (ipa_set_jf_unknown): Likewise.
30128 (ipa_get_ipa_bits_for_value): New function.
30129 (ipa_set_jfunc_bits): Likewise.
30130 (ipa_get_value_range): New overloaded functions.
30131 (ipa_set_jfunc_vr): Likewise.
30132 (ipa_compute_jump_functions_for_edge): Use the above functions to
30133 construct bits and vr parts of jump functions.
30134 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30135 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30137 (ipcp_grow_transformations_if_necessary): Also allocate
30138 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30140 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30141 them. Fix too long lines.
30142 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30143 vr_known being removed.
30144 (ipa_read_jump_function): Use new setter functions to construct bits
30145 and vr parts of jump functions or set them to NULL.
30146 (write_ipcp_transformation_info): Adjust for bits being pointers.
30147 (read_ipcp_transformation_info): Likewise.
30148 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30150 Include gt-ipa-prop.h.
30151 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30153 (ipcp_store_bits_results): Likewise.
30154 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30155 Do not write to existing jump functions but use a temporary instead.
30157 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30160 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30161 attempt to use its first operand as BIT_FIELD_REF base.
30163 2017-03-01 Richard Biener <rguenther@suse.de>
30165 PR middle-end/79721
30166 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30167 interpolating formula in wrapping arithmetic.
30168 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30170 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30172 PR tree-optimization/79734
30173 * tree-vect-generic.c (expand_vector_condition): Optimize
30174 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30175 Handle VEC_COND_EXPR where comparison has different inner width from
30176 type's inner width.
30178 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30180 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30181 markup, and similar issues. Remove @opindex entries for things
30182 that aren't options. Add missing -mmpy-option entries.
30184 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30186 PR tree-optimization/79737
30187 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30188 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30189 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30190 instead of byte_size. Formatting fix.
30191 (shift_bytes_in_array_right): Formatting fix.
30193 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30196 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30197 condition on optimize for the leaf function test.
30199 2017-02-28 Martin Liska <mliska@suse.cz>
30202 * read-rtl-function.c (function_reader::handle_unknown_directive):
30203 Bail out when one uses -flto.
30205 2017-02-28 Martin Liska <mliska@suse.cz>
30207 * common.opt: Replace space with tabular for options of <number>
30209 * config/i386/i386.opt: Show <number> value for
30210 -mlarge-data-threshold.
30211 * opts.c (print_filtered_help): Do not display number in hexadecimal
30214 2017-02-28 Martin Liska <mliska@suse.cz>
30216 * common.opt: Fix --help=option -Q for options which are of
30219 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30221 * config/i386/i386.c (print_reg): Error out for values
30222 of 8-bit size in invalid integer register.
30224 2017-02-28 Martin Sebor <msebor@redhat.com>
30226 PR tree-optimization/79691
30227 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30229 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30232 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30233 gcc_unreachable with output_operand_lossage.
30235 2017-02-28 Richard Biener <rguenther@suse.de>
30237 PR tree-optimization/79740
30238 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30240 (visit_nary_op): Insert the nary into the hashtable if we
30241 pattern-matched sth.
30242 * tree-ssa-pre.c (eliminate_insert): Robustify.
30244 2017-02-28 Richard Biener <rguenther@suse.de>
30246 PR middle-end/79731
30247 * fold-const.c (decode_field_reference): Reject out-of-bound
30250 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30252 * config/i386/i386.c: Include intl.h.
30253 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30254 instead of just cond ? "..." : "...".
30255 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30256 * coverage.c (read_counts_file): Likewise.
30257 * omp-offload.c: Include intl.h.
30258 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30259 of just cond ? "..." : "...".
30260 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30261 of just cond ? "..." : "...".
30263 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30266 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30268 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30269 'tune for' CPU name.
30270 * config/arm/arm-cpu-data.h: Regenerated.
30272 2017-02-28 Richard Biener <rguenther@suse.de>
30274 PR tree-optimization/79732
30275 * tree-inline.c (expand_call_inline): Do not shadow var.
30277 2017-02-28 Richard Biener <rguenther@suse.de>
30279 PR tree-optimization/79723
30280 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30281 address-space properly.
30283 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30285 * doc/optinfo.texi (Optimization groups): Fix option used for
30287 * doc/invoke.texi (-fopt-info): Document "omp".
30288 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30289 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30290 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30291 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30292 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30294 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30296 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30297 OPTGROUP_OMP with "omp".
30299 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30302 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30303 for arithmetic shift of unsigned V2DI.
30305 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30307 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30308 arc/linux.h headers.
30309 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30310 (LINK_SPEC): Likewise.
30311 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30312 (EXTRA_SPECS): Likewise.
30313 (STARTFILE_SPEC): Likewise.
30314 (ENDFILE_SPEC): Likewise.
30315 (LIB_SPEC): Likewise.
30316 (TARGET_SDATA_DEFAULT): Likewise.
30317 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30318 (MULTILIB_DEFAULTS): Likewise.
30319 (DWARF2_UNWIND_INFO): Likewise.
30320 * config/arc/big.h: New file.
30321 * config/arc/elf.h: Likewise.
30322 * config/arc/linux.h: Likewise.
30323 * config/arc/t-uClibc: Remove.
30325 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30327 PR tree-optimization/77536
30328 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30329 (tree_transform_and_unroll_loop): Use above function to compute the
30330 estimated niter of unrolled loop and use it when scaling profile.
30331 Also use count info rather than frequency if it's non-zero.
30332 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30333 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30334 (vect_transform_loop): Call above function.
30336 2017-02-27 Richard Biener <rguenther@suse.de>
30338 PR tree-optimization/45397
30339 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30340 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30341 (visit_nary_op): Add pattern matching for CSEing sign-changed
30342 or truncated operations with wider ones.
30344 2017-02-27 Richard Biener <rguenther@suse.de>
30346 PR tree-optimization/79690
30347 * tree-vect-stmts.c (vectorizable_store): Use vector type
30348 built from the DR with address-space.
30350 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
30352 * doc/invoke.texi (Optimize Options): Refine the description
30353 of asan-use-after-return.
30355 2017-02-25 Alan Modra <amodra@gmail.com>
30357 PR rtl-optimization/79584
30358 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
30359 base, not ad->base_term, the reg within base. Remove assertion
30360 that ad->base == ad->base_term. Replace gen_int_mode using
30361 bogus mode with const0_rtx.
30363 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30365 PR middle-end/79396
30366 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30367 FMA_EXPR like tcc_binary or tcc_unary.
30369 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
30372 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
30374 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
30375 (output_loc_operands): Handle DW_OP_call_ref and
30376 DW_OP_GNU_variable_value.
30377 (struct variable_value_struct): New type.
30378 (struct variable_value_hasher): Likewise.
30379 (variable_value_hash): New variable.
30380 (string_types): Remove.
30381 (copy_loc_descr): New function.
30382 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
30383 (prepend_loc_descr_to_each): New function.
30384 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
30385 instead of add_loc_descr_to_each if the first argument is single
30386 location list and the second has multiple.
30387 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
30388 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
30389 when looking for variable value which doesn't have other location info.
30390 (loc_list_from_tree): Formatting fix.
30391 (gen_array_type_die): Simplify DW_AT_string_length handling.
30392 (adjust_string_types): Remove.
30393 (gen_subprogram_die): Don't call adjust_string_types nor test/set
30394 string_types. Call resolve_variable_values.
30395 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
30396 (resolve_addr_in_expr): Likewise. Add A argument.
30397 (copy_deref_exprloc): Remove deref argument. Adjust for the
30398 original expression being DW_OP_GNU_variable_value with optionally
30399 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
30400 optionally after it.
30401 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
30402 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
30403 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
30404 (variable_value_hasher::hash, variable_value_hasher::equal): New
30406 (resolve_variable_value_in_expr, resolve_variable_value,
30407 resolve_variable_values, note_variable_value_in_expr,
30408 note_variable_value): New functions.
30409 (dwarf2out_early_finish): Call note_variable_value on all toplevel
30412 2017-02-24 Jakub Jelinek <jakub@redhat.com>
30415 * opts.h (handle_generated_option): Add GENERATED_P argument.
30416 * opts-common.c (handle_option): Adjust function comment.
30417 (handle_generated_option): Add GENERATED_P argument, pass it to
30419 (control_warning_option): Pass false to handle_generated_option
30421 * opts.c (maybe_default_option): Pass true to handle_generated_option
30423 * optc-gen.awk: Likewise.
30425 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30427 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
30428 a REG, look at the REG it is a SUBREG of.
30429 (splitter for cmpeqsi_t): Ditto.
30431 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30433 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
30434 the special USEs with the pattern of the insn, not the insn itself.
30436 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
30439 * doc/invoke.texi: Document -mload-store-pairs.
30441 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30442 Sandra Loosemore <sandra@codesourcery.com>
30444 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
30445 argument isn't a CONST_INT.
30446 (nios2_alternate_compare_const): Assert op is a CONST_INT.
30447 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
30448 (nios2_validate_compare): Bypass alternate compare logic if *op2
30449 is not a CONST_INT.
30450 (ldstwm_operation_p): Return false if first_base is not a REG or
30451 if first_offset is not a CONST_INT.
30453 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30455 * config/cris/cris.md: Use correct operand in a define_peephole2.
30457 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30459 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
30461 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30463 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
30464 this_insn if it is an INSN or JUMP_INSN.
30465 (force_offsettable): Look at base, not at addr.
30466 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
30467 on things that aren't necessarily CONST_INTs.
30469 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
30471 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
30472 -mfpmath=sse is the default also for x86-32 targets with SSE2
30473 instruction set when @option{-ffast-math} is enabled
30475 2017-02-24 Jeff Law <law@redhat.com>
30477 PR rtl-optimizatoin/79286
30478 * ira.c (update_equiv_regs): Drop may_trap_p exception to
30481 2017-02-24 Richard Biener <rguenther@suse.de>
30483 PR tree-optimization/79389
30484 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
30487 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
30489 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
30490 function comment to reflect reality.
30491 (loop_exits_before_overflow): Fix typo in function description.
30493 2017-02-24 Richard Biener <rguenther@suse.de>
30495 PR tree-optimization/79389
30496 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
30497 properly that a threading opportunity exists. Detect conditional
30498 copy/constant propagation opportunities.
30500 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
30502 * config/visium/visium.md (type): Add trap.
30503 (b): New mode attribute.
30504 (*btst): Rename into...
30505 (*btst<mode>): ...this and adjust.
30506 (*cbranchsi4_btst_insn): Rename into...
30507 (*cbranch<mode>4_btst_insn): ...this and adjust.
30508 (trap): New define_insn.
30510 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30512 PR tree-optimization/79389
30513 * ifcvt.c (struct noce_if_info): Add rev_cond field.
30514 (noce_reversed_cond_code): New function.
30515 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
30516 reversed_comparison_code. Formatting fix.
30517 (noce_try_store_flag): Test rev_cond != NULL in addition to
30518 reversed_comparison_code.
30519 (noce_try_store_flag_constants): Likewise.
30520 (noce_try_store_flag_mask): Likewise.
30521 (noce_try_addcc): Use rev_cond if non-NULL instead of
30522 reversed_comparison_code.
30523 (noce_try_cmove_arith): Likewise. Formatting fixes.
30524 (noce_try_minmax, noce_try_abs): Clear rev_cond.
30525 (noce_find_if_block): Initialize rev_cond.
30526 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
30527 instead of false as last argument never attempt to reverse it
30530 2017-02-23 Bin Cheng <bin.cheng@arm.com>
30532 PR tree-optimization/79663
30533 * tree-predcom.c (combine_chains): Process refs in reverse order
30534 only for ZERO length chains, and add explaining comment.
30536 2017-02-23 Jeff Law <law@redhat.com>
30538 PR tree-optimization/79578
30539 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
30540 in call to operand_equal_p.
30542 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
30545 * config/i386/cpuid.h: Fix another undefined behavior.
30547 2017-02-23 Richard Biener <rguenther@suse.de>
30549 PR tree-optimization/79683
30550 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
30551 vector types for data-refs.
30553 2017-02-23 Martin Liska <mliska@suse.cz>
30555 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
30557 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30559 PR middle-end/79665
30560 * internal-fn.c (get_range_pos_neg): Moved to ...
30561 * tree.c (get_range_pos_neg): ... here. No longer static.
30562 * tree.h (get_range_pos_neg): New prototype.
30563 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
30564 are known to be in between 0 and signed maximum inclusive, try to
30565 expand both unsigned and signed divmod and use the cheaper one from
30568 2017-02-22 Jeff Law <law@redhat.com>
30570 PR tree-optimization/79578
30571 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
30572 to compare base operands.
30574 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
30577 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
30578 gpc_reg_operand instead of fpr_reg_operand.
30580 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
30582 * config/mips/mips.c (mips_return_in_memory): Force FP
30583 vector types to be returned in memory for o32 ABI.
30585 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30587 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
30588 instead of DW_TAG_member for static data member declarations and don't
30589 set no_linkage_name for static inline data members.
30590 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
30593 2017-02-22 Martin Liska <mliska@suse.cz>
30595 * doc/invoke.texi: Replace inequality signs with square brackets
30598 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
30601 * lra-constraints.c (simplify_operand_subreg): Handle
30602 WORD_REGISTER_OPERATIONS targets.
30604 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30607 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
30608 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
30609 elimination by swapping fld*.
30611 2017-02-22 Richard Biener <rguenther@suse.de>
30613 PR tree-optimization/79673
30614 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
30615 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
30616 irrelevant address-space qualifiers and avoiding a
30617 ADDR_SPACE_CONVERT_EXPR from fold_convert.
30619 2017-02-22 Richard Biener <rguenther@suse.de>
30621 PR tree-optimization/79666
30622 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
30623 to not symbolically negate if that may introduce undefined
30626 2017-02-22 Martin Liska <mliska@suse.cz>
30629 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
30630 * data-streamer-out.c (streamer_write_gcov_count_stream):
30632 * value-prof.c (stream_out_histogram_value): Make assert more
30633 precise based on type of counter.
30635 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
30638 * config/i386/i386.md (standard_x87sse_constant_load splitter):
30639 Use nonimmediate_operand instead of memory_operand for operand 1.
30640 (float-extend standard_x87sse_constant_load splitter): Ditto.
30642 2017-02-21 Jeff Law <law@redhat.com>
30644 PR tree-optimization/79621
30645 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
30646 blocks with edges to themselves.
30648 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30651 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
30652 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
30653 Use gimple_call_builtin_p.
30656 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
30657 on temporarily removed DEBUG_INSNs.
30659 PR tree-optimization/79649
30660 * tree-loop-distribution.c (classify_partition): Give up on
30661 non-generic address space loads/stores.
30663 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
30665 * doc/loop.texi (Loop manipulation): Remove nonexistent
30666 tree_ssa_loop_version from the documentation.
30667 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
30669 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30672 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
30673 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
30674 * config/rs6000/rs6000.c: Include except.h.
30675 (rs6000_expand_split_stack_prologue): Call
30676 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
30678 2017-02-21 Martin Jambor <mjambor@suse.cz>
30681 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
30682 have been analyzed.
30684 2017-02-21 Martin Jambor <mjambor@suse.cz>
30686 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
30687 for backward compatibility only.
30688 * doc/invoke.texi (Option Summary): Remove all references to
30689 -fipa-cp-alignment.
30691 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
30695 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30697 * lra-constraints.c (curr_insn_transform): Handle
30698 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30700 2017-02-21 Martin Liska <mliska@suse.cz>
30702 * config/i386/i386.opt: Replace -masm-dialect with -masm.
30704 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
30706 PR translation/79638
30707 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
30709 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
30712 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
30713 (arm_function_ok_for_sibcall): Return false for an indirect call by
30714 descriptor if all the argument registers are used.
30715 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
30716 alignment of the function.
30718 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30720 PR tree-optimization/61441
30721 * simplify-rtx.c (simplify_const_unary_operation): For
30722 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
30723 the sNaN unmodified.
30725 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30727 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30728 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30729 instead of SYSTEM_HEADER_DIR.
30731 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
30732 Martin Liška <mliska@suse.cz>
30734 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
30735 Fix typos and grammar, use active voice, and clarify.
30737 2017-02-20 Marek Polacek <polacek@redhat.com>
30739 PR middle-end/79537
30740 * gimplify.c (gimplify_expr): Handle unused *&&L;.
30743 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
30745 2017-02-20 Jakub Jelinek <jakub@redhat.com>
30748 * config/i386/i386.c (ix86_expand_builtin): Handle
30749 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
30750 ix86_builtins_isa[fcode].isa as a requirement of those
30751 flags and any other flag in the bitmask.
30752 (ix86_init_mmx_sse_builtins): Use 0 instead of
30753 ~OPTION_MASK_ISA_64BIT as mask.
30754 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
30755 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
30756 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
30757 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
30759 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30762 * lra-constraints.c (split_reg): Check requested split mode
30763 is supported by the register.
30765 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30767 * lra-constraints.c (simplify_operand_subreg): Remove early
30770 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30773 * lra-constraints.c (curr_insn_transform): Tighten condition
30774 for converting SUBREG reloads from OP_OUT to OP_INOUT.
30776 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30779 * lra-constraints.c (curr_insn_transform): Handle
30780 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30782 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
30785 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
30787 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
30789 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
30792 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
30795 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30797 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
30800 2017-02-18 Jakub Jelinek <jakub@redhat.com>
30802 * final.c (last_columnnum, override_columnnum): New variables.
30803 (final_start_function): Set last_columnnum, pass it to begin_prologue
30804 hook and pass 0 to dwarf2out_begin_prologue.
30805 (final_scan_insn): Update override_columnnum. Pass last_columnnum
30806 to source_line debug hook.
30807 (notice_source_line): Compute last_columnnum and for debug_column_info
30808 return true on column changes.
30809 * debug.h (struct gcc_debug_hooks): Add column argument to
30810 source_line and begin_prologue hooks.
30811 (debug_nothing_int_charstar_int_bool): Remove prototype.
30812 (debug_nothing_int_int_charstar,
30813 debug_nothing_int_int_charstar_int_bool): New prototypes.
30814 (dwarf2out_begin_prologue): Add column argument.
30815 * debug.c (do_nothing_debug_hooks): Adjust source_line and
30816 begin_prologue hooks.
30817 (debug_nothing_int_charstar_int_bool): Remove.
30818 (debug_nothing_int_int_charstar,
30819 debug_nothing_int_int_charstar_int_bool): New functions.
30820 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
30821 through to dwarf2out_source_line.
30822 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
30823 (dwarf2out_source_line): Add column argument, emit it if requested.
30824 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
30826 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30827 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30828 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
30829 through to dwarf2out_begin_prologue.
30830 (vmsdbgout_source_line): Add column argument, pass it through to
30831 dwarf2out_source_line.
30832 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
30833 dbxout_source_line caller.
30834 (dbxout_source_line): Add column argument.
30836 * common.opt (gno-column-info, gcolumn-info): New options.
30837 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
30838 (check_die): Also test for multiple DW_AT_decl_column attributes.
30839 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
30840 DW_AT_decl_column if requested.
30841 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
30843 (gen_variable_die): Likewise.
30844 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
30845 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
30848 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
30849 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
30850 (ix86_handle_option): Handle OPT_m3dnowa.
30851 * doc/invoke.texi (-m3dnowa): Document.
30852 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
30853 -m3dnowa instead of -m3dnow -march=athlon.
30856 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
30857 instead of gcc_assert for K, r and R code checks. Formatting fixes.
30859 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30862 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
30863 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
30864 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
30865 generator for vsx_xxpermdi_<mode>_be.
30866 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
30867 force big-endian semantics.
30868 (vsx_xxpermdi_<mode>_be): New define_expand with same
30869 implementation as previous version of vsx_xxpermdi_<mode>.
30871 2017-02-17 Jakub Jelinek <jakub@redhat.com>
30873 PR tree-optimization/79327
30874 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
30875 variable, its initialization and use.
30877 2017-02-17 Julia Koval <julia.koval@intel.com>
30879 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
30880 (OPTION_MASK_ISA_PKU_UNSET): New.
30881 (ix86_handle_option): Handle -mrdpid.
30882 * config/i386/cpuid.h (bit_RDPID): New.
30883 * config/i386/driver-i386.c (host_detect_local_cpu):
30884 Detect RDPID feature.
30885 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
30886 * config/i386/i386-c.c (ix86_target_macros_internal):
30888 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
30889 (ix86_valid_target_attribute_inner_p): Add "rdpid".
30890 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
30891 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
30892 * config/i386/i386.md (define_insn "rdpid"): New.
30893 * config/i386/i386.opt Add -mrdpid.
30894 * config/i386/immintrin.h (_rdpid_u32): New.
30896 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
30898 PR rtl-optimization/79541
30899 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
30900 instead of transforming it into USE.
30902 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
30904 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
30905 If HONOR_SNANS (SFmode) force the input to a register.
30906 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
30907 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
30908 an frsp or similar insn.
30910 2017-02-17 Martin Liska <mliska@suse.cz>
30912 PR rtl-optimization/79577
30913 * params.def (selsched-max-sched-times): Increase minimum to 1.
30915 2017-02-17 Martin Liska <mliska@suse.cz>
30917 PR rtl-optimization/79574
30918 * gcse.c (want_to_gcse_p): Prevent integer overflow.
30920 2017-02-17 Martin Liska <mliska@suse.cz>
30922 PR tree-optimization/79529
30923 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
30924 ssa_defined_default_def_p to handle cases which are implicitly
30926 * tree-ssa.c (ssa_defined_default_def_p): New function.
30927 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
30928 which are implicitly defined.
30929 * tree-ssa.h (ssa_defined_default_def_p): Declare.
30931 2017-02-17 Richard Biener <rguenther@suse.de>
30933 PR middle-end/79576
30934 * params.def (max-ssa-name-query-depth): Limit to 10.
30936 2017-02-17 Richard Biener <rguenther@suse.de>
30938 PR tree-optimization/79552
30939 * tree-ssa-structalias.c (visit_loadstore): Properly verify
30942 2017-02-17 Richard Biener <rguenther@suse.de>
30945 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
30947 2017-02-17 Marek Polacek <polacek@redhat.com>
30949 PR middle-end/79536
30950 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
30951 (fold_negate_expr): New wrapper.
30953 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
30955 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
30956 Correct terminology and de-emphasize pre-standard behavior.
30958 2017-02-16 Alan Modra <amodra@gmail.com>
30960 PR rtl-optimization/79286
30961 * ira.c (def_dominates_uses): New function.
30962 (update_equiv_regs): Don't create an equivalence for insns that
30963 may trap where the register def does not dominate the use.
30965 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
30967 PR rtl-optimization/78127
30968 * lra.c (lra): Call lra_eliminate before finish the loop after
30971 2017-02-16 Richard Biener <rguenther@suse.de>
30973 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
30975 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
30976 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
30977 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
30978 (isl_val_int_from_wi): New function.
30979 (extract_affine_gmp): Rename to ...
30980 (extract_affine_wi): ... this, take a widest_int.
30981 (extract_affine_int): Just wrap extract_affine_wi.
30982 (add_param_constraints): Use isl_val_int_from_wi.
30983 (add_loop_constraints): Likewise, and extract_affine_wi.
30985 2017-02-15 Jeff Law <law@redhat.com>
30987 PR middle-end/79521
30988 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
30989 ira_init_register_move_cost_if_necessary.
30991 2017-02-15 Martin Sebor <msebor@redhat.com>
30993 PR middle-end/32003
30994 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
30995 removed in a prior commit.
30997 2017-02-15 Bin Cheng <bin.cheng@arm.com>
30999 PR tree-optimization/79347
31000 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31001 counters during peeling.
31003 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31005 * Makefile.in (site.exp): Remove "set ISLVER".
31007 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31010 * real.c (real_from_integer): Call real_convert even for decimal.
31012 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31015 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31017 2017-02-14 Andrew Pinski <apinski@cavium.com>
31019 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31020 cores and change the partno/implementer to be correct.
31021 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31022 the 'B" as the implementer.
31023 * config/aarch64/aarch64-tune.md: Regenerate.
31025 2017-02-14 Carl Love <cel@us.ibm.com>
31027 * config/rs6000/rs6000.c: Add case statement entry to make the
31028 xvcvuxdsp built-in argument unsigned.
31029 * config/rs6000/vsx.md: Fix the source and return operand types so they
31030 match the instruction definitions from the ISA document. Fix typo
31031 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31034 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31037 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31038 member early_clobber_alts.
31039 * lra-lives.c (reg_early_clobber_p): New.
31040 (process_bb_lives): Use it.
31041 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31042 (debug_operand_data): Initialize early_clobber_alts.
31043 (setup_operand_alternative): Set up early_clobber_alts.
31044 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31045 alternatives to new_insn_reg.
31046 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31048 (lra_update_insn_regno_info): Pass the new arg.
31050 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31052 PR middle-end/79505
31053 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31054 (new_oacc_loop_raw): Don't clear already cleared fields.
31057 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31058 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31059 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31061 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31064 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31066 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31069 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31070 the extra instruction to the right place to store 128-bit constant
31073 2017-02-14 Martin Sebor <msebor@redhat.com>
31075 PR middle-end/79448
31076 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31077 warning for strings of unknown length.
31079 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31081 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31083 2017-02-14 Jeff Law <law@redhat.com>
31086 * ira-costs.c (scan_one_insn): Initialize register move costs
31087 for pseudos seen in USE/CLOBBER insns.
31089 PR tree-optimization/79095
31090 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31091 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31092 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31093 if the operands are known to be not equal, then the resulting range
31095 (intersect_ranges): If the new range is ~[0,0] and the old range is
31096 wide, then prefer ~[0,0].
31097 * tree-vrp.c (overflow_comparison_p_1): New function.
31098 (overflow_comparison_p): New function.
31099 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31100 if NAME is used in an overflow test.
31101 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31102 overflow check that can be expressed as an equality test, then adjust
31103 ops to be that equality test.
31105 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31107 * config/s390/s390-builtin-types.def: Remove flags argument.
31108 * config/s390/s390.c (s390_init_builtins): Likewise.
31110 2017-02-14 Martin Liska <mliska@suse.cz>
31112 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31113 vector. Fix trailing white spaces.
31115 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31117 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31120 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31122 PR rtl-optimization/68664
31123 * config/arm/arm.c (arm_sched_can_speculate_insn):
31124 New function. Declare prototype.
31125 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31127 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31129 PR rtl-optimization/68664
31130 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31132 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31134 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31136 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31137 max skip bytes for function, loop and jump.
31139 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31141 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31142 ABS_EXPR for gimple dump.
31144 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31147 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31149 PR tree-optimization/79408
31150 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31151 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31152 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31153 also if rhs1 is INTEGER_CST.
31155 2017-02-14 Richard Biener <rguenther@suse.de>
31157 PR middle-end/79432
31158 * tree-into-ssa.c (insert_phi_nodes): When the function can
31159 have abnormal edges rewrite SSA names with broken use-def
31160 dominance out of SSA and register them for PHI insertion.
31162 2017-02-13 Martin Sebor <msebor@redhat.com>
31164 PR middle-end/79496
31165 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31166 clearing info.nowrite flag when snprintf size argument is a range.
31168 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31170 * cprop.c (cprop_jump): Add missing space in string literal.
31171 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31172 (get_constraint_for_component_ref): Likewise.
31173 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31174 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31175 * lra-constraints.c (process_alt_operands): Likewise.
31176 * ipa-inline.c (inline_small_functions): Likewise.
31177 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31178 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31179 * trans-mem.c (diagnose_tm_1_op): Likewise.
31180 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31181 (grid_parallel_clauses_gridifiable): Likewise.
31183 * config/nvptx/mkoffload.c (process): Add space in between
31186 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31187 "MOD4_SSE_REGS" and "ALL_REGS".
31189 * spellcheck.c (test_data): Add , in between "foo" and "food".
31191 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31194 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31195 boundary crossing check and subsequent code generation agree.
31197 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31199 * config/aarch64/aarch64.c (has_memory_op): Delete.
31200 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31203 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31205 PR rtl-optimization/79388
31206 PR rtl-optimization/79450
31207 * combine.c (distribute_notes): When removing TEM_INSN for which
31208 corresponding dest has last value recorded, invalidate that last
31211 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31213 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31214 of explicit '@'. Add missing assembly comment marker on branch costs
31217 2017-02-13 Nathan Sidwell <nathan@acm.org>
31219 * gengtype-lex.l (<in_struct>): Add '/'.
31221 2017-02-13 Martin Liska <mliska@suse.cz>
31224 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31226 2017-02-13 Richard Biener <rguenther@suse.de>
31228 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31230 * configure: Re-generate.
31231 * config.in: Likewise.
31232 * graphite-dependences.c: Simplify as if
31233 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31234 * graphite-isl-ast-to-gimple.c: Likewise.
31235 * graphite-optimize-isl.c: Likewise.
31236 * graphite-poly.c: Likewise.
31237 * graphite-sese-to-poly.c: Likewise.
31238 * graphite.h: Likewise.
31239 * toplev.c: Include isl/version.h and use isl_version () for
31240 printing the ISL version.
31241 * doc/install.texi: Update ISL requirement.
31243 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31245 * doc/standards.texi (Standards): Update reference to
31248 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31250 * doc/extend.texi (Named Address Spaces): sourceware.org now
31252 * doc/install.texi (Binaries): Ditto.
31255 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31257 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31258 standard terminology "stringize"/"stringizing" throughout.
31259 * doc/cppinternals.texi: Likewise.
31261 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31263 * doc/extend.texi: Fix some spelling mistakes and typos.
31264 * doc/invoke.texi: Likewise.
31266 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31269 * params.def (inline-min-speedup) Change from 10 to 8.
31271 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31273 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31276 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31279 * ipa-inline-analysis.c (get_minimal_bb): New function.
31280 (record_modified): Use it.
31281 (remap_edge_change_prob): Handle also ancestor functions.
31283 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31285 * doc/contrib.texi (Contributors): Remove broken link into
31286 the Mauve CVS repository.
31288 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31290 PR middle-end/79454
31291 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31292 result computation whenever lhs doesn't have vector mode, not
31293 just when it has BLKmode.
31295 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31297 * doc/makefile.texi (profiledbootstrap): Refer to the
31298 installation instructions only in textual form.
31300 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31303 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31305 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31307 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31308 (Specific): Update mingw-w64 reference.
31310 (Specific): Remove broken link to Renesas RX processor.
31312 2017-02-10 Richard Biener <rguenther@suse.de>
31314 * toplev.c (process_options): Do not mention obsolete graphite
31315 options when printing sorry message about missing graphite support.
31316 Mention -floop-nest-optimize.
31318 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31320 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31321 (vtst_p16): Likewise.
31322 (vtstq_p8): Likewise.
31323 (vtstq_p16): Likewise.
31325 (vtstq_p64): Likewise.
31326 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31327 (vset_lane_p64): New.
31328 (vsetq_lane_p64): New.
31330 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31332 PR tree-optimization/79411
31333 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31334 stmt operands are SSA_NAMEs used in abnormal phis.
31335 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31338 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31341 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
31344 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31346 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
31348 2017-02-09 Jakub Jelinek <jakub@redhat.com>
31350 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
31354 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
31355 not arbitrary TREE_CONSTANT.
31358 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
31359 "omp declare target link" attribute unless is_global_var.
31360 * omp-offload.c (find_link_var_op): Likewise.
31362 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
31363 Chung-Lin Tang <cltang@codesourcery.com>
31365 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31367 (gimplify_adjust_omp_clauses): Don't delete TILE.
31368 (gimplify_omp_for): Deal with TILE.
31369 * internal-fn.c (expand_GOACC_TILE): New function.
31370 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
31372 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
31373 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
31375 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
31376 avoid DIV for outermost collapse var.
31377 (expand_oacc_for): Insert tile element loop as needed. Adjust.
31378 Remove out of date comments, fix whitespace.
31379 * omp-general.c (omp_extract_for_data): Deal with tiling.
31380 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
31381 adjust OLF_DIM_BASE value.
31382 (struct omp_for_data): Add tiling field.
31383 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
31384 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
31385 for auto loops. Remove default auto determining, moved to
31386 oacc_loop_fixed_partitions.
31387 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
31388 stmts, add e_mask field.
31389 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
31390 (oacc_thread_numbers): Use oacc_dim_call.
31391 (oacc_xform_tile): New.
31392 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
31393 (finish_oacc_loop): Adjust for ifns vector.
31394 (oacc_loop_discover_walk): Append loop abstraction sites to list,
31395 add case for GOACC_TILE fns.
31396 (oacc_loop_xform_loop): Delete.
31397 (oacc_loop_process): Iterate over call list directly, and add
31398 handling for GOACC_TILE fns.
31399 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
31401 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
31402 vector partitioning to outer loops. Assign 2 partitions to loops
31403 when available. Add TILE handling.
31404 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
31405 (execite_oacc_device_lower): Process GOACC_TILE fns,
31406 ignore unknown specs.
31407 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
31408 * tree.c (omp_clause_num_ops): Adjust TILE ops.
31409 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
31411 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
31413 * configure.ac (ACX_BUGURL): Update.
31414 * configure: Regenerate.
31416 2017-02-09 Richard Biener <rguenther@suse.de>
31418 PR tree-optimization/69823
31419 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
31420 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
31422 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31424 * config/arc/arc-c.def: Add __NPS400__ definition.
31425 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
31426 (TARGET_NPS400): Define.
31428 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31430 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
31432 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
31433 pointer, arch_info.
31434 (arc_cpu_types): Fill the arch_info field with a pointer into the
31435 arc_arch_types table.
31436 (arc_selected_cpu): Declare.
31437 * config/arc/arc.c (arc_selected_cpu): Make global.
31438 (arc_selected_arch): Delete.
31439 (arc_base_cpu): Delete.
31440 (arc_override_options): Remove references to deleted variables,
31441 update access to arch information.
31442 (ARC_OPT): Update access to arch information.
31443 (ARC_OPTX): Likewise.
31444 * config/arc/arc.h (arc_base_cpu): Remove declaration.
31445 (TARGET_ARC600): Update access to arch information.
31446 (TARGET_ARC601): Likewise.
31447 (TARGET_ARC700): Likewise.
31448 (TARGET_EM): Likewise.
31449 (TARGET_HS): Likewise.
31450 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
31453 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
31456 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
31457 condition/operands for integer GE/LE/GEU/LEU operations.
31459 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
31461 PR translation/79397
31462 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
31465 2017-02-08 Martin Jambor <mjambor@suse.cz>
31468 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
31469 whether allocation happened.
31470 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
31471 nothing was allocated.
31473 2017-02-08 Jakub Jelinek <jakub@redhat.com>
31475 PR tree-optimization/79408
31476 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
31477 constant, but SSA_NAME with a known integer range, use the minimum
31478 of that range instead of op1 to determine if modulo can be replaced
31479 with its first operand.
31481 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31483 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
31485 2017-02-08 Richard Biener <rguenther@suse.de>
31487 PR tree-optimization/71824
31488 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31489 Check all loops contained in the merged region.
31491 2017-02-07 Andrew Pinski <apinski@cavium.com>
31493 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
31495 2017-02-07 Andrew Pinski <apinski@cavium.com>
31497 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
31498 (thunderxt88): Likewise.
31499 (thunderxt81): Disable LSE and change v8.1 to v8.
31500 (thunderxt83): Likewise.
31502 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31503 Richard Biener <rguenther@suse.de>
31505 PR middle-end/79399
31506 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
31507 type from int to size_t.
31508 * ira-costs.c (struct_costs_size): Change type from int to size_t.
31510 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31512 PR rtl-optimization/79386
31513 * cprop.c (bypass_conditional_jumps): Initialize
31514 bypass_last_basic_block already before splitting bbs after
31515 unconditional traps...
31516 (bypass_conditional_jumps): ... rather than here.
31519 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
31520 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
31521 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
31522 fix -masm=intel patterns.
31524 2017-02-07 Richard Biener <rguenther@suse.de>
31526 PR tree-optimization/79256
31527 PR middle-end/79278
31528 * builtins.c (get_object_alignment_2): Use min_align_of_type
31529 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
31530 and ADJUST_FIELD_ALIGN.
31532 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
31534 * doc/tm.texi: Regenerate.
31535 * stor-layout.c (layout_decl): Adjust.
31536 (update_alignment_for_field): Likewise.
31537 (place_field): Likewise.
31538 (min_align_of_type): Likewise.
31539 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
31540 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
31541 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
31542 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
31543 * config/frv/frv.c (frv_adjust_field_align): Likewise.
31544 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
31545 * config/i386/i386.c (x86_field_alignment): Likewise.
31546 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
31547 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
31548 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
31549 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
31550 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
31551 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
31555 2017-01-30 Richard Biener <rguenther@suse.de>
31557 PR tree-optimization/79256
31558 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31559 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31562 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
31564 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
31565 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
31566 builtins to SImode and emit a zero-extend, if necessary.
31568 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31570 * docs/invoke.texi (RISC-V Options): Alphabetize.
31572 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31574 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
31577 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31579 * config/riscv/riscv.c: New file.
31580 * common/config/riscv/riscv-common.c: Likewise.
31581 * config.gcc: Likewise.
31582 * config/riscv/constraints.md: Likewise.
31583 * config/riscv/elf.h: Likewise.
31584 * config/riscv/generic.md: Likewise.
31585 * config/riscv/linux.h: Likewise.
31586 * config/riscv/multilib-generator: Likewise.
31587 * config/riscv/peephole.md: Likewise.
31588 * config/riscv/pic.md: Likewise.
31589 * config/riscv/predicates.md: Likewise.
31590 * config/riscv/riscv-builtins.c: Likewise.
31591 * config/riscv/riscv-c.c: Likewise.
31592 * config/riscv/riscv-ftypes.def: Likewise.
31593 * config/riscv/riscv-modes.def: Likewise.
31594 * config/riscv/riscv-opts.h: Likewise.
31595 * config/riscv/riscv-protos.h: Likewise.
31596 * config/riscv/riscv.h: Likewise.
31597 * config/riscv/riscv.md: Likewise.
31598 * config/riscv/riscv.opt: Likewise.
31599 * config/riscv/sync.md: Likewise.
31600 * config/riscv/t-elf-multilib: Likewise.
31601 * config/riscv/t-linux: Likewise.
31602 * config/riscv/t-linux-multilib: Likewise.
31603 * config/riscv/t-riscv: Likewise.
31604 * configure.ac: Likewise.
31605 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
31606 Waterman as RISC-V maintainers.
31607 * doc/install.texi: Add RISC-V entries.
31608 * doc/invoke.texi: Add RISC-V options section.
31609 * doc/md.texi: Add RISC-V constraints section.
31610 * configure: Regenerated.
31612 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
31615 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
31616 false values to be constant vectors with all 0 or all 1 bits set.
31617 (vcondu<mode><mode>): Likewise.
31618 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
31620 (fpmask_comparison_operator): Update comment.
31621 (vecint_comparison_operator): New predicate.
31622 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
31623 vector conditionals when the true and false values are constant
31624 vectors with all 0 bits or all 1 bits set.
31626 2017-02-06 Martin Sebor <msebor@redhat.com>
31628 PR tree-optimization/79376
31629 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
31631 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
31633 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
31634 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
31635 to simplify split condition.
31637 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31639 * omp-expand.c (oxpand_omp_atomic_fetch_op,
31640 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
31643 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
31645 PR rtl-optimization/68664
31646 * target.def (can_speculate_insn): New hook.
31647 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
31648 * doc/tm.texi: Regenerate.
31649 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
31650 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
31651 (rs6000_sched_can_speculate_insn): New function.
31653 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31655 PR tree-optimization/79284
31656 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
31657 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
31658 vectorizable_mask_load_store, vectorizable_operation,
31659 vect_is_simple_cond, get_same_sized_vectype): Use it instead
31660 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
31661 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
31662 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
31663 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
31664 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
31665 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
31666 is_gimple_assign (stmt). Replace another such test with
31667 is_gimple_assign (stmt).
31669 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31672 * config/avr/avr.c (rtl-iter.h): Include it.
31673 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
31674 (avr_legitimate_combined_insn): ...and implementation.
31676 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
31678 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
31679 * config/s390/s390.c (s390_const_operand_ok)
31680 (s390_canonicalize_comparison, s390_extract_part)
31681 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
31682 (s390_contiguous_bitmask_p, s390_rtx_costs)
31683 (legitimize_pic_address): Likewise.
31684 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
31685 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
31686 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
31687 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
31688 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
31690 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31692 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
31693 REGNO($0) == REGNO($1).
31695 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31697 * config/s390/linux.h(SIZE_TYPE): Add comment.
31699 2017-02-06 Julian Brown <julian@codesourcery.com>
31700 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31701 Virendra Pathak <virendra.pathak@broadcom.com>
31703 * config/aarch64/aarch64-cores.def: Change the scheduler
31705 * config/aarch64/aarch64.md: Include thunderx2t99.md.
31706 * config/aarch64/thunderx2t99.md: New file.
31708 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31710 * doc/standards.texi (Go Language): Update link to language
31713 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31715 * tree-eh.c (lower_resx): Sanitize profile.
31716 (cleanup_empty_eh_move_lp): Likewise.
31718 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31721 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
31723 * cfgloopmanip.h (loop_version): Update prototype.
31724 * modulo-sched.c (sms_schedule): Update call of loop_version.
31725 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
31726 * tree-parloops.c (gen_parallel_loop): Likewise.
31727 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
31728 * tree-ssa-loop-split.c (split_loop): Likewise.
31729 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
31730 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
31732 2017-02-05 Martin Liska <mliska@suse.cz>
31735 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
31737 (print_operand_address): Initialize a struct to zero.
31739 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31741 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
31742 garbage collector only in textual form.
31744 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31746 * doc/extend.texi (x86 specific memory model extensions for
31747 transactional memory): Simplify a phrase.
31749 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
31752 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
31753 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
31754 (atomic_storedi_1): Likewise.
31756 2017-02-04 Jakub Jelinek <jakub@redhat.com>
31758 PR tree-optimization/79338
31759 * tree-parloops.c (gather_scalar_reductions): Don't call
31760 vect_analyze_loop_form for loop->inner before destroying loop's
31763 2017-02-03 Martin Sebor <msebor@redhat.com>
31765 PR tree-optimization/79327
31766 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
31767 when precision has resulted in leading zeros.
31768 (format_integer): Adjust the likely counter to assume an unknown
31769 argument that may be zero is non-zero.
31771 2017-02-03 Jason Merrill <jason@redhat.com>
31774 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
31775 avoid copying non-taken branch.
31777 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31779 PR tree-optimization/79340
31780 * tree-vect-loop.c (vectorizable_reduction): Release
31781 vec_defs elements after safe_splicing them into other vectors.
31784 PR tree-optimization/79327
31785 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
31786 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
31788 (format_integer): Use wide_int_to_tree instead of build_int_cst
31789 + to_?hwi. If argmin is NULL, just set argmin and argmax to
31790 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
31791 of shortest and longest sequence.
31793 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
31795 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
31796 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
31798 2017-02-03 Walter Lee <walt@tilera.com>
31801 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
31802 after initial stackframe link reg save.
31803 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
31805 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31808 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
31809 wu for stxssp alternative.
31811 2017-02-03 Martin Sebor <msebor@redhat.com>
31813 PR tree-optimization/79352
31814 * gimple-fold.c (get_range_strlen): Add argument.
31815 (get_range_strlen): Change return type to bool.
31816 (get_maxval_strlen): Pass in a dummy argument.
31817 * gimple-fold.h (get_range_strlen): Change return type to bool.
31818 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
31819 * tree.h (array_at_struct_end_p): Add argument.
31820 * tree.c (array_at_struct_end_p): Handle it.
31822 2017-02-03 Martin Liska <mliska@suse.cz>
31825 * multiple_target.c (create_dispatcher_calls): Redirect edge
31826 from a caller of a dispatcher.
31827 (expand_target_clones): Make the clones local.
31828 (ipa_target_clone): Do both target clones and resolvers.
31829 (ipa_dispatcher_calls): Remove the pass.
31830 (pass_dispatcher_calls::gate): Likewise.
31831 (make_pass_dispatcher_calls): Likewise.
31832 * passes.def (pass_target_clone): Put as very first IPA early
31835 2017-02-03 Martin Liska <mliska@suse.cz>
31837 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
31838 in case of a function with ifunc attribute.
31840 2017-02-03 Martin Liska <mliska@suse.cz>
31842 * cgraph.c (cgraph_node::dump): Dump function version info.
31843 * symtab.c (symtab_node::dump_base): Add missing new line.
31845 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31847 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
31848 (ifcombine_ifandif): Use it.
31850 2017-02-03 Martin Liska <mliska@suse.cz>
31852 * doc/invoke.texi: Document default value for
31853 use-after-scope-direct-emission-threshold.
31855 2017-02-03 Martin Liska <mliska@suse.cz>
31857 PR tree-optimization/79339
31858 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
31859 (format_floating): Likewise.
31861 2017-02-03 Martin Liska <mliska@suse.cz>
31864 * ipa-prop.c (ipa_node_params_t::insert): Remove current
31866 (ipa_node_params_t::remove): Likewise.
31867 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
31868 initialization from removed ipa_node_params_t::insert.
31869 (ipa_node_params::~ipa_node_params): Move from removed
31870 ipa_node_params_t::release.
31871 * symbol-summary.h (symbol_summary::m_released): New member.
31872 Do not release a summary twice. Do not allow to call finalizer
31873 for types of a summary that live in GGC memory.
31875 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31877 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
31880 2017-02-02 Martin Sebor <msebor@redhat.com>
31882 PR middle-end/79275
31883 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
31884 (format_string): Tighten up the range of output for non-constant
31885 strings and correct the expected range for wide non-constant strings.
31887 2017-02-02 Martin Sebor <msebor@redhat.com>
31889 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
31891 PR middle-end/32003
31892 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
31894 (-fdump-tree-@var): Add to index and document how to come up
31895 with pass-specific option and dump file names.
31896 (-fdump-passes): Clarify where to look for output.
31898 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31900 PR middle-end/77445
31901 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
31902 statistics of the analyzed path; allow threading for speed when
31903 any of BBs along the path are optimized for speed.
31905 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
31907 PR middle-end/78468
31908 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
31909 settings of the virtual registers.
31912 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
31914 * explow.c (get_dynamic_stack_size): Take known alignment of stack
31915 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
31918 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31920 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
31921 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
31923 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31925 * config/s390/s390.md: Add missing comments with the expanded
31927 * config/s390/vector.md: Likewise.
31928 * config/s390/vx-builtins.md: Likewise.
31930 2017-02-02 Jakub Jelinek <jakub@redhat.com>
31933 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
31934 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
31935 conditions on a single line.
31937 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31939 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31940 __S390_VX__ to __VX__.
31942 2017-02-01 Andrew Pinski <apinski@cavium.com>
31944 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
31945 stmt_info to record_stmt_cost.
31946 (vect_get_known_peeling_cost): Pass stmt_info if known to
31948 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
31949 cpu_vector_cost field into
31950 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
31951 field into vec_int_stmt_cost and vec_fp_stmt_cost.
31952 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
31953 splitting of scalar_stmt_cost and vec_stmt_cost.
31954 (thunderx_vector_cost): Likewise.
31955 (cortexa57_vector_cost): LIkewise.
31956 (exynosm1_vector_cost): Likewise.
31957 (xgene1_vector_cost): Likewise.
31958 (thunderx2t99_vector_cost): Improve after the splitting of the two
31960 (aarch64_builtin_vectorization_cost): Update for the splitting of
31961 scalar_stmt_cost and vec_stmt_cost.
31963 2017-02-01 Torvald Riegel <triegel@redhat.com>
31964 Richard Henderson <rth@redhat.com>
31966 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
31967 conditional on existance of a fast atomic load.
31968 * optabs-query.c (can_atomic_load_p): New function.
31969 * optabs-query.h (can_atomic_load_p): Declare it.
31970 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
31971 no fast atomic load is available for the particular size of access.
31972 (expand_atomic_compare_and_swap): Likewise.
31973 (expand_atomic_load): Likewise.
31974 (expand_atomic_store): Likewise.
31975 (expand_atomic_fetch_op): Likewise.
31976 * testsuite/lib/target-supports.exp
31977 (check_effective_target_sync_int_128): Remove x86 because it provides
31978 no fast atomic load.
31979 (check_effective_target_sync_int_128_runtime): Likewise.
31981 2017-02-01 Richard Biener <rguenther@suse.de>
31983 * graphite.c: Include tree-vectorizer.h for find_loop_location.
31984 (graphite_transform_loops): Provide opt-info for optimized nests.
31985 * tree-parloop.c (parallelize_loops): Provide opt-info for
31986 parallelized loops.
31988 2017-02-01 Richard Biener <rguenther@suse.de>
31990 PR middle-end/79315
31991 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
31992 was not set before.
31994 2017-02-01 Richard Biener <rguenther@suse.de>
31996 PR tree-optimization/71824
31997 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31998 Verify the loops are valid in the merged SESE region.
31999 (scop_detection::can_represent_loop_1): Check analyzing the
32000 evolution of the number of iterations in the region succeeds.
32002 2017-01-31 Ian Lance Taylor <iant@golang.org>
32004 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32005 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32007 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32009 PR preprocessor/79210
32010 * input.c (get_substring_ranges_for_loc): Replace line_width
32011 assertion with error-handling.
32013 2017-01-31 Richard Biener <rguenther@suse.de>
32015 PR tree-optimization/77318
32016 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32017 (create_pw_aff_from_tree): Take loop parameter.
32018 (add_condition_to_pbb): Pass loop of the condition to
32019 create_pw_aff_from_tree.
32021 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32023 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32024 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32026 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32030 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32032 (convert_int_to_float128): Likewise.
32033 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32034 (convert_int_to_float128): Likewise.
32035 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32036 (UNSPEC_IEEE128_CONVERT): Likewise.
32037 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32038 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32039 Use local variables for IBM extended format.
32040 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32041 (fix_trunc<mode>si2_fprs): Likewise.
32042 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32043 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32044 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32045 to know that we can now have integers of all sizes in vector
32047 (fix<uns>_<mode>di2_hw): Likewise.
32048 (float<uns>_<mode>si2_hw): Likewise.
32049 (fix_<mode>si2_hw): Likewise.
32050 (fixuns_<mode>si2_hw): Likewise.
32051 (float<uns>_<mode>di2_hw): Likewise.
32052 (float_<mode>di2_hw): Likewise.
32053 (float_<mode>si2_hw): Likewise.
32054 (floatuns_<mode>di2_hw): Likewise.
32055 (floatuns_<mode>si2_hw): Likewise.
32056 (xscvqp<su>wz_<mode>): Delete, no longer used.
32057 (xscvqp<su>dz_<mode>): Likewise.
32058 (xscv<su>dqp_<mode>): Likewise.
32059 (ieee128_mfvsrd_64bit): Likewise.
32060 (ieee128_mfvsrd_32bit): Likewise.
32061 (ieee128_mfvsrwz): Likewise.
32062 (ieee128_mtvsrw): Likewise.
32063 (ieee128_mtvsrd_64bit): Likewise.
32064 (ieee128_mtvsrd_32bit): Likewise.
32066 2017-01-31 Martin Liska <mliska@suse.cz>
32069 * ipa-prop.c (ipa_free_all_node_params): Call release method
32070 instead of ~sumbol_summary to not to trigger double times
32073 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32075 PR tree-optimization/71691
32076 * bitmap.h (class auto_bitmap): New.
32077 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32078 is_maybe_undefined instead of ssa_undefined_value_p.
32080 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32082 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32083 __S390_ARCH_LEVEL__ to __ARCH__.
32085 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32087 PR tree-optimization/79267
32088 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32089 if should_remove_lhs_p is true.
32091 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32094 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32095 (add_alignment_attribute): New.
32096 (base_type_die): Add alignment attribute.
32097 (subrange_type_die): Likewise.
32098 (modified_type_die): Likewise.
32099 (gen_array_type_die): Likewise.
32100 (gen_descr_array_type_die: Likewise.
32101 (gen_enumeration_type_die): Likewise.
32102 (gen_subprogram_die): Likewise.
32103 (gen_variable_die): Likewise.
32104 (gen_field_die): Likewise.
32105 (gen_ptr_to_mbr_type_die): Likewise.
32106 (gen_struct_or_union_type_die): Likewise.
32107 (gen_subroutine_type_die): Likewise.
32108 (gen_typedef_die): Likewise.
32109 (base_type_cmp): Compare alignment attribute.
32111 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32114 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32115 (setb_unsigned) New pattern for setb with CCUNS.
32116 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32117 subfc./subfe sequence to avoid overflow problems. Generate a
32118 shorter sequence with cmpld/setb for power9.
32119 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32120 for generating subfc. instruction.
32121 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32122 now uses this instruction.
32124 2017-01-30 Ian Lance Taylor <iant@google.com>
32127 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32128 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32130 2017-01-30 Martin Sebor <msebor@redhat.com>
32132 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32133 Move constant to the right of a relational operator.
32134 (get_mpfr_format_length, format_character, format_string): Ditto.
32135 (should_warn_p, maybe_warn): Same.
32137 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32139 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32142 * asan.c (get_translation_unit_decl): Remove function.
32143 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32145 2017-01-30 Martin Liska <mliska@suse.cz>
32147 PR gcov-profile/79259
32148 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32149 -fprofile-generate.
32151 2017-01-30 Martin Liska <mliska@suse.cz>
32154 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32155 Initialize variables with NULL value.
32157 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32160 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32162 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32164 2017-01-30 Richard Biener <rguenther@suse.de>
32166 PR tree-optimization/79276
32167 * tree-vrp.c (process_assert_insertions): Properly adjust common
32168 when removing a duplicate.
32170 2017-01-30 Richard Biener <rguenther@suse.de>
32172 PR tree-optimization/79256
32173 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32174 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32176 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32178 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32181 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32182 ("*r<noxa>sbg_<mode>_sll_bitmask")
32183 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32184 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32185 Use contiguous_bitmask_nowrap_operand.
32187 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32190 * config/rs6000/altivec.h (vec_xl): Revise #define.
32191 (vec_xst): Likewise.
32193 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32195 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32197 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32199 PR rtl-optimization/79194
32200 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32201 traps before call to bypass_conditional_jumps.
32203 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32205 PR tree-optimization/71374
32206 * lra-constraints.c (check_conflict_input_operands): New.
32207 (match_reload): Use it.
32209 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32212 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32213 account to calculate conflict_set.
32215 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32217 PR rtl-optimization/78559
32218 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32219 other_insn in combine.
32221 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32223 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32224 uint16_type_node for BT_UINT16.
32226 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32228 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32229 "RTL Tests" to menu.
32230 (GIMPLE Tests): New node.
32231 (RTL Tests): New node.
32233 2017-01-27 Richard Biener <rguenther@suse.de>
32235 PR tree-optimization/79245
32236 * tree-loop-distribution.c (distribute_loop): Apply cost
32237 modeling also to detected patterns.
32239 2017-01-27 Richard Biener <rguenther@suse.de>
32241 PR tree-optimization/71433
32242 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32243 (compare_assert_loc): New function.
32244 (process_assert_insertions): Sort and optimize assert locations
32245 to remove duplicates and push down identical assertions on
32246 edges to their destination block.
32248 2017-01-27 Richard Biener <rguenther@suse.de>
32250 PR tree-optimization/79244
32251 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32252 out SSA names even if abnormal.
32254 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32256 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32257 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32258 instead of MPFR_RNDN.
32260 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32263 * arm.c (arm_option_override): Don't call build_target_option_node
32264 until after doing all option overrides.
32265 (arm_valid_target_attribute_tree): Likewise.
32267 2017-01-27 Martin Liska <mliska@suse.cz>
32269 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32270 for {cd}tors and C++ {cd}tors.
32272 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32274 * config/s390/s390.md ("*setmem_long_and")
32275 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32277 2017-01-26 Martin Sebor <msebor@redhat.com>
32279 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32282 2017-01-26 Martin Sebor <msebor@redhat.com>
32284 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32285 HAVE_DFmode before using XFmode or DFmode.
32286 (parse_directive): Avoid using the z length modifier to avoid
32287 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32289 PR middle-end/78703
32290 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32291 to accept adjustment as an array.
32292 (get_int_range): New function.
32293 (struct directive): Make width and prec arrays.
32294 (directive::set_width, directive::set_precision): Call get_int_range.
32295 (format_integer, format_floating): Handle width and precision ranges.
32296 (format_string, parse_directive): Same.
32298 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32301 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32302 comdat_type_p set, just clone them, but keep the children in the
32306 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32307 which have direct callers with -fvar-tracking-assignments enabled
32309 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32310 inside of type units.
32312 2017-01-26 Martin Sebor <msebor@redhat.com>
32314 PR middle-end/78703
32315 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32317 (struct format_result): Replace number_chars, number_chars_min,
32318 and number_chars_max with a single member of struct result_range.
32320 (format_result::operator+=): Adjust.
32321 (struct fmtresult): Remove bounded. Handle likely and unlikely
32323 (fmtresult::adjust_for_width_or_precision): New function.
32324 (fmtresult:type_max_digits): New function.
32325 (bytes_remaining): Handle likely and unlikely counters.
32326 (min_bytes_remaining): Remove.
32327 (format_percent): Simplify.
32328 (format_integer, format_floating): Set likely and unlikely counters.
32329 (get_string_length, format_character, format_string): Same.
32330 (format_plain, should_warn_p): New function.
32331 (maybe_warn): Call should_warn_p. Update diagnostic messages
32332 and handle those for all directives, including plain strings.
32333 (format_directive): Handle likely and unlikely counters.
32334 Remove unnecessary quoting from diagnostics. Add an informational
32336 (add_bytes): Remove.
32337 (pass_sprintf_length::compute_format_length): Simplify.
32338 (try_substitute_return_value): Handle likely and unlikely counters.
32340 2017-01-26 Carl Love <cel@us.ibm.com>
32342 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
32343 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
32345 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
32348 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
32349 endianess for subregs into account.
32350 * lra-constraints.c (lra_constraints): Do risky transformations
32351 always on the first iteration.
32352 * lra-lives.c (check_pseudos_live_through_calls): Add arg
32353 last_call_used_reg_set.
32354 (process_bb_lives): Define and use last_call_used_reg_set.
32355 * lra.c (lra): Always continue after lra_constraints on the first
32358 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
32360 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
32362 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32364 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32366 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32367 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32368 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32369 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
32370 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
32371 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
32372 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
32373 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
32374 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
32376 2017-01-26 Marek Polacek <polacek@redhat.com>
32379 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
32380 for the third operand.
32382 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32384 PR middle-end/79236
32385 * omp-low.c (struct omp_context): Add simt_stmt field.
32386 (scan_omp_for): Return omp_context *.
32387 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
32388 context to the _simt_ SIMD stmt.
32389 (lower_omp_for): For combined SIMD with sibling _simt_
32390 SIMD, make sure to use the same decls in _looptemp_
32391 clauses as in the sibling.
32393 2017-01-26 David Sherwood <david.sherwood@arm.com>
32395 PR middle-end/79212
32396 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
32399 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32402 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
32403 emit fld b; fld a; if possible.
32405 * brig-builtins.def: Update copyright years.
32406 * config/arm/arm_acle_builtins.def: Update copyright years.
32408 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
32411 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
32412 constraint instead of o for the stxsd instruction.
32414 2017-01-25 Carl Love <cel@us.ibm.com>
32416 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
32417 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
32419 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
32421 * doc/invoke.texi (C++ Dialect Options): Fix typo.
32423 2017-01-25 Richard Biener <rguenther@suse.de>
32425 PR tree-optimization/69264
32426 * target.def (vector_alignment_reachable): Improve documentation.
32427 * doc/tm.texi: Regenerate.
32428 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
32430 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
32431 earlier changes with respect to TYPE_USER_ALIGN.
32432 (vector_alignment_reachable_p): Likewise. Improve dumping.
32434 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32437 * config/arm/arm.md (xordi3): Force constant operand into a register
32440 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32442 * doc/invoke.texi (-fstore-merging): Correct default optimization
32443 levels at which it is enabled.
32444 (-O): Move -fstore-merging from list to...
32447 2017-01-25 Richard Biener <rguenther@suse.de>
32450 * omp-expand.c: Include debug.h.
32451 (expand_omp_taskreg): Make sure to generate early debug before
32452 outlining anything from a function.
32453 (expand_omp_target): Likewise.
32454 (grid_expand_target_grid_body): Likewise.
32456 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
32459 * asan.c (get_translation_unit_decl): New function.
32460 (asan_add_global): Extract modules file name from globals
32461 TRANSLATION_UNIT_DECL name.
32463 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
32466 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
32467 for long calls with APCS frame and VFP.
32469 2017-01-24 David Malcolm <dmalcolm@redhat.com>
32471 * cfg.c (original_copy_tables_initialized_p): New function.
32472 * cfg.h (original_copy_tables_initialized_p): New decl.
32473 * cfgrtl.c (relink_block_chain): Guard the call to
32474 free_original_copy_tables with a call to
32475 original_copy_tables_initialized_p.
32476 * cgraph.h (symtab_node::native_rtl_p): New decl.
32477 * cgraphunit.c (symtab_node::native_rtl_p): New function.
32478 (symtab_node::needed_p): Don't assert for early assembly output
32479 for __RTL functions.
32480 (cgraph_node::finalize_function): Set "force_output" for __RTL
32482 (cgraph_node::analyze): Bail out early for __RTL functions.
32483 (analyze_functions): Update assertion to support __RTL functions.
32484 (cgraph_node::expand): Bail out early for __RTL functions.
32485 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
32487 * function.h (struct function): Update comment for field
32489 * gimple-expr.c: Include "tree-pass.h".
32490 (gimple_has_body_p): Return false for __RTL functions.
32491 * Makefile.in (OBJS): Add run-rtl-passes.o.
32492 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
32494 (gcc::pass_manager::get_clean_slate): New accessor.
32495 * passes.c: Include "insn-addr.h".
32496 (should_skip_pass_p): Add logging. Update logic for running
32497 "expand" to be compatible with both __GIMPLE and __RTL. Guard
32498 property-provider override so it is only done for gimple passes.
32500 (skip_pass): New function.
32501 (execute_one_pass): Call skip_pass when skipping passes.
32502 * read-md.c (md_reader::read_char): Support filtering
32503 the input to a subset of line numbers.
32504 (md_reader::md_reader): Initialize fields
32505 m_first_line and m_last_line.
32506 (md_reader::read_file_fragment): New function.
32507 * read-md.h (md_reader::read_file_fragment): New decl.
32508 (md_reader::m_first_line): New field.
32509 (md_reader::m_last_line): New field.
32510 * read-rtl-function.c (function_reader::create_function): Only
32511 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
32512 curr_properties. Set DECL_INITIAL to a dummy block.
32513 (read_rtl_function_body_from_file_range): New function.
32514 * read-rtl-function.h (read_rtl_function_body_from_file_range):
32516 * run-rtl-passes.c: New file.
32517 * run-rtl-passes.h: New file.
32519 2017-01-24 Jeff Law <law@redhat.com>
32521 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
32524 2017-01-24 Bin Cheng <bin.cheng@arm.com>
32526 PR tree-optimization/79159
32527 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
32528 (record_nonwrapping_iv): Improve boundary using above function if no
32529 value range information.
32531 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
32532 Martin Jambor <mjambor@suse.cz>
32534 * brig-builtins.def: New file.
32535 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
32536 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
32537 (DEF_HSAIL_SAT_BUILTIN): Likewise.
32538 (DEF_HSAIL_INTR_BUILTIN): Likewise.
32539 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
32540 * builtin-types.def (BT_INT8): New.
32541 (BT_INT16): Likewise.
32542 (BT_UINT8): Likewise.
32543 (BT_UINT16): Likewise.
32544 (BT_FN_ULONG): Likewise.
32545 (BT_FN_UINT_INT): Likewise.
32546 (BT_FN_UINT_ULONG): Likewise.
32547 (BT_FN_UINT_LONG): Likewise.
32548 (BT_FN_UINT_PTR): Likewise.
32549 (BT_FN_ULONG_PTR): Likewise.
32550 (BT_FN_INT8_FLOAT): Likewise.
32551 (BT_FN_INT16_FLOAT): Likewise.
32552 (BT_FN_UINT32_FLOAT): Likewise.
32553 (BT_FN_UINT16_FLOAT): Likewise.
32554 (BT_FN_UINT8_FLOAT): Likewise.
32555 (BT_FN_UINT64_FLOAT): Likewise.
32556 (BT_FN_UINT16_UINT32): Likewise.
32557 (BT_FN_UINT32_UINT16): Likewise.
32558 (BT_FN_UINT16_UINT16_UINT16): Likewise.
32559 (BT_FN_INT_PTR_INT): Likewise.
32560 (BT_FN_UINT_PTR_UINT): Likewise.
32561 (BT_FN_LONG_PTR_LONG): Likewise.
32562 (BT_FN_ULONG_PTR_ULONG): Likewise.
32563 (BT_FN_VOID_UINT64_UINT64): Likewise.
32564 (BT_FN_UINT8_UINT8_UINT8): Likewise.
32565 (BT_FN_INT8_INT8_INT8): Likewise.
32566 (BT_FN_INT16_INT16_INT16): Likewise.
32567 (BT_FN_INT_INT_INT): Likewise.
32568 (BT_FN_UINT_FLOAT_UINT): Likewise.
32569 (BT_FN_FLOAT_UINT_UINT): Likewise.
32570 (BT_FN_ULONG_UINT_UINT): Likewise.
32571 (BT_FN_ULONG_UINT_PTR): Likewise.
32572 (BT_FN_ULONG_ULONG_ULONG): Likewise.
32573 (BT_FN_UINT_UINT_UINT): Likewise.
32574 (BT_FN_VOID_UINT_PTR): Likewise.
32575 (BT_FN_UINT_UINT_PTR: Likewise.
32576 (BT_FN_UINT32_UINT64_PTR): Likewise.
32577 (BT_FN_INT_INT_UINT_UINT): Likewise.
32578 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
32579 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
32580 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
32581 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
32582 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
32583 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
32584 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
32585 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
32586 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
32587 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
32588 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
32589 * doc/frontends.texi: List BRIG FE.
32590 * doc/install.texi (Testing): Add BRIG tesring requirements.
32591 * doc/invoke.texi (Overall Options): Mention BRIG.
32592 * doc/standards.texi (Standards): Doucment BRIG HSA version.
32594 2017-01-24 Richard Biener <rguenther@suse.de>
32596 PR translation/79208
32597 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
32599 2017-01-24 Martin Jambor <mjambor@suse.cz>
32602 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
32603 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
32604 and known_contexts.
32606 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
32608 PR middle-end/79123
32609 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
32610 casts from signed to unsigned really don't have a range.
32612 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
32614 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
32615 GMP_RNDx for compatiblity.
32617 2017-01-24 Martin Liska <mliska@suse.cz>
32620 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
32621 that would prevent us to call alloca with -1 as argument.
32623 2017-01-24 Jakub Jelinek <jakub@redhat.com>
32625 * dwarf2out.c (output_compilation_unit_header, output_file_names):
32626 Avoid -Wformat-security warning.
32628 2017-01-23 Andrew Pinski <apinski@cavium.com>
32630 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
32633 2017-01-23 Martin Sebor <msebor@redhat.com>
32635 PR middle-end/78703
32636 * gimple-ssa-sprintf.c (warn_level): New global.
32637 (format_integer): Use it here and throughout the rest of the file.
32638 Use the same switch to compute sign as base.
32639 (maybe_warn): New function.
32640 (format_directive): Factor out warnings into maybe_warn.
32641 Add debugging output. Use warn_level.
32642 (add_bytes): Use warn_level.
32643 (pass_sprintf_length::compute_format_length): Add debugging output.
32644 (try_substitute_return_value): Same.
32645 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
32647 PR middle-end/78703
32648 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
32649 (struct fmtresult, format_integer, format_floating): Adjust.
32650 (fmtresult::fmtresult): Set max correctly in two argument ctor.
32651 (get_string_length, format_string,format_directive): Same.
32652 (pass_sprintf_length::compute_format_length): Same.
32653 (try_substitute_return_value): Simplify slightly.
32655 PR middle-end/78703
32656 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
32657 (fmtresult::operator+=): Outlined.
32658 (struct fmtresult): Add ctors.
32659 (struct conversion_spec): Rename...
32660 (struct directive): ...to this. Add and remove data members.
32661 (directive::set_width, directive::set_precision): New functions.
32662 (format_percent): Use fmtresult ctor.
32663 (get_width_and_precision): Remove.
32664 (format_integer): Make naming changes. Avoid computing width and
32666 (format_floating): Same. Adjust indentation.
32667 (format_character, format_none): New functions.
32668 (format_string): Moved character handling to format_character.
32669 (format_directive): Remove arguments, change return type.
32670 (parse_directive): New function.
32671 (pass_sprintf_length::compute_format_length): Move directive
32672 parsing to parse_directive.
32674 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32676 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
32677 (assign_assembler_name_if_needed): ... this.
32678 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
32679 (assign_assembler_name_if_needed): ... this.
32680 (free_lang_data_in_cgraph): Adjust callers.
32681 * cgraphunit.c (cgraph_node::analyze): Likewise.
32682 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
32685 2017-01-23 Richard Biener <rguenther@suse.de>
32687 PR tree-optimization/79088
32688 PR tree-optimization/79188
32689 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
32690 resetting loop bounds after last path deletion. Reset loop
32691 bounds of the target loop, make code match the comments.
32692 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
32693 Make sure loops need no fixups.
32695 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32697 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
32698 exponent support with double type for first argument.
32699 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
32700 type returned by __builtin_vec_extract_sig,
32701 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
32702 functions from "vector int" to "vector unsigned int" or from
32703 "vector long long int" to "vector unsigned long long int".
32704 Changed type returned by __builtin_vec_extract_exp,
32705 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
32706 functions from "vector int" to "vector unsigned int" or from
32707 "vector long long int" to "vector unsigned long long int".
32708 Changed return type of __builtin_vec_test_data_class,
32709 __builtin_vec_test_data_class_sp, and
32710 __builtin_vec_test_data_class_dp from "vector int" to
32711 "vector bool int" or from "vector long long int" to "vector bool
32712 long long int" and changed second argument type from "unsigned
32713 int" to "int". Added new overloaded function forms "vector float
32714 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
32715 "vector float __builtin_vec_insert_exp_sp (vector float, vector
32716 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
32717 double, vector unsigned long long int)" and "vector double
32718 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
32719 long int)". Changed return type of
32720 __builtin_scalar_test_data_class and
32721 __builtin_scalar_test_data_class_sp and
32722 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
32723 int" and changed second argument from "unsigned int" to "int".
32724 Changed type returned by __builtin_scalar_test_neg,
32725 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
32726 from "int" to "bool int". Added new overloaded function form
32727 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
32728 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
32729 exponent double-precision with floating point first argument.
32730 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
32731 documentation of scalar_test_data_class, scalar_test_neg,
32732 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
32733 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
32734 vec_test_data_class built-in functions to reflect refinements in
32735 their type signatures.
32737 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
32739 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
32741 (aarch64_elf_asm_destructor): Likewise.
32743 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
32745 PR rtl-optimization/78634
32746 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
32747 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
32748 * ifcvt.c (noce_try_cmove): Add missing cost check.
32750 PR rtl-optimization/71724
32751 * combine.c (if_then_else_cond): Look for situations where it is
32752 beneficial to undo the work of one of the recursive calls.
32754 2017-01-23 Bin Cheng <bin.cheng@arm.com>
32756 PR tree-optimization/70754
32757 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
32758 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
32759 combined stmt before it if not NULL.
32760 (combine_chains): Process refs reversely and compute dominance point
32763 2017-01-23 Martin Liska <mliska@suse.cz>
32765 PR tree-optimization/79196
32766 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
32767 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
32769 (strlen_optimize_stmt): Call the renamed function.
32771 2017-01-23 Michael Matz <matz@suse.de>
32773 PR tree-optimization/78384
32774 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
32776 2017-01-23 Richard Biener <rguenther@suse.de>
32778 PR tree-optimization/79186
32779 * tree-vrp.c (register_new_assert_for): Make sure we've seen
32780 both incoming edges before moving an assert.
32782 2017-01-23 Martin Jambor <mjambor@suse.cz>
32784 * ipa-prop.c (load_from_param_1): Removed.
32785 (load_from_unmodified_param): Bits from load_from_param_1 put back
32787 (load_from_param): Removed.
32788 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
32789 with stmt. Reverted back to use of load_from_unmodified_param.
32791 2017-01-23 Martin Jambor <mjambor@suse.cz>
32794 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
32795 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
32796 field a pointer to garbage collected vector, mark lattices and
32797 ipcp_orig_node with GTY((skip)).
32798 (ipa_get_param_count): Adjust to descriptors being a pointer.
32799 (ipa_get_param): Likewise.
32800 (ipa_get_type): Likewise.
32801 (ipa_get_param_move_cost): Likewise.
32802 (ipa_set_param_used): Likewise.
32803 (ipa_get_controlled_uses): Likewise.
32804 (ipa_set_controlled_uses): Likewise.
32805 (ipa_is_param_used): Likewise.
32806 (ipa_node_params_t): Move into garbage collector. New methods insert
32808 (ipa_node_params_sum): Annotate wth GTY(()).
32809 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
32811 (ipa_load_from_parm_agg): Adjust declaration.
32812 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
32813 * ipa-profile.c (ipa_profile): Likewise.
32814 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
32815 (ipa_populate_param_decls): Make descriptors parameter garbage
32817 (ipa_dump_param): Adjust to descriptors being a pointer.
32818 (ipa_alloc_node_params): Likewise.
32819 (ipa_initialize_node_params): Likewise.
32820 (load_from_param_1): Make descriptors parameter garbage collected.
32821 (load_from_unmodified_param): Likewise.
32822 (load_from_param): Likewise.
32823 (ipa_load_from_parm_agg): Likewise.
32824 (ipa_node_params::~ipa_node_params): Removed.
32825 (ipa_free_all_node_params): Remove call to delete operator.
32826 (ipa_node_params_t::insert): New.
32827 (ipa_node_params_t::remove): Likewise.
32828 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
32829 copy known_csts and known_contexts vectors.
32830 (ipa_read_node_info): Adjust to descriptors being a pointer.
32831 (ipcp_modif_dom_walker): Make m_descriptors field garbage
32833 (ipcp_transform_function): Make descriptors variable garbage
32836 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
32838 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
32839 * config/i386/avx512dqintrin.h: Ditto.
32840 * config/i386/avx512fintrin.h: Ditto.
32841 * config/i386/i386.c: Handle new builtins.
32842 * config/i386/i386-builtin.def: Add new builtins.
32843 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
32844 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
32846 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32847 Martin Liska <mliska@suse.cz>
32849 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
32850 * asan.c (asan_expand_poison_ifn): Support stores and use
32851 appropriate ASAN report function.
32852 * internal-fn.c (expand_ASAN_POISON_USE): New function.
32853 * internal-fn.def (ASAN_POISON_USE): Declare.
32854 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
32855 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
32856 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
32857 ASAN_POISON calls w/o LHS.
32858 * tree-ssa.c (execute_update_addresses_taken): Create clobber
32859 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
32860 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
32861 * gimplify.c (asan_poison_variables): Add attribute
32862 use_after_scope_memory to variables that really needs to live
32864 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
32865 having the attribute.
32867 2017-01-23 Martin Liska <mliska@suse.cz>
32869 * asan.c (create_asan_shadow_var): New function.
32870 (asan_expand_poison_ifn): Likewise.
32871 * asan.h (asan_expand_poison_ifn): New declaration.
32872 * internal-fn.c (expand_ASAN_POISON): Likewise.
32873 * internal-fn.def (ASAN_POISON): New builtin.
32874 * sanopt.c (pass_sanopt::execute): Expand
32875 asan_expand_poison_ifn.
32876 * tree-inline.c (copy_decl_for_dup_finish): Make function
32878 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
32879 * tree-ssa.c (is_asan_mark_p): New function.
32880 (execute_update_addresses_taken): Rewrite local variables
32881 (identified just by use-after-scope as addressable) into SSA.
32883 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32885 * doc/install.texi (Specific): opensource.apple.com uses https
32886 now. Remove trailing slash.
32888 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32890 * README.Portability: Remove note on an Irix compatibility issue.
32892 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
32894 * gcov.c (INCLUDE_ALGORITHM): Define.
32895 (INCLUDE_VECTOR): Define.
32896 No longer include <vector> and <algorithm> directly.
32898 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
32900 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
32902 * doc/invoke.texi (Code Gen Options): Ditto.
32904 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
32907 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
32909 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
32911 rtl-optimization/79125
32912 * cprop.c (local_cprop_pass): Handle cases where we make an
32913 unconditional trap.
32915 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
32919 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
32920 read from, for big endian.
32922 2017-01-20 Jiong Wang <jiong.wang@arm.com>
32924 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
32925 register pauth builtins for LP64 only.
32927 2017-01-20 Marek Polacek <polacek@redhat.com>
32930 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
32933 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32935 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
32937 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
32938 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
32939 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
32941 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32944 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
32945 in big-endian mode when they are not a single duplicated value.
32947 2017-01-20 Richard Biener <rguenther@suse.de>
32949 * BASE-VER: Bump to 7.0.1.
32951 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32953 * omp-low.c (omplow_simd_context): New struct. Use it...
32954 (lower_rec_simd_input_clauses): ...here and...
32955 (lower_rec_input_clauses): ...here to hold common data. Adjust all
32956 references to idx, lane, max_vf, is_simt.
32958 2017-01-20 Graham Markall <graham.markall@embecosm.com>
32960 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
32963 2017-01-20 Martin Jambor <mjambor@suse.cz>
32965 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
32966 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
32968 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
32969 (GTFILES): Rename hsa.c to hsa-common.c.
32970 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
32971 * hsa-dump.c: Likewise.
32972 * hsa-gen.c: Likewise.
32973 * hsa-regalloc.c: Likewise.
32974 * ipa-hsa.c: Likewise.
32975 * omp-expand.c: Likewise.
32976 * omp-low.c: Likewise.
32977 * toplev.c: Likewise.
32979 2017-01-20 Marek Polacek <polacek@redhat.com>
32982 * doc/invoke.texi: Document -Wduplicated-branches.
32983 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
32984 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
32985 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
32986 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
32987 return 0 only when not OEP_LEXICOGRAPHIC.
32988 (fold_build_cleanup_point_expr): Use the expression
32989 location when building CLEANUP_POINT_EXPR.
32990 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
32991 * tree.c (add_expr): Handle error_mark_node.
32993 2017-01-20 Martin Liska <mliska@suse.cz>
32996 * tree-profile.c (init_ic_make_global_vars): Do not call
32998 (gimple_init_gcov_profiler): Likewise.
33000 2017-01-20 Martin Liska <mliska@suse.cz>
33003 * cgraph.h (maybe_create_reference): Remove argument and
33005 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33007 * ipa-cp.c (create_specialized_node): Likewise.
33008 * symtab.c (symtab_node::maybe_create_reference): Handle
33009 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33011 2017-01-20 Martin Liska <mliska@suse.cz>
33013 * read-rtl-function.c (function_reader::create_function): Use
33014 build_decl instread of build_decl_stat.
33016 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33018 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33019 * config/i386/avx512dqintrin.h: Ditto.
33020 * config/i386/avx512fintrin.h: Ditto.
33021 * config/i386/i386-builtin-types.def: Add new types.
33022 * config/i386/i386.c: Handle new types.
33023 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33024 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33025 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33026 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33027 (__builtin_ia32_kshiftridi): New.
33028 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33030 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33034 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33035 define to rs6000_init_stack_protect_guard.
33036 (rs6000_init_stack_protect_guard): New function.
33038 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33039 Yunqiang Su <yunqiang.su@imgtec.com>
33041 * config.gcc (supported_defaults): Add madd4.
33042 (with_madd4): Add validation.
33043 (all_defaults): Add madd4.
33044 * config/mips/mips.opt (mmadd4): New option.
33045 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33047 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33049 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33050 (ISA_HAS_FUSED_MADD4): Likewise.
33051 * doc/invoke.texi (-mmadd4): Document the new option.
33052 * doc/install.texi (--with-madd4): Document the new option.
33054 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33056 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33057 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33058 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33059 (aarch64_init_pauth_hint_builtins): New.
33060 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33061 (aarch64_expand_builtin): Expand new builtins.
33063 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33065 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33066 * combine-stack-adj.c (no_unhandled_cfa): Handle
33067 REG_CFA_TOGGLE_RA_MANGLE.
33068 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33069 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33070 info for return address signing.
33071 (aarch64_expand_epilogue): Likewise.
33073 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33075 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33076 * config/aarch64/aarch64-protos.h
33077 (aarch64_return_address_signing_enabled): New declaration.
33078 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33080 (aarch64_expand_prologue): Sign return address before it's pushed onto
33082 (aarch64_expand_epilogue): Authenticate return address fetched from
33084 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33085 (aarch64_attributes): New function attributes for "sign-return-address".
33086 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33087 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33088 ("*do_return"): Generate combined instructions according to key index.
33089 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33090 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33092 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33093 * config/aarch64/aarch64.opt (msign-return-address=): New.
33094 * doc/extend.texi (AArch64 Function Attributes): Documents
33095 "sign-return-address=".
33096 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33098 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33100 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33101 overall option summary.
33103 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33105 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33106 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33107 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33108 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33110 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33112 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33113 -mpower9-minmax by default for -mcpu=power9.
33114 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33115 128-bit floating point.
33117 2017-01-20 Alan Modra <amodra@gmail.com>
33119 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33120 optimizing for size.
33122 2017-01-20 Alan Modra <amodra@gmail.com>
33125 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33126 for strcmp and strncmp from corresponding builtin decl.
33128 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33130 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33131 instead of i386/rtems-64.h.
33132 * config/i386/rtems-64.h: Remove.
33134 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33138 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33140 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33142 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33144 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33145 Change int to HOST_WIDE_INT.
33146 * config/aarch64/aarch64-protos.h
33147 (aarch64_simd_gen_const_vector_dup): Likewise.
33148 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33150 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33152 * langhooks-def.h (lhd_type_for_size): New decl.
33153 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33154 * langhooks.c (lhd_type_for_size): New function, taken from
33157 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33159 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33160 define_bypass for CR latency.
33161 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33162 (power9-alu2): Add define_bypass for CR latency.
33164 (power9-mul): Update insn latency.
33165 (power9-mul-compare): Update insn latency, bypass latency and remove
33168 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33170 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33172 * config/aarch64/aarch64.md
33173 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33174 aarch64_nopcrelative_literal_loads.
33175 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33177 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33179 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33180 TARGET_LOONGSON_3A.
33181 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33183 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33186 * config.gcc (supported_defaults): Add lxc1-sxc1.
33187 (with_lxc1_sxc1): Add validation.
33188 (all_defaults): Add lxc1-sxc1.
33189 * config/mips/mips.opt (mlxc1-sxc1): New option.
33190 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33192 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33193 __mips_no_lxc1_sxc1.
33194 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33195 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33196 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33198 2017-01-19 Richard Biener <rguenther@suse.de>
33200 PR tree-optimization/72488
33201 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33202 sure to restore SSA info.
33203 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33205 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33207 PR rtl-optimization/79121
33208 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33209 of the inner type when shifting an extended value.
33211 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33214 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33215 interposable aliases.
33217 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33220 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33221 Use the evmergelohi instruction.
33222 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33223 (mov_sitf_e500_subreg8_2_be): Likewise.
33224 (mov_sitf_e500_subreg12_2_le): Likewise.
33225 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33226 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33227 (mov_sitf_e500_subreg8_2_le): Likewise.
33228 (mov_sitf_e500_subreg12_2_be): Likewise.
33230 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33232 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33233 attribute from vecsimple to vecperm.
33234 (altivec_vbpermq2): Likewise.
33236 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33239 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33241 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33242 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33243 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33244 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33245 case where N arg is SIZE_MAX.
33246 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33247 (cmpstrsi): Add pattern.
33249 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33251 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33252 __builtin_vec_revb builtins.
33253 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33254 built-in functions to support generation of the ISA 3.0 XXBR<x>
33255 vector byte reverse instructions.
33256 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33257 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33258 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33259 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33260 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33261 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33262 (P9V_BUILTIN_VEC_REVB): Likewise.
33263 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33264 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33265 (p9_xxbrq_v16qi): Likewise.
33266 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33267 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33268 (p9_xxbrh_v8hi): Likewise.
33269 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33270 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33271 vec_revb built-in functions.
33273 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33275 PR rtl-optimization/78952
33276 * config/i386/i386.md (any_extract): New code iterator.
33277 (*insvqi_2): Use any_extract for source operand.
33278 (*insvqi_3): Use any_shiftrt for source operand.
33280 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33282 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33284 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33286 2017-01-18 Matthias Klose <doko@ubuntu.com>
33288 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33290 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33292 * config/rs6000/altivec.h (vec_bperm): Change #define.
33293 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33294 (altivec_vbpermq2): New define_insn.
33295 (altivec_vbpermd): Likewise.
33296 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33297 function interface.
33298 (VBPERMD): Likewise.
33299 (VBPERM): New polymorphic function interface.
33300 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33301 Add entries for P9V_BUILTIN_VEC_VBPERM.
33302 * doc/extend.texi: Add interfaces for vec_bperm.
33304 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33306 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33307 first letter of error messages.
33308 (s390_resolve_overloaded_builtin): Likewise.
33309 * config/s390/s390.c (s390_expand_builtin): Likewise.
33310 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33311 (s390_valid_target_attribute_inner_p): Likewise.
33312 * config/s390/s390.md ("tabort"): Likewise.
33314 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33316 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33317 (ISA_AVOID_DIV_HILO): New macro.
33318 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33319 (ISA_HAS_DDIV): Likewise.
33321 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33323 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33325 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33327 * doc/invoke.texi (fabi-version): Spelling fix.
33329 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33332 * doc/invoke.texi (fabi-version): Mention mangling fix for
33335 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33338 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33340 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
33343 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
33344 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
33346 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
33348 (stack_protect_test): Ditto.
33349 * config/rs6000/rs6000.opt (mstack-protector-guard=,
33350 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
33352 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
33353 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
33354 -mstack-protector-guard-offset=.
33355 (RS/6000 and PowerPC Options): Ditto.
33357 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33359 * config/i386/i386.h (MASK_CLASS_P): New define.
33360 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
33361 there are no registers from different register sets also when
33362 mask registers are used. Update function comment.
33363 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33364 to (*k/*r) and (*k/*km) alternatives.
33366 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33368 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33369 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
33370 (EH_RETURN_HANDLER_RTX): New define.
33371 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
33372 Force frame pointer in EH return functions.
33373 (aarch64_expand_epilogue): Add barrier for eh_return.
33374 (aarch64_final_eh_return_addr): Remove.
33375 (aarch64_eh_return_handler_rtx): New function.
33376 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
33378 (aarch64_eh_return_handler_rtx): New prototype.
33380 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33382 * config/rs6000/altivec.h (vec_rlmi): New #define.
33383 (vec_vrlnm): Likewise.
33384 (vec_rlnm): Likewise.
33385 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
33386 (UNSPEC_VRLNM): Likewise.
33387 (VIlong): New mode iterator.
33388 (altivec_vrl<VI_char>mi): New define_insn.
33389 (altivec_vrl<VI_char>nm): Likewise.
33390 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
33392 (VRLDNM): Likewise.
33393 (RLNM): New polymorphic function entry.
33394 (VRLWMI): New monomorphic function entry.
33395 (VRLDMI): Likewise.
33396 (RLMI): New polymorphic function entry.
33397 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
33398 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
33399 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
33402 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33405 * dwarf2out.c (field_byte_offset): Restore the
33406 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
33407 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
33408 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
33411 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33414 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
33416 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33419 * dwarf2out.c (add_data_member_location_attribute): For constant
33420 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
33421 instead of DW_AT_data_member_location, DW_AT_bit_offset and
33422 DW_AT_byte_size attributes.
33424 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33426 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
33427 after forcing to constant memory when the code model is medium.
33429 2017-01-17 Julia Koval <julia.koval@intel.com>
33432 * config/i386/avx512fintrin.h
33433 (_mm512_i32gather_ps): Change __addr type to void const*.
33434 (_mm512_mask_i32gather_ps): Ditto.
33435 (_mm512_i32gather_pd): Ditto.
33436 (_mm512_mask_i32gather_pd): Ditto.
33437 (_mm512_i64gather_ps): Ditto.
33438 (_mm512_mask_i64gather_ps): Ditto.
33439 (_mm512_i64gather_pd): Ditto.
33440 (_mm512_mask_i64gather_pd): Ditto.
33441 (_mm512_i32gather_epi32): Ditto.
33442 (_mm512_mask_i32gather_epi32): Ditto.
33443 (_mm512_i32gather_epi64): Ditto.
33444 (_mm512_mask_i32gather_epi64): Ditto.
33445 (_mm512_i64gather_epi32): Ditto.
33446 (_mm512_mask_i64gather_epi32): Ditto.
33447 (_mm512_i64gather_epi64): Ditto.
33448 (_mm512_mask_i64gather_epi64): Ditto.
33449 (_mm512_i32scatter_ps): Change __addr type to void*.
33450 (_mm512_mask_i32scatter_ps): Ditto.
33451 (_mm512_i32scatter_pd): Ditto.
33452 (_mm512_mask_i32scatter_pd): Ditto.
33453 (_mm512_i64scatter_ps): Ditto.
33454 (_mm512_mask_i64scatter_ps): Ditto.
33455 (_mm512_i64scatter_pd): Ditto.
33456 (_mm512_mask_i64scatter_pd): Ditto.
33457 (_mm512_i32scatter_epi32): Ditto.
33458 (_mm512_mask_i32scatter_epi32): Ditto.
33459 (_mm512_i32scatter_epi64): Ditto.
33460 (_mm512_mask_i32scatter_epi64): Ditto.
33461 (_mm512_i64scatter_epi32): Ditto.
33462 (_mm512_mask_i64scatter_epi32): Ditto.
33463 (_mm512_i64scatter_epi64): Ditto.
33464 (_mm512_mask_i64scatter_epi64): Ditto.
33465 * config/i386/avx512pfintrin.h
33466 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
33467 (_mm512_mask_prefetch_i32gather_ps): Ditto.
33468 (_mm512_mask_prefetch_i64gather_pd): Ditto.
33469 (_mm512_mask_prefetch_i64gather_ps): Ditto.
33470 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
33471 (_mm512_prefetch_i32scatter_ps): Ditto.
33472 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
33473 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
33474 (_mm512_prefetch_i64scatter_pd): Ditto.
33475 (_mm512_prefetch_i64scatter_ps): Ditto.
33476 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
33477 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
33478 * config/i386/avx512vlintrin.h
33479 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
33480 (_mm_mmask_i32gather_ps): Ditto.
33481 (_mm256_mmask_i32gather_pd): Ditto.
33482 (_mm_mmask_i32gather_pd): Ditto.
33483 (_mm256_mmask_i64gather_ps): Ditto.
33484 (_mm_mmask_i64gather_ps): Ditto.
33485 (_mm256_mmask_i64gather_pd): Ditto.
33486 (_mm_mmask_i64gather_pd): Ditto.
33487 (_mm256_mmask_i32gather_epi32): Ditto.
33488 (_mm_mmask_i32gather_epi32): Ditto.
33489 (_mm256_mmask_i32gather_epi64): Ditto.
33490 (_mm_mmask_i32gather_epi64): Ditto.
33491 (_mm256_mmask_i64gather_epi32): Ditto.
33492 (_mm_mmask_i64gather_epi32): Ditto.
33493 (_mm256_mmask_i64gather_epi64): Ditto.
33494 (_mm_mmask_i64gather_epi64): Ditto.
33495 (_mm256_i32scatter_ps): Change __addr type to void*.
33496 (_mm256_mask_i32scatter_ps): Ditto.
33497 (_mm_i32scatter_ps): Ditto.
33498 (_mm_mask_i32scatter_ps): Ditto.
33499 (_mm256_i32scatter_pd): Ditto.
33500 (_mm256_mask_i32scatter_pd): Ditto.
33501 (_mm_i32scatter_pd): Ditto.
33502 (_mm_mask_i32scatter_pd): Ditto.
33503 (_mm256_i64scatter_ps): Ditto.
33504 (_mm256_mask_i64scatter_ps): Ditto.
33505 (_mm_i64scatter_ps): Ditto.
33506 (_mm_mask_i64scatter_ps): Ditto.
33507 (_mm256_i64scatter_pd): Ditto.
33508 (_mm256_mask_i64scatter_pd): Ditto.
33509 (_mm_i64scatter_pd): Ditto.
33510 (_mm_mask_i64scatter_pd): Ditto.
33511 (_mm256_i32scatter_epi32): Ditto.
33512 (_mm256_mask_i32scatter_epi32): Ditto.
33513 (_mm_i32scatter_epi32): Ditto.
33514 (_mm_mask_i32scatter_epi32): Ditto.
33515 (_mm256_i32scatter_epi64): Ditto.
33516 (_mm256_mask_i32scatter_epi64): Ditto.
33517 (_mm_i32scatter_epi64): Ditto.
33518 (_mm_mask_i32scatter_epi64): Ditto.
33519 (_mm256_i64scatter_epi32): Ditto.
33520 (_mm256_mask_i64scatter_epi32): Ditto.
33521 (_mm_i64scatter_epi32): Ditto.
33522 (_mm_mask_i64scatter_epi32): Ditto.
33523 (_mm256_i64scatter_epi64): Ditto.
33524 (_mm256_mask_i64scatter_epi64): Ditto.
33525 (_mm_i64scatter_epi64): Ditto.
33526 (_mm_mask_i64scatter_epi64): Ditto.
33527 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
33528 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
33529 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
33530 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
33531 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
33532 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
33533 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
33534 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
33535 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
33536 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
33537 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
33538 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
33539 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
33540 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
33541 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
33542 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
33543 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
33544 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
33545 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
33546 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
33547 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
33548 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
33549 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
33550 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
33551 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
33552 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
33553 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
33554 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
33555 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
33556 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
33557 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
33558 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
33559 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
33560 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
33561 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
33562 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
33563 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
33564 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
33565 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
33566 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
33567 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
33568 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
33569 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
33570 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
33571 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
33572 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
33573 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
33574 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
33575 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
33576 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
33577 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
33578 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
33579 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
33580 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
33581 definitions accordingly.
33583 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
33584 Kuan-Lin Chen <kuanlinchentw@gmail.com>
33587 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
33590 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
33593 * ira-conflicts.c (ira_build_conflicts): Update total conflict
33594 hard regs for inner regno.
33596 2017-01-17 Martin Liska <mliska@suse.cz>
33599 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
33600 assumption and add comment.
33602 2017-01-17 Nathan Sidwell <nathan@acm.org>
33604 * ipa-visibility.c (localize_node): New function, broken out of ...
33605 (function_and_variable_visibility): ... here. Call it.
33607 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33609 PR middle-end/77445
33610 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
33611 correctly set frequency of oudgoing edge.
33612 (duplicate_thread_path): Fix profile updating.
33614 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33617 * configure.ac: Add GCC_BASE_VER.
33618 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
33619 version from BASE-VER file.
33620 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
33621 (gcc.o): Depend on $(BASEVER).
33622 * common.opt (dumpfullversion): New option.
33623 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
33624 * doc/invoke.texi: Document -dumpfullversion.
33625 * doc/install.texi: Document --with-gcc-major-version-only.
33626 * configure: Regenerated.
33628 2017-01-17 Richard Biener <rguenther@suse.de>
33630 PR tree-optimization/71433
33631 * tree-vrp.c (register_new_assert_for): Merge same asserts
33632 on all incoming edges.
33633 (process_assert_insertions_for): Handle insertions at the
33636 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
33638 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
33639 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
33641 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
33644 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
33647 2017-01-17 Alan Modra <amodra@gmail.com>
33650 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
33651 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
33652 symbolic stack limit when pic.
33654 2017-01-16 Martin Sebor <msebor@redhat.com>
33656 PR tree-optimization/78608
33657 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
33659 2017-01-16 Jeff Law <law@redhat.com>
33662 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
33663 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
33664 for several include directories that may be relative to sysroot.
33665 * config/i386/x-mingw32 (gplus_includedir): Define.
33666 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
33667 (native_system_includedir): Likewise.
33668 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
33669 override if TARGET_SYSTEM_ROOT is defined.
33670 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
33672 PR tree-optimization/79090
33673 PR tree-optimization/33562
33674 PR tree-optimization/61912
33675 PR tree-optimization/77485
33676 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
33677 and computed trims into the dump file.
33679 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33681 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
33683 2017-01-16 Jakub Jelinek <jakub@redhat.com>
33686 * gimplify.c (gimplify_init_constructor): If want_value and
33687 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
33691 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
33692 sequence. Formatting fixes.
33693 (doloop_optimize): Formatting fixes.
33696 * gcc.c (debug_level_greater_than_spec_func): New function.
33697 (static_spec_functions): Add debug-level-gt spec function.
33698 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
33700 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33701 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33702 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
33703 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
33704 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
33705 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
33707 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
33709 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
33710 QImode fixups to general and mask registers only.
33712 2017-01-16 Carl Love <cel@us.ibm.com>
33714 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
33715 for built-in functions
33716 vector signed char vec_nabs (vector signed char)
33717 vector signed short vec_nabs (vector signed short)
33718 vector signed int vec_nabs (vector signed int)
33719 vector signed long long vec_nabs (vector signed long long)
33720 vector float vec_nabs (vector float)
33721 vector double vec_nabs (vector double)
33722 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
33724 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
33725 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
33726 * doc/extend.texi: Update the documentation file for the new built-in
33729 2017-01-16 Martin Sebor <msebor@redhat.com>
33731 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
33734 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33736 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
33737 UNSPEC_VSX__XXSPLTD to require special splat handling.
33739 2017-01-16 David Malcolm <dmalcolm@redhat.com>
33742 * system.h: Poison strndup.
33744 2017-01-16 Alan Modra <amodra@gmail.com>
33747 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
33750 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
33752 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
33754 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
33756 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
33757 call recog here. Assert that INSN_CODE (insn) is non-negative.
33759 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33762 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
33764 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
33765 in the currently scheduled RTL fragment.
33767 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33769 PR rtl-optimization/78751
33770 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
33773 2017-01-14 Jeff Law <law@redhat.com>
33775 PR tree-optimization/79090
33776 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
33777 variable length stores.
33778 (compute_trims): Delete dead assignment to *trim_tail.
33779 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
33782 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
33784 PR rtl-optimization/78626
33785 PR rtl-optimization/78727
33786 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
33787 of a block, and split such blocks after everything else is finished.
33789 2017-01-14 Alan Modra <amodra@gmail.com>
33792 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
33793 target legitimate_combined_insn.
33794 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
33795 (rs6000_legitimate_combined_insn): New function.
33796 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
33798 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
33799 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
33800 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
33802 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
33804 * doc/frontends.texi (G++ and GCC): Remove references to Java.
33806 2017-01-13 Jeff Law <law@redhat.com>
33808 PR tree-optimization/33562
33809 PR tree-optimization/61912
33810 PR tree-optimization/77485
33811 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
33813 (delete_dead_assignment): Likewise.
33814 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
33815 statement to delete_dead_call and delete_dead_assignment.
33817 2017-01-13 David Malcolm <dmalcolm@redhat.com>
33820 * substring-locations.c (format_warning_va): Strengthen case 1 so
33821 that both endpoints of the substring must be within the format
33822 range for just the substring to be printed.
33824 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
33826 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
33827 * config/i386/i386.c (ix86_target_string): Add missing options
33828 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
33829 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
33830 flags_other and ix86_target_other to flags2_other. Display unknown
33832 (ix86_valid_target_attribute_inner_p): Add missing options and
33833 reorder options by implied ISAs, as in ix86_target_string.
33835 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33837 * hash-table.h (hash_table::too_empty_p): New function.
33838 (hash_table::expand): Use it.
33839 (hash_table::traverse): Likewise.
33840 (hash_table::empty_slot): Use sizeof (value_type) instead of
33841 sizeof (PTR) to convert bytes to elements. Shrink the table
33842 if the current size is excessive for the current number of
33845 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33847 * ira-costs.c (record_reg_classes): Break from the inner loop
33848 early once alt_fail is known to be true. Update outer loop
33849 handling accordingly.
33851 2017-01-13 Jeff Law <law@redhat.com>
33853 * tree-ssa-dse.c (decrement_count): New function.
33854 (increment_start_addr, maybe_trim_memstar_call): Likewise.
33855 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
33856 when we know the partially dead statement is a mem* function.
33858 PR tree-optimization/61912
33859 PR tree-optimization/77485
33860 * tree-ssa-dse.c: Include expr.h.
33861 (maybe_trim_constructor_store): New function.
33862 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
33864 PR tree-optimization/33562
33865 PR tree-optimization/61912
33866 PR tree-optimization/77485
33867 * doc/invoke.texi: Document new dse-max-object-size param.
33868 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
33869 * tree-ssa-dse.c: Include params.h.
33870 (dse_store_status): New enum.
33871 (initialize_ao_ref_for_dse): New, partially extracted from
33873 (valid_ao_ref_for_dse, normalize_ref): New.
33874 (setup_live_bytes_from_ref, compute_trims): Likewise.
33875 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
33876 (maybe_trim_partially_dead_store): Likewise.
33877 (maybe_trim_complex_store): Likewise.
33878 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
33879 Track what bytes live from the original store. Return tri-state
33880 for dead, partially dead or live.
33881 (dse_dom_walker): Add constructor, destructor and new private members.
33882 (delete_dead_call, delete_dead_assignment): New extracted from
33884 (dse_optimize_stmt): Make a member of dse_dom_walker.
33885 Use initialize_ao_ref_for_dse.
33887 PR tree-optimization/33562
33888 PR tree-optimization/61912
33889 PR tree-optimization/77485
33890 * sbitmap.h (bitmap_count_bits): Prototype.
33891 (bitmap_clear_range, bitmap_set_range): Likewise.
33892 * sbitmap.c (bitmap_clear_range): New function.
33893 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
33895 2017-01-13 Martin Liska <mliska@suse.cz>
33898 * function.c (set_cfun): Add new argument force.
33899 * function.h (set_cfun): Likewise.
33900 * ipa-inline-transform.c (inline_call): Use the function when
33901 strict alising from is dropped for function we inline to.
33903 2017-01-13 Richard Biener <rguenther@suse.de>
33905 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
33906 for dumping GIMPLE INTEGER_CSTs.
33908 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33910 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
33911 to 201112L since C++17.
33913 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
33916 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
33917 if -fsanitize=kernel-address is present.
33919 2017-01-13 Richard Biener <rguenther@suse.de>
33921 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
33922 as _Literal ( type ) number in case usual suffixes do not
33923 preserve all information.
33925 2017-01-13 Richard Biener <rguenther@suse.de>
33927 PR tree-optimization/77283
33928 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
33929 and ssa-iterators.h.
33930 (is_feasible_trace): Implement a cost model based on joiner
33933 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
33936 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
33937 char or short to __float128/_Float128 directly.
33939 2017-01-12 Martin Sebor <msebor@redhat.com>
33941 to -Wformat-overflow.
33942 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
33943 (min_bytes_remaining): Same.
33944 (get_string_length): Same.
33945 (format_string): Same.
33946 (format_directive): Same.
33948 (pass_sprintf_length::handle_gimple_call): Same.
33950 2017-01-12 Jakub Jelinek <jakub@redhat.com>
33952 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
33953 info.nowrite calls with no lhs that can't throw. Return bool
33954 whether gsi_remove has been called or not.
33955 (pass_sprintf_length::handle_gimple_call): Return bool whether
33956 try_substitute_return_value called gsi_remove. Formatting fix.
33957 (pass_sprintf_length::execute): Don't use gsi_remove if
33958 handle_gimple_call returned true.
33961 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
33962 be removed due to side-effects, don't remove following barrier nor
33963 turn the successor edge into fallthru edge.
33965 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33968 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
33969 element-reversing loads and stores as not swappable.
33971 2017-01-12 Nathan Sidwell <nathan@acm.org>
33972 Nicolai Stange <nicstange@gmail.com>
33974 * combine.c (try_combine): Don't ignore result of overlap checking
33975 loop. Combine overlap & asm check into single loop.
33977 2017-01-12 Richard Biener <rguenther@suse.de>
33979 * tree-pretty-print.c (dump_generic_node): Provide -gimple
33980 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
33982 2017-01-12 Richard Biener <rguenther@suse.de>
33984 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
33985 and TS_TARGET_OPTION directly derive from TS_BASE.
33986 * tree-core.h (tree_optimization_option): Derive from tree_base.
33987 (tree_target_option): Likewise.
33989 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
33991 * config/i386/i386.c (memory_address_length): Increase len
33992 only when rip_relative_addr_p returns false.
33994 2017-01-11 Julia Koval <julia.koval@intel.com>
33996 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
33997 (OPTION_MASK_ISA_SGX_SET): New.
33998 (ix86_handle_option): Handle OPT_msgx.
33999 * config.gcc: Added sgxintrin.h.
34000 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34001 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34002 * config/i386/i386.c (ix86_target_string): Add -msgx.
34004 (ix86_option_override_internal): Handle new options.
34005 (ix86_valid_target_attribute_inner_p): Add sgx.
34006 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34007 * config/i386/i386.opt: Add msgx.
34008 * config/i386/sgxintrin.h: New file.
34009 * config/i386/x86intrin.h: Add sgxintrin.h.
34011 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34014 * fold-const.c (maybe_nonzero_address): Return 1 for function
34016 (tree_single_nonzero_warnv_p): Don't handle function local objects
34020 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34023 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34026 * opts.c: Include "spellcheck.h"
34027 (struct string_fragment): New struct.
34028 (struct edit_distance_traits<const string_fragment &>): New
34030 (get_closest_sanitizer_option): New function.
34031 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34033 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34035 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34037 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34038 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34039 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34040 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34041 for initial die_offset if dwarf_split_debug_info.
34042 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34043 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34045 (output_skeleton_debug_sections): Formatting fix. Use
34046 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34047 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34049 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34051 * config/arm/cortex-a53.md: Add bypasses for
34052 cortex_a53_r2f_cvt.
34053 (cortex_a53_r2f): Only use for transfers.
34054 (cortex_a53_f2r): Likewise.
34055 (cortex_a53_r2f_cvt): Add reservation for conversions.
34056 (cortex_a53_f2r_cvt): Likewise.
34058 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34060 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34061 to all inlined functions, change static to extern.
34063 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34066 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34068 (arm_assemble_integer): Likewise.
34070 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34072 * config.gcc: Use new awk script to check CPU, FPU and architecture
34073 parameters for --with-... options.
34074 * config/arm/parsecpu.awk: New file
34075 * config/arm/arm-cpus.in: New file.
34076 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34078 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34080 * config/arm/t-arm: Update dependency rules.
34081 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34082 of processing .def files.
34083 * config/arm/genopt.sh: Deleted.
34084 * config/arm/gentune.sh: Deleted.
34085 * config/arm/arm-cores.def: Deleted.
34086 * config/arm/arm-arches.def: Deleted.
34087 * config/arm/arm-fpus.def: Deleted.
34088 * config/arm/arm-tune.md: Regenerated.
34089 * config/arm/arm-tables.opt: Regenerated.
34090 * config/arm/arm-cpu.h: New generated file.
34091 * config/arm/arm-cpu-data.h: New generated file.
34092 * config/arm/arm-cpu-cdata.h: New generated file.
34094 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34097 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34099 (input_varpool_node): Unpack dynamically_initialized bit.
34101 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34103 PR rtl-optimization/79032
34104 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34105 the alignment of the adjusted memory reference against that of MODE,
34106 instead of the alignment of the original memory reference.
34108 2017-01-11 Martin Jambor <mjambor@suse.cz>
34110 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34112 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34113 decorated functions.
34115 2017-01-11 Richard Biener <rguenther@suse.de>
34117 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34118 set range/nonnull info for PHI results. Do not set it on
34119 stmts marked for removal.
34121 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34123 * expr.c (store_field): In the bitfield case, fetch the return value
34124 from the registers before applying a single big-endian adjustment.
34125 Always do a final load for a BLKmode value not larger than a word.
34127 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34130 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34131 that we correctly handle column numbers greater than
34132 LINE_MAP_MAX_COLUMN_NUMBER.
34134 2017-01-10 Martin Sebor <msebor@redhat.com>
34136 PR middle-end/78245
34137 * gimple-ssa-sprintf.c (get_destination_size): Call
34138 {init,fini}object_sizes.
34139 * tree-object-size.c (addr_object_size): Adjust.
34140 (pass_through_call): Adjust.
34141 (pass_object_sizes::execute): Adjust.
34142 * tree-object-size.h (fini_object_sizes): Declare.
34144 2017-01-10 Martin Sebor <msebor@redhat.com>
34146 PR tree-optimization/78775
34147 * builtins.c (get_size_range): Move...
34148 * calls.c: ...to here.
34149 (alloc_max_size): Accept zero argument.
34150 (operand_signed_p): Remove.
34151 (maybe_warn_alloc_args_overflow): Call get_size_range.
34152 * calls.h (get_size_range): Declare.
34154 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34156 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34157 from TI's devices.csv file as of September 2016.
34158 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34160 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34162 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34163 * doc/invoke.texi: Likewise.
34164 * doc/md.texi: Likewise.
34165 * doc/objc.texi: Likewise.
34167 2017-01-10 Joshua Conner <joshconner@google.com>
34169 * config/arm/fuchsia-elf.h: New file.
34170 * config/fuchsia.h: New file.
34171 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34172 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34174 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34176 2016-01-10 Richard Biener <rguenther@suse.de>
34178 PR tree-optimization/79034
34179 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34180 Propagate out degenerate PHIs in the joiner.
34182 2017-01-10 Martin Liska <mliska@suse.cz>
34184 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34185 (sort_congruence_classes_by_decl_uid): Likewise.
34186 (sort_congruence_class_groups_by_decl_uid): Likewise.
34187 (sem_item_optimizer::merge_classes): Sort class, groups in these
34188 classes and members in the groups by DECL_UID of declarations.
34189 This would make merge operations stable.
34191 2017-01-10 Martin Liska <mliska@suse.cz>
34193 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34194 usage of m_classes_vec.
34195 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34196 (sem_item_optimizer::get_group_by_hash): Likewise.
34197 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34198 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34199 (sem_item_optimizer::verify_classes): Likewise.
34200 (sem_item_optimizer::process_cong_reduction): Likewise.
34201 (sem_item_optimizer::dump_cong_classes): Likewise.
34202 (sem_item_optimizer::merge_classes): Likewise.
34203 * ipa-icf.h (congruence_class_hash): Rename from
34204 congruence_class_group_hash. Remove declaration of m_classes_vec.
34206 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34208 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34209 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34210 * config.gcc: Add avx512vpopcntdqintrin.h.
34211 * config/i386/avx512vpopcntdqintrin.h: New.
34212 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34213 * config/i386/i386-builtin-types.def: Add new types.
34214 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34215 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34216 __builtin_ia32_vpopcountq_v8di_mask): New.
34217 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34218 __AVX512VPOPCNTDQ__.
34219 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34220 (PTA_AVX512VPOPCNTDQ): Define.
34221 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34222 TARGET_AVX512VPOPCNTDQ_P): Define.
34223 * config/i386/i386.opt: Add mavx512vpopcntdq.
34224 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34225 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34227 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34229 PR middle-end/77484
34230 * predict.def (PRED_CALL): Set to 67.
34232 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34234 * expr.c (store_field): In the bitfield case, if the value comes from
34235 a function call and is of an aggregate type returned in registers, do
34236 not modify the field mode; extract the value in all cases if the mode
34237 is BLKmode and the size is not larger than a word.
34239 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34242 * config/i386/cpuid.h: Fix undefined behavior.
34244 2017-01-04 Jeff Law <law@redhat.com>
34246 PR tree-optimization/79007
34247 PR tree-optimization/67955
34248 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34249 conservative for pt.null when flag_non_call_exceptions is on.
34251 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34253 PR translation/79019
34254 PR translation/79020
34255 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34256 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34257 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34259 * config/avr/avr.opt (maccumulate-args): Likewise.
34260 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34261 * common.opt (freport-bug): Likewise.
34262 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34263 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34264 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34265 translatable string.
34266 * config/i386/i386.c (function_value_32): Likewise.
34267 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34268 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34270 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34271 * common/config/msp430/msp430-common.c (msp430_handle_option):
34273 * symtab.c (symtab_node::verify_base): Likewise.
34274 * opts.c (set_debug_level): Likewise.
34275 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34276 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34277 missing whitespace to translatable strings.
34278 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34279 * config/sh/superh.h: Likewise.
34280 * config/i386/xopintrin.h: Likewise.
34281 * config/i386/znver1.md: Likewise.
34282 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34283 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34284 * double-int.h (struct double_int): Likewise.
34285 * double-int.c (div_and_round_double): Likewise.
34286 * wide-int.cc: Likewise.
34287 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34288 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34289 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34290 (crossjumps_occurred): ... this.
34291 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34294 PR tree-optimization/78899
34295 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34296 returning bool return struct loop *, NULL for failure and the new
34298 (versionable_outer_loop_p): Don't version outer loop if it has
34299 dont_vectorized bit set.
34300 (tree_if_conversion): When versioning outer loop, ensure
34301 tree_if_conversion is performed also on the inner loop of the
34302 non-vectorizable outer loop copy.
34303 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34304 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34305 prevent vectorization of it.
34306 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34307 the outer loop vectorization of the non-scalar version is attempted
34308 before vectorization of the inner loop in scalar version. If
34309 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34310 vectorization of its inner loop.
34311 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34312 has 2 inner loops, rename also on edges from bb whose single pred
34313 is outer_loop->header. Fix typo in function comment.
34315 2017-01-09 Martin Sebor <msebor@redhat.com>
34318 * asan.c (asan_emit_stack_protection): Increase local buffer size
34319 to avoid snprintf truncation warning.
34321 2017-01-09 Andrew Pinski <apinski@cavium.com>
34323 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34324 to reference thunderx2t99 for the tuning structure
34325 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34327 (thunderx2t99_extra_costs): This.
34328 * config/aarch64/aarch64-tune.md: Regenerate.
34329 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34330 (vulcan_addrcost_table): This.
34331 (vulcan_regmove_cost): Rename to ...
34332 (thunderx2t99_regmove_cost): This.
34333 (vulcan_vector_cost): Rename to ...
34334 (thunderx2t99_vector_cost): this.
34335 (vulcan_branch_cost): Rename to ...
34336 (thunderx2t99_branch_cost): This.
34337 (vulcan_tunings): Rename to ...
34338 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34339 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34341 2017-01-09 Martin Jambor <mjambor@suse.cz>
34345 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
34346 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
34347 (propagate_vr_accross_jump_function): Use the above function for all
34348 value range computations for pass-through jump functions and type
34349 converasion from explicit value range values.
34350 (ipcp_propagate_stage): Do not attempt to deduce types of formal
34351 parameters from TYPE_ARG_TYPES.
34352 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
34353 (ipa_write_node_info): Stream type of the actual argument.
34354 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
34356 2017-01-09 Martin Liska <mliska@suse.cz>
34359 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
34360 (lookup_compiler): Do not show error message with have_E.
34362 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34364 PR tree-optimization/78938
34365 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34366 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34367 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34368 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
34371 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34373 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
34376 2017-01-09 Richard Biener <rguenther@suse.de>
34378 PR tree-optimization/78997
34379 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
34380 name condition properly.
34382 2017-01-09 Richard Biener <rguenther@suse.de>
34385 * dwarf2out.c (is_cxx): New overload with context.
34386 (is_naming_typedef_decl): Use it.
34388 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34390 * invoke.texi (Option Summary): Correct spacing in option lists
34391 and add line breaks to fix over-long lines.
34393 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34395 PR middle-end/17660
34397 * extend.texi (Common Variable Attributes): Add xref to GCC
34398 Internals manual to explain mode attribute keywords.
34400 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34403 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
34404 and Preprocessor Options.
34405 (Options for Linking): Document -pthread here....
34406 (RS/6000 and PowerPC Options): ...not here.
34407 (Solaris 2 Options): ...or here.
34408 * doc/cppopts.texi: Document -pthread.
34410 2017-01-08 Martin Sebor <msebor@redhat.com>
34412 PR middle-end/77708
34413 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
34414 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
34415 New member functions.
34416 (format_directive): Used them.
34418 (pass_sprintf_length::handle_gimple_call): Same.
34419 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
34420 to avoid truncation for any argument.
34421 (extract_affine_mul): Same.
34422 * tree.c (get_file_function_name): Same.
34424 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34426 PR middle-end/77484
34427 * predict.def (PRED_INDIR_CALL): Set to 86.
34429 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34431 PR preprocessor/54124
34432 * doc/cppopts.texi: Reformat -d subtable to list the full name
34433 of the options. Add cross-reference to the docs for the general
34434 compiler -d options.
34435 * doc/invoke.texi (Developer Options): Add cross-reference to the
34436 preprocessor-specific -d option documentation.
34438 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34440 PR preprocessor/13498
34441 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
34442 redudant material, and reflect new command-line options.
34443 (System Headers): Likewise.
34445 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34447 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
34448 -isystem, and -idirafter. Copy-edit.
34449 * doc/cppopts.texi: Copy-edit. Remove contradiction about
34450 default for -ftrack-macro-expansion. Delete obsolete and
34451 badly-formatted implementation details about -fdebug-cpp output.
34452 * doc/cppwarnopts.texi: Copy-edit.
34454 2017-01-07 David Malcolm <dmalcolm@redhat.com>
34457 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34458 that the transition from a max line width >= 1<<10 to narrower
34459 lines works correctly.
34461 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
34463 * doc/options.texi (PerFunction): New.
34464 * opt-functions.awk (switch_flags): Map both Optimization and
34465 PerFunction to CL_OPTIMIZATION.
34466 * opth-gen.awk: Test for PerFunction flag along with
34468 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
34469 it only when the latter is present. Skip those that don't in
34470 the hash function generator.
34471 * common.opt (fvar-tracking): Mark as PerFunction instead of
34473 (fvar-tracking-assignments): Likewise.
34474 (fvar-tracking-assignments-toggle): Likewise.
34475 (fvar-tracking-uninit): Likewise.
34477 2017-01-07 Jakub Jelinek <jakub@redhat.com>
34479 PR translation/79018
34480 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
34483 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
34486 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
34487 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
34488 TARGET_LONG_JUMP_TABLE_OFFSETS.
34489 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
34490 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
34491 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
34492 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
34493 * config/m68k/m68k.md (tablejump expander): Likewise.
34494 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
34495 TARGET_LONG_JUMP_TABLE_OFFSETS.
34496 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
34497 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
34499 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
34500 David Holsgrove <david.holsgrove@xilinx.com>
34502 * common/config/microblaze/microblaze-common.c
34503 (TARGET_EXCEPT_UNWIND_INFO): Remove.
34504 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
34506 * config/microblaze/microblaze.c (microblaze_must_save_register)
34507 (microblaze_expand_epilogue, microblaze_return_addr): Handle
34509 (microblaze_eh_return): New function.
34510 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
34511 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
34512 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
34513 * config/microblaze/microblaze.md (eh_return): New pattern.
34515 2017-01-06 Jakub Jelinek <jakub@redhat.com>
34517 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
34518 GCC_DIAGNOSTIC_STRINGIFY): Define.
34520 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
34522 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34524 * config/arm/arm.md (<mcrr>): New.
34526 * config/arm/arm.c (arm_arch5te): New.
34527 (arm_option_override): Set arm_arch5te.
34528 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
34530 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
34531 (arm_mcrr_qualifiers): ... this. New.
34532 (MRRC_QUALIFIERS): Define to...
34533 (arm_mrrc_qualifiers): ... this. New.
34534 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
34536 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
34537 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
34538 (MRRCI, mrrc, MRRC): New.
34539 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
34540 VUNSPEC_MRRC2): New.
34542 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34544 * config/arm/arm.md (<mcr>): New.
34546 * config/arm/arm.c (arm_coproc_builtin_available): Add
34547 support for mcr, mrc, mcr2 and mrc2.
34548 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
34549 (arm_mcr_qualifiers): ... this. New.
34550 (MRC_QUALIFIERS): Define to ...
34551 (arm_mrc_qualifiers): ... this. New.
34552 (MCR_QUALIFIERS): Define to ...
34553 (arm_mcr_qualifiers): ... this. New.
34554 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
34556 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
34557 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
34558 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
34559 VUNSPEC_MRC2): New.
34561 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34563 * config/arm/arm.md (*ldc): New.
34567 * config/arm/arm.c (arm_coproc_builtin_available): Add
34568 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
34569 (arm_coproc_ldc_stc_legitimate_address): New.
34570 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
34571 'qualifier_const_pointer'.
34572 (LDC_QUALIFIERS): Define to...
34573 (arm_ldc_qualifiers): ... this. New.
34574 (STC_QUALIFIERS): Define to...
34575 (arm_stc_qualifiers): ... this. New.
34576 * config/arm/arm-protos.h
34577 (arm_coproc_ldc_stc_legitimate_address): New.
34578 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
34579 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
34580 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
34581 stc2, stcl, stc2l): New.
34582 * config/arm/constraints.md (Uz): New.
34583 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
34584 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
34585 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
34586 VUNSPEC_STC2L): New.
34588 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34590 * config/arm/arm.md (<cdp>): New.
34591 * config/arm/arm.c (neon_const_bounds): Rename this ...
34592 (arm_const_bounds): ... this.
34593 (arm_coproc_builtin_available): New.
34594 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
34595 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
34596 (CDP_QUALIFIERS): Define to...
34597 (arm_cdp_qualifiers): ... this. New.
34599 (arm_expand_builtin_args): Add case for 6 arguments.
34600 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
34601 (arm_const_bounds): ... this.
34602 (arm_coproc_builtin_available): New.
34603 * config/arm/arm_acle.h (__arm_cdp): New.
34605 * config/arm/arm_acle_builtins.def (cdp): New.
34607 * config/arm/iterators.md (CDPI,CDP,cdp): New.
34608 * config/arm/neon.md: Rename all 'neon_const_bounds' to
34609 'arm_const_bounds'.
34610 * config/arm/types.md (coproc): New.
34611 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
34612 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
34613 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
34614 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
34616 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34618 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
34619 (UBINOP_QUALIFIERS): New.
34621 (acle_builtin_data): New. Change comment.
34622 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
34623 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
34624 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
34625 arm_acle_builtins.def.
34626 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
34627 (arm_init_acle_builtins): New.
34628 (CRC32_BUILTIN): Remove.
34629 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
34630 crc32cb, crc32ch and crc32cw.
34631 (arm_init_crc32_builtins): Remove.
34632 (arm_init_builtins): Use arm_init_acle_builtins rather
34633 than arm_init_crc32_builtins.
34634 (arm_expand_acle_builtin): New.
34635 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
34636 * config/arm/arm_acle_builtins.def: New.
34638 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34640 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
34641 (arm_builtin_datum): ... this.
34642 (arm_init_neon_builtin): Rename to ...
34643 (arm_init_builtin): ... this. Add a new parameters PREFIX
34644 and USE_SIG_IN_NAME.
34645 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
34646 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
34647 'arm_builtin_datum'.
34648 (arm_init_vfp_builtins): Likewise.
34649 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
34650 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
34651 (arm_expand_neon_args): Rename to ...
34652 (arm_expand_builtin_args): ... this. Rename builtin_arg
34653 enum values and differentiate between ARG_BUILTIN_MEMORY
34654 and ARG_BUILTIN_NEON_MEMORY.
34655 (arm_expand_neon_builtin_1): Rename to ...
34656 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
34657 values, arm_expand_builtin_args and add bool parameter NEON.
34658 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
34659 (arm_expand_vfp_builtin): Likewise.
34660 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
34662 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34664 PR middle-end/77484
34665 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
34666 * predict.c (tree_estimate_probability_bb): Reverse direction of
34667 polymorphic call predictor.
34669 2017-01-06 David Malcolm <dmalcolm@redhat.com>
34671 * passes.c (execute_one_pass): Split out pass-skipping logic into...
34672 (determine_pass_name_match): ...this new function and...
34673 (should_skip_pass_p): ...this new function.
34675 2017-01-06 Nathan Sidwell <nathan@acm.org>
34677 * ipa-visibility.c (function_and_variable_visibility): Reformat
34678 comments and long lines. Remove extrneous if.
34679 * symtab.c (symtab_node::make_decl_local): Fix code format.
34680 (symtab_node::set_section_for_node): Fix comment typo.
34682 2017-01-06 Martin Liska <mliska@suse.cz>
34685 * lra-constraints.c: Rename invariant to lra_invariant.
34686 * predict.c (set_even_probabilities): Initialize e to NULL.
34688 2017-01-05 Martin Sebor <msebor@redhat.com>
34690 PR tree-optimization/78910
34691 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
34692 (format_integer): Correct off-by-one error in the handling
34693 of precision with negative numbers in signed conversions..
34695 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
34697 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
34699 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34701 PR tree-optimization/71016
34702 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
34703 factor_out_conditional_conversion. Formatting fix.
34704 (factor_out_conditional_conversion): Add cond_stmt argument.
34705 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
34706 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
34709 2017-01-05 David Malcolm <dmalcolm@redhat.com>
34711 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
34712 read-rtl-function.o, and selftest-rtl.o.
34713 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
34714 (selftest::aarch64_test_loading_full_dump): New function.
34715 (selftest::aarch64_run_selftests): New function.
34716 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
34717 selftest::aarch64_run_selftests.
34718 * config/i386/i386.c
34719 (selftest::ix86_test_loading_dump_fragment_1): New function.
34720 (selftest::ix86_test_loading_call_insn): New function.
34721 (selftest::ix86_test_loading_full_dump): New function.
34722 (selftest::ix86_test_loading_unspec): New function.
34723 (selftest::ix86_run_selftests): Call the new functions.
34724 * emit-rtl.c (maybe_set_max_label_num): New function.
34725 * emit-rtl.h (maybe_set_max_label_num): New decl.
34726 * function.c (instantiate_decls): Guard call to
34727 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
34728 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
34730 * gensupport.c (gen_reader::gen_reader): Pass "false"
34731 for new "compact" param of rtx_reader.
34732 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
34733 rather than an empty string for NULL strings.
34734 * read-md.c: Potentially include config.h rather than bconfig.h.
34735 Wrap include of errors.h with #ifdef GENERATOR_FILE.
34736 (have_error): New global, copied from errors.c.
34737 (md_reader::read_name): Rename to...
34738 (md_reader::read_name_1): ...this, adding "out_loc" param,
34739 and converting "missing name or number" to returning false, rather
34741 (md_reader::read_name): Reimplement in terms of read_name_1.
34742 (md_reader::read_name_or_nil): New function.
34743 (md_reader::read_string): Handle "(nil)" by returning NULL.
34744 (md_reader::md_reader): Add new param "compact".
34745 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
34746 (md_reader::read_file): New method.
34747 * read-md.h (md_reader::md_reader): Add new param "compact".
34748 (md_reader::read_file): New method.
34749 (md_reader::is_compact): New accessor.
34750 (md_reader::read_name): Convert return type from void to file_location.
34751 (md_reader::read_name_or_nil): New decl.
34752 (md_reader::read_name_1): New decl.
34753 (md_reader::m_compact): New field.
34754 (noop_reader::noop_reader): Pass "false" for new "compact" param
34756 (rtx_reader::rtx_reader): Add new "compact" param.
34757 (rtx_reader::read_rtx_operand): Make virtual and convert return
34758 type from void to rtx.
34759 (rtx_reader::read_until): New decl.
34760 (rtx_reader::handle_any_trailing_information): New virtual function.
34761 (rtx_reader::postprocess): New virtual function.
34762 (rtx_reader::finalize_string): New virtual function.
34763 (rtx_reader::m_in_call_function_usage): New field.
34764 (rtx_reader::m_reuse_rtx_by_id): New field.
34765 * read-rtl-function.c: New file.
34766 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
34767 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
34768 (selftest::verify_three_block_rtl_cfg): New decl.
34769 * read-rtl-function.h: New file.
34770 * read-rtl.c: Potentially include config.h rather than bconfig.h.
34771 For host, include function.h, memmodel.h, and emit-rtl.h.
34772 (one_time_initialization): New function.
34773 (struct compact_insn_name): New struct.
34774 (compact_insn_names): New array.
34775 (find_code): Handle insn codes in compact dumps.
34776 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
34777 (bind_subst_iter_and_attr): Likewise.
34778 (add_condition_to_string): Likewise.
34779 (add_condition_to_rtx): Likewise.
34780 (apply_attribute_uses): Likewise.
34781 (add_current_iterators): Likewise.
34782 (apply_iterators): Likewise.
34783 (initialize_iterators): Guard usage of apply_subst_iterator with
34784 #ifdef GENERATOR_FILE.
34785 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
34786 (md_reader::read_mapping): Likewise.
34787 (add_define_attr_for_define_subst): Likewise.
34788 (add_define_subst_attr): Likewise.
34789 (read_subst_mapping): Likewise.
34790 (check_code_iterator): Likewise.
34791 (rtx_reader::read_rtx): Likewise. Move one-time initialization
34793 (one_time_initialization): New function.
34794 (rtx_reader::read_until): New method.
34795 (read_flags): New function.
34796 (parse_reg_note_name): New function.
34797 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
34798 Handle reuse_rtx ids.
34799 Wrap iterator lookup within #ifdef GENERATOR_FILE.
34800 Add parsing support for RTL dumps, mirroring the special-cases in
34801 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
34802 values, and calling handle_any_trailing_information.
34803 (rtx_reader::read_rtx_operand): Convert return type from void
34804 to rtx, returning return_rtx. Handle case 'e'. Call
34805 finalize_string on XSTR and XTMPL fields.
34806 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
34807 "(nil)" values were omitted. Call the postprocess vfunc on the
34809 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
34810 class ctor. Initialize m_in_call_function_usage. Call
34811 one_time_initialization.
34812 * rtl-tests.c (selftest::test_uncond_jump): Call
34813 set_new_first_and_last_insn.
34814 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
34815 * selftest-rtl.c: New file.
34816 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
34817 (selftest::get_insn_by_uid): New decl.
34818 * selftest-run-tests.c (selftest::run_tests): Call
34819 read_rtl_function_c_tests.
34820 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
34821 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
34824 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
34826 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
34827 operands in a special way. Assert that pos+len <= mode precision.
34829 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34831 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
34832 3 argument Alias with unlimited for the negative form.
34833 (fno-vect-cost-model): Removed.
34835 2017-01-05 Martin Liska <mliska@suse.cz>
34837 * hsa-gen.c (gen_hsa_divmod): New function.
34838 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
34840 2017-01-05 Martin Liska <mliska@suse.cz>
34843 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
34846 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34848 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
34849 small constant length operands.
34851 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34853 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
34854 between loop iterations.
34856 2017-01-05 Martin Liska <mliska@suse.cz>
34859 * gimplify.c (gimplify_decl_expr): Compare to
34860 asan_poisoned_variables instread of checking flags.
34861 (gimplify_target_expr): Likewise.
34862 (gimplify_expr): Likewise.
34863 (gimplify_function_tree): Conditionally initialize
34864 asan_poisoned_variables.
34866 2017-01-04 Jeff Law <law@redhat.com>
34868 PR tree-optimizatin/78812
34869 * rtl.h (contains_mem_rtx_p): Prototype.
34870 * ifcvt.c (containts_mem_rtx_p): Move from here to...
34871 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
34872 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
34873 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
34874 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
34876 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34878 * input.c (assert_char_at_range): Default-initialize actual_range.
34880 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34882 * df-scan.c (df_ref_create_structure): Make regno unsigned,
34883 to match the caller.
34885 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34887 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
34888 insns after final jump in test to emit dummy move.
34890 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34892 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
34893 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
34895 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34897 * multiple_target.c (create_dispatcher_calls): Init e_next.
34898 * tree-ssa-loop-split.c (split_loop): Init border.
34899 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
34902 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
34907 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
34908 (altivec_register_operand): Do not return true if the operand
34909 contains a SUBREG mixing SImode and SFmode.
34910 (vsx_register_operand): Likewise.
34911 (vsx_reg_sfsubreg_ok): New predicate.
34912 (vfloat_operand): Do not return true if the operand contains a
34913 SUBREG mixing SImode and SFmode.
34914 (vint_operand): Likewise.
34915 (vlogical_operand): Likewise.
34916 (gpc_reg_operand): Likewise.
34917 (int_reg_operand): Likewise.
34918 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
34919 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
34920 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
34922 (rs6000_emit_move_si_sf_subreg): New helper function.
34923 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
34924 fixup SUBREGs involving SImode and SFmode.
34925 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
34926 numbers for the new peephole2 optimization.
34927 (peephole2 for SFmode unions): New peephole2 to optimize cases in
34928 the GLIBC math library that do AND/IOR/XOR operations on single
34929 precision floating point.
34930 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
34931 target macros to say whether we need to avoid SUBREGs mixing
34933 (TARGET_ALLOW_SF_SUBREG): Likewise.
34934 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
34935 (UNSPEC_SI_FROM_SF): Likewise.
34936 (iorxor): Change spacing.
34937 (and_ior_xor): New iterator for AND, IOR, and XOR.
34938 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
34939 (movdi_from_sf_zero_ext): Likewise.
34940 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
34941 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
34942 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
34943 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
34944 (fms<mode>4): Likewise.
34945 (fnma<mode>4): Likewise.
34946 (fnms<mode>4): Likewise.
34947 (nfma<mode>4): Likewise.
34948 (nfms<mode>4): Likewise.
34950 2017-01-04 Marek Polacek <polacek@redhat.com>
34953 * doc/invoke.texi: Document -Wpointer-compare.
34955 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34957 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
34960 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
34961 descriptions for -gdwarf-5 and emit them as uleb128 instead of
34964 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
34967 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
34968 documentation of the powerpc_popcntb_ok attribute.
34969 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34970 code to issue warning messages if a requested CPU configuration is
34971 not supported by the binary (assembler and loader) toolchain.
34972 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
34973 made to define a built-in function that has been disabled.
34974 (paired_init_builtins): Add assertion to prevent ICE if attempt is
34975 made to define a built-in function that has been disabled.
34976 (altivec_init_builtins): Add comment explaining why definition
34977 of the DST built-in functions is not preceded by an assertion
34978 check. Add assertions to prevent ICE if attempts are made to
34979 define an altivec predicate or an abs* built-in function that has
34981 (htm_init_builtins): Add comment explaining why definition of the
34982 htm built-in functions is not preceded by an assertion check.
34984 2017-01-04 Jeff Law <law@redhat.com>
34986 PR tree-optimizatin/67955
34987 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
34988 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
34989 the points-to solution does not include pt_null. Use DECL_PT_UID
34992 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
34994 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
34995 Use gen_int_mode instead of gen_lopwart for const_int operands.
34997 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34999 PR tree-optimization/71563
35000 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35001 out of range value - has low bits known to be zero.
35003 2017-01-04 Alan Modra <amodra@gmail.com>
35005 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35006 * configure: Regenerate.
35007 * config.in: Regenerate.
35009 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35012 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35013 a substring of the message, but strcmp with the whole message. Ifdef
35014 ENABLE_NLS, translate the message first using dgettext.
35016 2017-01-03 Jeff Law <law@redhat.com>
35018 PR tree-optimizatin/78856
35019 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35020 (mark_threaded_blocks): Remove code to truncate thread paths that
35021 cross multiple loop headers. Instead invalidate the cached loop
35022 iteration information and handle case of a thread path walking
35023 into an irreducible region.
35025 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35028 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35029 assertions. Add support for doing the signbit if the IEEE 128-bit
35030 floating point value is in a GPR.
35031 * config/rs6000/rs6000.md (Fsignbit): Delete.
35032 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35033 Update the length attribute if the value is in a GPR.
35034 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35035 the sign or zero extension instruction, since the value is always 0/1.
35036 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35039 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35040 extracting SImode to a GPR register so that we can generate a
35041 store, limit the vector to be in a traditional Altivec register
35042 for the vextuwrx instruction.
35044 2017-01-03 Ian Lance Taylor <iant@google.com>
35046 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35048 2017-01-03 Martin Sebor <msebor@redhat.com>
35050 PR tree-optimization/78696
35051 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35052 precision. Use MPFR for %f for greater fidelity. Correct handling
35054 (pass_sprintf_length::compute_format_length): Set width and precision
35055 specified by asrerisk to void_node for vararg functions.
35056 (try_substitute_return_value): Adjust dump output.
35058 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35060 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35062 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35064 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35065 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35066 -mlra/-mno-lra was passed to the compiler.
35068 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35070 PR rtl-optimization/65618
35071 * emit-rtl.c (try_split): Move initialization of "before" and
35072 "after" to just before the call to emit_insn_after_setloc.
35074 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35076 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35078 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35080 * dwarf2out.c (gen_enumeration_type_die): When
35081 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35083 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35085 PR tree-optimization/78965
35086 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35087 Change first argument from const call_info & to call_info &. For %n
35088 set info.nowrite to false.
35090 PR middle-end/78901
35091 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35092 possibly throwing calls.
35094 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35095 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35096 and fns handling, rather than in a separate case SSA_NAME.
35098 2017-01-02 Jeff Law <law@redhat.com>
35100 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35101 fixes for first_period and second_period variables.
35103 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35106 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35107 (*insvqi_1): New insn pattern.
35108 (*insvqi_1_mem_rex64): Ditto.
35109 (*insvqi_2): Ditto.
35110 (*insvqi_3): Rename from *insvqi.
35112 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35114 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35116 * doc/cfg.texi (Edges): Remove reference to Java.
35117 (Maintaining the CFG): Ditto.
35119 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35121 PR middle-end/77674
35122 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35123 transparent aliases.
35125 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35127 PR middle-end/77484
35128 * predict.def (PRED_CALL): Update hitrate.
35129 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35130 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35131 into direct/indirect/polymorphic variants.
35133 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35135 Update copyright years.
35137 * gcc.c (process_command): Update copyright notice dates.
35138 * gcov-dump.c (print_version): Ditto.
35139 * gcov.c (print_version): Ditto.
35140 * gcov-tool.c (print_version): Ditto.
35141 * gengtype.c (create_file): Ditto.
35142 * doc/cpp.texi: Bump @copying's copyright year.
35143 * doc/cppinternals.texi: Ditto.
35144 * doc/gcc.texi: Ditto.
35145 * doc/gccint.texi: Ditto.
35146 * doc/gcov.texi: Ditto.
35147 * doc/install.texi: Ditto.
35148 * doc/invoke.texi: Ditto.
35150 Copyright (C) 2017 Free Software Foundation, Inc.
35152 Copying and distribution of this file, with or without modification,
35153 are permitted in any medium without royalty provided the copyright
35154 notice and this notice are preserved.