1 2017-10-30 Richard Biener <rguenther@suse.de>
3 PR tree-optimization/82762
5 2017-10-23 Richard Biener <rguenther@suse.de>
7 PR tree-optimization/82129
9 2017-08-01 Richard Biener <rguenther@suse.de>
11 PR tree-optimization/81181
12 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
13 (compute_antic): ... end of iteration here.
15 2017-10-30 Joseph Myers <joseph@codesourcery.com>
17 * doc/invoke.texi (C Dialect Options): Document -std=c17,
18 -std=iso9899:2017 and -std=gnu17.
19 * doc/standards.texi (C Language): Document C17 support.
20 * doc/cpp.texi (Overview): Mention -std=c17.
21 (Standard Predefined Macros): Document C11 and C17 values of
22 __STDC_VERSION__. Do not refer to C99 support as incomplete.
23 * doc/extend.texi (Inline): Do not list individual options for
24 standards newer than C99.
25 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
27 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
30 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32 * asan.c (asan_finish_file): Align asan globals array by shadow
35 2017-10-30 Jakub Jelinek <jakub@redhat.com>
38 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
39 (struct store_immediate_info): Add bitregion_start and bitregion_end
41 (store_immediate_info::store_immediate_info): Add brs and bre
42 arguments and initialize bitregion_{start,end} from those.
43 (struct merged_store_group): Add bitregion_start, bitregion_end,
44 align_base and mask fields. Drop unnecessary struct keyword from
45 struct store_immediate_info. Add do_merge method.
46 (clear_bit_region_be): Use memset instead of loop storing zeros.
47 (merged_store_group::do_merge): New method.
48 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
49 stores as long as the surrounding bitregions have no gaps.
50 (merged_store_group::merge_overlapping): Use do_merge.
51 (merged_store_group::apply_stores): Test that bitregion_{start,end}
52 is byte aligned, rather than requiring that start and width are
53 byte aligned. Drop unnecessary struct keyword from
54 struct store_immediate_info. Allocate and populate also mask array.
55 Make start of the arrays relative to bitregion_start rather than
56 start and size them according to bitregion_{end,start} difference.
57 (struct imm_store_chain_info): Drop unnecessary struct keyword from
58 struct store_immediate_info.
59 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
60 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
61 struct keyword from struct store_immediate_info.
62 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
63 between stores as long as the surrounding bitregions have no gaps.
65 (struct split_store): Add orig non-static data member.
66 (split_store::split_store): Initialize orig to false.
67 (find_constituent_stmts): Return store_immediate_info *, non-NULL
68 if there is exactly a single original stmt. Change stmts argument
69 to pointer from reference, if NULL, don't push anything to it. Add
70 first argument, use it to optimize skipping over orig stmts that
71 are known to be before bitpos already. Simplify.
72 (split_group): Return unsigned int count how many stores are or
73 would be needed rather than a bool. Add allow_unaligned argument.
74 Change split_stores argument from reference to pointer, if NULL,
75 only do a dry run computing how many stores would be produced.
76 Rewritten algorithm to use both alignment and misalign if
77 !allow_unaligned and handle bitfield stores with gaps.
78 (imm_store_chain_info::output_merged_store): Set start_byte_pos
79 from bitregion_start instead of start. Compute allow_unaligned
80 here, if true, do 2 split_group dry runs to compute which one
81 produces fewer stores and prefer aligned if equal. Punt if
82 new count is bigger or equal than original before emitting any
83 statements, rather than during that. Remove no longer needed
84 new_ssa_names tracking. Replace num_stmts with
85 split_stores.length (). Use 32-bit stack allocated entries
86 in split_stores auto_vec. Try to reuse original store lhs/rhs1
87 if possible. Handle bitfields with gaps.
88 (pass_store_merging::execute): Ignore bitsize == 0 stores.
89 Compute bitregion_{start,end} for the stores and construct
90 store_immediate_info with that. Formatting fixes.
92 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
95 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
96 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
98 2017-10-29 Jim Wilson <wilson@tuliptree.org>
100 * gimplify.c: Include tm_p.h.
102 * common.opt (gcoff): Re-add as ignored option.
103 (gcoff1, gcoff2, gcoff3): Likewise.
105 * Makefile.in (OBJS): Delete sdbout.o.
106 (GTFILES): Delete $(srcdir)/sdbout.c.
107 * debug.h: Delete sdb_debug_hooks.
108 * final.c: Delete sdbout.h include.
109 (final_scan_insn): Delete SDB_DEBUG check.
110 (rest_of_clean_state): Likewise.
111 * output.h: Delete sdb_begin_function_line.
114 * toplev.c: Delete sdbout.h include.
115 (process_options): Delete SDB_DEBUG check.
116 * tree-core.h (tree_type_common): Delete pointer field of
118 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
120 * tree.h (TYPE_SYMTAB_POINTER): Delete.
121 (TYPE_SYMTAB_IS_POINTER): Delete.
122 (TYPE_SYMTAB_IS_DIE): Renumber.
123 * xcoffout.c: Refer to former sdbout.c file.
124 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
126 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
127 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
128 (Debugging Options): Delete -gcoff.
129 (-gstabs): Delete SDB reference.
131 (-gcoff@var{level}): Delete.
132 * doc/passes.texi (Debugging information output): Delete SDB and
134 * doc/tm.texi: Regenerate.
135 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
136 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
138 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
139 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
140 (SDB_DEBUGGING_INFO): Delete.
141 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
142 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
143 * target.def (output_source_filename): Delete COFF reference.
145 * common.opt (gcoff): Delete.
146 (gxcoff+): Update Negative chain.
147 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
149 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
150 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
151 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
153 (expand_function_start): Change sdb reference to past tense.
154 (expand_function_end): Change sdb reference to past tense.
155 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
156 * opts.c (debug_type_names): Delete coff entry.
157 (common_handle_option): Delete OPT_gcoff case.
158 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
160 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
161 * config/cris/cris.h: Delete SDB reference in comment.
162 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
163 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
164 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
165 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
167 (ix86_expand_prologue): Likewise.
168 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
169 * config/ia64/ia64.h: Likewise.
170 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
171 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
173 * config/mmix/mmix.h: Likewise.
174 * config/nds32/nds32.c: Likewise.
175 * config/stormy/storym16.h: Likewise.
176 * config/visium/visium.h: Likewise.
177 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
179 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
181 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
182 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
183 FRAME_POINTER_REGNUM point at high end of local var area.
185 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
187 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
188 Move comment around. Do not reset best_edge for a copiable
189 destination if the copy would cause a partition change.
190 (better_edge_p): Remove redundant check.
192 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
194 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
196 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
198 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
199 for math functions that have _Float<N> and _Float<N>X variants.
200 (mathfn_built_in_2): Add support for math functions that have
201 _Float<N> and _Float<N>X variants.
202 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
203 (expand_builtin_mathfn_ternary): Add support for fma with
204 _Float<N> and _Float<N>X variants.
205 (expand_builtin): Likewise.
206 (fold_builtin_3): Likewise.
207 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
208 create math function _Float<N> and _Float<N>X variants as external
210 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
211 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
212 the __builtin_ prefix and if not strict ansi, without the prefix.
213 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
214 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
215 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
216 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
217 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
218 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
219 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
220 function signatures for fma _Float<N> and _Float<N>X variants.
221 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
222 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
223 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
224 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
225 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
226 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
227 * gencfn-macros.c (print_case_cfn): Add support for math functions
228 that have _Float<N> and _Float<N>X variants.
229 (print_define_operator_list): Likewise.
230 (fltfn_suffixes): Likewise.
232 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
233 for math functions that have _Float<N> and _Float<N>X variants.
234 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
235 and _Float<N>X variants.
236 (COPYSIGN): Likewise.
239 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
240 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
242 (integer_valued_read_call_p): Likewise.
243 * fold-const-call.c (fold_const_call_ss): Likewise.
244 (fold_const_call_sss): Add support for copysign, fmin, and fmax
245 _Float<N> and _Float<N>X variants.
246 (fold_const_call_ssss): Add support for fma _Float<N> and
248 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
249 support for copysign and fma _Float<N> and _Float<N>X variants.
250 (backprop::process_builtin_call_use): Likewise.
251 * tree-call-cdce.c (can_test_argument_range); Add support for
252 sqrt _Float<N> and _Float<N>X variants.
253 (edom_only_function): Likewise.
254 (get_no_error_domain): Likewise.
255 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
256 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
257 copysign _Float<N> and _Float<N>X variants.
258 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
259 handled by machine independent code.
261 * doc/cpp.texi (Common Predefined Macros): Document defining
262 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
263 fma _Float<N> and _Float<N>X variants.
265 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
268 * config/i386/i386-modes.def (CCFPU): Remove definition.
269 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
270 (ix86_cc_modes_compatible): Ditto.
271 (ix86_expand_carry_flag_compare): Ditto.
272 (ix86_expand_int_movcc): Ditto.
273 (ix86_expand_int_addcc): Ditto.
274 (ix86_reverse_condition): Ditto.
275 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
276 Return true/false for unordered/ordered fp comparisons.
277 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
278 (ix86_prepare_fp_compare_args): Update for rename.
279 (ix86_expand_fp_compare): Update for rename. Generate unordered
280 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
281 (ix86_expand_sse_compare_and_jump): Ditto.
282 * config/i386/predicates.md (fcmov_comparison_operator):
283 Remove CCFPU mode handling.
284 (ix86_comparison_operator): Ditto.
285 (ix86_carry_flag_operator): Ditto.
286 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
287 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
288 (*cmpu<mode>_cc_i387): Ditto.
289 (FPCMP): Remove mode iterator.
290 (unord): Remove mode attribute.
291 (unord_subst): New define_subst transformation
292 (unord): New define_subst attribute.
294 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
295 (*cmpi<unord>xf_i387): Ditto.
296 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
297 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
298 using unord_subst transformation.
299 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
300 (round_saeonly): Also handle CCFP mode.
301 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
302 Remove UNSPEC_SAHF unspec handling.
304 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
306 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
308 2017-10-27 Jeff Law <law@redhat.com>
310 * gimple-ssa-sprintf.c: Include domwalk.h.
311 (class sprintf_dom_walker): New class, derived from dom_walker.
312 (sprintf_dom_walker::before_dom_children): New function.
313 (struct call_info): Moved into sprintf_dom_walker class
314 (compute_formath_length, handle_gimple_call): Likewise.
315 (sprintf_length::execute): Call the dominator walker rather
316 than walking the statements.
318 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
319 gimple statement locations.
320 (check_array_bounds): Corresponding changes. Get the statement's
321 location directly from wi->stmt.
323 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
326 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
328 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
330 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
331 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
333 2017-10-27 Jakub Jelinek <jakub@redhat.com>
336 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
337 * config/i386/i386.c (maybe_get_pool_constant): Removed.
338 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
339 maybe_get_pool_constant.
340 * config/i386/predicates.md (zero_extended_scalar_load_operand):
343 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
345 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
346 2.26 caveat. Update gas and gld versions.
347 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
350 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
352 * cgraph.h (set_malloc_flag): Declare.
353 * cgraph.c (set_malloc_flag_1): New function.
354 (set_malloc_flag): Likewise.
355 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
356 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
358 (read_ipa_call_summary): Add support for reading is_return_callee.
359 (write_ipa_call_summary): Stream is_return_callee.
360 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
361 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
362 ipa-prop.h, ipa-fnsummary.h.
363 (pure_const_names): Change to static.
364 (malloc_state_e): Define.
365 (malloc_state_names): Define.
366 (funct_state_d): Add field malloc_state.
367 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
368 (check_retval_uses): New function.
369 (malloc_candidate_p): Likewise.
370 (analyze_function): Add support for malloc attribute.
371 (pure_const_write_summary): Stream malloc_state.
372 (pure_const_read_summary): Add support for reading malloc_state.
373 (dump_malloc_lattice): New function.
374 (propagate_malloc): New function.
375 (warn_function_malloc): New function.
376 (ipa_pure_const::execute): Call propagate_malloc and
378 (pass_local_pure_const::execute): Add support for malloc attribute.
379 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
380 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
382 2017-10-27 Martin Liska <mliska@suse.cz>
384 PR gcov-profile/82457
385 * doc/invoke.texi: Document that one needs a non-strict ISO mode
386 for fork-like functions to be properly instrumented.
388 2017-10-27 Richard Biener <rguenther@suse.de>
391 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
392 info when we redirected EH.
394 2017-10-26 Michael Collison <michael.collison@arm.com>
396 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
398 (<optab>_trunchf<GPI:mode>2: New pattern.
399 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
400 * config/aarch64/iterators.md (wv): New mode attribute.
401 (vf, VF): New mode attributes.
402 (vgp, VGP): New mode attributes.
403 (s): Update attribute with SImode and DImode prefixes.
405 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
407 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
408 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
409 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
410 (nios2_option_overide): Initialize it. Don't allow R0-relative
412 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
413 (nios2_symbolic_constant_p): Likewise.
414 (nios2_legitimate_address_p): Likewise.
415 (nios2_r0rel_section_name_p): New.
416 (nios2_symbol_ref_in_r0rel_data_p): New.
417 (nios2_emit_move_sequence): Handle r0rel_constant_p.
418 (r0rel_constant_p): New.
419 (nios2_print_operand_address): Handle r0rel_constant_p.
420 (nios2_cdx_narrow_form_p): Likewise.
421 * config/nios2/nios2.opt (mr0rel-sec=): New option.
422 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
423 (Nios II Options): Document -mr0rel-sec.
425 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
427 * config/nios2/nios2.c: Include xregex.h.
428 (nios2_gprel_sec_regex): New.
429 (nios2_option_overide): Initialize it. Don't allow GP-relative
431 (nios2_small_section_name_p): Check for regex match.
432 * config/nios2/nios2.opt (mgprel-sec=): New option.
433 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
434 (Nios II Options): Document -mgprel-sec.
436 2017-10-26 Jim Wilson <wilson@tuliptree.org>
438 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
440 2017-10-26 Tom de Vries <tom@codesourcery.com>
442 PR tree-optimization/82707
443 * gimple.c (gimple_copy): Fix unsharing of
444 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
446 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
448 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
449 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
450 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
451 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
452 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
453 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
454 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
455 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
456 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
457 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
458 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
459 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
460 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
461 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
462 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
463 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
464 _mm512_mask_cmpunord_ps_mask): New intrinsics.
466 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
468 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
470 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
471 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
472 warning to rs6000.c. Remove the Undocumented flag, since it has
474 (-mabi=ibmlongdouble): Likewise.
475 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
476 already set, set the default format for long double.
477 (rs6000_debug_reg_global): Print whether long double is IBM or
479 (rs6000_option_override_internal): Rework setting long double
480 format. Only warn if the user is changing the long double default
481 and they did not use -Wno-psabi.
482 * doc/invoke.texi (PowerPC options): Update the documentation for
483 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
485 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
486 Alan Hayward <alan.hayward@arm.com>
487 David Sherwood <david.sherwood@arm.com>
489 * rtl.h (wider_subreg_mode): New function.
490 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
491 rather than an unsigned int *.
492 * ira-color.c (regno_max_ref_width): Replace with...
493 (regno_max_ref_mode): ...this new variable.
494 (coalesced_pseudo_reg_slot_compare): Update accordingly.
495 Use wider_subreg_mode.
496 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
497 rather than an unsigned int *.
498 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
499 (process_alt_operands): Likewise.
500 (invariant_p): Likewise.
501 * lra-spills.c (assign_mem_slot): Likewise.
502 (add_pseudo_to_slot): Likewise.
503 * lra.c (collect_non_operand_hard_regs): Likewise.
504 (add_regs_to_insn_regno_info): Likewise.
505 * reload1.c (regno_max_ref_width): Replace with...
506 (regno_max_ref_mode): ...this new variable.
507 (reload): Update accordingly. Update call to
508 ira_sort_regnos_for_alter_reg.
509 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
510 (init_eliminable_invariants): Update to use regno_max_ref_mode.
511 (scan_paradoxical_subregs): Likewise.
513 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
515 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
516 (aarch64_frame): Add emit_frame_chain boolean.
517 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
518 Move eh_return case to aarch64_layout_frame.
519 (aarch64_layout_frame): Initialize emit_frame_chain.
520 (aarch64_expand_prologue): Use emit_frame_chain.
522 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
524 * config/aarch64/aarch64.c (aarch64_layout_frame):
525 Ensure LR is always stored at the bottom of the callee-saves.
526 Remove rarely used frame layout which saves callee-saves at top of
527 frame, so the store of LR can be used as a valid probe in all cases.
529 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
531 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
532 Improve unaligned TImode/TFmode base/offset split.
534 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
535 Alan Hayward <alan.hayward@arm.com>
536 David Sherwood <david.sherwood@arm.com>
538 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
539 * combine.c (find_single_use_1): Likewise.
540 (expand_field_assignment): Likewise.
541 (move_deaths): Likewise.
542 * lra-constraints.c (simplify_operand_subreg): Likewise.
543 (curr_insn_transform): Likewise.
544 * lra.c (collect_non_operand_hard_regs): Likewise.
545 (add_regs_to_insn_regno_info): Likewise.
546 * rtlanal.c (reg_referenced_p): Likewise.
547 (covers_regno_no_parallel_p): Likewise.
549 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
551 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
552 Don't print any bits outside the precision of the value.
553 * wide-int.cc (test_printing): Add some new tests.
555 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
557 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
558 supports -xbrace_comment option.
559 * configure: Regenerate.
560 * config.in: Regenerate.
561 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
562 (ASM_CPU_SPEC): Use it.
564 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
566 * target.def (static_rtx_alignment): New hook.
567 * targhooks.h (default_static_rtx_alignment): Declare.
568 * targhooks.c (default_static_rtx_alignment): New function.
569 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
570 * doc/tm.texi: Regenerate.
571 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
572 instead of targetm.constant_alignment. Remove call to
574 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
575 (cris_preferred_mininum_alignment): New function, split out from...
576 (cris_constant_alignment): ...here.
577 (cris_static_rtx_alignment): New function.
578 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
580 (ix86_constant_alignment): ...here.
581 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
582 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
583 (mmix_static_rtx_alignment): New function.
584 * config/spu/spu.c (spu_static_rtx_alignment): New function.
585 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
587 2017-10-26 Tamar Christina <tamar.christina@arm.com>
590 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
591 Add flag_trapping_math and flag_fp_int_builtin_inexact.
593 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
595 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
596 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
597 mark as a sign-extending load.
598 (local_pic_load_u): Define.
600 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
603 * fold-const.c (operand_equal_for_comparison_p): Also return true
604 if ARG0 is a simple variant of ARG1 with narrower precision.
605 (fold_ternary_loc): Always pass unstripped operands to the predicate.
607 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
609 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
611 * i386.h (processor_costs): Add gather_static, gather_per_elt,
612 scatter_static, scatter_per_elt.
613 * x86-tune-costs.h: Add new cost entries.
615 2017-10-25 Richard Biener <rguenther@suse.de>
617 * tree-ssa-sccvn.h (vn_eliminate): Declare.
618 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
619 class pass_fre): Move to ...
620 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
621 class pass_fre): ... here and adjust for statistics.
623 2017-10-25 Jakub Jelinek <jakub@redhat.com>
626 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
627 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
628 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
629 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
632 2017-10-25 Richard Biener <rguenther@suse.de>
634 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
635 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
636 eliminate_push_avail, eliminate_insert): Move inside...
637 (class eliminate_dom_walker): ... this class in preparation
639 (fini_eliminate): Remove by merging with ...
640 (eliminate): ... this function. Adjust for class changes.
641 (pass_pre::execute): Remove fini_eliminate call.
642 (pass_fre::execute): Likewise.
644 2017-10-24 Jakub Jelinek <jakub@redhat.com>
647 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
648 (VPERMI2, VPERMI2I): New mode iterators.
649 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
650 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
652 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
653 mode iterator. Remove 3 old define_insn patterns.
654 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
655 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
656 VPERMI2 mode iterator, remove the other two expanders.
657 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
658 to use VPERMI2 mode iterator, add another alternative for vpermi2*
659 instructions, remove the other two patterns.
660 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
661 mode iterator, remove the other two patterns.
662 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
663 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
664 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
665 and adjust argument order accordingly.
666 (ix86_expand_vec_perm): Adjust caller.
667 (expand_vec_perm_1): Likewise.
668 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
669 (expand_vec_perm_vpermt2_vpshub2): ... this.
670 (ix86_expand_vec_perm_const_1): Adjust caller.
671 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
674 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
675 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
676 (vec_shl_<mode>): Remove unused expander.
677 (avx512bw_<shift_insn><mode>3): New define_insn.
678 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
679 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
681 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
684 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
687 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
689 PR rtl-optimization/82396
690 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
691 (autopref_multipass_init): Simplify initialization.
692 (autopref_rank_data): Simplify sort order.
693 * gcc/sched-int.h (autopref_multipass_data_): Remove
694 multi_mem_insn_p, min_offset and max_offset.
696 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
699 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
700 Check special value of flag_omit_frame_pointer.
701 (aarch64_can_eliminate): Likewise.
702 (aarch64_override_options_after_change_1): Simplify handling of
703 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
705 2017-10-24 Richard Biener <rguenther@suse.de>
707 PR tree-optimization/82697
708 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
709 zero for conditional load and unconditional store.
711 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
713 * doc/install.texi: Document bootstrap-cet.
715 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
718 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
719 ENDBR instruction at function entrance if function is only
722 2017-10-24 Jakub Jelinek <jakub@redhat.com>
725 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
726 patterns to better describe from which operation the CF is computed.
727 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
728 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
729 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
730 is 0, use _0 suffixed expanders instead of emitting a comparison
733 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
735 * config/i386/i386.md(*movsf_internal, *movdf_internal):
736 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
738 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
741 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
742 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
743 * loop-iv.c (iv_get_reaching_def): Likewise.
744 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
745 variable is promoted and the partition contains undefined values.
747 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
749 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
751 (nios2_address_cost): Define.
752 (nios2_legitimize_address): Recognize (exp + constant) directly.
753 (TARGET_ADDRESS_COST): Define.
755 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
757 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
758 (nios2_symbolic_memory_operand_p): Declare.
759 (nios2_split_large_constant): Declare.
760 (nios2_split_symbolic_memory_operand): Declare.
761 * config/nios2/nios2.c: Adjust includes.
762 (nios2_symbolic_constant_allowed): New.
763 (nios2_symbolic_constant_p): New.
764 (nios2_plus_symbolic_constant_p): New.
765 (nios2_valid_addr_expr_p): Recognize addresses involving
767 (nios2_legitimate_address_p): Likewise, also LO_SUM.
768 (nios2_symbolic_memory_operand_p): New.
769 (nios2_large_constant_p): New.
770 (nios2_split_large_constant): New.
771 (nios2_split_plus_large_constant): New.
772 (nios2_split_symbolic_memory_operand): New.
773 (nios2_legitimize_address): Code refactoring. Handle addresses
774 involving symbolic constants.
775 (nios2_emit_move_sequence): Likewise.
776 (nios2_print_operand): Improve error output.
777 (nios2_print_operand_address): Handle LO_SUM.
778 (nios2_cdx_narrow_form_p): Likewise.
779 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
780 operands involving symbolic constants.
781 (movhi_internal, movsi_internal): Likewise.
782 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
783 (extendhisi2, extendqi<mode>2): Likewise.
785 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
787 * tree-pass.h (PROP_rtl_split_insns): Define.
788 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
790 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
792 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
794 2017-10-23 Jakub Jelinek <jakub@redhat.com>
797 * target.def (const_not_ok_for_debug_p): Default to
798 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
799 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
800 * targhooks.c (default_const_not_ok_for_debug_p): New function.
801 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
802 which targetm.const_not_ok_for_debug_p returned true.
803 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
805 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
807 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
808 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
809 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
811 (ix86_const_not_ok_for_debug_p): New function.
812 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
813 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
815 2017-10-23 David Malcolm <dmalcolm@redhat.com>
818 * system.h: Conditionally include "unique-ptr.h" if
819 INCLUDE_UNIQUE_PTR is defined.
820 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
821 of defining INCLUDE_UNIQUE_PTR before including "system.h".
823 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
825 * config/rl78/rl78.md: New define_expand "subdi3".
827 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
830 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
831 DF_REF_INSN if DF_REF_INSN_INFO is false.
833 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
835 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
836 xmm_move instead of sse_move.
837 (sse_store_index): New function.
838 (ix86_register_move_cost): Be more sensible about mismatch stall;
839 model AVX moves correctly; make difference between sse->integer and
841 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
842 moves; make difference between SSE and AVX.
843 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
844 and zmm_move. Increase size of sse load and store tables;
845 add unaligned load and store tables; add ssemmx_to_integer.
846 * x86-tune-costs.h: Update all entries according to real
847 move latencies from Agner Fog's manual and chip documentation.
849 2017-10-23 Jakub Jelinek <jakub@redhat.com>
852 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
853 * config/i386/constraints.md (Wf): New constraint.
854 * config/i386/i386.md (UNSPEC_SBB): New unspec.
855 (cmp<dwi>_doubleword): Removed.
856 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
857 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
858 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
859 expand with cmp<dwi>_doubleword. For LTU and GEU use
860 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
862 * common.opt (gcolumn-info): Enable by default.
863 * doc/invoke.texi (gcolumn-info): Document new default.
865 2017-10-23 Richard Biener <rguenther@suse.de>
867 PR tree-optimization/82672
868 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
869 Fold the stmt if we propagated into it.
871 2017-10-23 Richard Biener <rguenther@suse.de>
873 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
874 (bitmap_remove_expr_from_set): ... this. All callers call this
875 for non-constant values.
876 (bitmap_set_subtract): Rename to...
877 (bitmap_set_subtract_expressions): ... this. Adjust and
879 (bitmap_set_contains_value): Remove superfluous check.
880 (bitmap_set_replace_value): Inline into single caller ...
881 (bitmap_value_replace_in_set): ... here and simplify.
882 (dependent_clean): Merge into ...
883 (clean): ... this using an overload. Adjust.
884 (prune_clobbered_mems): Adjust.
885 (compute_antic_aux): Likewise.
886 (compute_partial_antic_aux): Likewise.
888 2017-10-23 Richard Biener <rguenther@suse.de>
890 PR tree-optimization/82129
892 2017-08-01 Richard Biener <rguenther@suse.de>
894 PR tree-optimization/81181
895 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
896 (compute_antic): ... end of iteration here.
898 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
900 * target.def (starting_frame_offset): New hook.
901 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
902 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
903 * doc/tm.texi.in: Regenerate.
904 * hooks.h (hook_hwi_void_0): Declare.
905 * hooks.c (hook_hwi_void_0): New function.
906 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
907 STARTING_FRAME_OFFSET.
908 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
909 * reload1.c (reload): Likewise.
910 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
911 instead of STARTING_FRAME_OFFSET.
912 * function.c (try_fit_stack_local): Likewise.
913 (assign_stack_local_1): Likewise
914 (instantiate_virtual_regs): Likewise.
915 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
916 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
917 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
918 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
919 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
920 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
921 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
922 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
923 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
924 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
925 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
926 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
927 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
928 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
929 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
930 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
931 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
932 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
933 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
934 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
935 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
936 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
937 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
938 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
939 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
940 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
941 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
942 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
943 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
944 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
945 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
946 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
947 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
948 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
949 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
950 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
951 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
952 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
953 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
954 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
955 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
956 * config/avr/avr.c (avr_starting_frame_offset): Make static and
957 return a HOST_WIDE_INT.
958 (avr_builtin_setjmp_frame_value): Use it instead of
959 STARTING_FRAME_OFFSET.
960 (TARGET_STARTING_FRAME_OFFSET): Redefine.
961 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
962 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
964 (TARGET_STARTING_FRAME_OFFSET): Redefine.
965 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
966 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
967 (TARGET_CONSTANT_ALIGNMENT): Redefine.
968 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
969 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
970 (TARGET_STARTING_FRAME_OFFSET): Redefine.
971 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
972 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
973 (TARGET_STARTING_FRAME_OFFSET): Redefine.
974 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
975 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
977 (TARGET_STARTING_FRAME_OFFSET): Redefine.
978 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
979 * config/mips/mips.c (mips_compute_frame_info): Refer to
980 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
981 (mips_starting_frame_offset): New function.
982 (TARGET_STARTING_FRAME_OFFSET): Redefine.
983 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
984 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
985 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
986 and return a HOST_WIDE_INT.
987 (TARGET_STARTING_FRAME_OFFSET): Redefine.
988 (mmix_initial_elimination_offset): Refer to
989 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
990 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
991 * config/pa/pa.c (pa_starting_frame_offset): New function.
992 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
993 (pa_expand_prologue): Likewise.
994 (TARGET_STARTING_FRAME_OFFSET): Redefine.
995 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
996 !FRAME_GROWS_DOWNWARD handling to...
997 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
998 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
999 !FRAME_GROWS_DOWNWARD handling to...
1000 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1001 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
1002 !FRAME_GROWS_DOWNWARD handling to...
1003 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1004 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
1006 (rs6000_starting_frame_offset): New function.
1007 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
1008 !FRAME_GROWS_DOWNWARD handling to...
1009 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1010 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
1011 !FRAME_GROWS_DOWNWARD handling to...
1012 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1013 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
1014 !FRAME_GROWS_DOWNWARD handling to...
1015 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1016 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
1017 (rs6000_starting_frame_offset): New function.
1018 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
1019 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
1020 * config/vax/vax.c (vax_starting_frame_offset): New function.
1021 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
1022 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1023 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
1024 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
1025 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1026 * system.h (STARTING_FRAME_OFFSET): Poison.
1028 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1030 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
1031 SCALAR_TYPE_MODE instead of TYPE_MODE.
1033 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1034 Alan Hayward <alan.hayward@arm.com>
1035 David Sherwood <david.sherwood@arm.com>
1037 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
1039 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1040 Alan Hayward <alan.hayward@arm.com>
1041 David Sherwood <david.sherwood@arm.com>
1043 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
1045 2017-10-23 Richard Biener <rguenther@suse.de>
1047 PR tree-optimization/82129
1048 * tree-ssa-pre.c (bitmap_set_and): Remove.
1049 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
1050 canonicalizing expressions in the set to those with lowest
1051 ID rather than taking that from the first edge.
1053 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1055 * combine.c (rtx_equal_for_field_assignment_p): Use
1056 byte_lowpart_offset.
1058 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1059 Alan Hayward <alan.hayward@arm.com>
1060 David Sherwood <david.sherwood@arm.com>
1062 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
1063 to a SUBREG_PROMOTED_VAR.
1065 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1066 Alan Hayward <alan.hayward@arm.com>
1067 David Sherwood <david.sherwood@arm.com>
1069 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
1070 (expand_debug_source_expr): Likewise.
1071 * combine.c (combine_simplify_rtx): Likewise.
1072 * cse.c (fold_rtx): Likewise.
1073 * optabs.c (expand_float): Likewise.
1074 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1075 (simplify_binary_operation_1): Likewise.
1077 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1078 Alan Hayward <alan.hayward@arm.com>
1079 David Sherwood <david.sherwood@arm.com>
1081 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
1082 (record_promoted_value): Likewise.
1083 * expr.c (expand_expr_real_2): Likewise.
1084 * ree.c (update_reg_equal_equiv_notes): Likewise.
1085 (combine_set_extension): Likewise.
1086 * rtlanal.c (low_bitmask_len): Likewise.
1087 * simplify-rtx.c (neg_const_int): Likewise.
1088 (simplify_binary_operation_1): Likewise.
1090 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1091 Alan Hayward <alan.hayward@arm.com>
1092 David Sherwood <david.sherwood@arm.com>
1094 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
1095 * regcprop.c (maybe_mode_change): Likewise.
1096 * reload1.c (alter_reg): Likewise.
1098 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1100 * inchash.h (inchash::hash::add_wide_int): New function.
1101 * lto-streamer-out.c (hash_tree): Use it.
1103 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1105 * inchash.h (inchash::hash::add_wide_int): Rename to...
1106 (inchash::hash::add_hwi): ...this.
1107 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
1108 (polymorphic_call_target_hasher::hash): Likewise.
1109 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
1110 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
1111 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
1112 * lto-streamer-out.c (hash_tree): Likewise.
1113 * optc-save-gen.awk: Likewise.
1114 * tree.c (add_expr): Likewise.
1116 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
1119 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
1120 for ordered inequality comparisons even with TARGET_IEEE_FP.
1122 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
1125 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
1126 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
1127 Expand with cmp<dwi>_doubleword.
1129 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1131 * extend.texi: Add x86 specific to 'nocf_check' attribute.
1132 List CET intrinsics.
1133 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
1134 specific to -fcf-protection option.
1136 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1138 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
1139 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
1140 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
1141 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
1142 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
1143 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
1144 (extra_objs): Add cet.o for Linux/x86 targets.
1145 (tmake_file): Add i386/t-cet for Linux/x86 targets.
1146 * config/i386/cet.c: New file.
1147 * config/i386/cetintrin.h: Likewise.
1148 * config/i386/t-cet: Likewise.
1149 * config/i386/cpuid.h (bit_SHSTK): New.
1150 (bit_IBT): Likewise.
1151 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
1152 pass IBT and SHSTK bits.
1153 * config/i386/i386-builtin-types.def
1154 (VOID_FTYPE_UNSIGNED_PVOID): New.
1155 (VOID_FTYPE_UINT64_PVOID): Likewise.
1156 * config/i386/i386-builtin.def: Add CET intrinsics.
1157 * config/i386/i386-c.c (ix86_target_macros_internal): Add
1158 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
1159 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
1160 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
1162 * config/i386/i386.c (rest_of_insert_endbranch): New.
1163 (pass_data_insert_endbranch): Likewise.
1164 (pass_insert_endbranch): Likewise.
1165 (make_pass_insert_endbranch): Likewise.
1166 (ix86_notrack_prefixed_insn_p): Likewise.
1167 (ix86_target_string): Add -mibt, -mshstk flags.
1168 (ix86_option_override_internal): Add flag_cf_protection
1170 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
1171 (ix86_print_operand): Add 'notrack' prefix output.
1172 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
1173 (ix86_expand_builtin): Expand CET intrinsics.
1174 (x86_output_mi_thunk): Add 'endbranch' instruction.
1175 * config/i386/i386.h (TARGET_IBT): New.
1176 (TARGET_IBT_P): Likewise.
1177 (TARGET_SHSTK): Likewise.
1178 (TARGET_SHSTK_P): Likewise.
1179 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
1180 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
1181 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
1182 (builtin_setjmp_setup): New pattern.
1183 (builtin_longjmp): Likewise.
1184 (rdssp<mode>): Likewise.
1185 (incssp<mode>): Likewise.
1186 (saveprevssp): Likewise.
1187 (rstorssp): Likewise.
1188 (wrss<mode>): Likewise.
1189 (wruss<mode>): Likewise.
1190 (setssbsy): Likewise.
1191 (clrssbsy): Likewise.
1192 (nop_endbr): Likewise.
1193 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
1195 * config/i386/immintrin.h: Include <cetintrin.h>.
1196 * config/i386/linux-common.h
1197 (file_end_indicate_exec_stack_and_cet): New prototype.
1198 (TARGET_ASM_FILE_END): New.
1200 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
1202 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
1203 latencies instead of having separate table; make difference between
1204 integer and float costs.
1205 * i386.h (processor_costs): Remove scalar_stmt_cost,
1206 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
1207 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
1209 * x86-tune-costs.h: Remove entries which has been removed in
1210 procesor_costs from all tables; make cond_taken_branch_cost
1211 and cond_not_taken_branch_cost COST_N_INSNS based.
1213 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
1215 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
1217 2017-10-20 Jakub Jelinek <jakub@redhat.com>
1219 * config/i386/i386.md (isa): Remove fma_avx512f.
1220 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
1221 <avx512>_fmadd_<mode>_mask3<round_name>,
1222 <avx512>_fmsub_<mode>_mask<round_name>,
1223 <avx512>_fmsub_<mode>_mask3<round_name>,
1224 <avx512>_fnmadd_<mode>_mask<round_name>,
1225 <avx512>_fnmadd_<mode>_mask3<round_name>,
1226 <avx512>_fnmsub_<mode>_mask<round_name>,
1227 <avx512>_fnmsub_<mode>_mask3<round_name>,
1228 <avx512>_fmaddsub_<mode>_mask<round_name>,
1229 <avx512>_fmaddsub_<mode>_mask3<round_name>,
1230 <avx512>_fmsubadd_<mode>_mask<round_name>,
1231 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
1232 (*vec_widen_umult_even_v16si<mask_name>,
1233 *vec_widen_smult_even_v16si<mask_name>): Likewise.
1234 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
1236 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1238 * extend.texi: Add 'nocf_check' documentation.
1239 * gimple.texi: Add second parameter to
1240 gimple_build_call_from_tree.
1241 * invoke.texi: Add -fcf-protection documentation.
1242 * rtl.texi: Add REG_CALL_NOTRACK documenation.
1244 2017-10-20 Richard Biener <rguenther@suse.de>
1246 PR tree-optimization/82473
1247 * tree-vect-loop.c (vectorizable_reduction): Properly get at
1248 the largest input type.
1250 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1252 * c-attribs.c (handle_nocf_check_attribute): New function.
1253 (c_common_attribute_table): Add 'nocf_check' handling.
1254 * gimple-parser.c: Add second argument NULL to
1255 gimple_build_call_from_tree.
1256 * attrib.c (comp_type_attributes): Check nocf_check attribute.
1257 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
1259 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
1260 * common.opt: Add fcf-protection flag.
1261 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
1262 * flag-types.h: Add enum cf_protection_level.
1263 * gimple.c (gimple_build_call_from_tree): Add second parameter.
1264 Add 'nocf_check' attribute propagation to gimple call.
1265 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
1266 (gimple_build_call_from_tree): Update prototype.
1267 (gimple_call_nocf_check_p): New function.
1268 (gimple_call_set_nocf_check): Likewise.
1269 * gimplify.c: Add second argument to gimple_build_call_from_tree.
1270 * ipa-icf.c: Add nocf_check attribute in statement hash.
1271 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
1272 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
1273 * toplev.c (process_options): Add flag_cf_protection handling.
1275 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
1277 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
1279 2017-10-20 Richard Biener <rguenther@suse.de>
1281 PR tree-optimization/82603
1282 * tree-if-conv.c (predicate_mem_writes): Make sure to only
1283 remove false predicated stores.
1285 2017-10-20 Richard Biener <rguenther@suse.de>
1287 * graphite-isl-ast-to-gimple.c
1288 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
1289 Remove return value and simplify, dump copied stmt after lhs
1291 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
1292 Reduce dump verbosity.
1293 (gsi_insert_earliest): Likewise.
1294 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
1295 * graphite.c (print_global_statistics): Adjust dumping.
1296 (print_graphite_scop_statistics): Likewise.
1297 (print_graphite_statistics): Do not dump loops here.
1298 (graphite_transform_loops): But here.
1300 2017-10-20 Nicolas Roche <roche@adacore.com>
1302 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
1303 * configure: Regenerate.
1305 2017-10-20 Jakub Jelinek <jakub@redhat.com>
1308 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
1309 functions when optimizing replace GIMPLE_RETURN stmts with
1310 calls to __builtin_unreachable ().
1313 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
1314 for -fsanitize=thread link of executables.
1315 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
1316 link of executables.
1319 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
1321 (<shift_insn><mode>3<mask_name>): Change the last of the 3
1322 define_insns for logical vector shifts to use VI248_AVX512BW
1323 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
1324 condition, useless isa and prefix attributes. Change the first
1325 2 of these define_insns to ...
1326 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
1327 define_insn for avx512vl.
1328 (<shift_insn><mode>3): ... and this, new define_insn without
1329 masking for non-avx512vl.
1332 * config/i386/sse.md (*andnot<mode>3,
1333 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
1334 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
1335 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
1336 not applied use empty suffix even for TARGET_AVX512VL.
1337 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
1338 is applied, supply evex,evex or evex,evex,evex instead of just
1341 2017-10-20 Julia Koval <julia.koval@intel.com>
1343 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
1344 (OPTION_MASK_ISA_GFNI_UNSET): New.
1345 (ix86_handle_option): Handle OPT_mgfni.
1346 * config/i386/cpuid.h (bit_GFNI): New.
1347 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
1348 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
1349 * config/i386/i386.c (ix86_target_string): Add -mgfni.
1350 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
1351 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
1352 * config/i386/i386.opt: Add mgfni.
1354 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
1356 * config/msp430/msp430.c (msp430_option_override): Disable
1357 -fdelete-null-pointer-checks.
1358 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
1360 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
1362 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
1363 of x87 and SSE instructions.
1365 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
1367 * asan.c (create_cond_insert_point): Do not update edge count.
1368 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
1369 (afdo_propagate_circuit): Likewise.
1370 (afdo_calculate_branch_prob): Likewise.
1371 (afdo_annotate_cfg): Likewise.
1372 * basic-block.h (struct edge_def): Remove count.
1373 (edge_def::count): New accessor.
1374 * bb-reorder.c (rotate_loop): Update.
1375 (find_traces_1_round): Update.
1376 (connect_traces): Update.
1377 (sanitize_hot_paths): Update.
1378 * cfg.c (unchecked_make_edge): Update.
1379 (make_single_succ_edge): Update.
1380 (check_bb_profile): Update.
1381 (dump_edge_info): Update.
1382 (update_bb_profile_for_threading): Update.
1383 (scale_bbs_frequencies_int): Update.
1384 (scale_bbs_frequencies_gcov_type): Update.
1385 (scale_bbs_frequencies_profile_count): Update.
1386 (scale_bbs_frequencies): Update.
1387 * cfganal.c (connect_infinite_loops_to_exit): Update.
1388 * cfgbuild.c (compute_outgoing_frequencies): Update.
1389 (find_many_sub_basic_blocks): Update.
1390 * cfgcleanup.c (try_forward_edges): Update.
1391 (try_crossjump_to_edge): Update
1392 * cfgexpand.c (expand_gimple_cond): Update
1393 (expand_gimple_tailcall): Update
1394 (construct_exit_block): Update
1395 * cfghooks.c (verify_flow_info): Update
1396 (redirect_edge_succ_nodup): Update
1397 (split_edge): Update
1398 (make_forwarder_block): Update
1399 (duplicate_block): Update
1400 (account_profile_record): Update
1401 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
1402 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
1403 * cfgloopmanip.c (scale_loop_profile): Update.
1405 (lv_adjust_loop_entry_edge): Update.
1406 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
1407 (force_nonfallthru_and_redirect): Update.
1408 (purge_dead_edges): Update.
1409 (rtl_flow_call_edges_add): Update.
1410 * cgraphunit.c (init_lowered_empty_function): Update.
1411 (cgraph_node::expand_thunk): Update.
1412 * gimple-pretty-print.c (dump_probability): Update.
1413 (dump_edge_probability): Update.
1414 * gimple-ssa-isolate-paths.c (isolate_path): Update.
1415 * haifa-sched.c (sched_create_recovery_edges): Update.
1416 * hsa-gen.c (convert_switch_statements): Update.
1417 * ifcvt.c (dead_or_predicable): Update.
1418 * ipa-inline-transform.c (inline_transform): Update.
1419 * ipa-split.c (split_function): Update.
1420 * ipa-utils.c (ipa_merge_profiles): Update.
1421 * loop-doloop.c (add_test): Update.
1422 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
1423 * lto-streamer-in.c (input_cfg): Update.
1424 (input_function): Update.
1425 * lto-streamer-out.c (output_cfg): Update.
1426 * modulo-sched.c (sms_schedule): Update.
1427 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
1428 * predict.c (maybe_hot_edge_p): Update.
1429 (unlikely_executed_edge_p): Update.
1430 (probably_never_executed_edge_p): Update.
1431 (dump_prediction): Update.
1432 (drop_profile): Update.
1433 (propagate_unlikely_bbs_forward): Update.
1434 (determine_unlikely_bbs): Update.
1435 (force_edge_cold): Update.
1436 * profile.c (compute_branch_probabilities): Update.
1437 * reg-stack.c (better_edge): Update.
1438 * shrink-wrap.c (handle_simple_exit): Update.
1439 * tracer.c (better_p): Update.
1440 * trans-mem.c (expand_transaction): Update.
1441 (split_bb_make_tm_edge): Update.
1442 * tree-call-cdce.c: Update.
1443 * tree-cfg.c (gimple_find_sub_bbs): Update.
1444 (gimple_split_edge): Update.
1445 (gimple_duplicate_sese_region): Update.
1446 (gimple_duplicate_sese_tail): Update.
1447 (gimple_flow_call_edges_add): Update.
1448 (insert_cond_bb): Update.
1449 (execute_fixup_cfg): Update.
1450 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
1451 * tree-complex.c (expand_complex_div_wide): Update.
1452 * tree-eh.c (lower_resx): Update.
1453 (unsplit_eh): Update.
1454 (cleanup_empty_eh_move_lp): Update.
1455 * tree-inline.c (copy_edges_for_bb): Update.
1456 (freqs_to_counts): Update.
1457 (copy_cfg_body): Update.
1458 * tree-ssa-dce.c (remove_dead_stmt): Update.
1459 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
1460 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
1461 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
1462 (unloop_loops): Update.
1463 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
1464 * tree-ssa-loop-split.c (connect_loops): Update.
1465 (split_loop): Update.
1466 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
1467 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
1468 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
1469 * tree-ssa-reassoc.c (branch_fixup): Update.
1470 * tree-ssa-tail-merge.c (replace_block_by): Update.
1471 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
1472 (compute_path_counts): Update.
1473 (update_profile): Update.
1474 (recompute_probabilities): Update.
1475 (update_joiner_offpath_counts): Update.
1476 (estimated_freqs_path): Update.
1477 (freqs_to_counts_path): Update.
1478 (clear_counts_path): Update.
1479 (ssa_fix_duplicate_block_edges): Update.
1480 (duplicate_thread_path): Update.
1481 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
1482 (case_bit_test_cmp): Update.
1483 (collect_switch_conv_info): Update.
1484 (gen_inbound_check): Update.
1485 (do_jump_if_equal): Update.
1486 (emit_cmp_and_jump_insns): Update.
1487 * tree-tailcall.c (decrease_profile): Update.
1488 (eliminate_tail_call): Update.
1489 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
1490 (vect_do_peeling): Update.
1491 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
1492 * ubsan.c (ubsan_expand_null_ifn): Update.
1493 (ubsan_expand_ptr_ifn): Update.
1494 * value-prof.c (gimple_divmod_fixed_value): Update.
1495 (gimple_mod_pow2): Update.
1496 (gimple_mod_subtract): Update.
1497 (gimple_ic): Update.
1498 (gimple_stringop_fixed_value): Update.
1500 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
1503 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
1505 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
1507 PR rtl-optimization/82395
1508 * ira-color.c (allocno_priority_compare_func): Fix comparison step
1509 based on non_spilled_static_chain_regno_p.
1511 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
1513 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
1514 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
1515 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
1517 2017-10-19 Martin Sebor <msebor@redhat.com>
1519 PR tree-optimization/82596
1520 * tree.c (array_at_struct_end_p): Handle STRING_CST.
1522 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
1524 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
1525 (get_mem_refs_of_builtin_call): Likewise.
1526 * builtins.c (expand_builtin_apply): Adjust call to
1527 allocate_dynamic_stack_space.
1528 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
1529 the third argument to allocate_dynamic_stack_space, otherwise -1.
1530 (expand_builtin): Deal with all alloca variants.
1531 (is_inexpensive_builtin): Likewise.
1532 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
1533 * calls.c (special_function_p): Deal with all alloca variants.
1534 (initialize_argument_information): Adjust call to
1535 allocate_dynamic_stack_space.
1536 (expand_call): Likewise.
1537 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
1538 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
1539 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
1540 use it for the stack usage computation.
1541 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
1542 * function.c (gimplify_parameters): Call build_alloca_call_expr.
1543 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
1544 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
1545 (in_loop_p): Remove first argument and useless check.
1546 (pass_walloca::execute): Remove useless test and adjust call to above.
1547 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
1548 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
1549 (gimplify_call_expr): Deal with all alloca variants.
1550 * hsa-gen.c (gen_hsa_alloca): Likewise.
1551 (gen_hsa_insns_for_call): Likewise.
1552 * ipa-pure-const.c (special_builtin_state): Likewise.
1553 * tree-chkp.c (chkp_build_returned_bound): Likewise.
1554 * tree-object-size.c (alloc_object_size): Likewise.
1555 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
1556 (call_may_clobber_ref_p_1): Likewise.
1557 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
1558 (ccp_fold_stmt): Likewise.
1559 (optimize_stack_restore): Likewise.
1560 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
1561 (mark_all_reaching_defs_necessary_1): Likewise.
1562 (propagate_necessity): Likewise.
1563 (eliminate_unnecessary_stmts): Likewise.
1564 * tree.c (build_common_builtin_nodes): Build
1565 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
1566 (build_alloca_call_expr): New function.
1567 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
1568 (CASE_BUILT_IN_ALLOCA): Likewise.
1569 (build_alloca_call_expr): Declare.
1570 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
1572 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
1575 * dwarf2out.c (new_die_raw): New static inline function.
1576 (new_die): Use it to create the DIE.
1577 (add_AT_external_die_ref): Likewise.
1578 (clone_die): Likewise.
1579 (clone_as_declaration): Likewise.
1580 (dwarf2out_vms_debug_main_pointer): Likewise.
1581 (base_type_die): Likewise. Remove early return for corner cases.
1582 Do not call add_pubtype on the DIE here.
1583 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
1584 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
1585 typedefs for base types with DW_AT_endianity. Make sure a DIE with
1586 native order exists for base types, attach the DIE manually and call
1587 add_pubtype on it. Do not equate a reverse order DIE to the type.
1589 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
1591 * config/arm/arm.c (align_ok_ldrd_strd): New function.
1592 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
1594 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
1596 2017-10-19 Jakub Jelinek <jakub@redhat.com>
1598 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
1599 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
1600 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
1601 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
1602 * opts.c (sanitizer_opts): Add builtin.
1603 * ubsan.c (instrument_builtin): New function.
1604 (pass_ubsan::execute): Call it.
1605 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
1606 * doc/invoke.texi: Document -fsanitize=builtin.
1608 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
1609 builtins, store max (log2 (align), 0) into uchar field instead of
1610 align into uptr field.
1611 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
1612 store uchar 0 field instead of uptr 0 field.
1613 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
1614 instead of passing one address of struct with 2 locations pass
1615 two addresses of structs with 1 location each.
1616 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
1617 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
1618 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
1619 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
1620 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
1621 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
1622 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
1623 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
1625 2017-10-19 Martin Liska <mliska@suse.cz>
1628 * file-find.c (remove_prefix): Remove.
1629 * file-find.h (remove_prefix): Likewise.
1630 * gcc-ar.c: Remove smartness of lookup.
1632 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
1634 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
1635 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
1636 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
1638 2017-10-19 Jakub Jelinek <jakub@redhat.com>
1641 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
1642 (setcc + and to xor + setcc): New peephole2.
1644 2017-10-19 Tom de Vries <tom@codesourcery.com>
1646 * doc/sourcebuild.texi (Test Directives, Variants of
1647 dg-require-support): Add dg-require-stack-size.
1649 2017-10-19 Martin Liska <mliska@suse.cz>
1652 * gimplify.c (gimplify_decl_expr): Do not instrument variables
1653 that have a large alignment.
1654 (gimplify_target_expr): Likewise.
1656 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
1658 PR rtl-optimization/82602
1659 * ira.c (rtx_moveable_p): Return false for volatile asm.
1661 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
1664 * config/i386/i386-modes.def (CCGZ): New CC mode.
1665 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
1666 * config/i386/predicates.md (ix86_comparison_operator):
1668 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
1669 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
1670 with double-word subtraction.
1671 (put_condition_code): Handle CCGZmode.
1673 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
1675 * wide-int.cc (debug (const wide_int &)): New.
1676 (debug (const wide_int *)): New.
1677 (debug (const widest_int &)): New.
1678 (debug (const widest_int *)): New.
1680 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
1683 * lra-constraints.c (curr_insn_transform): Use non-input operand
1684 instead of output one for matched reload.
1686 2017-10-18 Bin Cheng <bin.cheng@arm.com>
1688 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
1689 (tree-ssa-loop-ivopts.h): New header file.
1690 (struct builtin_info): New fields.
1691 (classify_builtin_1): Compute and record base and offset parts for
1692 memset builtin partition by calling strip_offset.
1693 (offset_cmp, fuse_memset_builtins): New functions.
1694 (finalize_partitions): Fuse adjacent memset partitions by calling
1696 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
1697 Expose the interface.
1698 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
1700 2017-10-18 Bin Cheng <bin.cheng@arm.com>
1702 PR tree-optimization/82574
1703 * tree-loop-distribution.c (find_single_drs): New parameter. Check
1704 that data reference must be executed exactly once per iteration
1705 against the outermost loop in nest.
1706 (classify_partition): Update call to above function.
1708 2017-10-18 Richard Biener <rguenther@suse.de>
1710 PR tree-optimization/82591
1711 * graphite.c (graphite_transform_loops): Move code gen message
1713 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1714 Here. Handle scop_to_isl_ast failing.
1715 (scop_to_isl_ast): Limit the number of ISL operations.
1717 2017-10-18 Richard Biener <rguenther@suse.de>
1719 * graphite-isl-ast-to-gimple.c
1720 (translate_isl_ast_to_gimple::set_rename): Simplify.
1721 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
1722 (graphite_copy_stmts_from_block): ... here.
1723 (copy_bb_and_scalar_dependences): Simplify.
1724 (add_parameters_to_ivs_params): Canonicalize.
1725 (generate_entry_out_of_ssa_copies): Simplify.
1726 * graphite-sese-to-poly.c (extract_affine_name): Simplify
1727 by passing in ISL dimension.
1728 (parameter_index_in_region_1): Rename to ...
1729 (parameter_index_in_region): ... this.
1730 (extract_affine): Adjust assert, pass down parameter index.
1731 (add_param_constraints): Use range-info when available.
1732 (build_scop_context): Adjust.
1733 * sese.c (new_sese_info): Adjust.
1734 (free_sese_info): Likewise.
1735 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
1736 Remove unused typedefs.
1737 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
1739 2017-10-18 Martin Liska <mliska@suse.cz>
1741 * combine.c (simplify_compare_const): Add gcc_fallthrough.
1743 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1745 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
1746 (s390_sched_init): Do not reset s390_sched_state if we entered the
1747 current basic block via a fallthru edge and all others are unlikely.
1749 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1751 * config/s390/s390.c (NUM_SIDES): New variable.
1752 (LONGRUNNING_THRESHOLD): New variable.
1753 (LATENCY_FACTOR): New variable.
1754 (s390_sched_score): Decrease score for long-running instructions on
1756 (s390_sched_variable_issue): Perform bookkeeping for long-running
1759 2017-10-18 Richard Biener <rguenther@suse.de>
1761 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
1762 Simplify with removal of the parameter rename map.
1763 (set_rename): Likewise.
1764 (should_copy_to_new_region): Likewise.
1765 (graphite_copy_stmts_from_block): Likewise.
1766 (copy_bb_and_scalar_dependences): Remove initialization of
1767 unused copied_bb_map.
1769 (copy_internal_parameters): Likewise.
1770 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
1771 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
1772 Use INTEGRAL_TYPE_P.
1773 (parameter_index_in_region_1): Rename to ...
1774 (assign_parameter_index_in_region): ... this. Assert we have
1775 a parameter we handle.
1776 (scan_tree_for_params): Adjust.
1777 * sese.h (parameter_rename_map_t): Remove.
1778 (struct sese_info_t): Remove unused parameter_rename_map and
1779 copied_bb_map members.
1780 * sese.c (new_sese_info): Adjust.
1781 (free_sese_info): Likewise.
1783 2017-10-18 Martin Liska <mliska@suse.cz>
1786 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
1787 on an abnormal edge.
1789 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
1791 * doc/invoke.texi (ffunction-sections and fdata-sections):
1794 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
1796 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
1797 the use statement can throw internally.
1799 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
1801 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
1802 any RTX present on the RHS of a SET.
1803 * compare-elim.c (try_eliminate_compare): Restore comment.
1805 2017-10-17 Jakub Jelinek <jakub@redhat.com>
1807 * langhooks.h (struct lang_hooks): Document that tree_size langhook
1808 may be also called on tcc_type nodes.
1809 * langhooks.c (lhd_tree_size): Likewise.
1811 2017-10-17 David Malcolm <dmalcolm@redhat.com>
1813 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
1814 format_warning_at_substring.
1815 (maybe_warn): Convert source_range * param to a location_t. Pass
1816 UNKNOWN_LOCATION rather than NULL to fmtwarn.
1817 (format_directive): Remove code to extract source_ranges and
1818 source_range * in favor of just a location_t.
1819 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
1821 * substring-locations.c (format_warning_va): Convert
1822 source_range * param to a location_t.
1823 (format_warning_at_substring): Likewise.
1824 * substring-locations.h (format_warning_va): Likewise.
1825 (format_warning_at_substring): Likewise.
1827 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
1829 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
1831 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
1832 and scatter/gather ops.
1834 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
1835 vec_gather_load and vec_scatter_store.
1836 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
1837 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
1838 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
1839 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
1840 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
1841 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
1843 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
1845 * reg-stack.c (compare_for_stack_reg): Add bool argument.
1846 Detect FTST instruction and handle its register pops. Only pop
1847 second operand if can_pop_second_op is true.
1848 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
1849 set can_pop_second_op to false in the compare_for_stack_reg call.
1851 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
1852 output_fp_compare for stack register operands.
1853 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
1854 instructions here. Do not emit stack register pops here. Assert
1855 that FCOMPP pops next to top stack register. Rewrite function.
1857 2017-10-17 Nathan Sidwell <nathan@acm.org>
1860 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
1861 use DECL_ASSEMBLER_NAME_RAW.
1864 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
1867 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
1868 Wilco Dijkstra <wilco.dijkstra@arm.com>
1870 * builtins.c (expand_builtin_update_setjmp_buf): Add a
1871 converstion to Pmode from the buf_addr.
1873 2017-10-17 Richard Biener <rguenther@suse.de>
1875 * graphite-dependences.c (scop_get_reads_and_writes): Change
1876 output parameters to references.
1878 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
1880 PR 71026/tree-optimization
1881 * fold-const.c (distribute_real_division): Removed.
1882 (fold_binary_loc): Remove calls to distribute_real_divison.
1884 2017-10-17 Richard Biener <rguenther@suse.de>
1886 * graphite-scop-detection.c
1887 (scop_detection::stmt_has_simple_data_refs_p): Always use
1888 the full nest as region.
1889 (try_generate_gimple_bb): Likewise.
1890 * sese.c (scalar_evolution_in_region): Simplify now that
1891 SCEV can handle instantiation in regions.
1892 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
1893 in the non-loop part of a function if requested.
1895 2017-10-17 Richard Biener <rguenther@suse.de>
1897 PR tree-optimization/82563
1898 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
1900 (graphite_regenerate_ast_isl): Call it.
1901 * graphite-scop-detection.c (build_scops): Remove entry edge split.
1903 2017-10-17 Jakub Jelinek <jakub@redhat.com>
1905 PR tree-optimization/82549
1906 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
1907 Formatting fixes. Instead of calling make_bit_field_ref with negative
1910 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
1912 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
1913 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
1914 _mm_maskz_reduce_ss): New.
1915 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
1916 __builtin_ia32_reducess_mask): Ditto..
1917 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
1918 * config/i386/sse.md (reduces<mode>): Renamed to ...
1919 (reduces<mode><mask_scalar_name>): ... this.
1920 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
1922 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
1923 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
1925 2017-10-16 David Malcolm <dmalcolm@redhat.com>
1927 * Makefile.in (OBJS): Add unique-ptr-tests.o.
1928 * selftest-run-tests.c (selftest::run_tests): Call
1929 selftest::unique_ptr_tests_cc_tests.
1930 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
1931 * unique-ptr-tests.cc: New file.
1933 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
1936 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1938 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1939 (make_hard_regno_born, make_hard_regno_dead): Update
1940 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
1942 2017-10-16 Jeff Law <law@redhat.com>
1944 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
1946 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
1948 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
1950 2017-10-16 Olivier Hainque <hainque@adacore.com>
1952 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
1953 with_cpu if we were configured for an e500v2 target cpu name.
1955 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
1957 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
1958 * doc/invoke.texi: Document +nodsp as a valid extension for
1961 2017-10-16 Martin Liska <mliska@suse.cz>
1963 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
1964 (test_set_range): Likewise.
1965 (test_range_functions): Rename to ...
1966 (test_bit_in_range): ... this.
1967 (sbitmap_c_tests): Add new test.
1969 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1971 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
1973 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
1974 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
1976 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1978 * config/aarch64/aarch64-builtins.c
1979 (aarch64_types_quadopu_lane_qualifiers): New.
1980 (TYPES_QUADOPU_LANE): New.
1981 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
1982 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
1983 (aarch64_<sur>dot_laneq<vsi2qi>): New.
1984 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
1985 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
1986 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
1987 (Vdottype, DOTPROD): New.
1988 (sur): Add SDOT and UDOT.
1990 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1992 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
1993 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
1994 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
1996 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
1997 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
1998 Enable TARGET_DOTPROD.
1999 (cortex-a75.cortex-a55): Likewise.
2000 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
2002 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2004 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
2005 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
2007 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
2009 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
2010 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
2011 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
2012 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
2013 * config/arm/types.md (neon_dot, neon_dot_q): New.
2014 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
2016 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2018 * config/arm/arm.h (TARGET_DOTPROD): New.
2019 * config/arm/arm.c (arm_arch_dotprod): New.
2020 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
2021 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
2022 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
2023 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
2024 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
2025 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
2026 * doc/invoke.texi (armv8.2-a): Document dotprod
2028 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
2030 * i386.c (ix86_vec_cost): New function.
2031 (ix86_rtx_costs): Handle vector operations better.
2032 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
2033 * x86-tune-costs.h: Add new costs to all tables.
2035 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
2037 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
2039 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
2040 divsd, sqrtss and sqrtsd
2041 * x86-tune-costs.h: Add new entries to all costs.
2042 (znver1_cost): Fix to match real instruction latencies.
2044 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2045 Michael Collison <michael.collison@arm.com>
2047 * compare-elim.c: Include emit-rtl.h.
2048 (can_merge_compare_into_arith): New function.
2049 (try_validate_parallel): Likewise.
2050 (try_merge_compare): Likewise.
2051 (try_eliminate_compare): Call the above when no previous clobber
2053 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
2056 2017-10-14 Jakub Jelinek <jakub@redhat.com>
2060 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
2061 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
2065 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
2066 to be any operand_equal_p operands. For & (B - 1) require
2067 B to be power of 2. Recognize
2068 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
2070 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2073 * optabs.c (expand_memory_blockage): Fix call of
2074 targetm.have_memory_blockage.
2076 2017-10-14 Jakub Jelinek <jakub@redhat.com>
2079 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
2080 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
2081 objects to extra_objs instead of overwriting it.
2083 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2085 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
2086 Use any_fp_register_operand as operand[3] predicate. Simplify
2087 equality test for operands[2] and operands[4] memory location.
2088 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
2089 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
2090 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
2091 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
2092 any_fp_register_operand as operand[1] predicate. Simplify
2093 equality test for operands[0] and operands[3] memory location.
2094 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
2095 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
2096 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
2098 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2100 * target-insns.def: Add memory_blockage.
2101 * optabs.c (expand_memory_blockage): New function.
2102 (expand_asm_memory_barrier): Rename ...
2103 (expand_asm_memory_blockage): ... to this.
2104 (expand_mem_thread_fence): Call expand_memory_blockage
2105 instead of expand_asm_memory_barrier.
2106 (expand_mem_singnal_fence): Ditto.
2107 (expand_atomic_load): Ditto.
2108 (expand_atomic_store): Ditto.
2109 * doc/md.texi (Standard Pattern Names For Generation):
2110 Document memory_blockage instruction pattern.
2112 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
2114 * config/rl78/rl78.c (rl78_emit_libcall): New function.
2115 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
2116 * config/rl78/rl78.md: New define_expand "adddi3".
2118 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
2120 * cfghooks.c (verify_flow_info): Disable check that all probabilities
2123 2017-10-13 Jeff Law <law@redhat.com>
2125 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
2127 2017-10-13 Jakub Jelinek <jakub@redhat.com>
2130 * internal-fn.c (expand_mul_overflow): If both operands have
2131 the same highpart of -1 or 0 and the topmost bit of lowpart
2132 is different, overflow is if res <= 0 rather than res < 0.
2134 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
2136 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
2137 TARGET_P9_VECTOR code for unaligned_load case.
2139 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
2141 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
2143 2017-10-13 Nathan Sidwell <nathan@acm.org>
2145 * tree-core.h (tree_contains_struct): Make bool.
2146 * tree.c (tree_contains_struct): Likewise.
2147 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
2148 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
2149 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
2150 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
2151 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
2153 2017-10-13 Richard Biener <rguenther@suse.de>
2155 * graphite-isl-ast-to-gimple.c
2156 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
2157 parameters and dominance check.
2158 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
2159 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
2160 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
2161 Do not update SSA form here or do intermediate IL verification.
2162 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
2163 (graphite_initialize): Remove check on the number of loops in
2164 the function and inline into graphite_transform_loops.
2165 (graphite_finalize): Inline into graphite_transform_loops.
2166 (graphite_transform_loops): Perform SSA update and IL verification
2168 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
2170 2017-10-13 Richard Biener <rguenther@suse.de>
2172 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
2173 graphite_expression_type_precision): Avoid global constructor
2175 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
2176 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
2177 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
2178 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
2179 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
2180 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
2182 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
2185 * config/i386/i386.h (ix86_red_zone_size): New.
2186 * config/i386/i386.md (push peephole2s): Replace
2187 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
2189 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2190 Alan Hayward <alan.hayward@arm.com>
2191 David Sherwood <david.sherwood@arm.com>
2193 * combine.c (can_change_dest_mode): Reject changes in
2194 REGMODE_NATURAL_SIZE.
2196 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2197 Alan Hayward <alan.hayward@arm.com>
2198 David Sherwood <david.sherwood@arm.com>
2200 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
2201 (expand_debug_source_expr): Likewise.
2202 * combine.c (combine_simplify_rtx): Likewise.
2203 * cse.c (fold_rtx): Likewise.
2204 * fwprop.c (canonicalize_address): Likewise.
2205 * targhooks.c (default_shift_truncation_mask): Likewise.
2207 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2208 Alan Hayward <alan.hayward@arm.com>
2209 David Sherwood <david.sherwood@arm.com>
2211 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
2212 (widened_mode): Likewise.
2213 (expand_unop): Likewise.
2214 * ree.c (transform_ifelse): Likewise.
2215 (merge_def_and_ext): Likewise.
2216 (combine_reaching_defs): Likewise.
2217 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2219 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2220 Alan Hayward <alan.hayward@arm.com>
2221 David Sherwood <david.sherwood@arm.com>
2223 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
2224 * combine.c (gen_lowpart_for_combine): Likewise.
2225 * dwarf2out.c (rtl_for_decl_location): Likewise.
2226 * final.c (alter_subreg): Likewise.
2227 * rtlhooks.c (gen_lowpart_general): Likewise.
2228 (gen_lowpart_if_possible): Likewise.
2230 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2231 Alan Hayward <alan.hayward@arm.com>
2232 David Sherwood <david.sherwood@arm.com>
2234 * calls.c (expand_call): Use subreg_lowpart_offset.
2235 * cse.c (cse_insn): Likewise.
2236 * regcprop.c (copy_value): Likewise.
2237 (copyprop_hardreg_forward_1): Likewise.
2239 2017-10-13 Jakub Jelinek <jakub@redhat.com>
2242 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
2243 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
2244 =Q constraints to +Q and into insn condition add check
2245 that operands[0] and operands[1] are equal.
2246 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
2247 =Q constraints to +Q and into insn condition add check
2248 that operands[0] is equal to either operands[1] or operands[2].
2251 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
2252 instead of handling MINUS_EXPR twice (once for each argument),
2253 canonicalize operand order and handle just once, use rtype where
2254 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
2257 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
2258 any values of __C while still being pattern recognizable as a simple
2261 2017-10-13 Richard Biener <rguenther@suse.de>
2263 PR tree-optimization/82451
2265 2017-10-02 Richard Biener <rguenther@suse.de>
2267 PR tree-optimization/82355
2268 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
2269 a mapping for the enclosing loop but avoid generating one for
2271 (copy_bb_and_scalar_dependences): Remove premature codegen
2272 error on PHIs in blocks duplicated into multiple places.
2273 * graphite-scop-detection.c
2274 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
2275 in the region use it as loop and nest to analyze the DR in.
2276 (try_generate_gimple_bb): Likewise.
2277 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
2278 (add_loop_constraints): For blocks in a loop not in the region
2279 create a dimension with a single iteration.
2280 * sese.h (gbb_loop_at_index): Remove assert.
2282 * cfgloop.c (loop_preheader_edge): For the loop tree root
2283 return the single successor of the entry block.
2284 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2285 Reset the SCEV hashtable and niters.
2286 * graphite-scop-detection.c
2287 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
2288 assert that we only have POLYNOMIAL_CHREC that vary in loops
2289 contained in the region.
2290 (scop_detection::graphite_can_represent_expr): Adjust.
2291 (scop_detection::stmt_has_simple_data_refs_p): For loops
2292 not in the region set loop to NULL. The nest is now the
2293 entry edge to the region.
2294 (try_generate_gimple_bb): Likewise.
2295 * sese.c (scalar_evolution_in_region): Adjust for
2296 instantiate_scev change.
2297 * tree-data-ref.h (graphite_find_data_references_in_stmt):
2298 Make nest parameter the edge into the region.
2299 (create_data_ref): Likewise.
2300 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
2301 entry edge into a region and adjust instantiate_scev calls.
2302 (create_data_ref): Likewise.
2303 (graphite_find_data_references_in_stmt): Likewise.
2304 (find_data_references_in_stmt): Pass the loop preheader edge
2305 from the nest argument.
2306 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
2307 parameter the edge into the region.
2308 (instantiate_parameters): Use the loop preheader edge as entry.
2309 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
2311 (get_instantiated_value_entry): Make instantiate_below parameter
2312 the edge into the region.
2313 (instantiate_scev_name): Likewise. Adjust dominance checks,
2314 when we cannot use loop-based instantiation instantiate by
2315 walking use-def chains.
2316 (instantiate_scev_poly): Adjust.
2317 (instantiate_scev_binary): Likewise.
2318 (instantiate_scev_convert): Likewise.
2319 (instantiate_scev_not): Likewise.
2320 (instantiate_array_ref): Remove.
2321 (instantiate_scev_3): Likewise.
2322 (instantiate_scev_2): Likewise.
2323 (instantiate_scev_1): Likewise.
2324 (instantiate_scev_r): Do not blindly handle N-operand trees.
2325 Do not instantiate array-refs. Handle all constants and invariants.
2326 (instantiate_scev): Make instantiate_below parameter
2327 the edge into the region.
2328 (resolve_mixers): Use the loop preheader edge for the region
2329 parameter to instantiate_scev_r.
2330 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
2332 2017-10-13 Richard Biener <rguenther@suse.de>
2334 PR tree-optimization/82525
2335 * graphite-isl-ast-to-gimple.c
2336 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
2338 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
2339 Fail code generation when we cannot represent the isl integer.
2340 (binary_op_to_tree): Elide modulo operations that are no-ops
2341 in the type we code generate. Remove now superfluous code
2344 2017-10-13 Richard Biener <rguenther@suse.de>
2346 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
2347 (scop_detection::harmful_loop_in_region): Remove premature
2348 IV type restriction.
2349 (scop_detection::graphite_can_represent_scev): We can handle
2350 pointer IVs just fine.
2352 2017-10-13 Alan Modra <amodra@gmail.com>
2354 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
2355 "Clobbers and Scratch Registers". Add paragraph on
2356 alternative to clobbers for scratch registers and OpenBLAS
2359 2017-10-13 Alan Modra <amodra@gmail.com>
2361 * doc/extend.texi (Clobbers): Correct vax example. Delete old
2362 example of a memory input for a string of known length. Move
2363 commentary out of table. Add a number of new examples
2364 covering array memory inputs.
2366 2017-10-12 Martin Liska <mliska@suse.cz>
2368 PR tree-optimization/82493
2369 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
2370 (test_range_functions): New function.
2371 (sbitmap_c_tests): Likewise.
2372 * selftest-run-tests.c (selftest::run_tests): Run new tests.
2373 * selftest.h (sbitmap_c_tests): New function.
2375 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
2377 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
2379 * config/rs6000/amo.h: Fix spacing issue.
2381 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2384 * config/i386/i386.md (*ashl<mode>3_mask_1,
2385 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
2386 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
2389 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
2391 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
2392 (profile_probability): Set max_probability
2393 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
2395 * profile-count.c (profile_probability::differs_from_p): Do not
2396 rely on max_probaiblity == 10000
2398 2017-10-12 Jeff Law <law@redhat.com>
2400 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
2403 2017-10-12 Martin Sebor <msebor@redhat.com>
2407 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
2408 (handle_alias_pairs): Call it.
2409 * common.opt (-Wattribute-alias): New option.
2410 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
2411 * doc/invoke.texi (-Wattribute-alias): Document.
2413 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
2416 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
2418 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2420 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2421 (make_hard_regno_born, make_hard_regno_dead): Update
2422 bb_killed_pseudos and bb_gen_pseudos.
2424 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
2426 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
2428 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
2430 * config/alpha/alpha.c (alpha_split_conditional_move):
2431 Use std::swap instead of manually swapping.
2432 (alpha_stdarg_optimize_hook): Ditto.
2433 (alpha_canonicalize_comparison): Ditto.
2435 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2437 * tree-loop-distribution.c (struct builtin_info): New struct.
2438 (struct partition): Refactor fields into struct builtin_info.
2439 (partition_free): Free struct builtin_info.
2440 (build_size_arg_loc, build_addr_arg_loc): Delete.
2441 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
2442 information from struct builtin_info.
2443 (find_single_drs): New function refactored from classify_partition.
2444 Also moved builtin validity checks to this function.
2445 (compute_access_range, alloc_builtin): New functions.
2446 (classify_builtin_st, classify_builtin_ldst): New functions.
2447 (classify_partition): Refactor code into functions find_single_drs,
2448 classify_builtin_st and classify_builtin_ldst.
2449 (distribute_loop): Don't do runtime alias check when distributing
2451 (find_seed_stmts_for_distribution): New function.
2452 (pass_loop_distribution::execute): Refactor code finding seed
2453 stmts into above function. Support distribution for the innermost
2454 two-level loop nest. Adjust dump information.
2456 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2458 * tree-loop-distribution.c: Adjust the general comment.
2459 (NUM_PARTITION_THRESHOLD): New macro.
2460 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
2461 (classify_partition): Skip builtin pattern of loop nest's inner loop.
2462 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
2463 in call to build_partition_graph.
2464 (finalize_partitions): New parameter. Make loop distribution more
2465 conservative by fusing more partitions.
2466 (distribute_loop): Don't do runtime alias check in case of loop nest
2468 (find_seed_stmts_for_distribution): New function.
2469 (prepare_perfect_loop_nest): New function.
2470 (pass_loop_distribution::execute): Refactor code finding seed stmts
2471 and loop nest into above functions. Support loop nest distribution.
2472 Adjust dump information accordingly.
2474 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2476 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
2477 and set PTYPE_SEQUENTIAL for merged partition.
2479 2017-10-12 Richard Biener <rguenther@suse.de>
2481 PR tree-optimization/69728
2483 2017-09-19 Richard Biener <rguenther@suse.de>
2485 PR tree-optimization/69728
2486 * graphite-sese-to-poly.c (schedule_error): New global.
2487 (add_loop_schedule): Handle empty domain by failing the
2489 (build_original_schedule): Handle schedule_error.
2491 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
2492 domain by returning an unchanged schedule.
2494 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2496 * genrecog.c (validate_pattern): For VEC_SELECT verify that
2497 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
2499 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
2501 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
2504 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2507 * expr.c (store_field): Don't optimize away bitsize == 0 store
2508 from CALL_EXPR with addressable return type.
2510 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
2512 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
2513 * config/rs6000/rs6000.md (sel): Delete mode attribute.
2514 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
2515 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
2516 TARGET_ISEL instead of TARGET_ISEL<sel>.
2518 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
2520 * config/rs6000/rs6000.c
2521 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
2523 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
2525 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
2527 (reg_or_cint_operand): Fix comment.
2528 (reg_or_zero_operand): New predicate.
2529 * config/rs6000/rs6000-protos.h (output_isel): Delete.
2530 * config/rs6000/rs6000.c (output_isel): Delete.
2531 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
2532 instead of reg_or_cint_operand. Output instruction directly (not via
2534 (isel_unsigned_<mode>): Ditto.
2535 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
2536 gpc_reg_operand. Add an instruction alternative for this. Output
2537 instruction directly.
2538 (*isel_reversed_unsigned_<mode>): Ditto.
2540 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2542 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
2543 (TARGET_CANONICALIZE_COMPARISON): Define.
2545 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
2548 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
2549 Check whether the dest is REG before adding REG_EQUIV note.
2551 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
2554 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2556 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2557 (make_hard_regno_born, make_hard_regno_dead): Update
2558 bb_killed_pseudos and bb_gen_pseudos.
2560 2017-10-11 Nathan Sidwell <nathan@acm.org>
2562 * incpath.h (enum incpath_kind): Name enum, prefix values.
2563 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
2564 * incpath.c (heads, tails): Use INC_MAX.
2565 (add_env_var_paths, add_standard_paths): Use incpath_kind.
2566 (merge_include_chains, split_quote_chain,
2567 register_include_chains): Update incpath_kind names.
2568 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
2569 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
2571 (add_framework_path, darwin_register_objc_includes): Likewise.
2572 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
2574 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2576 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
2577 Do not use float_operator operator predicate.
2578 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
2579 * config/i386/predicates.md (float_operator): Remove predicate.
2581 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2583 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
2584 (*jccxf_i387): Ditto.
2585 (*jcc<mode>_i387): Ditto.
2586 (*jccu<mode>_i387): Ditto.
2587 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
2588 (*jcc_*_i387 splitters): Remove.
2589 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
2590 * config/i386/i386.c (ix86_split_fp_branch): Remove.
2591 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
2594 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
2596 * profile-count.h (slow_safe_scale_64bit): New function.
2597 (safe_scale_64bit): New inline.
2598 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
2599 * profile-count.c: Include wide-int.h
2600 (slow_safe_scale_64bit): New.
2602 2017-10-11 Nathan Sidwell <nathan@acm.org>
2604 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
2605 HAS_DECL_ASSEMBLER_NAME_P.
2606 * gimple-expr.c (gimple_decl_printable_name: Check
2607 HAS_DECL_ASSEMBLER_NAME_P too.
2608 * ipa-utils.h (type_in_anonymous_namespace_p): Check
2609 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
2610 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
2611 * passes.c (rest_of_decl_compilation): Check
2612 HAS_DECL_ASSEMBLER_NAME_P too.
2613 * recog.c (verify_changes): Likewise.
2614 * tree-pretty-print.c (dump_decl_name): Likewise.
2615 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
2617 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
2618 (SET_DECL_ASSEMBLER_NAME): Use it.
2619 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
2620 (COPY_DECL_ASSEMBLER_NAME): Likewise.
2621 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
2623 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
2625 * config.gcc (i386, x86_64): Add extra objects.
2626 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
2627 (ix86_min_insn_size): Declare.
2628 (ix86_issue_rate): Declare.
2629 (ix86_adjust_cost): Declare.
2630 (ia32_multipass_dfa_lookahead): Declare.
2631 (ix86_macro_fusion_p): Declare.
2632 (ix86_macro_fusion_pair_p): Declare.
2633 (ix86_bd_has_dispatch): Declare.
2634 (ix86_bd_do_dispatch): Declare.
2635 (ix86_core2i7_init_hooks): Declare.
2636 (ix86_atom_sched_reorder): Declare.
2637 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
2638 (COSTS_N_BYTES): Move to x86-tune-costs.h.
2639 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
2640 (rip_relative_addr_p): Rename to ...
2641 (ix86_rip_relative_addr_p): ... this one; export.
2642 (memory_address_length): Update.
2643 (ix86_issue_rate): Move to x86-tune-sched.c.
2644 (ix86_flags_dependent): Move to x86-tune-sched.c.
2645 (ix86_agi_dependent): Move to x86-tune-sched.c.
2646 (exact_dependency_1): Move to x86-tune-sched.c.
2647 (exact_store_load_dependency): Move to x86-tune-sched.c.
2648 (ix86_adjust_cost): Move to x86-tune-sched.c.
2649 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
2650 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
2651 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
2652 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
2653 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
2654 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
2655 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
2656 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
2657 (min_insn_size): Rename to ...
2658 (ix86_min_insn_size): ... this one; export.
2659 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
2660 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
2661 (core2i7_first_cycle_multipass_backtrack): Move to
2662 x86-tune-sched-core.c.
2663 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
2664 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
2665 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
2666 (ix86_avoid_jump_mispredicts): Update.
2667 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
2668 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
2669 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
2670 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
2671 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
2672 (MAX_INSN): Move to ix86-tune-sched-bd.c.
2673 (MAX_IMM): Move to ix86-tune-sched-bd.c.
2674 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
2675 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
2676 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
2677 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
2678 (MAX_STORE): Move to ix86-tune-sched-bd.c.
2679 (BIG): Move to ix86-tune-sched-bd.c.
2680 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
2681 (enum insn_path): Move to ix86-tune-sched-bd.c.
2682 (get_mem_group): Move to ix86-tune-sched-bd.c.
2683 (is_cmp): Move to ix86-tune-sched-bd.c.
2684 (dispatch_violation): Move to ix86-tune-sched-bd.c.
2685 (is_branch): Move to ix86-tune-sched-bd.c.
2686 (is_prefetch): Move to ix86-tune-sched-bd.c.
2687 (init_window): Move to ix86-tune-sched-bd.c.
2688 (allocate_window): Move to ix86-tune-sched-bd.c.
2689 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
2690 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
2691 (process_end_window): Move to ix86-tune-sched-bd.c.
2692 (allocate_next_window): Move to ix86-tune-sched-bd.c.
2693 (find_constant): Move to ix86-tune-sched-bd.c.
2694 (get_num_immediates): Move to ix86-tune-sched-bd.c.
2695 (has_immediate): Move to ix86-tune-sched-bd.c.
2696 (get_insn_path): Move to ix86-tune-sched-bd.c.
2697 (get_insn_group): Move to ix86-tune-sched-bd.c.
2698 (count_num_restricted): Move to ix86-tune-sched-bd.c.
2699 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
2700 (add_insn_window): Move to ix86-tune-sched-bd.c.
2701 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
2702 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
2703 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
2704 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
2705 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
2706 (do_dispatch): Move to ix86-tune-sched-bd.c.
2707 (has_dispatch): Move to ix86-tune-sched-bd.c.
2708 * config/i386/t-i386: Add new object files.
2709 * config/i386/x86-tune-costs.h: New file.
2710 * config/i386/x86-tune-sched-atom.c: New file.
2711 * config/i386/x86-tune-sched-bd.c: New file.
2712 * config/i386/x86-tune-sched-core.c: New file.
2713 * config/i386/x86-tune-sched.c: New file.
2715 2017-10-11 Liu Hao <lh_mouse@126.com>
2717 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
2718 the generic version below instead.
2719 (should_colorize): Recognize Windows consoles as terminals
2721 * pretty-print.c [__MINGW32__] (write_all): New function.
2722 [__MINGW32__] (find_esc_head): Likewise.
2723 [__MINGW32__] (find_esc_terminator): Likewise.
2724 [__MINGW32__] (eat_esc_sequence): Likewise.
2725 [__MINGW32__] (mingw_ansi_fputs): New function that handles
2727 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
2730 2017-10-11 Richard Biener <rguenther@suse.de>
2732 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
2733 Properly call analyze_scalar_evolution with the loop of the stmt.
2735 2017-10-11 Richard Biener <rguenther@suse.de>
2737 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
2738 * tree-core.h (tree_base): Add chrec_var union member.
2739 * tree.h (CHREC_VAR): Remove.
2740 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
2741 * tree-chrec.h (build_polynomial_chrec): Adjust.
2742 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
2743 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
2745 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
2747 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
2748 * match.pd: ... here.
2749 ((T) X == (T) Y): Relax condition.
2751 2017-10-11 Bin Cheng <bin.cheng@arm.com>
2753 PR tree-optimization/82472
2754 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
2756 (break_alias_scc_partitions): Update postorder number.
2758 2017-10-11 Martin Liska <mliska@suse.cz>
2761 * opts.c (parse_no_sanitize_attribute): Do not use error_value
2763 * opts.h (parse_no_sanitize_attribute): Remove last argument.
2765 2017-10-11 Martin Liska <mliska@suse.cz>
2767 * print-rtl.c (print_insn): Move declaration of idbuf
2768 to same scope as name.
2770 2017-10-11 Martin Liska <mliska@suse.cz>
2775 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
2778 2017-10-11 Martin Liska <mliska@suse.cz>
2781 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
2784 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
2786 * config.gcc (default_gnu_indirect_function): Default to yes for
2787 arm*-*-linux* with glibc.
2789 2017-10-11 Richard Biener <rguenther@suse.de>
2791 * tree-scalar-evolution.c (get_scalar_evolution): Handle
2792 default-defs and types we do not want to analyze.
2793 (interpret_loop_phi): Replace unreachable code with an assert.
2794 (compute_scalar_evolution_in_loop): Remove and inline ...
2795 (analyze_scalar_evolution_1): ... here, replacing condition with
2796 what makes the intent clearer. Remove handling of cases
2797 get_scalar_evolution now handles.
2799 2017-10-10 Jim Wilson <wilson@tuliptree.org>
2801 PR rtl-optimization/81434
2802 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
2803 comment for main loop. In sched_group_found if, also add checks for
2804 pass and min_cost_group.
2806 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
2808 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
2809 (rs6000_insn_cost): New function.
2810 * config/rs6000/rs6000.md (cost): New attribute.
2812 2017-10-10 Jakub Jelinek <jakub@redhat.com>
2813 H.J. Lu <hongjiu.lu@intel.com>
2817 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
2818 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
2819 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
2820 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
2821 to OPTION_MASK_ISA_AVX512VL - builtins that have both
2822 OPTION_MASK_ISA_MMX and some other bit set require both
2823 mmx and the ISAs without the mmx bit.
2824 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
2825 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
2826 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
2827 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
2828 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
2829 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
2830 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
2831 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
2832 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
2833 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
2834 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
2835 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
2836 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
2837 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
2838 Add OPTION_MASK_ISA_MMX.
2840 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
2842 * config.gcc (armv7*-*-freebsd*): New target.
2843 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
2845 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
2847 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
2848 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
2851 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
2853 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
2854 a template parameter.
2855 (WIDE_INT_REF_FOR): Update accordingly.
2856 * tree.h (wi::int_traits <const_tree>): Delete.
2857 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
2858 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
2859 (wi::tree_to_wide_ref): New typedef.
2860 (wi::to_wide): New function.
2861 * calls.c (get_size_range): Use wi::to_wide when operating on
2863 * cgraph.c (cgraph_node::create_thunk): Likewise.
2864 * config/i386/i386.c (ix86_data_alignment): Likewise.
2865 (ix86_local_alignment): Likewise.
2866 * dbxout.c (stabstr_O): Likewise.
2867 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
2868 * expr.c (const_vector_from_tree): Likewise.
2869 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
2870 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
2871 (fold_negate_expr_1, int_const_binop_1, const_binop)
2872 (fold_convert_const_int_from_real, optimize_bit_field_compare)
2873 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
2874 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
2875 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
2876 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
2877 (fold_not_const, round_up_loc): Likewise.
2878 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
2879 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
2880 (alloca_call_type): Likewise.
2881 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
2882 * godump.c (go_output_typedef): Likewise.
2883 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
2884 * internal-fn.c (get_min_precision): Likewise.
2885 * ipa-cp.c (ipcp_store_vr_results): Likewise.
2886 * ipa-polymorphic-call.c
2887 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
2888 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
2889 (ipa_modify_call_arguments): Likewise.
2890 * match.pd: Likewise.
2891 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
2892 * print-tree.c (print_node_brief, print_node): Likewise.
2893 * stmt.c (expand_case): Likewise.
2894 * stor-layout.c (layout_type): Likewise.
2895 * tree-affine.c (tree_to_aff_combination): Likewise.
2896 * tree-cfg.c (group_case_labels_stmt): Likewise.
2897 * tree-data-ref.c (dr_analyze_indices): Likewise.
2898 (prune_runtime_alias_test_list): Likewise.
2899 * tree-dump.c (dequeue_and_dump): Likewise.
2900 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
2901 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
2902 * tree-pretty-print.c (dump_generic_node): Likewise.
2903 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
2904 (simple_iv_with_niters): Likewise.
2905 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2906 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
2907 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
2908 * tree-ssa-loop-niter.c (split_to_var_and_offset)
2909 (refine_value_range_using_guard, number_of_iterations_ne_max)
2910 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
2911 (get_cst_init_from_scev, record_nonwrapping_iv)
2912 (scev_var_range_cant_overflow): Likewise.
2913 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
2914 * tree-ssa-pre.c (compute_avail): Likewise.
2915 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
2916 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
2917 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
2918 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
2919 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
2920 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
2921 (dump_case_nodes, try_switch_expansion): Likewise.
2922 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
2923 (vect_do_peeling): Likewise.
2924 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
2925 * tree-vect-stmts.c (vectorizable_load): Likewise.
2926 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
2927 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
2928 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
2929 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
2930 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
2931 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
2932 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
2933 (evrp_dom_walker::before_dom_children): Likewise.
2934 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
2935 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
2936 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
2937 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
2938 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
2939 (get_range_pos_neg): Likewise.
2940 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
2941 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
2942 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
2943 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
2944 * config/avr/avr.c (avr_fold_builtin): Likewise.
2945 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
2946 * config/msp430/msp430.c (msp430_attr): Likewise.
2947 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
2948 * config/powerpcspe/powerpcspe-c.c
2949 (altivec_resolve_overloaded_builtin): Likewise.
2950 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
2951 (rs6000_expand_ternop_builtin): Likewise.
2952 * config/rs6000/rs6000-c.c
2953 (altivec_resolve_overloaded_builtin): Likewise.
2954 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
2955 (rs6000_expand_ternop_builtin): Likewise.
2956 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
2958 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2960 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
2961 when copying loop nest with only one inner loop.
2963 2017-10-10 Richard Biener <rguenther@suse.de>
2965 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
2966 blocks if SCEV is active.
2967 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
2969 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
2970 (scev_initialize): Assert we are not yet initialized.
2972 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2974 * tree-loop-distribution.c (generate_loops_for_partition): Remove
2975 inner loop's exit stmt by making it always exit the loop, otherwise
2976 we would generate an infinite empty loop.
2978 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2980 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
2981 renaming variables in new preheader if it's deleted.
2983 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2985 * tree-loop-distribution.c (struct partition): Remove unused field
2986 loops of the structure.
2987 (partition_alloc, partition_free): Ditto.
2988 (build_rdg_partition_for_vertex): Ditto.
2990 2017-10-09 Jeff Law <law@redhat.com>
2992 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
2993 return type to match prototype and documentation.
2995 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2997 * config/rs6000/rs6000.c (processor_costs): Move to ...
2998 * config/rs6000/rs6000.h: ... here.
2999 (rs6000_cost): Declare.
3001 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
3003 * except.c (setjmp_fn): New global variable.
3004 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
3005 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
3006 if DONT_USE_BUILTIN_SETJMP is defined.
3008 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3010 * target.def (insn_cost): New hook.
3011 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
3012 * doc/tm.texi: Regenerate.
3013 * rtlanal.c (insn_cost): Use the new hook.
3015 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3017 * combine.c (combine_validate_cost): Compute the new insn_cost,
3018 not just pattern_cost.
3019 (try_combine): Adjust comment.
3021 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3023 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
3025 * combine.c (uid_insn_cost): Adjust comment.
3026 (combine_validate_cost): Adjust comment. Use pattern_cost instead
3028 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
3029 * dse.c (find_shift_sequence): Ditto.
3030 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
3031 (bb_valid_for_noce_process_p): Use pattern_cost.
3032 * rtl.h (insn_rtx_cost): Delete.
3033 (pattern_cost): New prototype.
3034 (insn_cost): New prototype.
3035 * rtlanal.c (insn_rtx_cost): Rename to...
3036 (pattern_cost): ... this.
3039 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
3041 * config/i386/i386.md (*jcc_2): Remove insn pattern.
3042 (*jcc<mode>_0_r_i387): Ditto.
3043 (*jccxf_r_i387): Ditto.
3044 (*jcc<mode>_r_i387): Ditto.
3045 (*jccu<mode>_r_i387): Ditto.
3046 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
3047 (*jcc): Rename from *jcc_1.
3049 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3051 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
3052 deferred rescans after the lvx/stvx recombination pre-pass.
3054 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
3056 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
3057 memory operation instruction support.
3058 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
3059 (rs6000-ibm-aix[789]*): Likewise.
3060 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
3061 Document new functions.
3063 2017-10-09 Richard Biener <rguenther@suse.de>
3065 PR tree-optimization/82397
3066 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
3067 equality only for semantically equal trees.
3069 2017-10-09 Richard Biener <rguenther@suse.de>
3071 PR tree-optimization/82449
3072 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
3073 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
3074 allow constant addresses.
3075 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
3078 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3080 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
3083 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3086 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
3089 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3092 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
3094 2017-10-09 Jakub Jelinek <jakub@redhat.com>
3097 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
3098 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
3100 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
3102 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
3103 (WI_BINARY_PREDICATE_RESULT): Likewise.
3104 (wi::binary_traits::operator_result): New type.
3105 (wi::binary_traits::predicate_result): Likewise.
3106 (generic_wide_int::operator~, unary generic_wide_int::operator-)
3107 (generic_wide_int::operator==, generic_wide_int::operator!=)
3108 (generic_wide_int::operator&, generic_wide_int::and_not)
3109 (generic_wide_int::operator|, generic_wide_int::or_not)
3110 (generic_wide_int::operator^, generic_wide_int::operator+
3111 (binary generic_wide_int::operator-, generic_wide_int::operator*):
3113 (operator~, unary operator-, operator==, operator!=, operator&)
3114 (operator|, operator^, operator+, binary operator-, operator*): New
3116 * expr.c (get_inner_reference): Use wi::bit_and_not.
3117 * fold-const.c (fold_binary_loc): Likewise.
3118 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
3119 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
3120 (bit_value_binop): Likewise.
3121 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
3122 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
3123 (extract_range_from_binary_expr_1): Likewise.
3124 (masked_increment): Likewise.
3125 (simplify_bit_ops_using_ranges): Likewise.
3127 2017-10-09 Martin Jambor <mjambor@suse.cz>
3130 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
3131 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
3132 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
3134 (hsa_fixup_mov_insn_type): New function.
3135 (hsa_op_with_type::get_in_type): Use it.
3136 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
3137 immediates in an assert.
3138 (hsa_op_with_type::extend_int_to_32bit): New method.
3139 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
3140 types. Convert to dest type if necessary.
3141 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
3142 (reg_for_gimple_ssa): Pass false as min32int to
3143 hsa_type_for_scalar_tree_type.
3144 (gen_hsa_addr): Fixup type when creating addresable temporary.
3145 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
3146 (gen_hsa_unary_operation): Extend operands and convert to dest type if
3147 necessary. Call hsa_fixup_mov_insn_type.
3148 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
3149 extend operands and convert to dest type if necessary.
3150 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
3151 to dest type if necessary.
3152 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
3153 if conversion nt necessary and size matches.
3154 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
3155 to dest type if necessary.
3156 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
3157 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
3159 (gen_hsa_clrsb): Likewise.
3160 (gen_hsa_ffs): Likewise.
3161 (gen_hsa_divmod): Extend operands and convert to dest type if
3163 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
3165 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
3167 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
3168 Remove empty default arguments. Use a brace block as output
3170 (conditional return): Ditto.
3172 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
3173 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
3175 (group_ending_nop): Ditto.
3176 (doloop_end): Ditto.
3177 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
3178 (splitters for those): Ditto.
3180 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
3182 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
3183 a conditional jump (and the compare for it) so that pc_rtx is the
3185 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
3186 for the deleted and renamed ctr<mode>_internal[234] patterns.
3187 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
3188 Delete second conditional return pattern.
3189 (ctr<mode>_internal2): Delete this second bdnz pattern.
3190 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
3191 (ctr<mode>_internal4): Delete this second bdz pattern.
3193 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
3195 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
3196 (always_initialized_rtx_for_ssa_name_p): New predicate.
3197 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
3198 (finish_out_of_ssa): Free new field of SA.
3199 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
3200 * tree-ssa-coalesce.c: Include tree-ssa.h.
3201 (get_parm_default_def_partitions): Remove extern keyword.
3202 (get_undefined_value_partitions): New function.
3203 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
3204 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
3206 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
3208 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
3210 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
3212 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
3214 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
3215 for targets that preffer 128bit.
3217 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
3219 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
3221 2017-10-08 Olivier Hainque <hainque@adacore.com>
3223 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
3224 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
3226 (thumb_set_return_address): Likewise.
3228 2017-10-08 Olivier Hainque <hainque@adacore.com>
3230 * common/config/arm/arm-common.c (arm_except_unwind_info):
3231 Handle DWARF2_UNWIND_INFO.
3233 2017-10-07 Michael Collison <michael.collison@arm.com>
3235 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
3238 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
3240 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
3241 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
3242 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
3243 defined, force the creation of a new block for a dispatch label.
3245 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
3247 * invoke.texi (Wsuggest-attribute=cold): Document.
3248 * common.opt (Wsuggest-attribute=cold): New
3249 * ipa-pure-const.c (warn_function_cold): New function.
3250 * predict.c (compute_function_frequency): Use it.
3251 * predict.h (warn_function_cold): Declare.
3253 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
3255 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
3258 2017-10-06 Martin Liska <mliska@suse.cz>
3260 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
3261 keyword for member functions.
3262 (struct sanopt_tree_couple): New struct.
3263 (struct sanopt_tree_couple_hash): New function.
3264 (struct sanopt_ctx): Add new hash_map.
3265 (has_dominating_ubsan_ptr_check): New function.
3266 (record_ubsan_ptr_check_stmt): Likewise.
3267 (maybe_optimize_ubsan_ptr_ifn): Likewise.
3268 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
3269 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
3271 2017-10-06 Sudakshina Das <sudi.das@arm.com>
3274 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
3275 aarch64_simd_valid_immediate on CONST_VECTORs.
3276 (aarch64_reg_or_bic_imm): Likewise.
3278 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
3280 PR rtl-optimization/82396
3281 * haifa-sched.c (ready_sort_real): Disable qsort checking.
3283 2017-10-06 Sebastian Pop <sebpop@gmail.com>
3285 * graphite-dependences.c (scop_get_reads): Move code to...
3286 (scop_get_must_writes): Move code to...
3287 (scop_get_may_writes): Move code to...
3288 (scop_get_reads_and_writes): ... here.
3289 (scop_get_dependences): Call scop_get_reads_and_writes.
3291 2017-10-06 Jakub Jelinek <jakub@redhat.com>
3293 PR tree-optimization/82434
3294 * fold-const.h (can_native_encode_type_p,
3295 can_native_encode_string_p): Remove.
3296 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
3297 don't encode anything, just return what would be otherwise returned.
3298 (native_encode_fixed, native_encode_complex, native_encode_vector):
3300 (native_encode_string): Likewise. Inline by hand
3301 can_native_encode_string_p.
3302 (can_native_encode_type_p): Remove.
3303 (can_native_encode_string_p): Remove.
3304 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
3305 STRING_CSTs using can_native_encode_string_p, test all
3306 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
3307 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
3308 argument from native_encode_expr.
3309 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
3310 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
3313 2017-10-06 Richard Biener <rguenther@suse.de>
3315 PR tree-optimization/82397
3316 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
3317 operand_equal_p but rely on data_ref_compare_tree for detecting
3319 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
3320 to match up with dr_group_sort_cmp.
3322 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3325 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
3327 * config/s390/s390-builtin-types.def: Regenerate.
3329 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3332 * config/s390/s390-builtin-types.def: Regenerate.
3333 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
3334 Change flag from B_VXE to B_VX.
3335 (s390_vec_min_dbl): Remove B_VXE flag.
3337 2017-10-06 Richard Biener <rguenther@suse.de>
3339 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
3340 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
3341 translate_isl_ast_to_gimple::is_valid_rename,
3342 translate_isl_ast_to_gimple::get_rename,
3343 translate_isl_ast_to_gimple::get_def_bb_for_const,
3344 translate_isl_ast_to_gimple::get_new_name,
3345 translate_isl_ast_to_gimple::collect_all_ssa_names,
3346 translate_isl_ast_to_gimple::copy_loop_phi_args,
3347 translate_isl_ast_to_gimple::collect_all_ssa_names,
3348 translate_isl_ast_to_gimple::copy_loop_phi_args,
3349 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
3350 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
3351 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
3352 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
3353 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
3354 translate_isl_ast_to_gimple::copy_cond_phi_args,
3355 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
3356 translate_isl_ast_to_gimple::edge_for_new_close_phis,
3357 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
3358 translate_isl_ast_to_gimple::rename_uses,
3359 translate_isl_ast_to_gimple::rename_all_uses): Remove.
3360 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
3361 (set_rename_for_each_def): Likewise.
3362 (graphite_copy_stmts_from_block): Handle debug stmt resetting
3363 here. Handle rewriting SCEV analyzable uses here.
3364 (copy_bb_and_scalar_dependences): Generate code for PHI
3366 (graphite_regenerate_ast_isl): Adjust.
3367 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
3368 (add_write, add_read): New functions.
3369 (build_cross_bb_scalars_def): Use it and simplify.
3370 (build_cross_bb_scalars_use): Likewise.
3371 (graphite_find_cross_bb_scalar_vars): Inline into...
3372 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
3373 simulating out-of-SSA. Compute liveout and add dependencies.
3374 (build_scops): Force an empty entry block.
3375 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
3377 (sese_build_liveouts): Declare.
3378 (sese_trivially_empty_bb_p): Likewise.
3379 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
3380 compute liveout and debug_liveout.
3381 (sese_bad_liveouts_use): Remove.
3382 (sese_reset_debug_liveouts_bb): Likewise.
3383 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
3384 (sese_build_liveouts): Build liveout and debug_liveout and store
3386 (new_sese_info): Adjust.
3387 (free_sese_info): Likewise.
3388 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
3389 do not build liveout here.
3390 (move_sese_in_condition): Adjust region entry.
3391 (scev_analyzable_p): Match up with chrec_apply requirements.
3392 (sese_trivially_empty_bb_p): New.
3393 * tree-into-ssa.c (get_reaching_def): Properly support generating
3394 default-defs for incremental rewrite of anonymous names.
3396 2017-10-06 Richard Biener <rguenther@suse.de>
3398 * graphite-sese-to-poly.c (extract_affine): For casts increasing
3399 precision do not perform modulo reduction.
3401 2017-10-06 Richard Biener <rguenther@suse.de>
3403 PR tree-optimization/82436
3404 * tree-vect-slp.c (vect_supported_load_permutation_p): More
3405 conservatively choose the vectorization factor when checking
3406 whether we can perform the required load permutation.
3407 (vect_transform_slp_perm_load): Assert when we may not fail.
3409 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
3411 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
3412 message for incompatible -msdata=* and -mcall-* options.
3414 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3416 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
3417 rate for post-reload scheduling.
3419 2017-10-05 Tamar Christina <tamar.christina@arm.com>
3421 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
3423 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3425 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
3426 to improve monte carlo in scimark.
3428 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3430 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
3431 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
3432 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
3433 pentium4_cost, nocona_cost): Set reassociation width to 1.
3434 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
3435 width to 2 for fp operations and 1 otherwise.
3436 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
3438 (atom_cost): Set reassociation width to 2.
3439 (slm_cost, generic_cost): Set fp reassociation width
3440 to 2 and 1 otherwise.
3441 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
3442 (core_cost): Set fp reassociation width to 4 and vector to 2.
3443 (ix86_reassociation_width): Rewrite using cost table; special case
3444 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
3445 and TARGET_AVX128_OPTIMAL.
3446 * config/i386/i386.h (processor_costs): Add
3447 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
3448 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
3449 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
3450 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
3451 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
3452 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
3454 2017-10-05 Nathan Sidwell <nathan@acm.org>
3456 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
3458 2017-10-05 Tamar Christina <tamar.christina@arm.com>
3460 * config/arm/arm.c (arm_test_fpu_data): New.
3461 (arm_run_selftests): Call arm_test_fpu_data.
3463 2017-10-04 Nathan Sidwell <nathan@acm.org>
3465 * toplev.c (toplev::main): Remove excess parens on pretty_printer
3467 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
3469 2017-10-04 Sudakshina Das <sudi.das@arm.com>
3471 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
3472 check type for aarch64_simd_valid_immediate.
3473 (aarch64_output_simd_mov_immediate): Update prototype.
3474 (aarch64_simd_valid_immediate): Update prototype.
3475 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
3476 support for ORR-immediate.
3477 (and<mode>3): modified pattern to add support for BIC-immediate.
3478 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
3479 now checks for valid immediate for BIC and ORR based on new enum
3481 (aarch64_output_simd_mov_immediate): Function now used to output
3482 BIC/ORR imm as well based on new enum argument.
3483 * config/aarch64/constraints.md (Do): New vector immediate constraint.
3485 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
3486 (aarch64_reg_or_bic_imm): Likewise.
3488 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3490 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
3491 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
3493 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
3497 PR rtl-optimization/82396
3498 * haifa-sched.c (autopref_multipass_init): Simplify
3500 (autopref_rank_data): Simplify sort order.
3501 * sched-int.h (autopref_multipass_data_): Remove
3502 multi_mem_insn_p, min_offset and max_offset.
3504 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3506 * doc/sourcebuild.texi: Document vect_peeling_profitable.
3508 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3510 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
3513 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3515 * doc/sourcebuild.texi: Document vect_long_mult.
3517 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
3519 PR tree-optimization/82413
3520 * fold-const.c (build_range_check): Use widest_int when comparing
3521 the maximum ETYPE value with HIGH.
3523 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
3525 PR rtl-optimization/82396
3526 * haifa-sched.c (autopref_multipass_init): Simplify
3528 (autopref_rank_data): Simplify sort order.
3529 * sched-int.h (autopref_multipass_data_): Remove
3530 multi_mem_insn_p, min_offset and max_offset.
3532 2017-10-04 Jakub Jelinek <jakub@redhat.com>
3534 PR tree-optimization/82381
3535 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
3536 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
3539 PR tree-optimization/82374
3540 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
3541 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
3542 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
3543 current_function_decl to the new decl.
3545 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
3547 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
3548 helper macro for IEEE float128 hardware built-in functions.
3549 (SQRTF128_ODD): Add built-in functions with the round-to-odd
3551 (TRUNCF128_ODD): Likewise.
3552 (ADDF128_ODD): Likewise.
3553 (SUBF128_ODD): Likewise.
3554 (MULF128_ODD): Likewise.
3555 (DIVF128_ODD): Likewise.
3556 (FMAF128_ODD): Likewise.
3557 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
3558 UNSPEC_TRUNC_ROUND_TO_ODD.
3559 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
3560 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
3561 floating point round to odd instructions.
3562 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
3563 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
3564 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
3565 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
3566 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
3567 (trunc<mode>sf2_hw): Change the truncate with round to odd
3568 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
3569 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
3570 to odd hardware instructions.
3571 (sub<mode>3_odd): Likewise.
3572 (mul<mode>3_odd): Likewise.
3573 (div<mode>3_odd): Likewise.
3574 (sqrt<mode>2_odd): Likewise.
3575 (fma<mode>4_odd): Likewise.
3576 (fms<mode>4_odd): Likewise.
3577 (nfma<mode>4_odd): Likewise.
3578 (nfms<mode>4_odd): Likewise.
3579 (trunc<mode>df2_odd): Change the truncate with round to odd
3580 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
3582 * doc/extend.texi (PowerPC built-in functions): Update documentation
3583 for existing IEEE float128-bit built-in functions. Add built-in
3584 functions that generate the IEEE 128-bit floating point round to
3587 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
3589 PR rtl-optimization/77729
3590 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
3591 to (X&(C1&~C2))|C2 transformations.
3593 2017-10-03 Martin Jambor <mjambor@suse.cz>
3595 PR tree-optimization/82363
3596 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
3597 mismatch, mark lacc written regardless of racc.
3599 2017-10-03 Jakub Jelinek <jakub@redhat.com>
3601 PR tree-optimization/82381
3602 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
3603 stmt_to_insert nor wheather SSA_NAMEs are default defs.
3604 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
3605 fallthrough into reassoc_stmt_dominates_stmt_p.
3608 * combine.c (combine_instructions): Don't combine in unreachable
3611 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
3614 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
3615 function to not use the have_cpu variable. Do not set cpu_index,
3616 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
3618 (rs6000_valid_attribute_p): Remove duplicate initializations of
3619 old_optimize and func_optimize.
3620 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
3621 (rs6000_activate_target_options): Make global.
3622 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
3625 2017-10-02 Jakub Jelinek <jakub@redhat.com>
3627 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
3628 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
3629 Set *poffset to 0 and *psize and *pmax_size to -1 if
3630 *poffset + *psize overflows in HOST_WIDE_INT.
3632 PR tree-optimization/82387
3633 PR tree-optimization/82388
3634 PR tree-optimization/82389
3635 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
3636 instead of live_bytes non-NULL.
3638 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
3641 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
3644 2017-10-02 Richard Biener <rguenther@suse.de>
3646 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
3647 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
3648 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
3650 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
3652 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
3653 requested precision matches the type's.
3654 * calls.c (alloc_max_size): Calculate the new candidate size as
3655 a widest_int and use wi::to_widest when comparing it with the
3656 current candidate size.
3657 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
3658 zero rather than integer_zero_node.
3659 * match.pd: Check for a no-op conversion before using wi::add
3660 rather than after. Use tree_to_uhwi when summing small shift
3661 counts into an unsigned int.
3663 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
3664 Alan Hayward <alan.hayward@arm.com>
3665 David Sherwood <david.sherwood@arm.com>
3668 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
3669 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
3670 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
3671 POINTER_AND_FP_REGS.
3673 2017-10-02 Richard Biener <rguenther@suse.de>
3675 PR tree-optimization/82355
3676 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
3677 a mapping for the enclosing loop but avoid generating one for
3679 (copy_bb_and_scalar_dependences): Remove premature codegen
3680 error on PHIs in blocks duplicated into multiple places.
3681 * graphite-scop-detection.c
3682 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
3683 in the region use it as loop and nest to analyze the DR in.
3684 (try_generate_gimple_bb): Likewise.
3685 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
3686 (add_loop_constraints): For blocks in a loop not in the region
3687 create a dimension with a single iteration.
3688 * sese.h (gbb_loop_at_index): Remove assert.
3690 2017-10-01 Kevin Buettner <kevinb@redhat.com>
3692 * omp-expand.c (adjust_context_scope): New function.
3693 (expand_parallel_call): Call adjust_context_scope.
3695 2017-10-01 Jeff Law <law@redhat.com>
3697 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
3698 dom_opt_dom_walker class with direct access to private members.
3699 Add comments. Call test_for_singularity.
3700 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
3701 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
3702 m_dummy_cond anymore.
3703 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
3705 (pass_dominator:execute): Build the dummy_cond here and pass it
3706 to the dom_opt_dom_walker ctor.
3707 (test_for_singularity): New function.
3709 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
3710 Maya Rashish <coypu@sdf.org>
3712 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
3713 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
3714 (alpha*-*-netbsd*) Use nbsd_tm_file.
3715 (arm*-*-netbsdelf*) Likewise.
3716 (i[34567]86-*-netbsdelf*) Likewise.
3717 (x86_64-*-netbsd*) Likewise.
3718 (mips*-*-netbsd*) Likewise.
3719 (powerpc-*-netbsd*) Likewise.
3720 (sh*-*-netbsd*) Likewise.
3721 (sparc-*-netbsdelf*) Likewise.
3722 (sparc64-*-netbsd*) Likewise.
3723 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
3725 (vax-*-netbsdelf*) Likewise.
3726 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
3727 (UINT_FAST8_TYPE) Likewise.
3728 (INT_FAST16_TYPE) Check CHAR_FAST16.
3729 (UINT_FAST16_TYPE) Likewise.
3731 2017-09-30 Jakub Jelinek <jakub@redhat.com>
3734 * config/i386/i386.md
3735 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
3736 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
3737 *divmodsi4_zext_2): New define_insn_and_split.
3738 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
3739 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
3740 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
3741 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
3742 New define_insn_and_split.
3743 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
3744 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
3745 operands[1] having DImode when mode is SImode.
3747 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
3748 always SImode for DIV and MOD in REG_EQUAL notes.
3750 2017-09-29 Yury Gribov <tetra2005@gmail.com>
3753 * match.pd: Fix handling of NaNs in pattern.
3755 2017-09-29 Jeff Law <law@redhat.com>
3757 * sbitmap.c (bitmap_bit_in_range_p): New function.
3758 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
3759 * tree-ssa-dse.c (live_bytes_read): New function.
3760 (dse_classify_store): Ignore reads of dead bytes.
3762 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
3763 typos and whitespace errors.
3764 * config/i386/predicates.md (address_no_seg_operand): Likewise.
3765 * config/s390/s390.c (s390_emit_prologue): Likewise.
3767 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
3770 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
3771 with a symbol for LRA.
3773 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
3775 PR rtl-optimization/82338
3776 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
3778 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
3780 * genmodes.c (calc_wider_mode): Suppress qsort macro.
3781 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
3782 (qsort_chk): Declare.
3783 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
3784 (qsort_chk): New function.
3786 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3788 PR tree-optimization/82337
3789 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
3790 phi definition if the PHI result appears in an abnormal PHI.
3791 (find_basis_for_base_expr): Don't record a basis if the LHS of the
3792 basis appears in an abnormal PHI.
3794 2017-09-29 Richard Biener <rguenther@suse.de>
3796 * graphite-isl-ast-to-gimple.c
3797 (translate_isl_ast_to_gimple::set_codegen_error): New function.
3798 (binary_op_to_tree): Use it.
3799 (get_rename_from_scev): Likewise.
3800 (copy_loop_phi_nodes): Likewise.
3801 (copy_bb_and_scalar_dependences): Likewise.
3802 (translate_pending_phi_nodes): Likewise.
3804 2017-09-29 Jakub Jelinek <jakub@redhat.com>
3807 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
3808 for movabsq $(i32 << shift), r64.
3810 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
3812 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
3813 index when encoding %esp as %rsp to avoid 0x67 prefix.
3815 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
3817 * config/i386/i386.md (*movsf_internal, *movdf_internal):
3818 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
3820 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
3822 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
3823 Extensions with more than 16 double VFP registers.
3824 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
3825 to_clear_mask and all code related to it. Replace the remaining
3826 entry by a sbitmap and adapt code accordingly.
3828 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
3830 * brig-builtins.def: Change pure attributes to const.
3832 2017-09-28 Joseph Myers <joseph@codesourcery.com>
3834 * config.gcc (default_gnu_indirect_function): Default to yes for
3835 sparc*-*-linux* with glibc.
3837 2017-09-28 Joseph Myers <joseph@codesourcery.com>
3839 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
3840 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
3841 when creating .init_array and .fini_array sections with priority
3844 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
3847 * config/aarch64/aarch64.c
3848 (aarch64_builtin_support_vector_misalignment): Always return false
3849 when misalignment is unknown.
3851 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
3853 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
3854 this function to return false if the definition used by the swap
3855 instruction is artificial, or if the memory address from which the
3856 constant value is loaded is not represented by a base address held
3857 in a register or if the base address register is a frame or stack
3858 pointer. Additionally, return false if the base address of the
3859 loaded constant is a SYMBOL_REF but is not considered to be a
3861 (replace_swapped_load_constant): New function.
3862 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
3863 loaded constant vector with a load of a swapped constant vector.
3865 2017-09-27 Carl Love <cel@us.ibm.com>
3867 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
3868 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
3869 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
3872 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
3874 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
3875 first, always call autopref_rank_data otherwise.
3877 2017-09-27 Richard Biener <rguenther@suse.de>
3879 * graphite-scop-detection.c (find_scop_parameters): Move
3880 loop bound handling ...
3881 (gather_bbs::before_dom_children): ... here, avoiding the need
3882 to build scop_info->loop_nest.
3883 (record_loop_in_sese): Remove.
3884 * sese.h (sese_info_t::loop_nest): Remove.
3885 * sese.c (new_sese_info): Do not allocate loop_nest.
3886 (free_sese_info): Do not free loop_nest.
3888 2017-09-27 Jakub Jelinek <jakub@redhat.com>
3891 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
3892 lhs from calls if the lhs has addressable type.
3894 2017-09-27 Richard Biener <rguenther@suse.de>
3896 * graphite.h (scop::max_alias_set): New member.
3897 * graphite-scop-detection.c: Remove references to non-existing
3898 --param in comments.
3899 (build_alias_sets): Record the maximum alias set used for drs.
3900 (build_scops): Support zero as unlimited for
3901 --param graphite-max-arrays-per-scop.
3902 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
3904 (build_poly_sr_1): ... here. Compute alias set based on the
3905 maximum alias set used for drs rather than
3906 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
3908 2017-09-27 Richard Biener <rguenther@suse.de>
3910 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
3911 --param loop-block-tile-size=0 to disable tiling.
3913 2017-09-27 Richard Biener <rguenther@suse.de>
3915 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
3916 (graphite-max-nb-scop-params): Document special value zero.
3917 * domwalk.h (dom_walker::STOP): New symbolical constant.
3918 (dom_walker::dom_walker): Add optional parameter for bb to
3920 (dom_walker::~dom_walker): Declare.
3921 (dom_walker::before_dom_children): Document STOP return value.
3922 (dom_walker::m_user_bb_to_rpo): New member.
3923 (dom_walker::m_bb_to_rpo): Likewise.
3924 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
3925 mapping here if not provided by the user.
3926 (dom_walker::~dom_walker): Free bb to RPO mapping if not
3927 provided by the user.
3928 (dom_walker::STOP): Define.
3929 (dom_walker::walk): Do not compute bb to RPO mapping here.
3930 Support STOP return value from before_dom_children to stop
3932 * graphite-optimize-isl.c (optimize_isl): If the schedule
3933 is the same still generate code if -fgraphite-identity
3934 or -floop-parallelize-all are given.
3935 * graphite-scop-detection.c: Include cfganal.h.
3936 (gather_bbs::gather_bbs): Get and pass through bb to RPO
3938 (gather_bbs::before_dom_children): Return STOP for BBs
3940 (build_scops): Compute bb to RPO mapping and pass it to
3941 the domwalk. Treat --param graphite-max-nb-scop-params=0
3942 as not limiting the number of params.
3943 * graphite.c (graphite_initialize): Remove limit on the
3944 number of basic-blocks in a function.
3945 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
3946 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
3947 default value of 10.
3949 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3951 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
3952 Adjust code to eliminate needing to do the shift right 32-bits
3953 operation after XSCVDPSPN.
3955 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3957 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
3958 ((X / Y) != 0 -> X >= Y): Likewise.
3960 2017-09-26 Carl Love <cel@us.ibm.com>
3962 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
3963 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
3964 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
3965 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
3966 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
3967 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
3968 definitions and overloading.
3969 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
3970 statement for P9V_BUILTIN_XST_LEN_R.
3971 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
3972 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
3973 define_expand and define_insn for the instructions and builtins.
3974 * doc/extend.texi: Update the built-in documentation file for the new
3976 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
3977 define_insn for the instructions
3979 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
3982 * gcc/config/netbsd-protos.h: New file.
3983 * gcc/config/netbsd.c: New file.
3984 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
3985 * gcc/config/t-netbsd: New file.
3986 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
3987 (tmake_file) Add t-netbsd.
3988 (extra_objs) Add netbsd.o.
3990 2017-09-26 Janus Weil <janus@gcc.gnu.org>
3994 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
3996 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3998 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
3999 sign extension from a vector register to a GPR by doing a 32-bit
4000 direct move and then an EXTSW.
4001 (extendsi<mode>2 splitter): Likewise.
4002 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
4003 right or vector extract after doing XSCVDPSPN. Use
4004 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
4006 (movdi_from_sf_zero_ext): Likewise.
4007 (reload_gpr_from_vsxsf): Likewise.
4008 (p8_mfvsrd_4_disf): Delete, no longer used.
4009 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
4010 then needing to move the SFmode to a GPR to use the XSCVDPSP
4011 instruction instead of FRSP and XSCVDPSPN.
4012 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
4013 it is adjacent to the other XSCVSPDP insns.
4014 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
4015 SFmode to be in traditional Altivec registers.
4016 (vsx_xscvdpspn): Eliminate useless alternative constraint.
4017 (vsx_xscvspdpn): Likewise.
4018 (vsx_xscvspdpn_scalar): Likewise.
4020 2017-09-26 Martin Jambor <mjambor@suse.cz>
4022 * tree-sra.c (compare_access_positions): Put integral types first,
4023 stabilize sorting of integral types, remove conditions putting
4024 non-full-precision integers last.
4025 (sort_and_splice_var_accesses): Disable scalarization if a
4026 non-integert would be represented by a non-full-precision integer.
4028 2017-09-26 Joseph Myers <joseph@codesourcery.com>
4030 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
4031 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
4032 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
4033 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
4034 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
4035 conditionals inside the function instead of around it. Call
4036 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
4037 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
4039 2017-09-26 Richard Biener <rguenther@suse.de>
4041 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
4043 (scop_detection::build_scop_breadth): ... this. Removed.
4044 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
4045 (scop_detection::harmful_stmt_in_bb): Likewise.
4046 (scop_detection::graphite_can_represent_stmt): Likewise.
4047 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
4048 (scop_detection::can_represent_loop): Remove recursion, fold in ...
4049 (scop_detection::can_represent_loop_1): ... this. Removed.
4050 (scop_detection::harmful_loop_in_region): Simplify after inlining
4051 the above and remove more quadraticness.
4052 (build_scops): Adjust.
4053 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
4056 2017-09-26 Jakub Jelinek <jakub@redhat.com>
4059 * config/i386/i386.c (ix86_print_operand_address_as): Only test
4060 REGNO (base) == SP_REG if base is a REG.
4063 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
4064 if it is different SSA_NAME.
4065 (optimize_range_tests_cmp_bitwise): New function.
4066 (optimize_range_tests): Call it.
4068 2017-09-26 Richard Biener <rguenther@suse.de>
4070 PR tree-optimization/82321
4071 * graphite.c (canonicalize_loop_closed_ssa): Properly check
4072 for the def being inside the loop.
4074 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4076 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
4078 * config/s390/s390-builtins.def: Fix constraint on op4.
4080 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4082 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
4083 independent expanders.
4084 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
4085 ("vec_ordered", "vec_unordered"): New expanders.
4087 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4089 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
4092 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4094 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
4095 vec_unpacks_lo_v16qi.
4096 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
4098 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4100 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
4101 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
4102 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
4104 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4106 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
4108 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
4110 ("*vec_slb<mode>"): New insn pattern.
4111 ("vec_shr_<mode>"): New expander.
4112 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
4113 and force the shift count operand to V16QImode.
4114 ("vec_srb<mode>"): Set shift count mode to V16QI.
4116 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4118 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
4119 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
4120 ("vec_widen_smult_hi_<mode>"): New expander definitions.
4122 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
4125 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
4127 2017-09-26 Richard Biener <rguenther@suse.de>
4129 PR tree-optimization/82320
4130 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
4133 2017-09-25 Jeff Law <law@redhat.com>
4135 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
4136 prototype for new argument.
4137 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
4138 mostly extracted from rs6000_emit_allocate_stack.
4139 (rs6000_emit_probe_stack_range_stack_clash): New function.
4140 (rs6000_emit_allocate_stack): Call
4141 rs6000_emit_probe_stack_range_stack_clash as needed.
4142 (rs6000_emit_probe_stack_range): Add additional argument
4143 to call to gen_probe_stack_range{si,di}.
4144 (output_probe_stack_range): New.
4145 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
4146 (output_probe_stack_range_stack_clash): New.
4147 (rs6000_emit_prologue): Emit notes into dump file as requested.
4148 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
4149 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
4150 Add additional operand and pass it to output_probe_stack_range.
4152 2017-09-25 Bin Cheng <bin.cheng@arm.com>
4154 PR tree-optimization/82163
4155 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
4156 (checking_verify_loop_closed_ssa): New parameter.
4157 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
4158 (check_loop_closed_ssa_stmt): Delete.
4159 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
4160 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
4161 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
4164 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
4166 * brig-builtins.def: Treat HSAIL barrier builtins as
4167 setjmp/longjump style functions.
4169 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
4171 * target.def (constant_alignment): New hook.
4172 * defaults.h (CONSTANT_ALIGNMENT): Delete.
4173 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
4174 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
4175 * doc/tm.texi: Regenerate.
4176 * targhooks.h (default_constant_alignment): Declare.
4177 (constant_alignment_word_strings): Likewise.
4178 * targhooks.c (default_constant_alignment): New function.
4179 (constant_alignment_word_strings): Likewise.
4180 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
4181 instead of CONSTANT_ALIGNMENT.
4182 * varasm.c (align_variable, get_variable_align, build_constant_desc)
4183 (force_const_mem): Likewise.
4184 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
4185 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
4186 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
4187 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4188 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
4190 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
4191 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4192 constant_alignment_word_strings.
4193 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
4194 (CONSTANT_ALIGNMENT): Likewise.
4195 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4196 (arm_constant_alignment): New function.
4197 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
4198 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4199 constant_alignment_word_strings.
4200 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
4201 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4202 constant_alignment_word_strings.
4203 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
4204 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4205 (cris_constant_alignment): New function.
4206 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
4207 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4208 (epiphany_constant_alignment): New function.
4209 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
4210 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4211 constant_alignment_word_strings.
4212 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
4213 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4214 constant_alignment_word_strings.
4215 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
4216 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4217 constant_alignment_word_strings.
4218 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
4219 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
4220 * config/i386/i386.c (ix86_constant_alignment): Make static.
4221 Use the same interface as the target hook.
4222 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4223 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
4224 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4225 constant_alignment_word_strings.
4226 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
4227 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
4228 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4229 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
4230 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4231 constant_alignment_word_strings.
4232 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
4233 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4234 constant_alignment_word_strings.
4235 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
4236 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4237 constant_alignment_word_strings.
4238 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
4239 * config/microblaze/microblaze.c (microblaze_constant_alignment):
4241 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4242 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
4243 * config/mips/mips.c (mips_constant_alignment): New function.
4244 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4245 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
4246 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
4247 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4248 (mmix_constant_alignment): Make static. Use the same interface
4250 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
4251 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4252 constant_alignment_word_strings.
4253 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
4254 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4255 constant_alignment_word_strings.
4256 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
4257 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4258 constant_alignment_word_strings.
4259 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
4260 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4261 (rs6000_constant_alignment): New function.
4262 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
4263 * config/riscv/riscv.c (riscv_constant_alignment): New function.
4264 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4265 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
4266 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4267 (rs6000_constant_alignment): New function.
4268 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
4269 * config/s390/s390.c (s390_constant_alignment): New function.
4270 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4271 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
4272 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4273 constant_alignment_word_strings.
4274 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
4275 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4276 (sparc_constant_alignment): New function.
4277 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
4278 * config/spu/spu.c (spu_constant_alignment): New function.
4279 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4280 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
4281 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4282 constant_alignment_word_strings.
4283 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
4284 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4285 constant_alignment_word_strings.
4286 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
4287 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4288 constant_alignment_word_strings.
4289 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
4290 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4291 (visium_constant_alignment): New function.
4292 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
4293 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4294 (xtensa_constant_alignment): New function.
4295 * system.h (CONSTANT_ALIGNMENT): Poison.
4297 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
4299 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
4300 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
4301 (rs6000_builtin_valid_without_lhs): New helper function.
4302 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4303 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
4305 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
4307 * target.h (vec_perm_indices): Use unsigned short rather than
4309 (auto_vec_perm_indices): Likewise.
4310 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
4311 Use unsigned int rather than unsigned char.
4312 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
4314 2017-09-25 Richard Biener <rguenther@suse.de>
4316 * cfgloop.h (sort_sibling_loops): Declare.
4317 * cfgloop.c (sort_sibling_loops_cmp): New helper.
4318 (sort_sibling_loops): New function sorting the sibling loop list
4320 * graphite.c (graphite_transform_loops): Sort sibling loops.
4322 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
4324 * target.def (vec_perm_const_ok): Change sel parameter to
4326 * optabs-query.c (can_vec_perm_p): Update accordingly.
4327 * doc/tm.texi: Regenerate.
4328 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
4329 auto_vec_perm_indices and remove separate nelt field.
4330 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
4331 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
4332 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
4333 (aarch64_expand_vec_perm_const): Update accordingly.
4334 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
4335 to vec_perm_indices.
4336 * config/arm/arm.c (expand_vec_perm_d): Change perm to
4337 auto_vec_perm_indices and remove separate nelt field.
4338 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
4339 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
4340 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
4342 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
4343 to vec_perm_indices.
4344 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
4345 sel to vec_perm_indices.
4346 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
4347 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
4348 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
4350 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
4353 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
4356 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
4357 on the FUNCTION_DECL function context if it has a DIE that is a
4360 2017-09-25 Richard Biener <rguenther@suse.de>
4362 PR tree-optimization/82285
4363 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
4366 2017-09-25 Tom de Vries <tom@codesourcery.com>
4370 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
4373 2017-09-25 Richard Biener <rguenther@suse.de>
4375 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
4376 ISL errors other than isl_error_quota happen. Dump if the
4377 schedule is the same.
4378 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
4379 errors instead of aborting inside ISL.
4381 2017-09-25 Iain Sandoe <iain@codesourcery.com>
4384 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
4385 of libgcc_eh for m64.
4386 * config/i386/darwin64.h: Likewise.
4388 2017-09-25 Richard Biener <rguenther@suse.de>
4391 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
4392 attribute for incomplete types nor twice for complete ones.
4394 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
4397 * config/i386/i386.c (ix86_print_operand_address_as): Encode
4398 %esp as %rsp to avoid 0x67 prefix if there is no index or base
4401 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
4404 * config/i386/i386.opt (mprefer-avx256): Use
4405 ix86_target_flags variable.
4406 * config/i386/i386.c (ix86_target_string): Move
4407 -mprefer-avx256 to flag2_opts.
4409 2017-09-22 Jakub Jelinek <jakub@redhat.com>
4412 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
4413 and x != -1 | y != -1 into (x & y) != -1.
4415 2017-09-22 Steve Ellcey <sellcey@cavium.com>
4417 * config.gcc: Add new case statement to set
4418 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
4419 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
4420 s390x-*-linux* case statements. Added aarch64 to the list of
4421 supported architectures.
4423 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4425 PR tree-optimization/82289
4426 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
4427 STMT_VINFO_RELEVANT_P.
4429 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4430 Alan Hayward <alan.hayward@arm.com>
4431 David Sherwood <david.sherwood@arm.com>
4433 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
4434 for VR_RANGE only; don't allow VR_ANTI_RANGE.
4435 (extract_range_from_binary_expr_1): Don't call
4436 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
4438 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4439 Alan Hayward <alan.hayward@arm.com>
4440 David Sherwood <david.sherwood@arm.com>
4442 * target.def (preferred_vector_alignment): New hook.
4443 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
4445 * doc/tm.texi: Regenerate.
4446 * targhooks.h (default_preferred_vector_alignment): Declare.
4447 * targhooks.c (default_preferred_vector_alignment): New function.
4448 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
4450 (DR_TARGET_ALIGNMENT): New macro.
4451 (aligned_access_p): Update commentary.
4452 (vect_known_alignment_in_bytes): New function.
4453 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
4455 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
4456 Calculate the misalignment based on the target alignment rather than
4458 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
4459 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
4460 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
4461 the target alignment, rather than masking the element misalignment
4462 with the number of elements in a vector. Also use the target
4463 alignment when calculating the maximum number of peels.
4464 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
4465 instead of TYPE_ALIGN_UNIT.
4466 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
4467 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
4468 (vect_create_addr_base_for_vector_ref): Update call accordingly.
4469 (vect_create_data_ref_ptr): Likewise.
4470 (vect_setup_realignment): Realign by ANDing with
4471 -DR_TARGET_MISALIGNMENT.
4472 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
4473 the number of peels based on DR_TARGET_ALIGNMENT.
4474 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
4475 with the guaranteed alignment boundary when deciding whether
4477 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
4478 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
4479 (ensure_base_align): Remove stmt_info parameter. Get the
4480 target base alignment from DR_TARGET_ALIGNMENT.
4481 (vectorizable_store): Update call accordingly. Interpret
4482 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
4484 (vectorizable_load): Likewise.
4486 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4487 Alan Hayward <alan.hayward@arm.com>
4488 David Sherwood <david.sherwood@arm.com>
4490 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
4491 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
4492 (vect_enhance_data_refs_alignment): Likewise.
4494 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
4496 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
4497 error. Only quit immediately if parsing is complete.
4498 (BEGIN): Initialize fatal_err and parse_done.
4499 (begin fpu, end fpu): Check number of arguments.
4500 (begin arch, end arch): Likewise.
4501 (begin cpu, end cpu): Likewise.
4502 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
4503 (optalias): Likewise.
4505 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
4507 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
4508 * config/arm/arm-isa.h: Delete. Move definitions to ...
4509 * arm-cpus.in: ... here. Use new feature and fgroup values.
4510 * config/arm/arm.c (arm_option_override): Use lower case for feature
4512 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
4513 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
4514 * config/arm/parsecpu.awk (END): Add new command 'isa'.
4516 (print_isa_bits_for): New function.
4517 (gen_isa): New function.
4518 (gen_comm_data): Use print_isa_bits_for.
4519 (define feature): New keyword.
4520 (define fgroup): New keyword.
4521 * config/arm/t-arm (TM_H): Remove.
4522 (GTM_H): Add arm-isa.h.
4523 (arm-isa.h): Add rule to generate file.
4524 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
4525 case for feature bit names.
4527 2017-09-22 Richard Biener <rguenther@suse.de>
4529 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
4531 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
4532 print of no dependency loops ...
4533 * graphite.c (graphite_transform_loops): ... here.
4534 (canonicalize_loop_closed_ssa_form): Work from inner to outer
4536 (same_close_phi_node, remove_duplicate_close_phi,
4537 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
4538 (canonicalize_loop_closed_ssa): ... here and simplify.
4539 * graphite-optimize-isl.c: Include tree-vectorizer.h.
4540 (optimize_isl): Use dump_printf_loc to tell when we stopped
4541 optimizing because of an ISL timeout.
4543 2017-09-22 Richard Biener <rguenther@suse.de>
4545 PR tree-optimization/82291
4546 * tree-if-conv.c (predicate_mem_writes): Make sure to
4547 remove writes in blocks predicated with false.
4549 2017-09-22 Richard Biener <rguenther@suse.de>
4551 * sese.c: Include cfganal.h.
4552 (if_region_set_false_region): Remove.
4553 (create_if_region_on_edge): Likewise.
4554 (move_sese_in_condition): Re-implement without destroying
4557 2017-09-22 Richard Biener <rguenther@suse.de>
4559 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
4560 Verify both BBs contain loop PHI nodes before dispatching to
4562 (graphite_regenerate_ast_isl): Do not recompute dominators,
4563 do not verify three times. Restructure for clarity.
4564 * graphite-scop-detection.c (same_close_phi_node,
4565 remove_duplicate_close_phi, make_close_phi_nodes_unique,
4566 defined_in_loop_p, canonicalize_loop_closed_ssa,
4567 canonicalize_loop_closed_ssa_form): Simplify, remove excess
4568 checking and SSA rewrite, move to ...
4569 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
4570 (graphite_initialize): Do not pass in ctx, do not reset the
4571 SCEV cache, compute only dominators.
4572 (graphite_transform_loops): Allocate ISL ctx after
4573 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
4574 Maintain post-dominators only around build_scops.
4575 * sese.c (if_region_set_false_region): Make static. Free
4576 and recompute dominators.
4577 (move_sese_in_condition): Assert we don't get called with
4578 post-dominators computed.
4579 * sese.h (if_region_set_false_region): Remove.
4581 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
4583 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
4584 mode attribute for TARGET_AVX512VL.
4586 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
4588 * config/i386/i386.opt (mprefer-avx256): New option.
4589 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
4591 (ix86_preferred_simd_mode): Return 256-bit AVX modes
4592 for TARGET_PREFER_AVX256.
4593 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
4595 2017-09-21 Jeff Law <law@redhat.com>
4597 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
4598 Fix dump output if the only stack space is for pushed registers.
4600 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4602 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
4605 2017-09-21 Martin Sebor <msebor@redhat.com>
4608 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
4609 code (in C++) or code that triggers warnings.
4611 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
4613 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
4615 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
4617 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
4618 * sched-rgn.c: Ditto.
4619 * sel-sched-ir.c: Ditto.
4621 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
4623 * toplev.h (set_random_seed): Adjust return type.
4624 * toplev.c (init_local_tick): Move eager initialization of random_seed
4625 to get_random_seed. Adjust comment.
4626 (init_random_seed): Inline to get_random_seed, delete.
4627 (get_random_seed): Initialize random_seed lazily.
4628 (set_random_seed): Do not return previous value.
4629 (print_switch_value): Do not call get_random_seed.
4631 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
4633 * cgraph.c (delete_function_version): New, broken out from...
4634 (cgraph_node::delete_function_version): ...here. Rename to
4635 cgraph_node::delete_function_version_by_decl. Update all uses.
4636 (cgraph_node::remove): Call delete_function_version.
4638 2017-09-21 Jakub Jelinek <jakub@redhat.com>
4641 * tree-inline.c (expand_call_inline): Emit clobber stmts for
4642 VAR_DECLs to which addressable non-volatile parameters are mapped
4643 and for id->retvar after the return value assignment. Clear
4644 id->retval and id->retbnd after inlining.
4646 2017-09-21 Richard Biener <rguenther@suse.de>
4648 PR tree-optimization/82276
4649 PR tree-optimization/82244
4650 * tree-vrp.c (build_assert_expr_for): Set
4651 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
4653 (remove_range_assertions): Revert earlier change.
4655 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
4658 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
4660 2017-09-21 Richard Biener <rguenther@suse.de>
4662 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4663 Restore valid IL after code generation errors.
4664 * graphite.c (graphite_transform_loops): Diagnose code
4665 generation issues as MSG_MISSED_OPTIMIZATION and continue
4666 with processing SCOPs.
4668 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4669 Alan Hayward <alan.hayward@arm.com>
4670 David Sherwood <david.sherwood@arm.com>
4672 * calls.c (compute_argument_addresses): Use simplify_gen_binary
4673 rather than choosing between plus_constant and gen_rtx_<CODE>.
4674 * expr.c (emit_push_insn): Likewise.
4675 (expand_expr_real_2): Likewise.
4677 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4678 Alan Hayward <alan.hayward@arm.com>
4679 David Sherwood <david.sherwood@arm.com>
4681 * loop-unroll.c (split_iv): Call copy_rtx on the step.
4683 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4684 Alan Hayward <alan.hayward@arm.com>
4685 David Sherwood <david.sherwood@arm.com>
4687 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
4688 calling tree_to_uhwi.
4690 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4691 Alan Hayward <alan.hayward@arm.com>
4692 David Sherwood <david.sherwood@arm.com>
4694 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
4695 INTEGER_CST rather than a negative test for ADDR_EXPR.
4697 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4698 Alan Hayward <alan.hayward@arm.com>
4699 David Sherwood <david.sherwood@arm.com>
4701 * tree-vrp.c (extract_range_from_binary_expr_1): Check
4702 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
4704 2017-09-21 Richard Biener <rguenther@suse.de>
4706 PR tree-optimization/71351
4707 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
4708 graphite_create_new_loop_guard): Remove, fold remaining parts
4710 (translate_isl_ast_node_for): ... here and simplify.
4712 2017-09-21 Jakub Jelinek <jakub@redhat.com>
4715 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
4716 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
4717 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
4718 alternative always use QI mode, for -Os imov (=R,R) alternative
4719 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
4722 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4723 Jeff Law <law@redhat.com>
4725 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
4726 (allocate_stack_space): New function, partially extracted from
4728 (s390_emit_prologue): Track offset to most recent stack probe.
4729 Code to allocate space moved into allocate_stack_space.
4730 Dump actions when no stack is allocated.
4731 (s390_prologue_plus_offset): New function.
4732 (s390_emit_stack_probe): Likewise.
4734 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
4736 * common.opt (Wa, Wl, Wp, g, gz=): Add
4738 (gno-column-info): Remove.
4739 (gcolumn-info): Drop RejectNegative.
4741 (gno-record-gcc-switches): Remove.
4742 (grecord-gcc-switches): Drop RejectNegative.
4743 (gno-split-dwarf): Remove.
4744 (gsplit-dwarf): Drop RejectNegative.
4745 (gno-strict-dwarf): Remove.
4746 (gstrict-dwarf): Drop RejectNegative.
4747 * config/darwin.opt (gfull, gused): Add RejectNegative.
4748 * dwarf2out.c (gen_producer_string): Drop
4749 gno-record-gcc-switches handler.
4750 * optc-gen.awk: Add g to prefixes with negative forms.
4751 * opts-common.c (remapping_prefix_p): New.
4752 (find_opt): Check it.
4753 (generate_canonical_option): Test g prefix.
4754 (option_map): Add -gno- mapping.
4755 (add_misspelling_candidates): Check remapping_prefix_p.
4757 2017-09-20 Jeff Law <law@redhat.com>
4759 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
4760 thinko in stack clash protection support.
4762 * explow.c (compute_stack_clash_protection_loop_data): Use
4763 CONST_INT_P instead of explicit test. Verify object is a
4764 CONST_INT_P before looking at INTVAL.
4765 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
4766 instead of explicit test.
4768 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
4771 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
4772 address instead of to r1 and r11.
4774 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
4776 * config.gcc: Support "knm".
4777 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
4778 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4780 * config/i386/i386.c (m_KNM): Define.
4781 (processor_target_table): Add "knm".
4783 (ix86_option_override_internal): Add "knm".
4784 (ix86_issue_rate): Add PROCESSOR_KNM.
4785 (ix86_adjust_cost): Ditto.
4786 (ia32_multipass_dfa_lookahead): Ditto.
4787 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
4788 (fold_builtin_cpu): Add M_INTEL_KNM.
4789 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
4790 (processor_type): Add PROCESSOR_KNM.
4791 * config/i386/x86-tune.def: Add m_KNM.
4792 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
4794 2017-09-20 Richard Biener <rguenther@suse.de>
4796 PR tree-optimization/80213
4797 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
4798 are allowed in empty BBs as well.
4799 (canonicalize_loop_closed_ssa): Also look for other complex
4801 (scop_detection::get_sese): Include the loop-closed PHI block
4803 (scop_detection::merge_sese): Remove code adding extra blocks.
4804 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
4805 (build_scops): Assert the final returned scop is invalid.
4807 2017-09-20 Richard Biener <rguenther@suse.de>
4809 PR tree-optimization/82264
4810 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
4812 (vn_phi_lookup): Likewise.
4813 (vn_phi_insert): Likewise.
4815 2017-09-20 Jakub Jelinek <jakub@redhat.com>
4817 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
4818 that fits into uhwi or shwi, add DW_AT_const_value regardless
4819 of early_dwarf without going through RTL, using add_AT_unsigned
4822 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
4823 (DEBUG_LTO_ABBREV_SECTION): Likewise.
4824 (DEBUG_LTO_MACINFO_SECTION): Likewise.
4825 (DEBUG_MACRO_SECTION): Likewise.
4826 (DEBUG_LTO_MACRO_SECTION): Likewise.
4827 (DEBUG_STR_DWO_SECTION): Likewise.
4828 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
4829 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
4830 (DEBUG_LTO_DWO_LINE_SECTION): Define.
4831 (DEBUG_LTO_LINE_STR_SECTION): Define.
4832 (init_sections_and_labels): Initialize debug_line_str_section
4833 variable. Initialize debug_loc_section for -gdwarf-5 to
4834 DEBUG_LOCLISTS_SECTION. Formatting fixes.
4836 2017-09-20 Richard Biener <rguenther@suse.de>
4838 * graphite-sese-to-poly.c (extract_affine): Properly handle
4839 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
4841 2017-09-20 Richard Biener <rguenther@suse.de>
4843 PR tree-optimization/81373
4844 * graphite-scop-detection.c (build_cross_bb_scalars_def):
4845 Force SESE live-out defs to be handled even if they are
4848 2017-09-19 Jeff Law <law@redhat.com>
4850 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
4851 nothing for stack adjustments with REG_STACK_CHECK.
4852 * sched-deps.c (parse_add_or_inc): Reject insns with
4853 REG_STACK_CHECK from dependency breaking.
4854 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
4855 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
4856 * reg-notes.def (STACK_CHECK): New note.
4858 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
4859 (ix86_expand_prologue): Dump stack clash info as needed.
4860 Call ix86_adjust_stack_and_probe_stack_clash as needed.
4862 * function.c (dump_stack_clash_frame_info): New function.
4863 * function.h (dump_stack_clash_frame_info): Prototype.
4864 (enum stack_clash_probes): New enum.
4866 * config/alpha/alpha.c (alpha_expand_prologue): Also check
4867 flag_stack_clash_protection.
4868 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
4869 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
4870 (arm_frame_pointer_required): Likewise.
4871 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
4872 (ia64_expand_prologue): Likewise.
4873 * config/mips/mips.c (mips_expand_prologue): Likewise.
4874 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
4875 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
4876 (sparc_flat_expand_prologue): Likewise.
4877 * config/spu/spu.c (spu_expand_prologue): Likewise.
4879 * explow.c: Include "params.h".
4880 (anti_adjust_stack_and_probe_stack_clash): New function.
4881 (get_stack_check_protect): Likewise.
4882 (compute_stack_clash_protection_loop_data): Likewise.
4883 (emit_stack_clash_protection_loop_start): Likewise.
4884 (emit_stack_clash_protection_loop_end): Likewise.
4885 (allocate_dynamic_stack_space): Use get_stack_check_protect.
4886 Use anti_adjust_stack_and_probe_stack_clash.
4887 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
4888 (emit_stack_clash_protection_loop_start): Likewise.
4889 (emit_stack_clash_protection_loop_end): Likewise.
4890 * rtl.h (get_stack_check_protect): Prototype.
4891 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
4892 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
4893 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
4895 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
4897 * doc/tm.texi: Rebuilt.
4898 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
4899 get_stack_check_protect.
4900 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
4901 * config/arm/arm.c (arm_expand_prologue): Likewise.
4902 (arm_frame_pointer_required): Likewise.
4903 * config/i386/i386.c (ix86_expand_prologue): Likewise.
4904 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
4905 * config/mips/mips.c (mips_expand_prologue): Likewise.
4906 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
4907 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
4908 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
4909 (sparc_flat_expand_prologue): Likewise.
4911 * common.opt (-fstack-clash-protection): New option.
4912 * flag-types.h (enum stack_check_type): Note difference between
4913 -fstack-check= and -fstack-clash-protection.
4914 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
4915 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
4916 * toplev.c (process_options): Issue warnings/errors for cases
4917 not handled with -fstack-clash-protection.
4918 * doc/invoke.texi (-fstack-clash-protection): Document new option.
4919 (-fstack-check): Note additional problem with -fstack-check=generic.
4920 Note that -fstack-check is primarily for Ada and refer users
4921 to -fstack-clash-protection for stack-clash-protection.
4922 Document new params for stack clash protection.
4924 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
4926 * config/i386/i386.c (ix86_split_long_move): Do not handle
4927 address used for LEA in a special way.
4929 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
4931 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
4933 2017-09-19 Martin Sebor <msebor@redhat.com>
4936 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
4937 of incompatible types.
4939 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
4941 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
4942 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
4943 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4944 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
4946 2017-09-19 Richard Biener <rguenther@suse.de>
4948 PR tree-optimization/82244
4949 * tree-vrp.c (remove_range_assertions): Do not propagate
4950 a constant to abnormals but replace the assert with a copy.
4952 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
4954 PR rtl-optimization/57878
4955 PR rtl-optimization/68988
4956 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
4957 avoidance test involving non_reload_pseudos. Move frequency test
4958 below the general fragmentation avoidance test.
4960 2017-09-19 Richard Biener <rguenther@suse.de>
4962 PR tree-optimization/69728
4963 * graphite-sese-to-poly.c (schedule_error): New global.
4964 (add_loop_schedule): Handle empty domain by failing the
4966 (build_original_schedule): Handle schedule_error.
4968 2017-09-19 Richard Biener <rguenther@suse.de>
4970 * graphite-scop-detection.c (scop_detection::can_represent_loop):
4971 Do not iterate to sibling loops but only to siblings of inner
4974 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
4977 * config/m68k/m68k.md (moveq feeding equality comparison): Check
4978 that the registers are different.
4980 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
4982 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
4983 to processor_model and "amdfam17h" to arch_names_table.
4984 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
4986 2017-09-18 Jakub Jelinek <jakub@redhat.com>
4989 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
4991 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4992 Alan Hayward <alan.hayward@arm.com>
4993 David Sherwood <david.sherwood@arm.com>
4995 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
4997 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
4999 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
5000 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
5001 vect_schedule_slp_instance.
5002 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
5003 Update call to vect_slp_analyze_node_operations. Simplify return
5005 (vect_slp_analyze_bb_1): Update call accordingly.
5006 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
5007 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
5008 (vect_schedule_slp): Update call accordingly.
5010 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5011 Alan Hayward <alan.hayward@arm.com>
5012 David Sherwood <david.sherwood@arm.com>
5014 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
5015 with types that aren't in fact scalar.
5017 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5019 * tree-vect-slp.c (vect_record_max_nunits): New function,
5021 (vect_build_slp_tree_1): ...here.
5022 (vect_build_slp_tree_2): Call it for phis too.
5024 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5026 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
5027 to vect_get_vec_def_for_operand when getting the mask operand.
5029 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5030 Alan Hayward <alan.hayward@arm.com>
5031 David Sherwood <david.sherwood@arm.com>
5033 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
5036 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5037 Alan Hayward <alan.hayward@arm.com>
5038 David Sherwood <david.sherwood@arm.com>
5040 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
5041 calculation for vector booleans.
5043 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5044 Alan Hayward <alan.hayward@arm.com>
5045 David Sherwood <david.sherwood@arm.com>
5047 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
5049 (vect_transform_stmt): ...here.
5050 (vect_analyze_stmt): Use it instead of calling
5051 vectorizable_live_operation directly.
5053 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
5055 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
5056 non-SIMT targets in acc vector loops.
5058 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
5060 * configure.ac: Add arc and check if assembler supports gdwarf2.
5061 * configure: Regenerate.
5063 2017-09-18 Richard Biener <rguenther@suse.de>
5065 PR tree-optimization/82220
5066 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
5067 epilogue niters from the min_profitable_iters compute.
5069 2017-09-18 Jakub Jelinek <jakub@redhat.com>
5072 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
5073 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
5074 (ix86_init_pic_reg): Revert 2017-09-01 changes.
5076 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
5079 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
5080 switching to a new text section.
5082 2017-09-18 Richard Biener <rguenther@suse.de>
5084 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
5086 (build_alias_set): Reject aliases with no access function.
5088 2017-09-18 Richard Biener <rguenther@suse.de>
5090 PR tree-optimization/79622
5091 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
5093 (build_cross_bb_scalars_use): Likewise.
5095 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
5097 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
5099 2017-09-18 Alan Modra <amodra@gmail.com>
5102 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
5103 stack_pointer_rtx for count 0. Update comments. Break up
5104 large rtl expression.
5106 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
5108 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
5109 Increase to 20 bytes.
5110 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
5111 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
5112 or avx version of the stub.
5114 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
5117 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
5118 compute the minimum stack alignment. Also update preferred stack
5119 boundary for leaf functions.
5121 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
5123 PR tree-optimization/82228
5124 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
5127 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
5129 * common/config/nds32/nds32-common.c
5130 (nds32_option_optimization_table): Refine formatting.
5131 (nds32_option_optimization_table): Use -fsched-pressure and
5132 -fomit-frame-pointer for specific optimization level.
5134 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
5136 * config/nds32/nds32.c: Refine formatting and comments.
5137 * config/nds32/nds32.h: Likewise.
5138 * config/nds32/nds32.md: Likewise.
5139 * config/nds32/nds32-cost.c: Likewise.
5140 * config/nds32/nds32-isr.c: Likewise.
5141 * config/nds32/nds32-md-auxiliary.c: Likewise.
5142 * config/nds32/nds32-multiple.md: Likewise.
5143 * config/nds32/nds32-predicates.c: Likewise.
5145 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
5146 Jakub Jelinek <jakub@redhat.com>
5148 Add support for -std=c++2a.
5149 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
5151 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
5153 2017-09-15 Steve Ellcey <sellcey@cavium.com>
5156 * doc/extend.texi (Common Function Attributes): Add
5157 references to ARM, AArch64, and S/390 specific attributes.
5158 (Function Specific Option Pragmas): Add AArch64 and S/390
5159 to list of back ends that support the target pragma.
5161 2017-09-15 Nathan Sidwell <nathan@acm.org>
5163 * doc/standards.texi: Fix C++17 description. Update URLs for
5166 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
5168 * common.opt (Wcast-align=strict): New warning option.
5169 * doc/invoke.texi: Document -Wcast-align=strict.
5171 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
5173 * cgraph.h (cgraph_thunk_info): Add comments.
5174 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
5175 assert for VIRTUAL_* arguments stricter.
5177 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
5179 PR tree-optimization/71026
5180 * match.pd: Move RDIV patterns from fold-const.c
5181 * fold-const.c (distribute_real_division): Removed.
5182 (fold_binary_loc): Remove calls to distribute_real_divison.
5184 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5186 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
5187 c++1z and gnu++1z as deprecated. Change other references to
5188 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
5189 Change -Wc++1z-compat to -Wc++17-compat.
5190 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
5191 * dwarf2out.c (highest_c_language): Handle C++17.
5192 (gen_compile_unit_die): Likewise.
5194 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5196 PR rtl-optimization/82192
5197 * combine.c (make_extraction): Don't look through non-paradoxical
5198 SUBREGs or TRUNCATE if pos + len is or might be bigger than
5201 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5202 Alan Hayward <alan.hayward@arm.com>
5203 David Sherwood <david.sherwood@arm.com>
5205 * target.def (function_arg_offset): New hook.
5206 * targhooks.h (default_function_arg_offset): Declare.
5207 * targhooks.c (default_function_arg_offset): New function.
5208 * function.c (locate_and_pad_parm): Use
5209 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
5210 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
5211 (TARGET_FUNCTION_ARG_OFFSET): ...this.
5212 * doc/tm.texi: Regenerate.
5213 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
5214 * config/spu/spu.c (spu_function_arg_offset): New function.
5215 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
5216 * system.h (FUNCTION_ARG_OFFSET): Poison.
5218 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5219 Alan Hayard <alan.hayward@arm.com>
5220 David Sherwood <david.sherwood@arm.com>
5222 * target.def (truly_noop_truncation): New hook.
5223 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
5224 than TRULY_NOOP_TRUNCATION.
5225 * hooks.h (hook_bool_uint_uint_true): Declare.
5226 * hooks.c (hook_bool_uint_uint_true): New function.
5227 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
5228 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
5229 * doc/tm.texi: Regenerate.
5230 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
5231 rather than TRULY_NOOP_TRUNCATION in comments.
5232 (simplify_comparison): Likewise.
5233 (record_truncated_value): Likewise.
5234 * expmed.c (extract_bit_field_1): Likewise.
5235 (extract_split_bit_field): Likewise.
5236 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
5237 instead of TRULY_NOOP_TRUNCATION.
5238 * function.c (assign_parm_setup_block): Likewise.
5239 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
5240 * rtlhooks.c: Include target.h.
5241 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
5242 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
5243 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
5244 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
5245 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
5246 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
5247 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
5248 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
5249 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
5250 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
5251 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
5252 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
5253 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
5254 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
5255 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
5256 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
5257 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
5258 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
5259 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
5260 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
5261 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
5262 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
5263 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
5264 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
5265 * config/mips/mips.c (mips_truly_noop_truncation): New function.
5266 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
5267 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5268 rather than TRULY_NOOP_TRUNCATION in comments.
5269 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
5270 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
5271 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
5272 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
5273 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
5274 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
5275 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
5276 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
5277 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
5278 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
5279 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
5280 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5281 rather than TRULY_NOOP_TRUNCATION in comments.
5282 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
5283 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
5284 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
5285 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
5286 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
5287 TRULY_NOOP_TRUNCATION condition.
5288 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
5289 (TRULY_NOOP_TRUNCATION): Delete.
5290 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
5291 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
5292 * config/spu/spu.c (spu_truly_noop_truncation): New function.
5293 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
5294 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
5295 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
5296 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
5297 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
5298 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5299 rather than TRULY_NOOP_TRUNCATION in comments.
5300 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
5301 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
5302 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
5303 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
5304 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
5305 * system.h (TRULY_NOOP_TRUNCATION): Poison.
5307 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
5310 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
5311 (*cmp_ior): Likewise.
5312 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
5313 (*ior_scc_scc_cmp): Likewise.
5314 (*and_scc_scc): Likewise.
5315 (*and_scc_scc_cmp): Likewise.
5317 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5318 Alan Hayard <alan.hayward@arm.com>
5319 David Sherwood <david.sherwood@arm.com>
5321 * target.def (can_change_mode_class): New hook.
5322 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
5323 (hard_regno_nregs): Likewise.
5324 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
5325 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
5326 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
5327 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
5328 (LOAD_EXTEND_OP): Update accordingly.
5329 * doc/tm.texi: Regenerate.
5330 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
5331 CANNOT_CHANGE_MODE_CLASS.
5332 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
5333 (REG_CAN_CHANGE_MODE_P): ...this new macro.
5334 * combine.c (simplify_set): Update accordingly.
5335 * emit-rtl.c (validate_subreg): Likewise.
5336 * recog.c (general_operand): Likewise.
5337 * regcprop.c (mode_change_ok): Likewise.
5338 * reload1.c (choose_reload_regs): Likewise.
5339 (inherit_piecemeal_p): Likewise.
5340 * rtlanal.c (simplify_subreg_regno): Likewise.
5341 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
5342 instead of CANNOT_CHANGE_MODE_CLASS.
5343 (reload_cse_simplify_operands): Likewise.
5344 * reload.c (push_reload): Use targetm.can_change_mode_class
5345 instead of CANNOT_CHANGE_MODE_CLASS.
5346 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
5347 REG_CANNOT_CHANGE_MODE_P.
5348 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5349 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
5350 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5351 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5352 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5353 (arm_can_change_mode_class): New function.
5354 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
5355 than CANNOT_CHANGE_MODE_CLASS in comments.
5356 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5357 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
5358 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
5359 (ix86_can_change_mode_class): ...this new function, inverting the
5360 sense of the return value.
5361 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5362 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5363 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5364 (ia64_can_change_mode_class): New function.
5365 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5366 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
5367 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
5368 (m32c_can_change_mode_class): ...this new function, inverting the
5369 sense of the return value.
5370 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5371 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5372 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
5373 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
5374 (mips_can_change_mode_class): ...this new function, inverting the
5375 sense of the return value.
5376 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5377 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5378 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5379 (msp430_can_change_mode_class): New function.
5380 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5381 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
5382 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5383 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5384 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5385 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
5386 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5387 (pa_cannot_change_mode_class): Replace with...
5388 (pa_can_change_mode_class): ...this new function, inverting the
5389 sense of the return value.
5390 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
5391 than CANNOT_CHANGE_MODE_CLASS in comments.
5392 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5393 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
5394 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5395 (pdp11_cannot_change_mode_class): Replace with...
5396 (pdp11_can_change_mode_class): ...this new function, inverting the
5397 sense of the return value.
5398 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5399 * config/powerpcspe/powerpcspe-protos.h
5400 (rs6000_cannot_change_mode_class_ptr): Delete.
5401 * config/powerpcspe/powerpcspe.c
5402 (rs6000_cannot_change_mode_class_ptr): Delete.
5403 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5404 (rs6000_option_override_internal): Assign to
5405 targetm.can_change_mode_class instead of
5406 rs6000_cannot_change_mode_class_ptr.
5407 (rs6000_cannot_change_mode_class): Replace with...
5408 (rs6000_can_change_mode_class): ...this new function, inverting the
5409 sense of the return value.
5410 (rs6000_debug_cannot_change_mode_class): Replace with...
5411 (rs6000_debug_can_change_mode_class): ...this new function.
5412 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5413 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
5414 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5415 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5416 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
5418 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
5419 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5420 (rs6000_option_override_internal): Assign to
5421 targetm.can_change_mode_class instead of
5422 rs6000_cannot_change_mode_class_ptr.
5423 (rs6000_cannot_change_mode_class): Replace with...
5424 (rs6000_can_change_mode_class): ...this new function, inverting the
5425 sense of the return value.
5426 (rs6000_debug_cannot_change_mode_class): Replace with...
5427 (rs6000_debug_can_change_mode_class): ...this new function.
5428 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5429 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
5430 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
5431 (s390_can_change_mode_class): ...this new function, inverting the
5432 sense of the return value.
5433 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5434 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5435 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
5436 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5437 (sh_cannot_change_mode_class): Replace with...
5438 (sh_can_change_mode_class): ...this new function, inverting the
5439 sense of the return value.
5440 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5441 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5442 (sparc_can_change_mode_class): New function.
5443 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5444 * config/spu/spu.c (spu_can_change_mode_class): New function.
5445 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5446 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5447 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5448 (visium_can_change_mode_class): New function.
5449 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
5451 2017-09-15 Richard Biener <rguenther@suse.de>
5453 PR tree-optimization/82217
5454 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
5455 but not undefined case.
5457 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5460 * postreload.c (reload_cse_simplify_operands): Skip
5461 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
5463 2017-09-15 Richard Biener <rguenther@suse.de>
5465 PR tree-optimization/68823
5466 * graphite-scop-detection.c (build_alias_set): If we have a
5467 possible dependence check whether we can handle them by just
5468 looking at the DRs DR_ACCESS_FNs.
5469 (build_scops): If build_alias_set fails, fail the SCOP.
5471 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
5473 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
5474 to support float128 built-in functions that require the ISA 3.0
5476 (BU_FLOAT128_3_HW): Likewise.
5477 (SQRTF128): Add support for the IEEE 128-bit square root and fma
5479 (FMAF128): Likewise.
5481 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
5482 support for built-in functions that need the ISA 3.0 IEEE 128-bit
5483 floating point instructions.
5484 (rs6000_invalid_builtin): Likewise.
5485 (rs6000_builtin_mask_names): Likewise.
5486 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
5487 (RS6000_BTM_FLOAT128_HW): Likewise.
5488 (RS6000_BTM_COMMON): Likewise.
5489 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
5491 * doc/extend.texi (RS/6000 built-in functions): Document the
5492 IEEE 128-bit floating point square root and fused multiply-add
5495 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
5497 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
5498 reg (r2) isn't in the set of registers defined in the prologue.
5500 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5501 Alan Hayward <alan.hayward@arm.com>
5502 David Sherwood <david.sherwood@arm.com>
5504 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
5505 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
5506 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
5507 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
5508 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
5510 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
5511 max_vectorization_factor.
5512 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
5514 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5515 Alan Hayward <alan.hayward@arm.com>
5516 David Sherwood <david.sherwood@arm.com>
5518 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
5519 (vect_worthwhile_without_simd_p): Declare.
5520 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
5521 (vectorizable_reduction): Use it.
5522 * tree-vect-stmts.c (vectorizable_shift): Likewise.
5523 (vectorizable_operation): Likewise.
5525 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5526 Alan Hayward <alan.hayward@arm.com>
5527 David Sherwood <david.sherwood@arm.com>
5529 * tree-vectorizer.h (vect_get_num_copies): New function.
5530 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
5531 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5532 (vectorizable_induction): Likewise.
5533 (vectorizable_live_operation): Likewise.
5534 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5535 (vectorizable_bswap): Likewise.
5536 (vectorizable_call): Likewise.
5537 (vectorizable_conversion): Likewise.
5538 (vectorizable_assignment): Likewise.
5539 (vectorizable_shift): Likewise.
5540 (vectorizable_operation): Likewise.
5541 (vectorizable_store): Likewise.
5542 (vectorizable_load): Likewise.
5543 (vectorizable_condition): Likewise.
5544 (vectorizable_comparison): Likewise.
5545 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
5547 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5548 Alan Hayward <alan.hayward@arm.com>
5549 David Sherwood <david.sherwood@arm.com>
5551 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
5552 of vect_init_vector.
5554 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5555 Alan Hayward <alan.hayward@arm.com>
5556 David Sherwood <david.sherwood@arm.com>
5558 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
5559 an inline wrapper that provides a location.
5560 (gimple_build_vector): Likewise.
5561 * gimple-fold.c (gimple_build_vector_from_val): New function.
5562 (gimple_build_vector): Likewise.
5563 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
5564 functions to build the initial value. Always return a gimple value.
5565 (get_initial_defs_for_reduction): Likewise. Only compute
5567 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
5568 vect_init_vector on the results from get_initial_def(s)_for_reduction.
5569 (vectorizable_induction): Use gimple_build_vector rather than
5572 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5573 Alan Hayward <alan.hayward@arm.com>
5574 David Sherwood <david.sherwood@arm.com>
5576 * target.h (vec_perm_indices): New typedef.
5577 (auto_vec_perm_indices): Likewise.
5578 * optabs-query.h: Include target.h
5579 (can_vec_perm_p): Take a vec_perm_indices *.
5580 * optabs-query.c (can_vec_perm_p): Likewise.
5581 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
5582 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
5583 * tree-vect-generic.c (lower_vec_perm): Likewise.
5584 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
5585 (vect_grouped_load_supported): Likewise.
5586 (vect_shift_permute_load_chain): Likewise.
5587 (vect_permute_store_chain): Use auto_vec_perm_indices.
5588 (vect_permute_load_chain): Likewise.
5589 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
5590 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
5591 Update uses of can_vec_perm_p.
5592 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
5593 mode with a number of elements. Take a vec_perm_indices *.
5594 (vect_create_epilog_for_reduction): Update accordingly.
5595 Use auto_vec_perm_indices.
5596 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
5597 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
5598 (vect_transform_slp_perm_load): Likewise.
5599 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
5600 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
5601 (vect_gen_perm_mask_checked): Likewise.
5602 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
5603 (vect_gen_perm_mask_checked): Likewise.
5604 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
5605 (vectorizable_store): Likewise.
5606 (vectorizable_load): Likewise.
5607 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
5608 (vectorizable_bswap): Likewise.
5610 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5611 Alan Hayward <alan.hayward@arm.com>
5612 David Sherwood <david.sherwood@arm.com>
5614 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
5615 * tree.c (build_vector): Likewise.
5616 (build_vector_from_ctor): Update accordingly.
5617 (build_vector_from_val): Likewise.
5618 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
5619 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
5620 * tree-vect-generic.c (add_rshift): Likewise.
5621 (expand_vector_divmod): Likewise.
5622 (optimize_vector_constructor): Likewise.
5623 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
5624 (vect_transform_slp_perm_load): Likewise.
5625 (vect_schedule_slp_instance): Likewise.
5626 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
5627 (vectorizable_call): Likewise.
5628 (vect_gen_perm_mask_any): Likewise. Add elements in order.
5629 * expmed.c (make_tree): Likewise.
5630 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
5631 a vector passed to build_vector.
5632 (fold_convert_const): Likewise.
5633 (exact_inverse): Likewise.
5634 (fold_ternary_loc): Likewise.
5635 (fold_relational_const): Likewise.
5636 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
5637 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
5638 (const_unop): Likewise. Store the reduction accumulator in a
5639 variable rather than an array.
5640 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
5641 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
5642 the new vector, rather than constructing it after the input arrays.
5643 (native_interpret_vector): Use auto_vec<tree> when building
5644 a vector passed to build_vector. Add elements in order.
5645 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
5646 auto_vec<tree> when building a vector passed to build_vector.
5647 (vect_create_epilog_for_reduction): Likewise.
5648 (vectorizable_induction): Likewise.
5649 (get_initial_def_for_reduction): Likewise. Fix indentation of
5651 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
5653 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
5654 passed to build_vector.
5656 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5657 Alan Hayward <alan.hayward@arm.com>
5658 David Sherwood <david.sherwood@arm.com>
5660 * tree-core.h (tree_base::u): Add an "nelts" field.
5661 (tree_vector): Use VECTOR_CST_NELTS as the length.
5662 * tree.c (tree_size): Likewise.
5663 (make_vector): Initialize VECTOR_CST_NELTS.
5664 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
5665 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
5666 TYPE_VECTOR_SUBPARTS.
5667 * expr.c (const_vector_mask_from_tree): Consistently use "units"
5668 as the number of units, setting it from VECTOR_CST_NELTS.
5669 (const_vector_from_tree): Likewise.
5670 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
5671 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
5672 (fold_negate_expr_1): Likewise.
5673 (fold_convert_const): Likewise.
5674 (const_binop): Likewise. Differentiate the number of output and
5676 (const_unop): Likewise.
5677 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
5678 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
5679 in cases that did the opposite.
5681 2017-09-14 Richard Biener <rguenther@suse.de>
5683 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
5686 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
5688 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
5690 2017-09-14 Jakub Jelinek <jakub@redhat.com>
5693 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
5694 if and where to split a bb, except for splitting before debug insn
5695 sequences followed by non-label real insn. Delete debug insns
5696 in between basic blocks.
5698 * combine.c (make_compound_operation_int): Formatting fixes.
5700 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
5701 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5702 * config/netbsd.h (LINK_EH_SPEC): Likewise.
5703 * config/sol2.h (LINK_EH_SPEC): Likewise.
5704 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5705 * config/s390/linux.h (LINK_SPEC): Likewise.
5706 * config/freebsd.h (LINK_EH_SPEC): Likewise.
5707 * config/openbsd.h (LINK_EH_SPEC): Likewise.
5708 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5709 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
5710 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
5711 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5712 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
5713 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
5715 2017-09-13 Jakub Jelinek <jakub@redhat.com>
5717 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
5719 (ENDFILE_LINUX_SPEC): Likewise.
5720 (LINK_EH_SPEC): Likewise.
5721 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
5722 (LINK_OS_LINUX_SPEC32): Likewise.
5723 (LINK_OS_LINUX_SPEC64): Likewise.
5724 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
5725 (LINK_OS_LINUX_SPEC): Likewise.
5727 2017-09-13 Martin Liska <mliska@suse.cz>
5730 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
5731 CASE_HIGH is NULL_TREE.
5733 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
5734 Alan Hayward <alan.hayward@arm.com>
5735 David Sherwood <david.sherwood@arm.com>
5737 * target.def (secondary_memory_needed): New hook.
5738 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
5739 instead of SECONDARY_MEMORY_NEEDED.
5740 (secondary_memory_needed_mode): Likewise.
5741 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
5742 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
5743 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
5744 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
5745 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
5746 * doc/tm.texi: Regenerate.
5747 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
5748 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
5749 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5750 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
5751 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
5752 * config/i386/i386.c (inline_secondary_memory_needed): Put the
5753 mode argument first and change the reg_class arguments to reg_class_t.
5754 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
5755 Make static. Update the call to inline_secondary_memory_needed.
5756 (ix86_register_move_cost): Update the call to
5757 inline_secondary_memory_needed.
5758 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5759 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
5761 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
5762 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5764 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
5765 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
5766 * config/mips/mips.c (mips_secondary_memory_needed): Make static
5767 and match hook interface. Add comment from mips.h.
5768 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5769 * config/mmix/mmix.md (truncdfsf2): Refer to
5770 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5772 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
5773 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
5774 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5775 (pa_secondary_memory_needed): New function.
5776 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
5777 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
5778 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5779 (pdp11_secondary_memory_needed): Make static and match hook interface.
5780 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
5781 * config/powerpcspe/powerpcspe-protos.h
5782 (rs6000_secondary_memory_needed_ptr): Delete.
5783 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
5785 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5786 (rs6000_option_override_internal): Assign to
5787 targetm.secondary_memory_needed rather than
5788 rs6000_secondary_memory_needed_ptr.
5789 (rs6000_secondary_memory_needed): Match hook interface.
5790 (rs6000_debug_secondary_memory_needed): Likewise.
5791 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
5792 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
5793 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
5794 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5795 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
5796 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
5798 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
5799 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5800 (rs6000_option_override_internal): Assign to
5801 targetm.secondary_memory_needed rather than
5802 rs6000_secondary_memory_needed_ptr.
5803 (rs6000_secondary_memory_needed): Match hook interface.
5804 (rs6000_debug_secondary_memory_needed): Likewise.
5805 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
5806 * config/s390/s390.c (s390_secondary_memory_needed): New function.
5807 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5808 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
5809 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5810 (sparc_secondary_memory_needed): New function.
5811 * lra-constraints.c (check_and_process_move): Refer to
5812 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5814 (curr_insn_transform): Likewise.
5815 (process_alt_operands): Use targetm.secondary_memory_needed
5816 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5817 (check_secondary_memory_needed_p): Likewise.
5818 (choose_split_class): Likewise.
5819 * reload.c: Unconditionally include code that was previously
5820 conditional on SECONDARY_MEMORY_NEEDED.
5821 (push_secondary_reload): Use targetm.secondary_memory_needed
5822 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5823 (push_reload): Likewise.
5824 * reload1.c: Unconditionally include code that was previously
5825 conditional on SECONDARY_MEMORY_NEEDED.
5826 (choose_reload_regs): Use targetm.secondary_memory_needed
5827 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5828 (gen_reload): Likewise.
5829 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
5831 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
5832 Alan Hayward <alan.hayward@arm.com>
5833 David Sherwood <david.sherwood@arm.com>
5835 * target.def (secondary_memory_needed_mode): New hook:
5836 * targhooks.c (default_secondary_memory_needed_mode): Declare.
5837 * targhooks.h (default_secondary_memory_needed_mode): New function.
5838 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
5839 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
5840 * doc/tm.texi: Regenerate.
5841 * lra-constraints.c (check_and_process_move): Use
5842 targetm.secondary_memory_needed_mode instead of
5843 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
5844 (curr_insn_transform): Likewise.
5845 * reload.c (get_secondary_mem): Likewise.
5846 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5847 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
5849 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5850 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5851 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
5852 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5853 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
5855 * config/powerpcspe/powerpcspe-protos.h
5856 (rs6000_secondary_memory_needed_mode): Delete.
5857 * config/powerpcspe/powerpcspe.c
5858 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5859 (rs6000_secondary_memory_needed_mode): Make static.
5860 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5861 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
5863 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
5865 (rs6000_secondary_memory_needed_mode): Make static.
5866 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5867 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
5868 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5869 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5870 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
5872 (sparc_secondary_memory_needed_mode): New function.
5873 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
5875 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
5877 * config/aarch64/constraints.md (Umq): New constraint.
5878 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
5880 (mov<mode>): Update condition.
5882 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5884 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
5885 when bitposition is the same.
5887 2017-09-13 Richard Biener <rguenther@suse.de>
5889 * dwarf2out.c (output_die_symbol): Remove.
5890 (output_die): Do not output a DIEs symbol.
5892 2017-09-13 Richard Biener <rguenther@suse.de>
5895 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
5896 default-def to avoid breaking iterator update with the weird
5897 interaction with cgraph_update_edges_for_call_stmt_node.
5899 2017-09-13 Richard Biener <rguenther@suse.de>
5901 * tree-cfg.c (verify_gimple_assign_binary): Add verification
5902 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
5903 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
5904 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
5906 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
5908 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
5909 Disable pc relative literal load irrespective of
5910 TARGET_FIX_ERR_A53_84341 for default.
5912 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
5914 * config/sparc/sparc.c (output_return): Output the source location of
5915 the insn in the delay slot, if any.
5916 (output_sibcall): Likewise.
5918 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
5921 * common.opt (-static-pie): New alias.
5922 (shared): Negate static-pie.
5923 (-no-pie): Update help text.
5925 (static-pie): New option.
5926 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
5927 -static-pie support.
5928 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
5929 (LINK_EH_SPEC): Likewise.
5930 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5931 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
5932 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
5933 * gcc.c (LINK_COMMAND_SPEC): Likewise.
5934 (init_gcc_specs): Likewise.
5935 (init_spec): Likewise.
5936 (display_help): Update help message for -pie.
5937 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
5940 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
5942 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
5943 (movdi_aarch64): Likewise.
5944 (movti_aarch64): Likewise.
5946 2017-09-12 Simon Wright <simon@pushface.org>
5949 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
5950 calculation of the minor version, always output as 0.
5952 2017-09-12 Jakub Jelinek <jakub@redhat.com>
5955 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5956 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
5957 on it early, rather than manual conversion late. For
5958 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
5959 instead of performing manual conversion.
5961 2017-09-12 Carl Love <cel@us.ibm.com>
5963 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
5964 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
5966 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5967 VMULOSW): Add definitions.
5968 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5969 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5970 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5971 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5972 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5974 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
5976 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
5978 (movti_aarch64): Likewise.
5979 (movdf_aarch64): Likewise.
5980 (movtf_aarch64): Likewise.
5981 (load_pairdi): Likewise.
5982 (store_pairdi): Likewise.
5983 (load_pairdf): Likewise.
5984 (store_pairdf): Likewise.
5985 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
5986 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
5987 (ldr_got_small_<mode>): Likewise.
5988 (ldr_got_small_28k_<mode>): Likewise.
5989 (ldr_got_tiny): Likewise.
5990 * config/aarch64/iterators.md (ldst_sz): New.
5991 (ldpstp_sz): Likewise.
5992 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
5994 (thunderx_load): Split load_8 to load_16.
5995 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
5997 (thunderx2t99_storepair_basic): Split store_8 to store_16.
5998 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
5999 (xgene1_store_pair): Split store_8 to store_16.
6000 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
6001 (falkor_st_0_st_sd): Split store_8 to store_16.
6003 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
6005 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
6006 and store1/2/3/4 to store_4/8/12/16.
6007 * config/aarch64/aarch64.md: Update for rename.
6008 * config/arm/arm.md: Likewise.: Likewise.
6009 * config/arm/arm.c: Likewise.
6010 * config/arm/thumb1.md: Likewise.
6011 * config/arm/thumb2.md: Likewise.
6012 * config/arm/vfp.md: Likewise.
6013 * config/arm/arm-generic.md: Likewise.
6014 * config/arm/arm1020e.md: Likewise.
6015 * config/arm/arm1026ejs.md: Likewise.
6016 * config/arm/arm1136jfs.md: Likewise.
6017 * config/arm/arm926ejs.md: Likewise.
6018 * config/arm/cortex-a15.md: Likewise.
6019 * config/arm/cortex-a17.md: Likewise.
6020 * config/arm/cortex-a5.md: Likewise.
6021 * config/arm/cortex-a53.md: Likewise.
6022 * config/arm/cortex-a57.md: Likewise.
6023 * config/arm/cortex-a7.md: Likewise.
6024 * config/arm/cortex-a8.md: Likewise.
6025 * config/arm/cortex-a9.md: Likewise.
6026 * config/arm/cortex-m4.md: Likewise.
6027 * config/arm/cortex-m7.md: Likewise.
6028 * config/arm/cortex-r4.md: Likewise.
6029 * config/arm/exynos-m1.md: Likewise.
6030 * config/arm/fa526.md: Likewise.
6031 * config/arm/fa606te.md: Likewise.
6032 * config/arm/fa626te.md: Likewise.
6033 * config/arm/fa726te.md: Likewise.
6034 * config/arm/fmp626.md: Likewise.
6035 * config/arm/iwmmxt.md: Likewise.
6036 * config/arm/ldmstm.md: Likewise.
6037 * config/arm/marvell-pj4.md: Likewise.
6038 * config/arm/xgene1.md: Likewise.
6039 * config/aarch64/thunderx.md: Likewise.
6040 * config/aarch64/thunderx2t99.md: Likewise.
6041 * config/aarch64/falkor.md: Likewise.
6043 2017-09-12 Martin Liska <mliska@suse.cz>
6045 * attribs.c (private_lookup_attribute): New function.
6046 * attribs.h (private_lookup_attribute): Declared here.
6047 (lookup_attribute): Called from this place.
6049 2017-09-12 Richard Biener <rguenther@suse.de>
6051 PR tree-optimization/82157
6052 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
6053 stmts with side-effects.
6055 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6056 Alan Hayward <alan.hayward@arm.com>
6057 David Sherwood <david.sherwood@arm.com>
6059 * target.def (hard_regno_nregs): New hook.
6060 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
6061 * targhooks.h (default_hard_regno_nregs): Declare.
6062 * targhooks.c (default_hard_regno_nregs): New function.
6063 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
6064 (TARGET_HARD_REGNO_NREGS): ...this hook.
6065 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
6066 (CLASS_MAX_NREGS): Likewise.
6067 * doc/tm.texi: Regenerate.
6068 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
6069 instead of HARD_REGNO_NREGS.
6070 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
6071 HARD_REGNO_NREGS in the comment.
6072 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
6073 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
6074 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
6075 Return an unsigned int.
6076 (TARGET_HARD_REGNO_NREGS): Redefine.
6077 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
6078 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
6079 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
6080 (arc_hard_regno_nregs): New function.
6081 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
6082 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
6083 (arm_hard_regno_nregs): New function.
6084 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
6085 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
6086 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
6087 (TARGET_HARD_REGNO_NREGS): Redefine.
6088 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
6089 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
6090 (HARD_REGNO_NREGS): Delete.
6091 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
6092 (cr16_hard_regno_nregs): New function.
6093 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
6094 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
6095 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
6096 (cris_hard_regno_nregs): New function.
6097 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
6098 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
6099 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
6100 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
6101 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
6102 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
6103 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
6104 (frv_hard_regno_nregs): Make static. Take and return an
6106 (frv_class_max_nregs): Remove outdated copy of documentation.
6107 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
6108 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
6109 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
6110 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
6111 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
6112 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
6113 (TARGET_HARD_REGNO_NREGS): Redefine.
6114 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
6115 (CLASS_MAX_NREGS): Update comment.
6116 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
6117 (ia64_hard_regno_nregs): New function.
6118 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
6119 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
6120 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
6121 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
6122 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
6124 (m32c_hard_regno_nregs): Likewise. Make static.
6125 (TARGET_HARD_REGNO_NREGS): Redefine.
6126 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
6127 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
6128 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
6129 (m68k_hard_regno_nregs): New function.
6130 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
6131 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
6132 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
6133 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
6134 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
6135 Take and return an unsigned int.
6136 (TARGET_HARD_REGNO_NREGS): Redefine.
6137 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
6138 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
6139 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
6140 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
6141 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
6142 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
6143 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
6144 (msp430_hard_regno_nregs): Make static. Take and return an
6146 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
6147 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
6148 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
6149 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
6150 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
6151 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
6152 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
6153 (TARGET_HARD_REGNO_NREGS): Redefine.
6154 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
6155 (PA_HARD_REGNO_NREGS): ...this.
6156 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
6157 (PA_HARD_REGNO_NREGS): ...this.
6158 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
6159 (pa_hard_regno_nregs): New function.
6160 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
6161 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
6162 (pdp11_hard_regno_nregs): New function.
6163 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
6164 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
6165 (rs6000_hard_regno_nregs_hook): New function.
6166 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
6167 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
6168 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
6169 Take and return an unsigned int. Move earlier in file.
6170 (TARGET_HARD_REGNO_NREGS): Redefine.
6171 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
6172 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
6173 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
6174 (rl78_hard_regno_nregs): Make static. Take and return an
6176 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
6177 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
6178 (rs6000_hard_regno_nregs_hook): New function.
6179 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
6180 * config/rx/rx.c (rx_hard_regno_nregs): New function.
6181 (TARGET_HARD_REGNO_NREGS): Redefine.
6182 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
6183 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
6184 instead of HARD_REGNO_NREGS.
6185 (s390_hard_regno_nregs): New function.
6186 (s390_hard_regno_mode_ok): Add comment from s390.h.
6187 (TARGET_HARD_REGNO_NREGS): Redefine.
6188 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
6189 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
6190 (sh_hard_regno_nregs): New function.
6191 (sh_pass_in_reg_p): Use it.
6192 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
6193 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
6194 (sparc_hard_regno_nregs): New function.
6195 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
6196 * config/spu/spu.c (spu_hard_regno_nregs): New function.
6197 (spu_function_arg_advance): Use it, supplying a valid register number.
6198 (TARGET_HARD_REGNO_NREGS): Redefine.
6199 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
6200 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
6201 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
6202 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
6203 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
6204 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
6205 (CLASS_MAX_NREGS): Remove copy of old documentation.
6206 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
6207 (visium_hard_regno_nregs): New function.
6208 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
6209 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
6210 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
6211 xtensa_hard_regno_nregs): New function.
6212 * system.h (HARD_REGNO_NREGS): Poison.
6214 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6216 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
6217 hard_regno_nregs instead of HARD_REGNO_NREGS.
6218 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
6219 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
6220 (c6x_expand_epilogue): Likewise.
6221 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
6222 (frv_read_iacc_argument): Likewise.
6223 * config/sh/sh.c: Include regs.h.
6224 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
6225 (regs_used): Likewise.
6226 (output_stack_adjust): Likewise.
6227 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
6228 * expmed.c: Include regs.h.
6229 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
6230 * ree.c: Include regs.h.
6231 (combine_reaching_defs): Use hard_regno_nregs instead of
6233 (add_removable_extension): Likewise.
6235 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6237 * regs.h (hard_regno_nregs): Turn into a function.
6238 (end_hard_regno): Update accordingly.
6239 * caller-save.c (setup_save_areas): Likewise.
6240 (save_call_clobbered_regs): Likewise.
6241 (replace_reg_with_saved_mem): Likewise.
6242 (insert_restore): Likewise.
6243 (insert_save): Likewise.
6244 * combine.c (can_change_dest_mode): Likewise.
6245 (move_deaths): Likewise.
6246 (distribute_notes): Likewise.
6247 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
6248 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
6249 (rs6000_split_multireg_move): Likewise.
6250 (rs6000_register_move_cost): Likewise.
6251 (rs6000_memory_move_cost): Likewise.
6252 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
6253 (rs6000_split_multireg_move): Likewise.
6254 (rs6000_register_move_cost): Likewise.
6255 (rs6000_memory_move_cost): Likewise.
6256 * cselib.c (cselib_reset_table): Likewise.
6257 (cselib_lookup_1): Likewise.
6258 * emit-rtl.c (set_mode_and_regno): Likewise.
6259 * function.c (aggregate_value_p): Likewise.
6260 * ira-color.c (setup_profitable_hard_regs): Likewise.
6261 (check_hard_reg_p): Likewise.
6262 (calculate_saved_nregs): Likewise.
6263 (assign_hard_reg): Likewise.
6264 (improve_allocation): Likewise.
6265 (calculate_spill_cost): Likewise.
6266 * ira-emit.c (modify_move_list): Likewise.
6267 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
6268 (ira_hard_reg_in_set_p): Likewise.
6269 * ira.c (setup_reg_mode_hard_regset): Likewise.
6270 (clarify_prohibited_class_mode_regs): Likewise.
6271 (check_allocation): Likewise.
6272 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6273 (lra_setup_reg_renumber): Likewise.
6274 (setup_try_hard_regno_pseudos): Likewise.
6275 (spill_for): Likewise.
6276 (assign_hard_regno): Likewise.
6277 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
6278 * lra-constraints.c (in_class_p): Likewise.
6279 (lra_constraint_offset): Likewise.
6280 (simplify_operand_subreg): Likewise.
6281 (lra_constraints): Likewise.
6282 (split_reg): Likewise.
6283 (split_if_necessary): Likewise.
6284 (invariant_p): Likewise.
6285 (inherit_in_ebb): Likewise.
6286 * lra-lives.c (process_bb_lives): Likewise.
6287 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
6288 (get_hard_regs): Likewise.
6289 (do_remat): Likewise.
6290 * lra-spills.c (assign_spill_hard_regs): Likewise.
6291 * mode-switching.c (create_pre_exit): Likewise.
6292 * postreload.c (reload_combine_recognize_pattern): Likewise.
6293 * recog.c (peep2_find_free_register): Likewise.
6294 * regcprop.c (kill_value_regno): Likewise.
6295 (set_value_regno): Likewise.
6296 (copy_value): Likewise.
6297 (maybe_mode_change): Likewise.
6298 (find_oldest_value_reg): Likewise.
6299 (copyprop_hardreg_forward_1): Likewise.
6300 * regrename.c (check_new_reg_p): Likewise.
6301 (regrename_do_replace): Likewise.
6302 * reload.c (push_reload): Likewise.
6303 (combine_reloads): Likewise.
6304 (find_dummy_reload): Likewise.
6305 (operands_match_p): Likewise.
6306 (find_reloads): Likewise.
6307 (find_equiv_reg): Likewise.
6308 (reload_adjust_reg_for_mode): Likewise.
6309 * reload1.c (count_pseudo): Likewise.
6310 (count_spilled_pseudo): Likewise.
6311 (find_reg): Likewise.
6312 (clear_reload_reg_in_use): Likewise.
6313 (free_for_value_p): Likewise.
6314 (allocate_reload_reg): Likewise.
6315 (choose_reload_regs): Likewise.
6316 (reload_adjust_reg_for_temp): Likewise.
6317 (emit_reload_insns): Likewise.
6318 (delete_output_reload): Likewise.
6319 * rtlanal.c (subreg_get_info): Likewise.
6320 * sched-deps.c (sched_analyze_reg): Likewise.
6321 * sel-sched.c (init_regs_for_mode): Likewise.
6322 (mark_unavailable_hard_regs): Likewise.
6323 (choose_best_reg_1): Likewise.
6324 (verify_target_availability): Likewise.
6325 * valtrack.c (dead_debug_insert_temp): Likewise.
6326 * var-tracking.c (track_loc_p): Likewise.
6327 (emit_note_insn_var_location): Likewise.
6328 * varasm.c (make_decl_rtl): Likewise.
6329 * reginfo.c (choose_hard_reg_mode): Likewise.
6330 (init_reg_modes_target): Refer directly to
6331 this_target_regs->x_hard_regno_nregs.
6333 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6335 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
6336 instead of hard_regno_nregs.
6338 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6340 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
6341 end_hard_regno instead of hard_regno_nregs.
6342 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
6343 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
6344 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
6345 * ira-color.c (improve_allocation): Likewise.
6346 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6347 * lra-lives.c (mark_regno_live): Likewise.
6348 (mark_regno_dead): Likewise.
6349 * lra-remat.c (operand_to_remat): Likewise.
6350 * lra.c (collect_non_operand_hard_regs): Likewise.
6351 * postreload.c (reload_combine_note_store): Likewise.
6352 (move2add_valid_value_p): Likewise.
6353 * reload.c (regno_clobbered_p): Likewise.
6355 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6357 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
6359 * config/v850/v850.c (v850_reorg): Likewise.
6360 * reload.c (refers_to_regno_for_reload_p): Likewise.
6361 (find_equiv_reg): Likewise.
6362 * reload1.c (reload_reg_reaches_end_p): Likewise.
6364 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6366 * caller-save.c (add_used_regs): Use REG_NREGS instead of
6368 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
6369 * config/arm/arm.c (output_move_neon): Likewise.
6370 (arm_attr_length_move_neon): Likewise.
6371 (neon_split_vcombine): Likewise.
6372 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
6373 (c6x_mark_reg_written): Likewise.
6374 (c6x_dwarf_register_span): Likewise.
6375 * config/i386/i386.c (ix86_save_reg): Likewise.
6376 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
6377 (rws_access_reg): Likewise.
6378 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
6379 * mode-switching.c (create_pre_exit): Likewise.
6380 * ree.c (combine_reaching_defs): Likewise.
6381 (add_removable_extension): Likewise.
6382 * regcprop.c (find_oldest_value_reg): Likewise.
6383 (copyprop_hardreg_forward_1): Likewise.
6384 * reload.c (reload_inner_reg_of_subreg): Likewise.
6385 (push_reload): Likewise.
6386 (combine_reloads): Likewise.
6387 (find_dummy_reload): Likewise.
6388 (reload_adjust_reg_for_mode): Likewise.
6389 * reload1.c (find_reload_regs): Likewise.
6390 (forget_old_reloads_1): Likewise.
6391 (reload_reg_free_for_value_p): Likewise.
6392 (reload_adjust_reg_for_temp): Likewise.
6393 (emit_reload_insns): Likewise.
6394 (delete_output_reload): Likewise.
6395 * sel-sched.c (choose_best_reg_1): Likewise.
6396 (choose_best_pseudo_reg): Likewise.
6398 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6399 Alan Hayward <alan.hayward@arm.com>
6400 David Sherwood <david.sherwood@arm.com>
6402 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
6403 * target.def (slow_unaligned_access): New hook.
6404 * targhooks.h (default_slow_unaligned_access): Declare.
6405 * targhooks.c (default_slow_unaligned_access): New function.
6406 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
6407 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
6408 * doc/tm.texi: Regenerate.
6409 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
6410 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
6411 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
6413 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
6414 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
6416 (rs6000_slow_unaligned_access): New function.
6417 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
6418 (expand_block_compare): Likewise.
6419 (expand_strn_compare): Likewise.
6420 (rs6000_rtx_costs): Likewise.
6421 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
6422 (riscv_slow_unaligned_access): Likewise.
6423 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
6424 (riscv_slow_unaligned_access_p): ...this and make static.
6425 (riscv_option_override): Update accordingly.
6426 (riscv_slow_unaligned_access): New function.
6427 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
6428 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
6429 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
6430 (rs6000_slow_unaligned_access): New function.
6431 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
6432 (rs6000_rtx_costs): Likewise.
6433 * config/rs6000/rs6000-string.c (expand_block_compare)
6434 (expand_strn_compare): Use targetm.slow_unaligned_access instead
6435 of SLOW_UNALIGNED_ACCESS.
6436 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
6437 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
6438 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
6439 of SLOW_UNALIGNED_ACCESS.
6440 * expmed.c (simple_mem_bitfield_p): Likewise.
6441 * expr.c (alignment_for_piecewise_move): Likewise.
6442 (emit_group_load_1): Likewise.
6443 (emit_group_store): Likewise.
6444 (copy_blkmode_from_reg): Likewise.
6445 (emit_push_insn): Likewise.
6446 (expand_assignment): Likewise.
6447 (store_field): Likewise.
6448 (expand_expr_real_1): Likewise.
6449 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
6450 * lra-constraints.c (simplify_operand_subreg): Likewise.
6451 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
6452 * gimple-ssa-store-merging.c: Likewise in block comment at start
6454 * tree-ssa-strlen.c: Include target.h.
6455 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
6456 of SLOW_UNALIGNED_ACCESS.
6457 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
6459 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6461 PR rtl-optimization/82185
6462 * expmed.c (emit_store_flag_int): Only test tem if it has been
6465 2017-09-12 Richard Biener <rguenther@suse.de>
6468 * match.pd ((FTYPE) N CMP CST): Fix typo.
6470 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
6472 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
6474 (mips_near_type_p): Add 'short_call' attribute as a synonym
6476 * doc/extend.texi (short_call): Document new function attribute.
6478 2017-09-12 Jakub Jelinek <jakub@redhat.com>
6481 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
6482 assertion check that in the condition.
6483 (get_atomic_generic_size): Likewise. Before testing if parameter
6484 has pointer type, if it has array type, call for C++
6485 default_conversion to perform array-to-pointer conversion.
6487 2017-09-12 Richard Biener <rguenther@suse.de>
6489 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
6490 for operations we cannot scalarize.
6492 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
6494 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
6495 vectors heap vectors. Clean up comments.
6496 Make visited_bbs a reference.
6497 (profitable_jump_thread_path): Make GC
6498 vectors heap vectors. Clean up comments.
6500 (convert_and_register_jump_thread_path): Make GC vectors heap
6502 (check_subpath_and_update_thread_path): Same. Clean up comments.
6503 Make visited_bbs a reference.
6504 (handle_phi): Abstract common code to to
6505 register_jump_thread_path_if_profitable.
6506 Rename VAR_BB to DEF_BB.
6508 Make GC vectors heap vectors.
6509 Make visited_bbs a reference.
6510 (handle_assignment): Same.
6511 (register_jump_thread_path_if_profitable): New.
6512 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
6514 Make GC vectors heap vectors. Clean up comments.
6515 Make visited_bbs a reference.
6516 (find_jump_threads_backwards): Make visited_bbs live in the stack.
6517 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
6520 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
6523 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
6524 words of E_DImode object are reachable by xtensa_uimm8x4 access.
6526 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
6528 Revert r251800 and r251799.
6530 2017-09-11 Martin Jambor <mjambor@suse.cz>
6533 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
6534 arguments in advance.
6535 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
6536 use it to find predecessor edges.
6537 (naive_outof_ssa): Collect vector of predecessors.
6539 2017-09-08 Jason Merrill <jason@redhat.com>
6541 PR c++/70029 - ICE with ref-qualifier and -flto
6542 * langhooks.h (struct lang_hooks_for_types): Add
6543 copy_lang_qualifiers.
6544 * attribs.c (build_type_attribute_qual_variant): Use it.
6545 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
6547 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
6548 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
6550 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
6553 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
6554 (*mulsi3_sp64): New instruction.
6555 (mulsi3): New expander.
6557 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
6559 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
6561 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6563 * sancov.c: Include memmodel.h.
6565 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
6568 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
6571 2017-09-07 Carl Love <cel@us.ibm.com>
6573 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
6574 the sldi instruction.
6576 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
6578 * sancov.c: Include tm_p.h.
6580 2017-09-07 Jakub Jelinek <jakub@redhat.com>
6583 * output.h (switch_to_other_text_partition): New declaration.
6584 * varasm.c (switch_to_other_text_partition): New function.
6585 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
6586 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
6587 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
6588 to the other text partition before emitting LCL label and switch back
6589 after emitting the word after it.
6591 2017-09-07 Richard Biener <rguenther@suse.de>
6593 * passes.def (pass_split_crit_edges): Remove instance before PRE.
6594 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
6595 critical edges here, after loop init.
6596 (pass_data_pre): Remove PROP_no_crit_edges flags.
6597 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
6598 for valueization of call args to avoid leaking VN_TOP.
6599 (visit_use): Assert we do not visit default defs.
6600 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
6601 Use error_mark_node to more easily detect leaking VN_TOP.
6602 All default-defs are varying, not VN_TOP. Mark them visited.
6603 (run_scc_vn): Make code match comment.
6605 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6607 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
6608 OPTION_MASK_FLOAT128_KEYWORD.
6609 (POWERPC_MASKS): Likewise.
6610 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
6611 support for the -mfloat128-type option, and make -mfloat128
6612 default on PowerPC Linux systems. Define or undefine
6613 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
6614 Define __float128 to be __ieee128 if IEEE 128-bit support is
6615 enabled, or undefine it.
6616 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
6617 Delete defining __FLOAT128_TYPE__.
6618 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
6619 -mfloat128-type option and make -mfloat128 default on PowerPC
6621 (TARGET_FLOAT128_TYPE): Likewise.
6622 (-mfloat128-type): Likewise.
6623 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6624 Delete the -mfloat128-type option and make -mfloat128 default on
6625 PowerPC Linux systems. Always use __ieee128 to be the keyword for
6626 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
6627 128-bit floating point is enabled. Change tests from using
6628 -mfloat128-type to -mfloat128.
6629 (rs6000_mangle_type): Use the correct mangling for the __float128
6630 type even if normal long double is restricted to 64-bits.
6631 (floatn_mode): Enable the _Float128 type by default on VSX Linux
6633 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
6634 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
6635 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
6637 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
6638 documentation for -mfloat128.
6640 2017-09-06 Olivier Hainque <hainque@adacore.com>
6642 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
6644 2017-09-06 Wish Wu <wishwu007@gmail.com>
6645 Jakub Jelinek <jakub@redhat.com>
6647 * asan.c (initialize_sanitizer_builtins): Add
6648 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
6649 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
6650 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
6651 BT_FN_VOID_UINT64_PTR variables.
6652 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
6653 (BT_FN_VOID_UINT16_UINT16): Likewise.
6654 (BT_FN_VOID_UINT32_UINT32): Likewise.
6655 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
6656 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
6657 (BT_FN_VOID_UINT64_PTR): Likewise.
6658 * common.opt (flag_sanitize_coverage): New variable.
6659 (fsanitize-coverage=trace-pc): Remove.
6660 (fsanitize-coverage=): Add.
6661 * flag-types.h (enum sanitize_coverage_code): New enum.
6662 * fold-const.c (fold_range_test): Disable non-short-circuit
6663 optimization if flag_sanitize_coverage.
6664 (fold_truth_andor): Likewise.
6665 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
6666 * opts.c (COVERAGE_SANITIZER_OPT): Define.
6667 (coverage_sanitizer_opts): New array.
6668 (get_closest_sanitizer_option): Add OPTS argument, handle also
6669 OPT_fsanitize_coverage_.
6670 (parse_sanitizer_options): Adjusted to also handle
6671 OPT_fsanitize_coverage_.
6672 (common_handle_option): Add OPT_fsanitize_coverage_.
6673 * sancov.c (instrument_comparison, instrument_switch): New function.
6674 (sancov_pass): Add trace-cmp support.
6675 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
6676 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
6677 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
6678 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
6679 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
6680 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
6681 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
6682 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
6683 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
6684 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
6686 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
6688 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6689 error. Only quit immediately if parsing is complete.
6690 (BEGIN): Initialize fatal_err and parse_done.
6691 (begin fpu, end fpu): Check number of arguments.
6692 (begin arch, end arch): Likewise.
6693 (begin cpu, end cpu): Likewise.
6694 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6695 (optalias): Likewise.
6697 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
6699 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6700 * config/arm/arm-isa.h: Delete. Move definitions to ...
6701 * arm-cpus.in: ... here. Use new feature and fgroup values.
6702 * config/arm/arm.c (arm_option_override): Use lower case for feature
6704 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6705 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6706 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6708 (print_isa_bits_for): New function.
6709 (gen_isa): New function.
6710 (gen_comm_data): Use print_isa_bits_for.
6711 (define feature): New keyword.
6712 (define fgroup): New keyword.
6713 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
6714 (arm-isa.h): Add rule to generate file.
6715 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6716 case for feature bit names.
6718 2017-09-06 Richard Biener <rguenther@suse.de>
6720 * tree-ssa-pre.c (NECESSARY): Remove.
6721 (create_expression_by_pieces): Do not touch pass-local flags.
6722 (insert_into_preds_of_block): Likewise.
6723 (do_pre_regular_insertion): Likewise.
6724 (eliminate_insert): Likewise.
6725 (eliminate_dom_walker::before_dom_children): Likewise.
6726 (fini_eliminate): Do not look at inserted_exprs.
6727 (mark_operand_necessary): Remove.
6728 (remove_dead_inserted_code): Replace with simple work-list
6729 algorithm based on inserted_exprs and SSA uses.
6730 (pass_pre::execute): Re-order fini_eliminate and
6731 remove_dead_inserted_code.
6733 2017-09-06 Olivier Hainque <hainque@adacore.com>
6735 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
6736 for VxWorks 7. Adjust surrounding comments.
6738 2017-09-06 Richard Biener <rguenther@suse.de>
6740 * gimple-ssa-strength-reduction.c
6741 (find_candidates_dom_walker::before_dom_children): Also allow
6744 2017-09-06 Richard Biener <rguenther@suse.de>
6746 PR tree-optimization/82108
6747 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
6748 for gap in the non-permutation SLP case.
6750 2017-09-06 Martin Jambor <mjambor@suse.cz>
6752 PR tree-optimization/82078
6753 * tree-sra.c (sort_and_splice_var_accesses): Move call to
6754 add_access_to_work_queue...
6755 (build_accesses_from_assign): ...here.
6756 (propagate_all_subaccesses): Make sure racc is the group
6757 representative, if there is one.
6759 2017-09-06 Jakub Jelinek <jakub@redhat.com>
6762 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
6765 2017-09-06 Richard Biener <rguenther@suse.de>
6767 * gimple-ssa-strength-reduction.c
6768 (find_candidates_dom_walker::before_doom_children): Use a
6769 type and not a mode check.
6771 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
6774 * config/arm/predicates.md (arm_general_adddi_operand): Create new
6776 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
6778 2017-09-05 Jeff Law <law@redhat.com>
6780 PR tree-optimization/64910
6781 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
6782 cases where we have 3 or more operands.
6784 2017-09-05 Jakub Jelinek <jakub@redhat.com>
6787 * omp-low.c (lower_omp_for): Recompute tree invariant if
6788 gimple_omp_for_initial/final is ADDR_EXPR.
6791 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
6792 into gimple val before gimplification fo the COND_EXPR.
6794 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
6796 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
6797 REGION_COPY argument.
6798 (thread_through_all_blocks): Remove unused argument to
6799 duplicate_thread_path.
6801 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6802 Alan Hayward <alan.hayward@arm.com>
6803 David Sherwood <david.sherwood@arm.com>
6805 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
6806 Take a scalar_mode rather than a machine_mode.
6807 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6808 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
6809 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6810 (aarch64_gen_adjusted_ldpstp): Likewise.
6811 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
6813 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6814 Alan Hayward <alan.hayward@arm.com>
6815 David Sherwood <david.sherwood@arm.com>
6817 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
6818 Take a scalar_int_mode instead of a machine_mode.
6819 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
6820 (aarch64_output_scalar_simd_mov_immediate): Likewise.
6821 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
6822 (aarch64_simd_attr_length_rglist): Delete.
6823 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
6824 a scalar_int_mode instead of a machine_mode.
6825 (aarch64_add_offset): Likewise.
6826 (aarch64_internal_mov_immediate): Likewise
6827 (aarch64_add_constant_internal): Likewise.
6828 (aarch64_add_constant): Likewise.
6829 (aarch64_movw_imm): Likewise.
6830 (aarch64_rtx_arith_op_extract_p): Likewise.
6831 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
6832 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
6833 Remove assert that the mode isn't a vector.
6834 (aarch64_output_scalar_simd_mov_immediate): Likewise.
6835 (aarch64_expand_mov_immediate): Update calls after above changes.
6836 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
6837 (aarch64_and_bitmask_imm): Check for scalar integer modes.
6838 (aarch64_move_imm): Likewise.
6839 (aarch64_can_const_movi_rtx_p): Likewise.
6840 (aarch64_strip_extend): Likewise.
6841 (aarch64_extr_rtx_p): Likewise.
6842 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
6843 a CONST_INT when the mode parameter is VOIDmode.
6844 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
6846 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6848 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
6849 * stor-layout.c (bitwise_mode_for_mode): Likewise.
6850 (bitwise_type_for_mode): Update accordingly.
6852 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6854 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
6855 * stor-layout.c (mode_for_size_tree): Likewise.
6856 (mode_for_array): Update accordingly.
6857 (layout_decl): Likewise.
6858 (compute_record_mode): Likewise. Only set the mode once.
6860 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6862 * target.def (get_mask_mode): Change return type to opt_mode.
6864 * doc/tm.texi: Regenerate.
6865 * targhooks.h (default_get_mask_mode): Return an opt_mode.
6866 * targhooks.c (default_get_mask_mode): Likewise.
6867 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
6868 * optabs-query.c (can_vec_mask_load_store_p): Update use of
6869 targetm.get_mask_mode.
6870 * tree.c (build_truth_vector_type): Likewise.
6872 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6874 * machmode.h (mode_for_vector): Return an opt_mode.
6875 * stor-layout.c (mode_for_vector): Likewise.
6876 (mode_for_int_vector): Update accordingly.
6877 (layout_type): Likewise.
6878 * config/i386/i386.c (emit_memmov): Likewise.
6879 (ix86_expand_set_or_movmem): Likewise.
6880 (ix86_expand_vector_init): Likewise.
6881 (ix86_get_mask_mode): Likewise.
6882 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
6884 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
6885 * expmed.c (extract_bit_field_1): Likewise.
6886 * expr.c (expand_expr_real_2): Likewise.
6887 * optabs-query.c (can_vec_perm_p): Likewise.
6888 (can_vec_mask_load_store_p): Likewise.
6889 * optabs.c (expand_vec_perm): Likewise.
6890 * targhooks.c (default_get_mask_mode): Likewise.
6891 * tree-vect-stmts.c (vectorizable_store): Likewise.
6892 (vectorizable_load): Likewise.
6893 (get_vectype_for_scalar_type_and_size): Likewise.
6895 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6897 * machmode.h (mode_for_int_vector): New function.
6898 * stor-layout.c (mode_for_int_vector): Likewise.
6899 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
6900 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
6901 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
6902 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
6903 (s390_expand_vcond): Likewise.
6905 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6907 * machmode.h (opt_machine_mode): New type.
6908 (opt_mode<T>): Allow construction from anything that can be
6910 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
6911 (mode_for_size): Return an opt_machine_mode.
6912 * stor-layout.c (mode_for_size): Likewise.
6913 (mode_for_size_tree): Update call accordingly.
6914 (bitwise_mode_for_mode): Likewise.
6915 (make_fract_type): Likewise.
6916 (make_accum_type): Likewise.
6917 * caller-save.c (replace_reg_with_saved_mem): Update call
6919 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6920 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6921 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6922 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6923 * expmed.c (extract_bit_field_1): Likewise.
6924 * reload.c (get_secondary_mem): Likewise.
6925 * varasm.c (assemble_integer): Likewise.
6926 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
6929 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6931 * machmode.h (decimal_float_mode_for_size): New function.
6932 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
6933 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
6934 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
6935 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
6936 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
6937 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
6939 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6941 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
6942 (get_builtin_sync_mode): Likewise.
6943 (expand_ifn_atomic_compare_exchange): Likewise.
6944 (expand_builtin_atomic_clear): Likewise.
6945 (expand_builtin_atomic_test_and_set): Likewise.
6946 (fold_builtin_atomic_always_lock_free): Likewise.
6947 * calls.c (compute_argument_addresses): Likewise.
6948 (emit_library_call_value_1): Likewise.
6949 (store_one_arg): Likewise.
6950 * combine.c (combine_instructions): Likewise.
6951 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
6952 * config/arm/arm.c (arm_function_value): Likewise.
6953 (aapcs_allocate_return_reg): Likewise.
6954 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
6955 * config/i386/i386.c (construct_container): Likewise.
6956 (ix86_gimplify_va_arg): Likewise.
6957 (ix86_expand_sse_cmp): Likewise.
6958 (emit_memmov): Likewise.
6959 (emit_memset): Likewise.
6960 (expand_small_movmem_or_setmem): Likewise.
6961 (ix86_expand_pextr): Likewise.
6962 (ix86_expand_pinsr): Likewise.
6963 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
6964 * config/microblaze/microblaze.c (microblaze_block_move_straight):
6966 * config/mips/mips.c (mips_function_value_1) Likewise.
6967 (mips_block_move_straight): Likewise.
6968 (mips_expand_ins_as_unaligned_store): Likewise.
6969 * config/powerpcspe/powerpcspe.c
6970 (rs6000_darwin64_record_arg_advance_flush): Likewise.
6971 (rs6000_darwin64_record_arg_flush): Likewise.
6972 * config/rs6000/rs6000.c
6973 (rs6000_darwin64_record_arg_advance_flush): Likewise.
6974 (rs6000_darwin64_record_arg_flush): Likewise.
6975 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
6976 (sparc_function_value_1): Likewise.
6977 * config/spu/spu.c (adjust_operand): Likewise.
6978 (spu_emit_branch_or_set): Likewise.
6979 (arith_immediate_p): Likewise.
6980 * emit-rtl.c (gen_lowpart_common): Likewise.
6981 * expr.c (expand_expr_real_1): Likewise.
6982 * function.c (assign_parm_setup_block): Likewise.
6983 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6984 * reload1.c (alter_reg): Likewise.
6985 * stor-layout.c (mode_for_vector): Likewise.
6986 (layout_type): Likewise.
6988 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6990 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
6991 (spu_convert_move): Likewise.
6992 * lower-subreg.c (resolve_simple_move): Likewise.
6994 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6997 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
6998 define_insn to a define_expand.
6999 (altivec_vsum2sws_direct): New define_insn.
7000 (altivec_vsumsws): Convert from a define_insn to a define_expand.
7002 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
7004 * config/arm/arm.c (arm_option_params_internal): Improve setting of
7007 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
7012 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
7013 realign stack if stack alignment needed is less than incoming
7016 2017-09-05 Marek Polacek <polacek@redhat.com>
7019 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
7022 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
7024 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
7026 2017-09-05 Richard Biener <rguenther@suse.de>
7028 PR tree-optimization/82084
7029 * fold-const.c (can_native_encode_string_p): Handle wide characters.
7031 2017-09-05 Richard Biener <rguenther@suse.de>
7033 PR tree-optimization/82102
7034 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
7036 2017-09-05 Martin Liska <mliska@suse.cz>
7038 PR tree-optimization/82032
7039 * tree-cfg.c (generate_range_test): New function.
7040 * tree-cfg.h (generate_range_test): Declared here.
7041 * tree-cfgcleanup.c (convert_single_case_switch): New function.
7042 (cleanup_control_expr_graph): Use it.
7043 * tree-switch-conversion.c (try_switch_expansion): Remove
7045 (emit_case_nodes): Use generate_range_test.
7047 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
7050 * config/i386/i386.md (*<btsc><mode>_mask): Add
7051 TARGET_USE_BT to insn constraint.
7052 (*btr<mode>_mask): Ditto.
7054 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
7056 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
7057 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
7059 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
7062 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
7063 TARGET_NEON and TARGET_IWMMXT.
7064 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
7065 TARGET_NEON and TARGET_IWMMXT.
7066 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
7068 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
7070 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
7071 (ix86_rewrite_tls_address): Ditto.
7072 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
7073 (ix86_rewrite_tls_address_1): Ditto.
7074 (ix86_rewrite_tls_address): Ditto.
7075 * config/i386/predicates.md (tls_address_pattern): New predicate.
7076 * config/i386/i386.md (TLS address splitter): New splitter.
7078 2017-09-04 Richard Biener <rguenther@suse.de>
7080 PR tree-optimization/82084
7081 * fold-const.h (can_native_encode_string_p): Declare.
7082 * fold-const.c (can_native_encode_string_p): Factor out from ...
7083 (native_encode_string): ... here.
7084 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
7085 vectorizing stores from constants we later cannot handle.
7087 2017-09-04 Marek Polacek <polacek@redhat.com>
7090 * doc/invoke.texi: Update -Wtautological-compare documentation.
7092 2017-09-04 Jeff Law <law@redhat.com>
7094 PR tree-optimization/64910
7095 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
7096 swap the first and last operand if the last is a constant.
7098 2017-09-04 Marek Polacek <polacek@redhat.com>
7101 * convert.c (do_narrow): When sanitizing signed integer overflows,
7102 bail out for signed types.
7103 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
7105 2017-09-04 Richard Biener <rguenther@suse.de>
7107 PR tree-optimization/82060
7108 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
7109 Move devirtualization after stmt folding and before EH/AB/noreturn
7110 cleanup to get the stmt refs canonicalized. Use a bool instead
7111 of gimple_modified_p since that doesn't work for NOPs. Schedule
7112 NOPs generated by folding for removal.
7114 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7115 Alan Hayward <alan.hayward@arm.com>
7116 David Sherwood <david.sherwood@arm.com>
7118 * coretypes.h (pad_direction): New enum.
7119 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
7120 (FUNCTION_ARG_PADDING): Likewise.
7121 * target.def (function_arg_padding): New hook.
7122 * targhooks.h (default_function_arg_padding): Declare.
7123 * targhooks.c (default_function_arg_padding): New function.
7124 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
7125 (TARGET_FUNCTION_ARG_PADDING): ...this.
7126 * doc/tm.texi: Regenerate.
7127 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
7128 instead of direction.
7129 (compute_argument_addresses): Likewise.
7130 (load_register_parameters): Likewise.
7131 (emit_library_call_value_1): Likewise.
7132 (store_one_arg): Use targetm.calls.function_arg_padding instead
7133 of FUNCTION_ARG_PADDING.
7134 (must_pass_in_stack_var_size_or_pad): Likewise.
7135 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
7136 (emit_group_store): Likewise.
7137 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
7138 instead of FUNCTION_ARG_PADDING.
7139 (emit_push_insn): Likewise, and propagate enum change throughout
7141 * function.h (direction): Delete.
7142 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
7144 * function.c (assign_parm_find_stack_rtl): Likewise.
7145 (assign_parm_setup_block_p): Likewise.
7146 (assign_parm_setup_block): Likewise.
7147 (gimplify_parameters): Likewise.
7148 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
7149 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
7151 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
7152 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7153 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
7154 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
7155 (aarch64_function_arg_padding): ...this new function.
7156 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
7157 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7158 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
7159 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7160 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
7161 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7162 (arm_pad_arg_upward): Replace with...
7163 (arm_function_arg_padding): ...this new function.
7164 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
7166 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
7167 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
7168 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7169 (ia64_hpux_function_arg_padding): Replace with...
7170 (ia64_function_arg_padding): ...this new function. Use pad_direction
7171 instead of direction. Check for TARGET_HPUX.
7172 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
7173 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7174 (iq2000_function_arg_padding): New function.
7175 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
7176 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
7177 (mips_function_arg_padding): ...this new function.
7178 (mips_pad_reg_upward): Update accordingly.
7179 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7180 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
7181 targetm.calls.function_arg_padding.
7182 (FUNCTION_ARG_PADDING): Delete.
7183 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7184 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
7185 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7186 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
7187 (nios2_block_reg_padding): Return pad_direction instead of direction.
7188 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
7189 instead of direction.
7190 (nios2_function_arg_padding): Likewise. Make static.
7191 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7192 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
7193 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
7194 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
7195 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7196 (pa_function_arg_padding): Make static. Return pad_direction instead
7198 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
7199 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7200 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
7201 instead of direction. Use targetm.calls.function_arg_padding.
7202 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
7203 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
7204 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
7205 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
7206 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
7208 (function_arg_padding): Rename to...
7209 (rs6000_function_arg_padding): ...this. Make static. Return
7210 pad_direction instead of direction.
7211 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
7212 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
7213 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7214 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
7215 instead of direction. Use targetm.calls.function_arg_padding.
7216 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
7217 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
7218 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
7219 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
7220 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7221 (function_arg_padding): Rename to...
7222 (rs6000_function_arg_padding): ...this. Make static. Return
7223 pad_direction instead of direction.
7224 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
7225 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
7226 * config/s390/s390.c (s390_function_arg_padding): New function.
7227 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7228 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
7229 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
7230 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7231 (function_arg_padding): Rename to...
7232 (sparc_function_arg_padding): ...this. Make static. Return
7233 pad_direction instead of direction.
7234 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
7235 * config/spu/spu.c (spu_function_arg_padding): New function.
7236 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7237 * system.h (FUNCTION_ARG_PADDING): Poison.
7239 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7240 Alan Hayward <alan.hayward@arm.com>
7241 David Sherwood <david.sherwood@arm.com>
7243 * target.def (modes_tieable_p): New hook.
7244 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
7245 (TARGET_MODES_TIEABLE_P): ...this.
7246 * doc/tm.texi.in: Regenerate.
7247 * hooks.h (hook_bool_mode_mode_true): Declare.
7248 * hooks.c (hook_bool_mode_mode_true): New function.
7249 * combine.c (subst): Use targetm.modes_tieable_p instead of
7251 * dse.c (find_shift_sequence): Likewise.
7252 * expmed.c (extract_low_bits): Likewise.
7253 * lower-subreg.c: Include target.h.
7254 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
7256 * rtlanal.c (rtx_cost): Likewise.
7257 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
7258 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
7259 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
7260 (TARGET_MODES_TIEABLE_P): Redefine.
7261 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
7262 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
7263 (TARGET_MODES_TIEABLE_P): Redefine.
7264 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
7265 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
7266 (arc_modes_tieable_p): New function.
7267 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
7268 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
7269 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
7270 (arm_modes_tieable_p): Make static.
7271 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
7272 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
7273 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
7274 (TARGET_MODES_TIEABLE_P): Redefine.
7275 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
7276 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
7277 (TARGET_MODES_TIEABLE_P): Redefine.
7278 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
7279 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
7280 (cr16_modes_tieable_p): New function.
7281 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
7282 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
7283 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
7284 (TRULY_NOOP_TRUNCATION): Update comment.
7285 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
7286 (TRULY_NOOP_TRUNCATION): Update comment.
7287 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
7288 (frv_modes_tieable_p): New function.
7289 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
7290 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
7291 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
7292 (TARGET_MODES_TIEABLE_P): Redefine.
7293 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
7294 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
7295 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
7296 (TARGET_MODES_TIEABLE_P): Redefine.
7297 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
7298 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
7299 (ia64_modes_tieable_p): New function.
7300 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
7301 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
7302 (iq2000_modes_tieable_p): New function.
7303 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
7304 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
7305 (lm32_modes_tieable_p): New function.
7306 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
7307 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
7308 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
7309 (TARGET_MODES_TIEABLE_P): Redefine.
7310 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
7311 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
7312 (m32r_modes_tieable_p): New function.
7313 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
7314 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
7315 (m68k_modes_tieable_p): New function.
7316 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
7317 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
7318 (mcore_modes_tieable_p): New function.
7319 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
7320 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
7322 (TARGET_MODES_TIEABLE_P): Redefine.
7323 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
7324 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
7325 * config/mips/mips.c (mips_modes_tieable_p): Make static.
7326 (TARGET_MODES_TIEABLE_P): Redefine.
7327 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
7328 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
7329 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
7330 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
7331 (mn10300_modes_tieable_p): ...this and make static.
7332 (TARGET_MODES_TIEABLE_P): Redefine.
7333 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
7334 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
7335 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
7336 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
7337 (msp430_modes_tieable_p): Make static.
7338 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
7339 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
7340 (TARGET_MODES_TIEABLE_P): Redefine.
7341 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
7342 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
7343 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
7344 (TARGET_MODES_TIEABLE_P): Redefine.
7345 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
7346 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
7347 * config/pa/pa.c (pa_modes_tieable_p): Make static.
7348 (TARGET_MODES_TIEABLE_P): Redefine.
7349 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
7350 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
7351 (pdp11_modes_tieable_p): New function.
7352 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
7353 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
7354 (rs6000_modes_tieable_p): New function.
7355 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
7356 * config/powerpcspe/powerpcspe.md: Update comment.
7357 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
7358 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
7359 (TARGET_MODES_TIEABLE_P): Redefine.
7360 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
7361 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
7362 (rl78_modes_tieable_p): New function.
7363 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
7364 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
7365 (rs6000_modes_tieable_p): New function.
7366 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
7367 * config/rs6000/rs6000.md: Update comment.
7368 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
7369 * config/rx/rx.c (rx_modes_tieable_p): New function.
7370 (TARGET_MODES_TIEABLE_P): Redefine.
7371 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
7372 * config/s390/s390.c (s390_modes_tieable_p): New function.
7373 (TARGET_MODES_TIEABLE_P): Redefine.
7374 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
7375 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
7376 (sh_modes_tieable_p): New function.
7377 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
7378 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
7379 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
7380 (sparc_modes_tieable_p): Make static.
7381 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
7382 * config/spu/spu.c (spu_modes_tieable_p): New function.
7383 (TARGET_MODES_TIEABLE_P): Redefine.
7384 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
7385 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
7386 (TARGET_MODES_TIEABLE_P): Redefine.
7387 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
7388 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
7389 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
7390 * config/v850/v850.c (v850_modes_tieable_p): New function.
7391 (TARGET_MODES_TIEABLE_P): Redefine.
7392 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
7393 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
7394 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
7395 (visium_modes_tieable_p): New function.
7396 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
7397 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
7398 (xtensa_modes_tieable_p): New function.
7399 * system.h (MODES_TIEABLE_P): Poison.
7401 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7402 Alan Hayward <alan.hayward@arm.com>
7403 David Sherwood <david.sherwood@arm.com>
7405 * target.def (hard_regno_mode_ok): New hook.
7406 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
7407 (TARGET_HARD_REGNO_MODE_OK): ...this.
7408 * doc/tm.texi.in: Regenerate.
7409 * hooks.h (hook_bool_uint_mode_true): Declare.
7410 * hooks.c (hook_bool_uint_mode_true): New function.
7411 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
7413 * genpreds.c (write_insn_preds_c): Add an include of target.h.
7414 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
7415 instead of HARD_REGNO_MODE_OK.
7416 * caller-save.c: Include target.h.
7417 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
7419 * combine.c (can_combine_p): Likewise.
7420 (combinable_i3pat): Likewise.
7421 (can_change_dest_mode): Likewise.
7422 * expr.c (init_expr_target): Likewise.
7423 (convert_move): Likewise.
7424 (convert_modes): Likewise.
7425 * ira.c (setup_prohibited_class_mode_regs): Likewise.
7426 (setup_prohibited_mode_move_regs): Likewise.
7427 * ira.h (target_ira): Likewise.
7428 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7429 * lra-constraints.c (process_alt_operands): Likewise.
7430 (split_reg): Likewise.
7431 * recog.c (peep2_find_free_register): Likewise.
7432 * ree.c (combine_reaching_defs): Likewise.
7433 * regcprop.c (maybe_mode_change): Likewise.
7434 * reginfo.c (init_reg_sets_1): Likewise.
7435 (choose_hard_reg_mode): Likewise.
7436 (simplifiable_subregs): Likewise.
7437 * regrename.c (check_new_reg_p): Likewise.
7438 * reload.c (find_valid_class): Likewise.
7439 (find_valid_class_1): Likewise.
7440 (reload_inner_reg_of_subreg): Likewise.
7441 (push_reload): Likewise.
7442 (combine_reloads): Likewise.
7443 (find_dummy_reload): Likewise.
7444 (find_reloads): Likewise.
7445 * reload1.c (find_reg): Likewise.
7446 (set_reload_reg): Likewise.
7447 (allocate_reload_reg): Likewise.
7448 (choose_reload_regs): Likewise.
7449 (reload_adjust_reg_for_temp): Likewise.
7450 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
7451 (simplify_subreg_regno): Likewise.
7452 * sel-sched.c (init_regs_for_mode): Likewise.
7453 * varasm.c (make_decl_rtl): Likewise.
7454 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
7455 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
7457 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
7458 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
7459 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7460 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
7461 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
7462 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7463 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
7464 (arc_mode_class): Delete.
7465 (HARD_REGNO_MODE_OK): Delete.
7466 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7467 (arc_hard_regno_mode_ok): Rename old array to...
7468 (arc_hard_regno_mode_ok_modes): ...this.
7469 (arc_conditional_register_usage): Update accordingly.
7470 (arc_mode_class): Make static.
7471 (arc_hard_regno_mode_ok): New function.
7472 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
7473 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
7474 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7475 (arm_hard_regno_mode_ok): Make static.
7476 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
7478 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
7479 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
7480 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
7482 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7483 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
7484 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
7485 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
7486 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
7487 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7488 * config/bfin/predicates.md (valid_reg_operand): Use
7489 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
7490 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
7491 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
7492 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7493 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
7494 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
7495 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7496 (cr16_hard_regno_mode_ok): Make static and return a bool.
7497 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
7498 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7499 (cris_hard_regno_mode_ok): New function.
7500 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
7501 (epiphany_mode_class): Delete.
7502 (HARD_REGNO_MODE_OK): Delete.
7503 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
7504 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7505 (hard_regno_mode_ok): Rename to...
7506 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
7507 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
7508 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
7510 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
7511 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
7512 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7513 (frv_hard_regno_mode_ok): Make static and return a bool.
7514 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
7516 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
7517 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
7518 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
7519 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
7521 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7522 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
7523 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
7524 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
7526 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7527 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
7528 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7529 (ia64_hard_regno_mode_ok): New function.
7530 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
7531 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7532 (iq2000_hard_regno_mode_ok): New function.
7533 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
7534 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7535 (lm32_hard_regno_mode_ok): New function.
7536 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
7537 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
7538 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
7539 instead of HARD_REGNO_MODE_OK.
7540 (m32c_hard_regno_ok): Rename to...
7541 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
7542 (m32c_cannot_change_mode_class): Update accordingly.
7543 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7544 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
7545 (m32r_mode_class): Delete.
7546 (HARD_REGNO_MODE_OK): Delete.
7547 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7548 (m32r_hard_regno_mode_ok): Rename to...
7549 (m32r_hard_regno_modes): ...this.
7550 (m32r_mode_class): Make static.
7551 (m32r_hard_regno_mode_ok): New function.
7552 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
7553 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
7554 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7555 (m68k_hard_regno_mode_ok): Make static.
7556 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
7557 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7558 (mcore_hard_regno_mode_ok): New function.
7559 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
7560 (HARD_REGNO_MODE_OK): Delete.
7561 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
7563 (microblaze_hard_regno_mode_ok_p): ...this and make static.
7564 (microblaze_hard_regno_mode_ok): New function.
7565 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7566 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
7567 (mips_hard_regno_mode_ok): Delete.
7568 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
7569 (mips_hard_regno_mode_ok_p): ...this and make static.
7570 (mips_hard_regno_mode_ok_p): Rename to...
7571 (mips_hard_regno_mode_ok_uncached): ...this.
7572 (mips_hard_regno_mode_ok): New function.
7573 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
7574 of HARD_REGNO_MODE_OK.
7575 (mips_option_override): Update after above name changes.
7576 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7577 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
7578 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
7579 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
7580 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
7581 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7582 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
7583 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
7584 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
7585 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7586 (msp430_hard_regno_mode_ok): Make static and return a bool.
7587 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
7588 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
7589 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
7591 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7592 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
7593 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
7594 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
7595 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
7596 (PA_HARD_REGNO_MODE_OK): ...this
7597 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
7598 (PA_HARD_REGNO_MODE_OK): ...this.
7599 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7600 (pa_hard_regno_mode_ok): New function.
7601 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
7602 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7603 (pdp11_hard_regno_mode_ok): New function.
7604 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
7605 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
7607 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
7609 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7610 (rs6000_hard_regno_mode_ok): Rename to...
7611 (rs6000_hard_regno_mode_ok_uncached): ...this.
7612 (rs6000_init_hard_regno_mode_ok): Update accordingly.
7613 (rs6000_hard_regno_mode_ok): New function.
7614 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
7615 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
7616 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
7617 (riscv_hard_regno_mode_ok): ...this and make static.
7618 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7619 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
7620 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
7621 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7622 (rl78_hard_regno_mode_ok): Make static and return bool.
7623 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
7624 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
7626 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
7627 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7628 (rs6000_hard_regno_mode_ok): Rename to...
7629 (rs6000_hard_regno_mode_ok_uncached): ...this.
7630 (rs6000_init_hard_regno_mode_ok): Update accordingly.
7631 (rs6000_hard_regno_mode_ok): New function.
7632 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
7633 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
7634 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7635 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
7636 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
7637 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
7638 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7639 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
7640 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
7641 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7642 (sh_hard_regno_mode_ok): Make static.
7643 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
7644 instead of HARD_REGNO_MODE_OK.
7645 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
7646 (sparc_mode_class): Delete.
7647 (HARD_REGNO_MODE_OK): Delete.
7648 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7649 (hard_regno_mode_classes): Make static.
7650 (sparc_mode_class): Likewise.
7651 (sparc_hard_regno_mode_ok): New function.
7652 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
7653 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
7654 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
7656 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7657 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
7658 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
7659 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
7660 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
7661 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7662 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
7663 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
7664 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7665 (visium_hard_regno_mode_ok): New function.
7666 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
7667 instead of HARD_REGNO_MODE_OK.
7668 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
7669 (HARD_REGNO_MODE_OK): Delete.
7670 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
7671 (xtensa_hard_regno_mode_ok_p): ...this and make static.
7672 (xtensa_option_override): Update accordingly.
7673 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7674 (xtensa_hard_regno_mode_ok): New function.
7675 * system.h (HARD_REGNO_MODE_OK): Poison.
7677 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7678 Alan Hayward <alan.hayward@arm.com>
7679 David Sherwood <david.sherwood@arm.com>
7681 * target.def (hard_regno_call_part_clobbered): New hook.
7682 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
7683 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
7684 * doc/tm.texi: Regenerate.
7685 * hooks.h (hook_bool_uint_mode_false): Declare.
7686 * hooks.c (hook_bool_uint_mode_false): New function.
7687 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7688 * cselib.c (cselib_process_insn): Use
7689 targetm.hard_regno_call_part_clobbered instead of
7690 HARD_REGNO_CALL_PART_CLOBBERED.
7691 * ira-conflicts.c (ira_build_conflicts): Likewise.
7692 * ira-costs.c (ira_tune_allocno_costs): Likewise.
7693 * lra-constraints.c (need_for_call_save_p): Likewise.
7694 * lra-lives.c: Include target.h.
7695 (check_pseudos_live_through_calls): Use
7696 targetm.hard_regno_call_part_clobbered instead of
7697 HARD_REGNO_CALL_PART_CLOBBERED.
7698 * regcprop.c: Include target.h.
7699 (copyprop_hardreg_forward_1): Use
7700 targetm.hard_regno_call_part_clobbered instead of
7701 HARD_REGNO_CALL_PART_CLOBBERED.
7702 * reginfo.c (choose_hard_reg_mode): Likewise.
7703 * regrename.c (check_new_reg_p): Likewise.
7704 * reload.c (find_equiv_reg): Likewise.
7705 * reload1.c (emit_reload_insns): Likewise.
7706 * sched-deps.c (deps_analyze_insn): Likewise.
7707 * sel-sched.c (init_regs_for_mode): Likewise.
7708 (mark_unavailable_hard_regs): Likewise.
7709 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
7710 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7711 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
7713 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7714 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7715 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
7717 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
7719 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7720 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7721 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
7723 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7724 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7725 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
7727 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7728 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
7730 * config/powerpcspe/powerpcspe.c
7731 (rs6000_hard_regno_call_part_clobbered): New function.
7732 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7733 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7734 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
7736 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7737 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7738 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
7740 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7741 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7742 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
7744 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7745 Alan Hayward <alan.hayward@arm.com>
7746 David Sherwood <david.sherwood@arm.com>
7748 * rtl.h (subreg_memory_offset): Declare.
7749 * emit-rtl.c (subreg_memory_offset): New function.
7750 * expmed.c (store_bit_field_1): Use it.
7751 * expr.c (undefined_operand_subword_p): Likewise.
7752 * simplify-rtx.c (simplify_subreg): Likewise.
7754 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
7756 PR rtl-optimization/57448
7759 * optabs.c (expand_atomic_load): Place compiler memory barriers if
7760 using atomic_load pattern.
7761 (expand_atomic_store): Likewise.
7763 2017-09-04 Jakub Jelinek <jakub@redhat.com>
7766 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
7767 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
7768 handling. Use replace_call_with_value with NULL instead of
7769 gsi_replace, unlink_stmt_vdef and release_defs.
7771 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
7774 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
7776 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7779 * rtl.h (emit_library_call_value_1): Declare.
7780 (emit_library_call): Replace declaration with a series of overloads.
7781 Remove the parameter count argument.
7782 (emit_library_call_value): Likewise.
7783 * calls.c (emit_library_call_value_1): Make global. Replace varargs
7784 with an "rtx_mode_t *".
7785 (emit_library_call_value): Delete.
7786 (emit_library_call): Likewise.
7787 * asan.c (asan_emit_stack_protection): Update calls accordingly.
7788 (asan_emit_allocas_unpoison): Likewise.
7789 * builtins.c (expand_builtin_powi): Likewise.
7790 (expand_asan_emit_allocas_unpoison): Likewise.
7791 * cfgexpand.c (expand_main_function): Likewise.
7792 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
7793 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
7794 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
7795 * config/arm/arm.c (arm_trampoline_init): Likewise.
7796 (arm_call_tls_get_addr): Likewise.
7797 (arm_expand_divmod_libfunc): Likewise.
7798 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
7799 (smulsi3_highpart): Likewise.
7800 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
7801 (c6x_expand_compare): Likewise.
7802 (c6x_expand_movmem): Likewise.
7803 * config/frv/frv.c (frv_trampoline_init): Likewise.
7804 * config/i386/i386.c (ix86_trampoline_init): Likewise.
7805 (ix86_expand_divmod_libfunc): Likewise.
7806 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
7807 (ia64_expand_compare): Likewise.
7808 (ia64_profile_hook): Likewise.
7809 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
7810 (nonlocal_goto): Likewise.
7811 (restore_stack_nonlocal): Likewise.
7812 * config/m32r/m32r.c (block_move_call): Likewise.
7813 (m32r_trampoline_init): Likewise.
7814 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
7815 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
7816 (m68k_call_m68k_read_tp): Likewise.
7817 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
7818 (microblaze_expand_divide): Likewise.
7819 * config/mips/mips.h (mips_args): Likewise.
7820 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
7821 (MIPS_ICACHE_SYNC): Likewise.
7822 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
7823 (nios2_trampoline_init): Likewise.
7824 * config/pa/pa.c (hppa_tls_call): Likewise.
7825 (pa_trampoline_init): Likewise.
7826 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
7827 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
7828 (expand_strn_compare): Likewise.
7829 (rs6000_generate_compare): Likewise.
7830 (rs6000_expand_float128_convert): Likewise.
7831 (output_profile_hook): Likewise.
7832 (rs6000_trampoline_init): Likewise.
7833 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
7834 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
7835 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
7836 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
7837 (rs6000_generate_compare): Likewise.
7838 (rs6000_expand_float128_convert): Likewise.
7839 (output_profile_hook): Likewise.
7840 (rs6000_trampoline_init): Likewise.
7841 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
7842 * config/sh/sh.c (sh_trampoline_init): Likewise.
7843 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
7844 (sparc_emit_float_lib_cmp): Likewise.
7845 (sparc32_initialize_trampoline): Likewise.
7846 (sparc64_initialize_trampoline): Likewise.
7847 (sparc_profile_hook): Likewise.
7848 * config/spu/spu.c (ea_load_store): Likewise.
7849 * config/spu/spu.md (floatunssidf2): Likewise.
7850 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
7851 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
7852 * config/visium/visium.c (expand_block_move_4): Likewise.
7853 (expand_block_move_2): Likewise.
7854 (expand_block_move_1): Likewise.
7855 (expand_block_set_4): Likewise.
7856 (expand_block_set_2): Likewise.
7857 (expand_block_set_1): Likewise.
7858 (visium_trampoline_init): Likewise.
7859 (visium_profile_hook): Likewise.
7860 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
7861 (xtensa_setup_frame_addresses): Likewise.
7862 (xtensa_trampoline_init): Likewise.
7863 * except.c (sjlj_emit_function_enter): Likewise.
7864 (sjlj_emit_function_exit): Likewise.
7865 * explow.c (allocate_dynamic_stack_space): Likewise.
7866 (probe_stack_range): Likewise.
7867 * expr.c (convert_mode_scalar): Likewise.
7868 * optabs.c (expand_binop): Likewise.
7869 (expand_twoval_binop_libfunc): Likewise.
7870 (expand_unop): Likewise.
7871 (prepare_cmp_insn): Likewise.
7872 (prepare_float_lib_cmp): Likewise.
7873 (expand_float): Likewise.
7874 (expand_fix): Likewise.
7875 (expand_fixed_convert): Likewise.
7876 (maybe_emit_sync_lock_test_and_set): Likewise.
7877 (expand_atomic_compare_and_swap): Likewise.
7878 (expand_mem_thread_fence): Likewise.
7879 (expand_atomic_fetch_op): Likewise.
7881 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
7883 * doc/generic.texi (OpenACC): Adjust URL.
7884 * doc/invoke.texi (C Dialect Options): Ditto.
7886 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
7888 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
7889 predicate for operand 1. Add (m,<S>) constraint.
7890 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
7891 Prevent memory operand 1 with register operand 2.
7893 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
7895 PR rtl-optimization/82024
7896 * combine.c (try_combine): If the combination result is a PARALLEL,
7897 and we only need to retain the SET in there that would be placed
7898 at I2, check that we can place that at I3 instead, before doing so.
7900 2017-09-01 Jakub Jelinek <jakub@redhat.com>
7903 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
7905 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
7906 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
7909 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
7910 Jeff Law <law@redhat.com>
7912 * varasm.c (bss_initializer_p): Do not put constants into .bss
7913 (categorize_decl_for_section): Handle bss_initializer_p returning
7914 false when DECL_INITIAL is NULL.
7916 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7919 * config/s390/s390.c (s390_can_inline_p): New function.
7921 2017-09-01 Jeff Law <law@redhat.com>
7923 PR tree-optimization/82052
7924 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
7925 Always initialize the returned slot after a hash table miss
7926 when INSERT is true.
7928 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
7930 * config/s390/s390.md (mem_signal_fence): Remove.
7931 * doc/md.texi (mem_signal_fence): Remove.
7932 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
7934 * target-insns.def (mem_signal_fence): Remove.
7936 2017-09-01 Jakub Jelinek <jakub@redhat.com>
7939 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
7942 * asan.c (create_odr_indicator): Strip name encoding from assembler
7943 name before appending it after __odr_asan_.
7945 2017-09-01 Martin Liska <mliska@suse.cz>
7947 PR tree-optimization/82059
7948 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
7949 frequency only when an edge is redirected.
7951 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7953 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
7954 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
7955 (arc_conditional_register_usage): Remove ARC600 lp_count
7957 (arc_file_start): Emit Tag_ARC_CPU_variation.
7958 (arc_can_use_doloop_p): New conditions to use ZOLs.
7959 (hwloop_fail): New function.
7960 (hwloop_optimize): Likewise.
7961 (hwloop_pattern_reg): Likewise.
7962 (arc_doloop_hooks): New struct, to be used with reorg_loops.
7963 (arc_reorg_loops): New function, calls reorg_loops.
7964 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
7965 (arc600_corereg_hazard): Remove ZOL checking, case handled by
7967 (arc_loop_hazard): Remove function, functionality moved into
7969 (arc_hazard): Remove arc_loop_hazard call.
7970 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
7971 into hwloop_optimize.
7972 (arc_label_align): Remove ZOL handling.
7973 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
7974 * config/arc/arc.md (doloop_begin): Remove pattern.
7975 (doloop_begin_i): Likewise.
7976 (doloop_end_i): Likewise.
7977 (doloop_fallback): Likewise.
7978 (doloop_fallback_m): Likewise.
7979 (doloop_end): Reimplement expand.
7980 (arc_lp): New pattern for LP instruction.
7981 (loop_end): New pattern.
7982 (loop_fail): Likewise.
7983 (decrement_and_branch_until_zero): Likewise.
7984 * config/arc/arc.opt (mlpc-width): New option.
7985 * doc/invoke.texi (mlpc-width): Document option.
7987 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7989 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
7990 (arc_ccfsm_advance): Fix checking for delay slots.
7991 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
7993 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7995 * config/arc/arc.md (movqi_insn): Add stores to save constant long
7997 (movhi_insn): Update store instruction constraint which are saving
7998 6-bit short immediates.
7999 (movsi_insn): Consider also short scaled load operations.
8000 (zero_extendhisi2_i): Use Usd constraint instead of T.
8001 (extendhisi2_i): Add q constraint.
8002 (arc_clzsi2): Add type and length attributes.
8003 (arc_ctzsi2): Likewise.
8004 * config/arc/constraints.md (Usc): Update constraint, the
8005 assembler can parse two relocations for a single instruction.
8007 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8009 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
8010 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
8012 2017-08-31 Olivier Hainque <hainque@adacore.com>
8014 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
8015 match as powerpc-wrs-vxworks*.
8017 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
8019 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
8020 register constraint for by-element operand.
8021 (aarch64_mls_elt_merge<mode>): Likewise.
8023 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
8025 * config/arc/arc.c (arc_can_follow_jump): Check for short
8028 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
8030 * config.gcc: Use g.opt for arc.
8031 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
8032 functionality moved to ...
8033 (legitimate_scaled_address_p): New function, ...here.
8034 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
8035 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
8036 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
8038 (arc_override_options): Handle G option.
8039 (arc_output_pic_addr_const): Correct function definition.
8040 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
8041 (arc_decl_anon_ns_mem_p): Delete.
8042 (arc_in_small_data_p): Overhaul this function to take into
8043 consideration the value given via G option.
8044 (arc_rewrite_small_data_1): Renamed and corrected old
8045 arc_rewrite_small_data function.
8046 (arc_rewrite_small_data): New function.
8047 (small_data_pattern): Don't use pic_offset_table_rtx.
8048 * config/arc/arc.h (CC1_SPEC): Recognize G option.
8049 * config/arc/simdext.md (movmisalignv2hi): Use
8050 prepare_move_operands function.
8052 (movmisalign*): Likewise.
8053 * doc/invoke.texi (ARC options): Document -G option.
8055 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
8057 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
8059 * config/arc/arc.c (arc_print_operand): Output scalled address for
8060 sdata whenever is possible.
8061 (arc_in_small_data_p): Allow sdata for 64bit datum when double
8062 load/stores are available.
8063 (compact_sda_memory_operand): Check for the alignment required by
8064 code density instructions.
8065 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
8067 * config/arc/constraints.md (Usd): Update constraint.
8068 (Us0): New constraint.
8069 (Usc): Update constraint.
8071 2017-08-31 Richard Biener <rguenther@suse.de>
8074 * dwarf2out.c (dwarf2out_early_global_decl): Process each
8077 2017-08-31 Tamar Christina <tamar.christina@arm.com>
8079 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
8080 Resize type_signature.
8082 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8083 Alan Hayward <alan.hayward@arm.com>
8084 David Sherwood <david.sherwood@arm.com>
8086 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
8087 subregs whose inner modes can be stored in GPRs.
8088 (aarch64_classify_index): Likewise.
8090 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8091 Alan Hayward <alan.hayward@arm.com>
8092 David Sherwood <david.sherwood@arm.com>
8094 * config/aarch64/iterators.md (V_cmp_result): Rename to...
8095 (V_INT_EQUIV): ...this.
8096 (v_cmp_result): Rename to...
8097 (v_int_equiv): ...this.
8098 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
8099 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
8100 (copysign<mode>3): Likewise.
8101 (aarch64_simd_bsl<mode>_internal): Likewise.
8102 (aarch64_simd_bsl<mode>): Likewise.
8103 (vec_cmp<mode><mode>): Likewise.
8104 (vcond<mode><mode>): Likewise.
8105 (vcond<v_cmp_mixed><mode>): Likewise.
8106 (vcondu<mode><v_cmp_mixed>): Likewise.
8107 (aarch64_cm<optab><mode>): Likewise.
8108 (aarch64_cmtst<mode>): Likewise.
8109 (aarch64_fac<optab><mode>): Likewise.
8110 (vec_perm_const<mode>): Likewise.
8111 (vcond_mask_<mode><v_cmp_result>): Rename to...
8112 (vcond_mask_<mode><v_int_equiv>): ...this.
8113 (vec_cmp<mode><v_cmp_result>): Rename to...
8114 (vec_cmp<mode><v_int_equiv>): ...this.
8116 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8117 Alan Hayward <alan.hayward@arm.com>
8118 David Sherwood <david.sherwood@arm.com>
8120 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
8122 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
8123 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
8124 (UNSPEC_LD4_DREG): New unspecs.
8125 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
8126 (aarch64_ld2<mode>_dreg_be): Replace with...
8127 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
8129 (aarch64_ld3<mode>_dreg_le)
8130 (aarch64_ld3<mode>_dreg_be): Replace with...
8131 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
8133 (aarch64_ld4<mode>_dreg_le)
8134 (aarch64_ld4<mode>_dreg_be): Replace with...
8135 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
8138 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8140 PR tree-optimization/81987
8141 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
8142 insert an initializer in a location not dominated by the stride
8145 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
8147 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
8148 on the entire header of the finally block in the fallthru case.
8150 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
8152 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
8154 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
8156 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
8157 rs6000_emit_move_from_cr and call renamed function.
8158 (rs6000_emit_prologue): Call renamed functions.
8159 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
8160 movesi_from_cr, remove volatile CRs.
8162 2017-08-30 Jon Beniston <jon@beniston.com>
8163 Richard Biener <rguenther@suse.de>
8165 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
8166 of VECTOR_MODE_P check.
8167 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
8168 element vector types.
8170 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8172 * df.h (df_read_modify_subreg_p): Remove in favor of...
8173 * rtl.h (read_modify_subreg_p): ...this new function. Take a
8174 const_rtx instead of an rtx.
8175 * cprop.c (local_cprop_find_used_regs): Update accordingly.
8176 * df-problems.c (df_word_lr_mark_ref): Likewise.
8177 * ira-lives.c (mark_pseudo_reg_live): Likewise.
8178 (mark_pseudo_reg_dead): Likewise.
8179 (mark_ref_dead): Likewise.
8180 * reginfo.c (init_subregs_of_mode): Likewise.
8181 * sched-deps.c (sched_analyze_1): Likewise.
8182 * df-scan.c (df_def_record_1): Likewise.
8183 (df_uses_record): Likewise.
8184 (df_read_modify_subreg_p): Remove in favor of...
8185 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
8186 const_rtx instead of an rtx.
8188 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8189 Alan Hayward <alan.hayward@arm.com>
8190 David Sherwood <david.sherwood@arm.com>
8192 * rtl.h (partial_subreg_p): New function.
8193 * caller-save.c (save_call_clobbered_regs): Use it.
8194 * calls.c (expand_call): Likewise.
8195 * combine.c (combinable_i3pat): Likewise.
8196 (simplify_set): Likewise.
8197 (make_extraction): Likewise.
8198 (make_compound_operation_int): Likewise.
8199 (gen_lowpart_or_truncate): Likewise.
8200 (force_to_mode): Likewise.
8201 (make_field_assignment): Likewise.
8202 (reg_truncated_to_mode): Likewise.
8203 (record_truncated_value): Likewise.
8204 (move_deaths): Likewise.
8205 * cse.c (record_jump_cond): Likewise.
8206 (cse_insn): Likewise.
8207 * cselib.c (cselib_lookup_1): Likewise.
8208 * expmed.c (extract_bit_field_using_extv): Likewise.
8209 * function.c (assign_parm_setup_reg): Likewise.
8210 * ifcvt.c (noce_convert_multiple_sets): Likewise.
8211 * ira-build.c (create_insn_allocnos): Likewise.
8212 * lra-coalesce.c (merge_pseudos): Likewise.
8213 * lra-constraints.c (match_reload): Likewise.
8214 (simplify_operand_subreg): Likewise.
8215 (curr_insn_transform): Likewise.
8216 * lra-lives.c (process_bb_lives): Likewise.
8217 * lra.c (new_insn_reg): Likewise.
8218 (lra_substitute_pseudo): Likewise.
8219 * regcprop.c (mode_change_ok): Likewise.
8220 (maybe_mode_change): Likewise.
8221 (copyprop_hardreg_forward_1): Likewise.
8222 * reload.c (push_reload): Likewise.
8223 (find_reloads): Likewise.
8224 (find_reloads_subreg_address): Likewise.
8225 * reload1.c (alter_reg): Likewise.
8226 (eliminate_regs_1): Likewise.
8227 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8229 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
8231 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
8232 back to if statements, including unpack.
8234 2017-08-30 Martin Liska <mliska@suse.cz>
8237 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
8239 (func_checker::compare_asm_inputs_outputs): ... this function.
8240 (func_checker::compare_gimple_asm): Use the function to compare
8241 also ASM constrains.
8242 * ipa-icf-gimple.h: Rename the function.
8244 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8245 Alan Hayward <alan.hayward@arm.com>
8246 David Sherwood <david.sherwood@arm.com>
8248 * coretypes.h (complex_mode): New type.
8249 * gdbhooks.py (build_pretty_printer): Handle it.
8250 * machmode.h (complex_mode): New class.
8251 (complex_mode::includes_p): New function.
8252 (is_complex_int_mode): Likewise.
8253 (is_complex_float_mode): Likewise.
8254 * genmodes.c (get_mode_class): Handle complex mode classes.
8255 * function.c (expand_function_end): Use is_complex_int_mode.
8257 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8258 Alan Hayward <alan.hayward@arm.com>
8259 David Sherwood <david.sherwood@arm.com>
8261 * coretypes.h (scalar_mode_pod): New typedef.
8262 * gdbhooks.py (build_pretty_printer): Handle it.
8263 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
8264 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
8265 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
8266 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
8269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8270 Alan Hayward <alan.hayward@arm.com>
8271 David Sherwood <david.sherwood@arm.com>
8273 * machmode.h (mode_for_vector): Take a scalar_mode instead
8275 * stor-layout.c (mode_for_vector): Likewise.
8276 * explow.c (promote_mode): Use as_a <scalar_mode>.
8277 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
8279 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8280 Alan Hayward <alan.hayward@arm.com>
8281 David Sherwood <david.sherwood@arm.com>
8283 * target.def (preferred_simd_mode): Take a scalar_mode
8284 instead of a machine_mode.
8285 * targhooks.h (default_preferred_simd_mode): Likewise.
8286 * targhooks.c (default_preferred_simd_mode): Likewise.
8287 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
8288 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
8289 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
8290 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
8291 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
8292 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
8293 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
8294 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
8296 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
8297 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
8298 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
8299 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
8300 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
8301 * doc/tm.texi: Regenerate.
8302 * optabs-query.c (can_vec_mask_load_store_p): Return false for
8305 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8306 Alan Hayward <alan.hayward@arm.com>
8307 David Sherwood <david.sherwood@arm.com>
8309 * target.def (scalar_mode_supported_p): Take a scalar_mode
8310 instead of a machine_mode.
8311 * targhooks.h (default_scalar_mode_supported_p): Likewise.
8312 * targhooks.c (default_scalar_mode_supported_p): Likewise.
8313 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
8314 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
8315 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
8316 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
8317 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
8318 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
8319 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
8320 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
8321 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
8322 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
8323 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
8324 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
8326 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
8327 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
8328 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
8329 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
8330 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
8332 * doc/tm.texi: Regenerate.
8334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8335 Alan Hayward <alan.hayward@arm.com>
8336 David Sherwood <david.sherwood@arm.com>
8338 * coretypes.h (opt_scalar_mode): New typedef.
8339 * gdbhooks.py (build_pretty_printers): Handle it.
8340 * machmode.h (mode_iterator::get_2xwider): Add overload for
8342 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
8344 * expr.c (convert_mode_scalar): Likewise.
8345 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8346 * optabs.c (expand_float): Likewise.
8347 (expand_fix): Likewise.
8348 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8350 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8351 Alan Hayward <alan.hayward@arm.com>
8352 David Sherwood <david.sherwood@arm.com>
8354 * optabs.c (expand_float): Explicitly check for scalars before
8355 using a branching expansion.
8356 (expand_fix): Likewise.
8358 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8359 Alan Hayward <alan.hayward@arm.com>
8360 David Sherwood <david.sherwood@arm.com>
8362 * expr.c (convert_mode): Split scalar handling out into...
8363 (convert_mode_scalar): ...this new function. Treat the modes
8366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8367 Alan Hayward <alan.hayward@arm.com>
8368 David Sherwood <david.sherwood@arm.com>
8370 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
8372 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
8374 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8375 Alan Hayward <alan.hayward@arm.com>
8376 David Sherwood <david.sherwood@arm.com>
8378 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
8379 rather than a machine_mode.
8380 (fixed_from_string): Likewise.
8381 (fixed_convert): Likewise.
8382 (fixed_convert_from_int): Likewise.
8383 (fixed_convert_from_real): Likewise.
8384 (real_convert_from_fixed): Likewise.
8385 * fixed-value.c (fixed_from_double_int): Likewise.
8386 (fixed_from_string): Likewise.
8387 (fixed_convert): Likewise.
8388 (fixed_convert_from_int): Likewise.
8389 (fixed_convert_from_real): Likewise.
8390 (real_convert_from_fixed): Likewise.
8391 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
8393 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8394 Alan Hayward <alan.hayward@arm.com>
8395 David Sherwood <david.sherwood@arm.com>
8397 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
8398 of separate mode class checks. Do not allow vector modes here.
8399 (immed_wide_int_const): Use as_a <scalar_mode>.
8400 * explow.c (trunc_int_for_mode): Likewise.
8401 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
8402 (wi::shwi): Likewise.
8403 (wi::min_value): Likewise.
8404 (wi::max_value): Likewise.
8405 * dwarf2out.c (loc_descriptor): Likewise.
8406 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
8409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8410 Alan Hayward <alan.hayward@arm.com>
8411 David Sherwood <david.sherwood@arm.com>
8413 * tree.h (SCALAR_TYPE_MODE): New macro.
8414 * expr.c (expand_expr_addr_expr_1): Use it.
8415 (expand_expr_real_2): Likewise.
8416 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
8417 (fold_convert_const_fixed_from_int): Likewise.
8418 (fold_convert_const_fixed_from_real): Likewise.
8419 (native_encode_fixed): Likewise
8420 (native_encode_complex): Likewise
8421 (native_encode_vector): Likewise.
8422 (native_interpret_fixed): Likewise.
8423 (native_interpret_real): Likewise.
8424 (native_interpret_complex): Likewise.
8425 (native_interpret_vector): Likewise.
8426 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
8427 (simd_clone_adjust_argument_types): Likewise.
8428 (simd_clone_init_simd_arrays): Likewise.
8429 (simd_clone_adjust): Likewise.
8430 * stor-layout.c (layout_type): Likewise.
8431 * tree.c (build_minus_one_cst): Likewise.
8432 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
8433 * tree-inline.c (estimate_move_cost): Likewise.
8434 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
8435 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
8436 (vectorizable_reduction): Likewise.
8437 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
8438 (vect_recog_mixed_size_cond_pattern): Likewise.
8439 (check_bool_pattern): Likewise.
8440 (adjust_bool_pattern): Likewise.
8441 (search_type_for_mask_1): Likewise.
8442 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
8443 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8444 (vectorizable_load): Likewise.
8445 (vectorizable_store): Likewise.
8446 * ubsan.c (ubsan_encode_value): Likewise.
8447 * varasm.c (output_constant): Likewise.
8449 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8450 Alan Hayward <alan.hayward@arm.com>
8451 David Sherwood <david.sherwood@arm.com>
8453 * coretypes.h (scalar_mode): New class.
8454 * machmode.h (scalar_mode): Likewise.
8455 (scalar_mode::includes_p): New function.
8456 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
8457 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
8458 * genmodes.c (get_mode_class): Handle remaining scalar modes.
8459 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
8460 * expmed.c (store_bit_field_1): Likewise.
8461 (extract_bit_field_1): Likewise.
8462 * expr.c (write_complex_part): Likewise.
8463 (read_complex_part): Likewise.
8464 (emit_move_complex_push): Likewise.
8465 (expand_expr_real_2): Likewise.
8466 * function.c (assign_parm_setup_reg): Likewise.
8467 (assign_parms_unsplit_complex): Likewise.
8468 * optabs.c (expand_binop): Likewise.
8469 * rtlanal.c (subreg_get_info): Likewise.
8470 * simplify-rtx.c (simplify_immed_subreg): Likewise.
8471 * varasm.c (output_constant_pool_2): Likewise.
8473 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8474 Alan Hayward <alan.hayward@arm.com>
8475 David Sherwood <david.sherwood@arm.com>
8477 * expmed.c (extract_high_half): Use scalar_int_mode and remove
8479 (expmed_mult_highpart_optab): Likewise.
8480 (expmed_mult_highpart): Likewise.
8482 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8483 Alan Hayward <alan.hayward@arm.com>
8484 David Sherwood <david.sherwood@arm.com>
8486 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
8487 instead of a machine_mode.
8488 (builtin_memset_read_str): Likewise.
8489 * builtins.c (c_readstr): Likewise.
8490 (builtin_memcpy_read_str): Likewise.
8491 (builtin_strncpy_read_str): Likewise.
8492 (builtin_memset_read_str): Likewise.
8493 (builtin_memset_gen_str): Likewise.
8494 (expand_builtin_signbit): Use scalar_int_mode for local variables.
8495 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
8496 instead of a machine_mode.
8497 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
8499 (make_extraction): Likewise.
8500 (try_widen_shift_mode): Take and return scalar_int_modes instead
8502 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
8503 a scalar_int_mode instead of a machine_mode.
8504 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
8505 (avr_addr_space_pointer_mode): Likewise.
8506 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
8507 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
8508 (msp430_unwind_word_mode): Likewise.
8509 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
8510 (spu_addr_space_pointer_mode): Likewise.
8511 (spu_addr_space_address_mode): Likewise.
8512 (spu_libgcc_cmp_return_mode): Likewise.
8513 (spu_libgcc_shift_count_mode): Likewise.
8514 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
8515 (rl78_addr_space_pointer_mode): Likewise.
8516 (fl78_unwind_word_mode): Likewise.
8517 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
8519 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
8520 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
8521 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
8522 (mips_valid_pointer_mode): Likewise.
8523 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
8524 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
8525 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
8527 (ft32_addr_space_address_mode): Likewise.
8528 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
8529 scalar_int_mode instead of a machine_mode.
8530 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
8532 (m32c_addr_space_address_mode): Likewise.
8533 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
8534 (rs6000_eh_return_filter_mode): Likewise.
8535 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
8536 (rs6000_eh_return_filter_mode): Likewise.
8537 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
8538 (s390_libgcc_shift_count_mode): Likewise.
8539 (s390_unwind_word_mode): Likewise.
8540 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
8542 * target.def (mode_rep_extended): Likewise.
8543 (valid_pointer_mode): Likewise.
8544 (addr_space.valid_pointer_mode): Likewise.
8545 (eh_return_filter_mode): Return a scalar_int_mode rather than
8547 (libgcc_cmp_return_mode): Likewise.
8548 (libgcc_shift_count_mode): Likewise.
8549 (unwind_word_mode): Likewise.
8550 (addr_space.pointer_mode): Likewise.
8551 (addr_space.address_mode): Likewise.
8552 * doc/tm.texi: Regenerate.
8553 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
8555 (do_jump): Use scalar_int_mode for local variables.
8556 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
8557 rather than a machine_mode.
8558 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
8559 (scompare_loc_descriptor_wide): Likewise.
8560 (scompare_loc_descriptor_narrow): Likewise.
8561 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
8563 * except.c (sjlj_emit_dispatch_table): Likewise.
8564 (expand_builtin_eh_copy_values): Likewise.
8565 * explow.c (convert_memory_address_addr_space_1): Likewise.
8566 Take a scalar_int_mode rather than a machine_mode.
8567 (convert_memory_address_addr_space): Take a scalar_int_mode rather
8568 than a machine_mode.
8569 (memory_address_addr_space): Use scalar_int_mode for local variables.
8570 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
8571 rather than a machine_mode.
8572 * expmed.c (mask_rtx): Likewise.
8573 (init_expmed_one_conv): Likewise.
8574 (expand_mult_highpart_adjust): Likewise.
8575 (extract_high_half): Likewise.
8576 (expmed_mult_highpart_optab): Likewise.
8577 (expmed_mult_highpart): Likewise.
8578 (expand_smod_pow2): Likewise.
8579 (expand_sdiv_pow2): Likewise.
8580 (emit_store_flag_int): Likewise.
8581 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
8583 (extract_low_bits): Likewise.
8584 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
8586 * expr.c (pieces_addr::adjust): Likewise.
8587 (can_store_by_pieces): Likewise.
8588 (store_by_pieces): Likewise.
8589 (clear_by_pieces_1): Likewise.
8590 (expand_expr_addr_expr_1): Likewise.
8591 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
8592 (expand_expr_real_1): Likewise.
8593 (try_casesi): Likewise.
8594 * final.c (shorten_branches): Likewise.
8595 * fold-const.c (fold_convert_const_int_from_fixed): Change the
8596 type of "mode" to machine_mode.
8597 * internal-fn.c (expand_arith_overflow_result_store): Take a
8598 scalar_int_mode rather than a machine_mode.
8599 (expand_mul_overflow): Use scalar_int_mode for local variables.
8600 * loop-doloop.c (doloop_modify): Likewise.
8601 (doloop_optimize): Likewise.
8602 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
8603 than a machine_mode.
8604 (expand_doubleword_shift_condmove): Likewise.
8605 (expand_doubleword_shift): Likewise.
8606 (expand_doubleword_clz): Likewise.
8607 (expand_doubleword_popcount): Likewise.
8608 (expand_doubleword_parity): Likewise.
8609 (expand_absneg_bit): Use scalar_int_mode for local variables.
8610 (prepare_float_lib_cmp): Likewise.
8611 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
8612 rather than a machine_mode.
8613 (convert_memory_address_addr_space): Likewise.
8614 (get_mode_bounds): Likewise.
8615 (get_address_mode): Return a scalar_int_mode rather than a
8617 * rtlanal.c (get_address_mode): Likewise.
8618 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
8619 than a machine_mode.
8620 * targhooks.c (default_mode_rep_extended): Likewise.
8621 (default_valid_pointer_mode): Likewise.
8622 (default_addr_space_valid_pointer_mode): Likewise.
8623 (default_eh_return_filter_mode): Return a scalar_int_mode rather
8624 than a machine_mode.
8625 (default_libgcc_cmp_return_mode): Likewise.
8626 (default_libgcc_shift_count_mode): Likewise.
8627 (default_unwind_word_mode): Likewise.
8628 (default_addr_space_pointer_mode): Likewise.
8629 (default_addr_space_address_mode): Likewise.
8630 * targhooks.h (default_eh_return_filter_mode): Likewise.
8631 (default_libgcc_cmp_return_mode): Likewise.
8632 (default_libgcc_shift_count_mode): Likewise.
8633 (default_unwind_word_mode): Likewise.
8634 (default_addr_space_pointer_mode): Likewise.
8635 (default_addr_space_address_mode): Likewise.
8636 (default_mode_rep_extended): Take a scalar_int_mode rather than
8638 (default_valid_pointer_mode): Likewise.
8639 (default_addr_space_valid_pointer_mode): Likewise.
8640 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
8642 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
8643 rather than a machine_mode.
8644 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
8645 for local variables.
8646 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8647 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
8648 than a machine_mode.
8650 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8651 Alan Hayward <alan.hayward@arm.com>
8652 David Sherwood <david.sherwood@arm.com>
8654 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
8655 the mode argument to scalar_int_mode.
8656 (do_jump_by_parts_zero_rtx): Likewise.
8657 (do_jump_by_parts_equality_rtx): Likewise.
8658 (do_jump_by_parts_greater): Take a mode argument.
8659 (do_jump_by_parts_equality): Likewise.
8660 (do_jump_1): Update calls accordingly.
8662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8663 Alan Hayward <alan.hayward@arm.com>
8664 David Sherwood <david.sherwood@arm.com>
8666 * is-a.h (safe_dyn_cast): New function.
8667 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
8668 (jump_table_for_label): Likewise.
8669 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
8670 instead of an rtx_insn *.
8671 (shorten_branches): Use dyn_cast instead of LABEL_P and
8672 JUMP_TABLE_DATA_P. Use jump_table_for_label and
8673 rtx_jump_table_data::get_data_mode.
8674 (final_scan_insn): Likewise.
8676 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8677 Alan Hayward <alan.hayward@arm.com>
8678 David Sherwood <david.sherwood@arm.com>
8680 * combine.c (try_combine): Use is_a <scalar_int_mode> when
8681 trying to combine a full-register integer set with a subreg
8684 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8685 Alan Hayward <alan.hayward@arm.com>
8686 David Sherwood <david.sherwood@arm.com>
8688 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
8689 that is always either address_mode or pointer_mode.
8691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8692 Alan Hayward <alan.hayward@arm.com>
8693 David Sherwood <david.sherwood@arm.com>
8695 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
8696 when the two are known to be equal.
8698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8699 Alan Hayward <alan.hayward@arm.com>
8700 David Sherwood <david.sherwood@arm.com>
8702 * simplify-rtx.c (simplify_const_unary_operation): Use
8703 is_a <scalar_int_mode> instead of checking for a nonzero
8704 precision. Forcibly convert op_mode to a scalar_int_mode
8705 in that case. More clearly differentiate the operand and
8706 result modes and use the former when deciding what the value
8707 of a count-bits operation should be. Use is_int_mode instead
8708 of checking for a MODE_INT. Remove redundant check for whether
8709 this mode has a zero precision.
8711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8712 Alan Hayward <alan.hayward@arm.com>
8713 David Sherwood <david.sherwood@arm.com>
8715 * optabs.c (widen_leading): Change the type of the mode argument
8716 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
8717 (widen_bswap): Likewise.
8718 (expand_parity): Likewise.
8719 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
8720 (expand_ffs): Likewise.
8721 (epand_unop): Check for scalar integer modes before calling the
8724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8725 Alan Hayward <alan.hayward@arm.com>
8726 David Sherwood <david.sherwood@arm.com>
8728 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
8730 (expand_expr_real_1): Update call accordingly.
8732 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8733 Alan Hayward <alan.hayward@arm.com>
8734 David Sherwood <david.sherwood@arm.com>
8736 * expmed.c (store_bit_field_using_insv): Add op0_mode and
8737 value_mode arguments. Use scalar_int_mode internally.
8738 (store_bit_field_1): Rename the new integer mode from imode
8739 to op0_mode and use it instead of GET_MODE (op0). Update calls
8740 to store_split_bit_field, store_bit_field_using_insv and
8741 store_fixed_bit_field.
8742 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
8743 Use scalar_int_mode internally. Use a bit count rather than a mode
8744 when calculating the largest bit size for get_best_mode.
8745 Update calls to store_split_bit_field and store_fixed_bit_field_1.
8746 (store_fixed_bit_field_1): Add mode and value_mode arguments.
8747 Remove assertion that OP0 has a scalar integer mode.
8748 (store_split_bit_field): Add op0_mode and value_mode arguments.
8749 Update calls to extract_fixed_bit_field.
8750 (extract_bit_field_using_extv): Add an op0_mode argument.
8751 Use scalar_int_mode internally.
8752 (extract_bit_field_1): Rename the new integer mode from imode to
8753 op0_mode and use it instead of GET_MODE (op0). Update calls to
8754 extract_split_bit_field, extract_bit_field_using_extv and
8755 extract_fixed_bit_field.
8756 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
8757 to extract_split_bit_field and extract_fixed_bit_field_1.
8758 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
8759 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
8761 (extract_split_bit_field): Add an op0_mode argument. Update call
8762 to extract_fixed_bit_field.
8764 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8765 Alan Hayward <alan.hayward@arm.com>
8766 David Sherwood <david.sherwood@arm.com>
8768 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
8769 * explow.c (hard_function_value): Likewise.
8770 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
8771 convert_to_mode call outside the loop.
8772 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
8773 for the mode iterator. Require the mode specified by max_pieces
8775 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
8777 (copy_blkmode_to_reg): Likewise.
8778 (set_storage_via_setmem): Likewise.
8779 * optabs.c (prepare_cmp_insn): Likewise.
8780 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8781 * stor-layout.c (finish_bitfield_representative): Likewise.
8783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8784 Alan Hayward <alan.hayward@arm.com>
8785 David Sherwood <david.sherwood@arm.com>
8787 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
8788 * expr.c (convert_move): Use them.
8789 (convert_modes): Likewise.
8790 (store_expr_with_bounds): Likewise.
8792 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8793 Alan Hayward <alan.hayward@arm.com>
8794 David Sherwood <david.sherwood@arm.com>
8796 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
8797 parameter for the mode of "x". Remove the "known_x", "known_mode"
8798 and "known_ret" arguments. Change the type of the mode argument
8800 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
8801 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
8802 (reg_num_sign_bit_copies_for_combine): Likewise.
8803 * rtlanal.c (nonzero_bits1): Likewise.
8804 (num_sign_bit_copies1): Likewise.
8805 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
8806 (reg_num_sign_bit_copies_general): Likewise.
8807 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
8808 (reg_nonzero_bits_general): Likewise.
8810 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8811 Alan Hayward <alan.hayward@arm.com>
8812 David Sherwood <david.sherwood@arm.com>
8814 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
8815 than in subroutines. Return 1 for non-integer modes.
8816 (cached_num_sign_bit_copies): Change the type of the mode parameter
8818 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
8819 classes. Handle CONST_INT_P first and then check whether X also
8820 has a scalar integer mode. Check the same thing for inner registers
8821 of a SUBREG and for values that are being extended or truncated.
8823 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8824 Alan Hayward <alan.hayward@arm.com>
8825 David Sherwood <david.sherwood@arm.com>
8827 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
8828 in subroutines. Return the mode mask for non-integer modes.
8829 (cached_nonzero_bits): Change the type of the mode parameter
8831 (nonzero_bits1): Likewise. Remove early exit for other mode
8832 classes. Handle CONST_INT_P first and then check whether X
8833 also has a scalar integer mode.
8835 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8836 Alan Hayward <alan.hayward@arm.com>
8837 David Sherwood <david.sherwood@arm.com>
8839 * expr.c (widest_int_mode_for_size): Make the comment match the code.
8840 Return a scalar_int_mode and assert that the size is greater than
8842 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
8843 (op_by_pieces_d::op_by_pieces_d): Likewise.
8844 (op_by_pieces_d::run): Likewise.
8845 (can_store_by_pieces): Likewise.
8847 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8848 Alan Hayward <alan.hayward@arm.com>
8849 David Sherwood <david.sherwood@arm.com>
8851 * combine.c (extract_left_shift): Add a mode argument and update
8853 (make_compound_operation_int): Change the type of the mode parameter
8854 to scalar_int_mode and update the call to extract_left_shift.
8856 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8857 Alan Hayward <alan.hayward@arm.com>
8858 David Sherwood <david.sherwood@arm.com>
8860 * combine.c (simplify_and_const_int): Change the type of the mode
8861 parameter to scalar_int_mode.
8862 (simplify_and_const_int_1): Likewise. Update recursive call.
8864 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8865 Alan Hayward <alan.hayward@arm.com>
8866 David Sherwood <david.sherwood@arm.com>
8868 * combine.c (simplify_compare_const): Check that the mode is a
8869 scalar_int_mode (rather than VOIDmode) before testing its
8871 (simplify_comparison): Move COMPARISON_P handling out of the
8872 loop and restrict the latter part of the loop to scalar_int_modes.
8873 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
8874 and when considering SUBREG_REGs. Use is_int_mode instead of
8875 checking GET_MODE_CLASS against MODE_INT.
8877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8878 Alan Hayward <alan.hayward@arm.com>
8879 David Sherwood <david.sherwood@arm.com>
8881 * combine.c (try_widen_shift_mode): Move check for equal modes to...
8882 (simplify_shift_const_1): ...here. Use scalar_int_mode for
8883 shift_unit_mode and for modes involved in scalar shifts.
8885 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8886 Alan Hayward <alan.hayward@arm.com>
8887 David Sherwood <david.sherwood@arm.com>
8889 * combine.c (force_int_to_mode): New function, split out from...
8890 (force_to_mode): ...here. Keep xmode up-to-date and use it
8891 instead of GET_MODE (x).
8893 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8894 Alan Hayward <alan.hayward@arm.com>
8895 David Sherwood <david.sherwood@arm.com>
8897 * optabs-query.h (extraction_insn::struct_mode): Change type to
8898 opt_scalar_int_mode and update comment.
8899 (extraction_insn::field_mode): Change type to scalar_int_mode.
8900 (extraction_insn::pos_mode): Likewise.
8901 * combine.c (make_extraction): Update accordingly.
8902 * optabs-query.c (get_traditional_extraction_insn): Likewise.
8903 (get_optab_extraction_insn): Likewise.
8904 * recog.c (simplify_while_replacing): Likewise.
8905 * expmed.c (narrow_bit_field_mem): Change the type of the mode
8906 parameter to opt_scalar_int_mode.
8908 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8909 Alan Hayward <alan.hayward@arm.com>
8910 David Sherwood <david.sherwood@arm.com>
8912 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
8913 to a scalar_int_mode instead of a machine_mode.
8914 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
8915 (get_best_mode): Return a boolean and use a pointer argument to store
8916 the selected mode. Replace the limit mode parameter with a bit limit.
8917 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
8918 for the values returned by bit_field_mode_iterator::next_mode.
8919 (store_bit_field): Update call to get_best_mode.
8920 (store_fixed_bit_field): Likewise.
8921 (extract_fixed_bit_field): Likewise.
8922 * expr.c (optimize_bitfield_assignment_op): Likewise.
8923 * fold-const.c (optimize_bit_field_compare): Likewise.
8924 (fold_truth_andor_1): Likewise.
8925 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
8926 Update for new type of m_mode.
8927 (get_best_mode): As above.
8929 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8930 Alan Hayward <alan.hayward@arm.com>
8931 David Sherwood <david.sherwood@arm.com>
8933 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
8934 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
8935 (store_bit_field): Check is_a <scalar_int_mode> before calling
8936 strict_volatile_bitfield_p.
8937 (extract_bit_field): Likewise.
8939 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8940 Alan Hayward <alan.hayward@arm.com>
8941 David Sherwood <david.sherwood@arm.com>
8943 * target.def (cstore_mode): Return a scalar_int_mode.
8944 * doc/tm.texi: Regenerate.
8945 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
8946 * targhooks.h (default_cstore_mode): Likewise.
8947 * targhooks.c (default_cstore_mode): Likewise, using a forced
8949 * expmed.c (emit_cstore): Expect the target of the cstore to be
8952 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8953 Alan Hayward <alan.hayward@arm.com>
8954 David Sherwood <david.sherwood@arm.com>
8956 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
8958 (niter_desc): Likewise mode.
8959 (iv_analyze): Add a mode parameter.
8961 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
8962 and change its type to scalar_int_mode.
8963 * loop-iv.c: Update commentary at head of file.
8964 (iv_constant): Pass the mode paraeter before the rtx it describes
8965 and change its type to scalar_int_mode. Remove VOIDmode handling.
8966 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
8967 (iv_extend): Likewise.
8968 (shorten_into_mode): Likewise.
8969 (iv_add): Use scalar_int_mode.
8970 (iv_mult): Likewise.
8971 (iv_shift): Likewise.
8972 (canonicalize_iv_subregs): Likewise.
8973 (get_biv_step_1): Pass the outer_mode parameter before the rtx
8974 it describes and change its mode to scalar_int_mode. Also change
8975 the type of the returned inner_mode to scalar_int_mode.
8976 (get_biv_step): Likewise, turning outer_mode from a pointer
8977 into a direct parameter. Update call to get_biv_step_1.
8978 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
8979 iv_constant and get_biv_step.
8980 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
8981 and change its type to scalar_int_mode. Don't initialise iv->mode
8982 to VOIDmode and remove later checks for its still being VOIDmode.
8983 Update calls to iv_analyze_op and iv_analyze_expr. Check
8984 is_a <scalar_int_mode> when changing the mode under consideration.
8985 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
8986 Update call to iv_analyze_expr.
8987 (iv_analyze_op): Add a mode parameter. Reject subregs whose
8988 inner register is not also a scalar_int_mode. Update call to
8990 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
8991 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
8992 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
8993 separate mode class checks. Update calls to iv_analyze. Remove
8994 fix-up of VOIDmodes after iv_analyze_biv.
8995 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
8996 don't have a scalar_int_mode. Update call to biv_p.
8998 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8999 Alan Hayward <alan.hayward@arm.com>
9000 David Sherwood <david.sherwood@arm.com>
9002 * cfgexpand.c (convert_debug_memory_address): Use
9003 as_a <scalar_int_mode>.
9004 * combine.c (expand_compound_operation): Likewise.
9005 (make_extraction): Likewise.
9006 (change_zero_ext): Likewise.
9007 (simplify_comparison): Likewise.
9008 * cse.c (cse_insn): Likewise.
9009 * dwarf2out.c (minmax_loc_descriptor): Likewise.
9010 (mem_loc_descriptor): Likewise.
9011 (loc_descriptor): Likewise.
9012 * expmed.c (init_expmed_one_mode): Likewise.
9013 (synth_mult): Likewise.
9014 (emit_store_flag_1): Likewise.
9015 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
9016 of a comparison with size.
9017 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
9018 (reduce_to_bit_field_precision): Likewise.
9019 * function.c (expand_function_end): Likewise.
9020 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
9021 * loop-doloop.c (doloop_modify): Likewise.
9022 * optabs.c (expand_binop): Likewise.
9023 (expand_unop): Likewise.
9024 (expand_copysign_absneg): Likewise.
9025 (prepare_cmp_insn): Likewise.
9026 (maybe_legitimize_operand): Likewise.
9027 * recog.c (const_scalar_int_operand): Likewise.
9028 * rtlanal.c (get_address_mode): Likewise.
9029 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9030 (simplify_cond_clz_ctz): Likewise.
9031 * tree-nested.c (get_nl_goto_field): Likewise.
9032 * tree.c (build_vector_type_for_mode): Likewise.
9033 * var-tracking.c (use_narrower_mode): Likewise.
9035 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9036 Alan Hayward <alan.hayward@arm.com>
9037 David Sherwood <david.sherwood@arm.com>
9039 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
9040 * builtins.c (expand_builtin_signbit): Use it.
9041 * cfgexpand.c (expand_debug_expr): Likewise.
9042 * dojump.c (do_jump): Likewise.
9043 (do_compare_and_jump): Likewise.
9044 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
9045 * expmed.c (make_tree): Likewise.
9046 * expr.c (expand_expr_real_2): Likewise.
9047 (expand_expr_real_1): Likewise.
9048 (try_casesi): Likewise.
9049 * fold-const-call.c (fold_const_call_ss): Likewise.
9050 * fold-const.c (unextend): Likewise.
9051 (extract_muldiv_1): Likewise.
9052 (fold_single_bit_test): Likewise.
9053 (native_encode_int): Likewise.
9054 (native_encode_string): Likewise.
9055 (native_interpret_int): Likewise.
9056 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
9057 * internal-fn.c (expand_addsub_overflow): Likewise.
9058 (expand_neg_overflow): Likewise.
9059 (expand_mul_overflow): Likewise.
9060 (expand_arith_overflow): Likewise.
9061 * match.pd: Likewise.
9062 * stor-layout.c (layout_type): Likewise.
9063 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
9064 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
9065 * tree-ssanames.c (get_range_info): Likewise.
9066 * tree-switch-conversion.c (array_value_type) Likewise.
9067 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
9068 (vect_recog_divmod_pattern): Likewise.
9069 (vect_recog_mixed_size_cond_pattern): Likewise.
9070 * tree-vrp.c (extract_range_basic): Likewise.
9071 (simplify_float_conversion_using_ranges): Likewise.
9072 * tree.c (int_fits_type_p): Likewise.
9073 * ubsan.c (instrument_bool_enum_load): Likewise.
9074 * varasm.c (mergeable_string_section): Likewise.
9075 (narrowing_initializer_constant_valid_p): Likewise.
9076 (output_constant): Likewise.
9078 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9079 Alan Hayward <alan.hayward@arm.com>
9080 David Sherwood <david.sherwood@arm.com>
9082 * machmode.h (NARROWEST_INT_MODE): New macro.
9083 * expr.c (alignment_for_piecewise_move): Use it instead of
9084 GET_CLASS_NARROWEST_MODE (MODE_INT).
9085 (push_block): Likewise.
9086 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
9088 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9090 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9091 Alan Hayward <alan.hayward@arm.com>
9092 David Sherwood <david.sherwood@arm.com>
9094 * postreload.c (move2add_valid_value_p): Change the type of the
9095 mode parameter to scalar_int_mode.
9096 (move2add_use_add2_insn): Add a mode parameter and use it instead
9098 (move2add_use_add3_insn): Likewise.
9099 (reload_cse_move2add): Update accordingly.
9101 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9102 Alan Hayward <alan.hayward@arm.com>
9103 David Sherwood <david.sherwood@arm.com>
9105 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
9107 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
9108 * optabs.c (expand_unop): Likewise.
9110 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9111 Alan Hayward <alan.hayward@arm.com>
9112 David Sherwood <david.sherwood@arm.com>
9114 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
9115 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
9116 (popcount_loc_descriptor): Likewise.
9117 (bswap_loc_descriptor): Likewise.
9118 (rotate_loc_descriptor): Likewise.
9119 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
9120 calling the functions above.
9122 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9123 Alan Hayward <alan.hayward@arm.com>
9124 David Sherwood <david.sherwood@arm.com>
9126 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
9128 (try_combine): Likewise.
9129 (simplify_if_then_else): Likewise.
9130 * cse.c (cse_insn): Likewise.
9131 * dwarf2out.c (mem_loc_descriptor): Likewise.
9132 * emit-rtl.c (gen_lowpart_common): Likewise.
9133 * simplify-rtx.c (simplify_truncation): Likewise.
9134 (simplify_binary_operation_1): Likewise.
9135 (simplify_const_relational_operation): Likewise.
9136 (simplify_ternary_operation): Likewise.
9137 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
9139 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9140 Alan Hayward <alan.hayward@arm.com>
9141 David Sherwood <david.sherwood@arm.com>
9143 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
9144 * reload.c (push_reload): Likewise.
9145 (find_reloads): Likewise.
9147 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9148 Alan Hayward <alan.hayward@arm.com>
9149 David Sherwood <david.sherwood@arm.com>
9151 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
9152 (make_compound_operation_int): Likewise.
9153 (change_zero_ext): Likewise.
9154 * expr.c (convert_move): Likewise.
9155 (convert_modes): Likewise.
9156 * fwprop.c (forward_propagate_subreg): Likewise.
9157 * loop-iv.c (get_biv_step_1): Likewise.
9158 * optabs.c (widen_operand): Likewise.
9159 * postreload.c (move2add_valid_value_p): Likewise.
9160 * recog.c (simplify_while_replacing): Likewise.
9161 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9162 (simplify_binary_operation_1): Likewise. Remove redundant
9163 mode equality check.
9165 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9166 Alan Hayward <alan.hayward@arm.com>
9167 David Sherwood <david.sherwood@arm.com>
9169 * combine.c (combine_simplify_rtx): Add checks for
9170 is_a <scalar_int_mode>.
9171 (simplify_if_then_else): Likewise.
9172 (make_field_assignment): Likewise.
9173 (simplify_comparison): Likewise.
9174 * ifcvt.c (noce_try_bitop): Likewise.
9175 * loop-invariant.c (canonicalize_address_mult): Likewise.
9176 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9178 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9179 Alan Hayward <alan.hayward@arm.com>
9180 David Sherwood <david.sherwood@arm.com>
9182 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
9183 is_a <scalar_int_mode> instead of != BLKmode.
9185 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9186 Alan Hayward <alan.hayward@arm.com>
9187 David Sherwood <david.sherwood@arm.com>
9189 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
9190 instead of != VOIDmode.
9191 * combine.c (if_then_else_cond): Likewise.
9192 (change_zero_ext): Likewise.
9193 * dwarf2out.c (mem_loc_descriptor): Likewise.
9194 (loc_descriptor): Likewise.
9195 * rtlanal.c (canonicalize_condition): Likewise.
9196 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
9198 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9199 Alan Hayward <alan.hayward@arm.com>
9200 David Sherwood <david.sherwood@arm.com>
9202 * simplify-rtx.c (simplify_binary_operation_1): Use
9203 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
9205 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9206 Alan Hayward <alan.hayward@arm.com>
9207 David Sherwood <david.sherwood@arm.com>
9209 * wide-int.h (int_traits<unsigned char>) New class.
9210 (int_traits<unsigned short>) Likewise.
9211 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
9212 Use GET_MODE_UNIT_PRECISION and remove redundant test for
9214 * combine.c (set_nonzero_bits_and_sign_copies): Use
9215 is_a <scalar_int_mode>.
9216 (find_split_point): Likewise.
9217 (combine_simplify_rtx): Likewise.
9218 (simplify_logical): Likewise.
9219 (expand_compound_operation): Likewise.
9220 (expand_field_assignment): Likewise.
9221 (make_compound_operation): Likewise.
9222 (extended_count): Likewise.
9223 (change_zero_ext): Likewise.
9224 (simplify_comparison): Likewise.
9225 * dwarf2out.c (scompare_loc_descriptor): Likewise.
9226 (ucompare_loc_descriptor): Likewise.
9227 (minmax_loc_descriptor): Likewise.
9228 (mem_loc_descriptor): Likewise.
9229 (loc_descriptor): Likewise.
9230 * expmed.c (init_expmed_one_mode): Likewise.
9231 * lra-constraints.c (lra_constraint_offset): Likewise.
9232 * optabs.c (prepare_libcall_arg): Likewise.
9233 * postreload.c (move2add_note_store): Likewise.
9234 * reload.c (operands_match_p): Likewise.
9235 * rtl.h (load_extend_op): Likewise.
9236 * rtlhooks.c (gen_lowpart_general): Likewise.
9237 * simplify-rtx.c (simplify_truncation): Likewise.
9238 (simplify_unary_operation_1): Likewise.
9239 (simplify_binary_operation_1): Likewise.
9240 (simplify_const_binary_operation): Likewise.
9241 (simplify_const_relational_operation): Likewise.
9242 (simplify_subreg): Likewise.
9243 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9244 * var-tracking.c (adjust_mems): Likewise.
9245 (prepare_call_arguments): Likewise.
9247 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9248 Alan Hayward <alan.hayward@arm.com>
9249 David Sherwood <david.sherwood@arm.com>
9251 * machmode.h (is_int_mode): New fuction.
9252 * combine.c (find_split_point): Use it.
9253 (combine_simplify_rtx): Likewise.
9254 (simplify_if_then_else): Likewise.
9255 (simplify_set): Likewise.
9256 (simplify_shift_const_1): Likewise.
9257 (simplify_comparison): Likewise.
9258 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
9259 * cse.c (notreg_cost): Likewise.
9260 (cse_insn): Likewise.
9261 * cselib.c (cselib_lookup_1): Likewise.
9262 * dojump.c (do_jump_1): Likewise.
9263 (do_compare_rtx_and_jump): Likewise.
9264 * dse.c (get_call_args): Likewise.
9265 * dwarf2out.c (rtl_for_decl_init): Likewise.
9266 (native_encode_initializer): Likewise.
9267 * expmed.c (emit_store_flag_1): Likewise.
9268 (emit_store_flag): Likewise.
9269 * expr.c (convert_modes): Likewise.
9270 (store_field): Likewise.
9271 (expand_expr_real_1): Likewise.
9272 * fold-const.c (fold_read_from_constant_string): Likewise.
9273 * gimple-ssa-sprintf.c (get_format_string): Likewise.
9274 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
9275 * optabs.c (expand_binop): Likewise.
9276 (expand_unop): Likewise.
9277 (expand_abs_nojump): Likewise.
9278 (expand_one_cmpl_abs_nojump): Likewise.
9279 * simplify-rtx.c (mode_signbit_p): Likewise.
9280 (val_signbit_p): Likewise.
9281 (val_signbit_known_set_p): Likewise.
9282 (val_signbit_known_clear_p): Likewise.
9283 (simplify_relational_operation_1): Likewise.
9284 * tree.c (vector_type_mode): Likewise.
9286 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9287 Alan Hayward <alan.hayward@arm.com>
9288 David Sherwood <david.sherwood@arm.com>
9290 * machmode.h (smallest_mode_for_size): Fix formatting.
9291 (smallest_int_mode_for_size): New function.
9292 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
9293 instead of smallest_mode_for_size.
9294 * combine.c (make_extraction): Likewise.
9295 * config/arc/arc.c (arc_expand_movmem): Likewise.
9296 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
9297 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
9298 * config/s390/s390.c (s390_expand_insv): Likewise.
9299 * config/sparc/sparc.c (assign_int_registers): Likewise.
9300 * config/spu/spu.c (spu_function_value): Likewise.
9301 (spu_function_arg): Likewise.
9302 * coverage.c (get_gcov_type): Likewise.
9303 (get_gcov_unsigned_t): Likewise.
9304 * dse.c (find_shift_sequence): Likewise.
9305 * expmed.c (store_bit_field_1): Likewise.
9306 * expr.c (convert_move): Likewise.
9307 (store_field): Likewise.
9308 * internal-fn.c (expand_arith_overflow): Likewise.
9309 * optabs-query.c (get_best_extraction_insn): Likewise.
9310 * optabs.c (expand_twoval_binop_libfunc): Likewise.
9311 * stor-layout.c (layout_type): Likewise.
9312 (initialize_sizetypes): Likewise.
9313 * targhooks.c (default_get_mask_mode): Likewise.
9314 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
9316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9317 Alan Hayward <alan.hayward@arm.com>
9318 David Sherwood <david.sherwood@arm.com>
9320 * machmode.h (opt_mode::else_blk): New function.
9321 (int_mode_for_mode): Declare.
9322 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
9323 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
9325 * cfgexpand.c (expand_debug_expr): Likewise.
9326 * combine.c (gen_lowpart_or_truncate): Likewise.
9327 (gen_lowpart_for_combine): Likewise.
9328 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
9329 * config/avr/avr.c (avr_to_int_mode): Likewise.
9330 (avr_out_plus_1): Likewise.
9331 (avr_out_plus): Likewise.
9332 (avr_out_round): Likewise.
9333 * config/i386/i386.c (ix86_split_to_parts): Likewise.
9334 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
9335 (s390_expand_vcond): Likewise.
9336 * config/spu/spu.c (spu_split_immediate): Likewise.
9337 (spu_expand_mov): Likewise.
9338 * dse.c (get_stored_val): Likewise.
9339 * expmed.c (store_bit_field_1): Likewise.
9340 (convert_extracted_bit_field): Use int_mode_for_mode instead of
9342 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
9343 (extract_low_bits): Likewise.
9344 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
9345 handling rather than repeating the check.
9346 (emit_group_store): Likewise.
9347 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
9348 * optabs.c (expand_absneg_bit): Likewise.
9349 (expand_copysign_absneg): Likewise.
9350 (expand_copysign_bit): Likewise.
9351 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
9352 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
9353 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
9354 * var-tracking.c (prepare_call_arguments): Likewise.
9355 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
9356 int_mode_for_mode instead of mode_for_size.
9357 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
9359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9360 Alan Hayward <alan.hayward@arm.com>
9361 David Sherwood <david.sherwood@arm.com>
9363 * machmode.h (int_mode_for_size): New function.
9364 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
9365 instead of mode_for_size.
9366 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
9368 * combine.c (expand_field_assignment): Use int_mode_for_size
9369 instead of mode_for_size.
9370 (make_extraction): Likewise.
9371 (simplify_shift_const_1): Likewise.
9372 (simplify_comparison): Likewise.
9373 * dojump.c (do_jump): Likewise.
9374 * dwarf2out.c (mem_loc_descriptor): Likewise.
9375 * emit-rtl.c (init_derived_machine_modes): Likewise.
9376 * expmed.c (flip_storage_order): Likewise.
9377 (convert_extracted_bit_field): Likewise.
9378 * expr.c (copy_blkmode_from_reg): Likewise.
9379 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
9380 * internal-fn.c (expand_mul_overflow): Likewise.
9381 * lower-subreg.c (simple_move): Likewise.
9382 * optabs-libfuncs.c (init_optabs): Likewise.
9383 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9384 * tree.c (vector_type_mode): Likewise.
9385 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
9386 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
9387 * tree-vect-generic.c (expand_vector_parallel): Likewise.
9388 * tree-vect-stmts.c (vectorizable_load): Likewise.
9389 (vectorizable_store): Likewise.
9391 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9392 Alan Hayward <alan.hayward@arm.com>
9393 David Sherwood <david.sherwood@arm.com>
9395 * coretypes.h (pod_mode): New type.
9396 (scalar_int_mode_pod): New typedef.
9397 * machmode.h (pod_mode): New class.
9398 (int_n_data_t::m): Change type to scalar_int_mode_pod.
9399 * genmodes.c (emit_mode_int_n): Update accordingly.
9400 * lower-subreg.h (target_lower_subreg): Change type to
9401 scalar_int_mode_pod.
9402 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
9403 scalar_int_mode_pod.
9405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9406 Alan Hayward <alan.hayward@arm.com>
9407 David Sherwood <david.sherwood@arm.com>
9409 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
9410 machine_mode to scalar_int_mode.
9411 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
9412 (rs6000_option_override_internal): Remove cast to int.
9413 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
9414 machine_mode to scalar_int_mode.
9415 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
9416 (rs6000_option_override_internal): Remove cast to int.
9417 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
9418 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
9420 * config/s390/s390.c (s390_expand_builtin): Likewise.
9421 * coretypes.h (scalar_int_mode): New type.
9422 (opt_scalar_int_mode): New typedef.
9423 * machmode.h (scalar_int_mode): New class.
9424 (scalar_int_mode::includes_p): New function.
9425 (byte_mode): Change type to scalar_int_mode.
9426 (word_mode): Likewise.
9427 (ptr_mode): Likewise.
9428 * emit-rtl.c (byte_mode): Likewise.
9429 (word_mode): Likewise.
9430 (ptr_mode): Likewise.
9431 (init_derived_machine_modes): Update accordingly.
9432 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
9433 and MODE_PARTIAL_INT.
9434 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
9435 opt_scalar_int_mode.
9437 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9438 Alan Hayward <alan.hayward@arm.com>
9439 David Sherwood <david.sherwood@arm.com>
9441 * target.def (libgcc_floating_mode_supported_p): Take a
9443 * doc/tm.texi: Regenerate.
9444 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
9446 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
9447 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
9450 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9451 Alan Hayward <alan.hayward@arm.com>
9452 David Sherwood <david.sherwood@arm.com>
9454 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
9455 * doc/tm.texi: Regenerate.
9456 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
9457 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
9458 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
9459 * targhooks.h (default_floatn_mode): Likewise.
9460 * targhooks.c (default_floatn_mode): Likewise.
9461 * tree.c (build_common_tree_nodes): Update accordingly.
9463 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9464 Alan Hayward <alan.hayward@arm.com>
9465 David Sherwood <david.sherwood@arm.com>
9467 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
9468 (mode_iterator::iterate_p): Likewise.
9469 (mode_iterator::get_wider): Likewise.
9470 * expr.c (init_expr_target): Use opt_scalar_float_mode.
9472 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9473 Alan Hayward <alan.hayward@arm.com>
9474 David Sherwood <david.sherwood@arm.com>
9476 * coretypes.h (opt_scalar_float_mode): New typedef.
9477 * machmode.h (float_mode_for_size): New function.
9478 * emit-rtl.c (double_mode): Delete.
9479 (init_emit_once): Use float_mode_for_size.
9480 * stor-layout.c (layout_type): Likewise.
9481 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
9483 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9484 Alan Hayward <alan.hayward@arm.com>
9485 David Sherwood <david.sherwood@arm.com>
9487 * output.h (assemble_real): Take a scalar_float_mode.
9488 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
9489 * config/arm/arm.md (consttable_4): Likewise.
9490 (consttable_8): Likewise.
9491 (consttable_16): Likewise.
9492 * config/mips/mips.md (consttable_float): Likewise.
9493 * config/s390/s390.c (s390_output_pool_entry): Likewise.
9494 * varasm.c (assemble_real): Take a scalar_float_mode.
9495 (output_constant_pool_2): Update accordingly.
9496 (output_constant): Likewise.
9498 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9499 Alan Hayward <alan.hayward@arm.com>
9500 David Sherwood <david.sherwood@arm.com>
9502 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
9503 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
9504 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
9505 (native_encode_real): Likewise.
9506 (native_interpret_real): Likewise.
9507 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
9508 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9510 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9511 Alan Hayward <alan.hayward@arm.com>
9512 David Sherwood <david.sherwood@arm.com>
9514 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
9515 <scalar_float_mode>. Simplify.
9516 (gen_extend_conv_libfunc): Likewise.
9518 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9519 Alan Hayward <alan.hayward@arm.com>
9520 David Sherwood <david.sherwood@arm.com>
9522 * coretypes.h (scalar_float_mode): New type.
9523 * machmode.h (mode_traits::from_int): Use machine_mode if
9524 USE_ENUM_MODES is defined.
9525 (is_a): New function.
9527 (dyn_cast): Likewise.
9528 (scalar_float_mode): New class.
9529 (scalar_float_mode::includes_p): New function.
9530 (is_float_mode): Likewise.
9531 * gdbhooks.py (MachineModePrinter): New class.
9532 (build_pretty_printer): Use it for scalar_float_mode.
9533 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
9534 (format_helper::format_helper): Turn into a template.
9535 * genmodes.c (get_mode_class): New function.
9536 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
9537 or machine_mode if none.
9538 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
9539 as_a <scalar_float_mode>.
9540 * dwarf2out.c (mem_loc_descriptor): Likewise.
9541 (insert_float): Likewise.
9542 (add_const_value_attribute): Likewise.
9543 * simplify-rtx.c (simplify_immed_subreg): Likewise.
9544 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
9545 (expand_unop): Update accordingly.
9546 (expand_abs_nojump): Likewise.
9547 (expand_copysign_absneg): Take a scalar_float_mode.
9548 (expand_copysign_bit): Likewise.
9549 (expand_copysign): Update accordingly.
9551 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9552 Alan Hayward <alan.hayward@arm.com>
9553 David Sherwood <david.sherwood@arm.com>
9555 * coretypes.h (opt_mode): New class.
9556 * machmode.h (opt_mode): Likewise.
9557 (opt_mode::else_void): New function.
9558 (opt_mode::require): Likewise.
9559 (opt_mode::exists): Likewise.
9560 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
9561 (GET_MODE_2XWIDER_MODE): Likewise.
9562 (mode_iterator::get_wider): Update accordingly.
9563 (mode_iterator::get_2xwider): Likewise.
9564 (mode_iterator::get_known_wider): Likewise, turning into a template.
9565 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
9566 forcing a wider mode to exist.
9567 * config/cr16/cr16.h (LONG_REG_P): Likewise.
9568 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9569 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
9570 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
9571 * lower-subreg.c (init_lower_subreg): Likewise.
9572 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
9573 on the final iteration.
9574 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
9575 a wider mode exists before asking for a move pattern.
9576 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
9577 forcing a wider mode to exist.
9578 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
9579 returning false if no such mode exists.
9580 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
9581 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
9582 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
9583 Avoid checking for a MODE_INT if we already know the mode is not a
9585 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
9586 forcing a wider mode to exist.
9587 (expmed_mult_highpart_optab): Likewise.
9588 (expmed_mult_highpart): Likewise.
9589 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
9591 * lto-streamer-in.c (lto_input_mode_table): Likewise.
9592 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
9593 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9594 * internal-fn.c (expand_mul_overflow): Update use of
9595 GET_MODE_2XWIDER_MODE.
9596 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9597 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
9598 GET_MODE_WIDER_MODE.
9599 (convert_plusminus_to_widen): Likewise.
9600 * tree-switch-conversion.c (array_value_type): Likewise.
9601 * var-tracking.c (emit_note_insn_var_location): Likewise.
9602 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9603 Return false inside rather than outside the loop if no wider mode
9605 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
9606 and GET_MODE_2XWIDER_MODE
9607 (can_compare_p): Use else_void.
9608 * gdbhooks.py (OptMachineModePrinter): New class.
9609 (build_pretty_printer): Use it for opt_mode.
9611 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9612 Alan Hayward <alan.hayward@arm.com>
9613 David Sherwood <david.sherwood@arm.com>
9615 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
9616 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
9618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9619 Alan Hayward <alan.hayward@arm.com>
9620 David Sherwood <david.sherwood@arm.com>
9622 * machmode.h (mode_traits): New structure.
9623 (get_narrowest_mode): New function.
9624 (mode_iterator::start): Likewise.
9625 (mode_iterator::iterate_p): Likewise.
9626 (mode_iterator::get_wider): Likewise.
9627 (mode_iterator::get_known_wider): Likewise.
9628 (mode_iterator::get_2xwider): Likewise.
9629 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
9630 (FOR_EACH_MODE): Likewise.
9631 (FOR_EACH_MODE_FROM): Likewise.
9632 (FOR_EACH_MODE_UNTIL): Likewise.
9633 (FOR_EACH_WIDER_MODE): Likewise.
9634 (FOR_EACH_2XWIDER_MODE): Likewise.
9635 * builtins.c (expand_builtin_strlen): Use new mode iterators.
9636 * combine.c (simplify_comparison): Likewise
9637 * config/i386/i386.c (type_natural_mode): Likewise.
9638 * cse.c (cse_insn): Likewise.
9639 * dse.c (find_shift_sequence): Likewise.
9640 * emit-rtl.c (init_derived_machine_modes): Likewise.
9641 (init_emit_once): Likewise.
9642 * explow.c (hard_function_value): Likewise.
9643 * expmed.c (extract_fixed_bit_field_1): Likewise.
9644 (extract_bit_field_1): Likewise.
9645 (expand_divmod): Likewise.
9646 (emit_store_flag_1): Likewise.
9647 * expr.c (init_expr_target): Likewise.
9648 (convert_move): Likewise.
9649 (alignment_for_piecewise_move): Likewise.
9650 (widest_int_mode_for_size): Likewise.
9651 (emit_block_move_via_movmem): Likewise.
9652 (copy_blkmode_to_reg): Likewise.
9653 (set_storage_via_setmem): Likewise.
9654 (compress_float_constant): Likewise.
9655 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9656 * optabs-query.c (get_best_extraction_insn): Likewise.
9657 * optabs.c (expand_binop): Likewise.
9658 (expand_twoval_unop): Likewise.
9659 (expand_twoval_binop): Likewise.
9660 (widen_leading): Likewise.
9661 (widen_bswap): Likewise.
9662 (expand_parity): Likewise.
9663 (expand_unop): Likewise.
9664 (prepare_cmp_insn): Likewise.
9665 (prepare_float_lib_cmp): Likewise.
9666 (expand_float): Likewise.
9667 (expand_fix): Likewise.
9668 (expand_sfix_optab): Likewise.
9669 * postreload.c (move2add_use_add2_insn): Likewise.
9670 * reg-stack.c (reg_to_stack): Likewise.
9671 * reginfo.c (choose_hard_reg_mode): Likewise.
9672 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9673 * stor-layout.c (mode_for_size): Likewise.
9674 (smallest_mode_for_size): Likewise.
9675 (mode_for_vector): Likewise.
9676 (finish_bitfield_representative): Likewise.
9677 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
9678 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
9679 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9680 * var-tracking.c (prepare_call_arguments): Likewise.
9682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9683 Alan Hayward <alan.hayward@arm.com>
9684 David Sherwood <david.sherwood@arm.com>
9686 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
9687 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
9688 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
9689 * machmode.h (mode_size): Move earlier in file.
9690 (mode_precision): Likewise.
9691 (mode_inner): Likewise.
9692 (mode_nunits): Likewise.
9693 (mode_unit_size): Likewise.
9694 (unit_unit_precision): Likewise.
9695 (mode_wider): Likewise.
9696 (mode_2xwider): Likewise.
9697 (machine_mode): New class.
9698 (mode_to_bytes): New function.
9699 (mode_to_bits): Likewise.
9700 (mode_to_precision): Likewise.
9701 (mode_to_inner): Likewise.
9702 (mode_to_unit_size): Likewise.
9703 (mode_to_unit_precision): Likewise.
9704 (mode_to_nunits): Likewise.
9705 (GET_MODE_SIZE): Use mode_to_bytes.
9706 (GET_MODE_BITSIZE): Use mode_to_bits.
9707 (GET_MODE_PRECISION): Use mode_to_precision.
9708 (GET_MODE_INNER): Use mode_to_inner.
9709 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
9710 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
9711 (GET_MODE_NUNITS): Use mode_to_nunits.
9712 * system.h (ALWAYS_INLINE): New macro.
9713 * config/powerpcspe/powerpcspe-c.c
9714 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
9715 int for arg1_mode and arg2_mode.
9717 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9718 Alan Hayward <alan.hayward@arm.com>
9719 David Sherwood <david.sherwood@arm.com>
9721 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
9722 Prefix mode names with E_ in case statements.
9723 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9724 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
9725 (aarch64_split_simd_move): Likewise.
9726 (aarch64_gen_storewb_pair): Likewise.
9727 (aarch64_gen_loadwb_pair): Likewise.
9728 (aarch64_gen_store_pair): Likewise.
9729 (aarch64_gen_load_pair): Likewise.
9730 (aarch64_get_condition_code_1): Likewise.
9731 (aarch64_constant_pool_reload_icode): Likewise.
9732 (get_rsqrte_type): Likewise.
9733 (get_rsqrts_type): Likewise.
9734 (get_recpe_type): Likewise.
9735 (get_recps_type): Likewise.
9736 (aarch64_gimplify_va_arg_expr): Likewise.
9737 (aarch64_simd_container_mode): Likewise.
9738 (aarch64_emit_load_exclusive): Likewise.
9739 (aarch64_emit_store_exclusive): Likewise.
9740 (aarch64_expand_compare_and_swap): Likewise.
9741 (aarch64_gen_atomic_cas): Likewise.
9742 (aarch64_emit_bic): Likewise.
9743 (aarch64_emit_atomic_swap): Likewise.
9744 (aarch64_emit_atomic_load_op): Likewise.
9745 (aarch64_evpc_trn): Likewise.
9746 (aarch64_evpc_uzp): Likewise.
9747 (aarch64_evpc_zip): Likewise.
9748 (aarch64_evpc_ext): Likewise.
9749 (aarch64_evpc_rev): Likewise.
9750 (aarch64_evpc_dup): Likewise.
9751 (aarch64_gen_ccmp_first): Likewise.
9752 (aarch64_gen_ccmp_next): Likewise.
9753 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
9754 (alpha_emit_xfloating_libcall): Likewise.
9755 (emit_insxl): Likewise.
9756 (alpha_arg_type): Likewise.
9757 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
9758 (arc_preferred_simd_mode): Likewise.
9759 (arc_secondary_reload): Likewise.
9760 (get_arc_condition_code): Likewise.
9761 (arc_print_operand): Likewise.
9762 (arc_legitimate_constant_p): Likewise.
9763 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9764 * config/arc/arc.md (casesi_load): Likewise.
9765 (casesi_compact_jump): Likewise.
9766 * config/arc/predicates.md (proper_comparison_operator): Likewise.
9767 (cc_use_register): Likewise.
9768 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9769 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
9770 (arm_init_iwmmxt_builtins): Likewise.
9771 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
9772 (neon_expand_vector_init): Likewise.
9773 (arm_attr_length_move_neon): Likewise.
9774 (maybe_get_arm_condition_code): Likewise.
9775 (arm_emit_vector_const): Likewise.
9776 (arm_preferred_simd_mode): Likewise.
9777 (arm_output_iwmmxt_tinsr): Likewise.
9778 (thumb1_output_casesi): Likewise.
9779 (thumb2_output_casesi): Likewise.
9780 (arm_emit_load_exclusive): Likewise.
9781 (arm_emit_store_exclusive): Likewise.
9782 (arm_expand_compare_and_swap): Likewise.
9783 (arm_evpc_neon_vuzp): Likewise.
9784 (arm_evpc_neon_vzip): Likewise.
9785 (arm_evpc_neon_vrev): Likewise.
9786 (arm_evpc_neon_vtrn): Likewise.
9787 (arm_evpc_neon_vext): Likewise.
9788 (arm_validize_comparison): Likewise.
9789 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
9790 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
9791 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
9792 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
9793 (c6x_preferred_simd_mode): Likewise.
9794 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
9795 (epiphany_rtx_costs): Likewise.
9796 * config/epiphany/predicates.md (proper_comparison_operator):
9798 * config/frv/frv.c (condexec_memory_operand): Likewise.
9799 (frv_emit_move): Likewise.
9800 (output_move_single): Likewise.
9801 (output_condmove_single): Likewise.
9802 (frv_hard_regno_mode_ok): Likewise.
9803 (frv_matching_accg_mode): Likewise.
9804 * config/h8300/h8300.c (split_adds_subs): Likewise.
9805 (h8300_rtx_costs): Likewise.
9806 (h8300_print_operand): Likewise.
9807 (compute_mov_length): Likewise.
9808 (output_logical_op): Likewise.
9809 (compute_logical_op_length): Likewise.
9810 (compute_logical_op_cc): Likewise.
9811 (h8300_shift_needs_scratch_p): Likewise.
9812 (output_a_shift): Likewise.
9813 (compute_a_shift_length): Likewise.
9814 (compute_a_shift_cc): Likewise.
9815 (expand_a_rotate): Likewise.
9816 (output_a_rotate): Likewise.
9817 * config/i386/i386.c (classify_argument): Likewise.
9818 (function_arg_advance_32): Likewise.
9819 (function_arg_32): Likewise.
9820 (function_arg_64): Likewise.
9821 (function_value_64): Likewise.
9822 (ix86_gimplify_va_arg): Likewise.
9823 (ix86_legitimate_constant_p): Likewise.
9824 (put_condition_code): Likewise.
9825 (split_double_mode): Likewise.
9826 (ix86_avx256_split_vector_move_misalign): Likewise.
9827 (ix86_expand_vector_logical_operator): Likewise.
9828 (ix86_split_idivmod): Likewise.
9829 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
9830 (ix86_build_const_vector): Likewise.
9831 (ix86_build_signbit_mask): Likewise.
9832 (ix86_match_ccmode): Likewise.
9833 (ix86_cc_modes_compatible): Likewise.
9834 (ix86_expand_branch): Likewise.
9835 (ix86_expand_sse_cmp): Likewise.
9836 (ix86_expand_sse_movcc): Likewise.
9837 (ix86_expand_int_sse_cmp): Likewise.
9838 (ix86_expand_vec_perm_vpermi2): Likewise.
9839 (ix86_expand_vec_perm): Likewise.
9840 (ix86_expand_sse_unpack): Likewise.
9841 (ix86_expand_int_addcc): Likewise.
9842 (ix86_split_to_parts): Likewise.
9843 (ix86_vectorize_builtin_gather): Likewise.
9844 (ix86_vectorize_builtin_scatter): Likewise.
9845 (avx_vpermilp_parallel): Likewise.
9846 (inline_memory_move_cost): Likewise.
9847 (ix86_tieable_integer_mode_p): Likewise.
9848 (x86_maybe_negate_const_int): Likewise.
9849 (ix86_expand_vector_init_duplicate): Likewise.
9850 (ix86_expand_vector_init_one_nonzero): Likewise.
9851 (ix86_expand_vector_init_one_var): Likewise.
9852 (ix86_expand_vector_init_concat): Likewise.
9853 (ix86_expand_vector_init_interleave): Likewise.
9854 (ix86_expand_vector_init_general): Likewise.
9855 (ix86_expand_vector_set): Likewise.
9856 (ix86_expand_vector_extract): Likewise.
9857 (emit_reduc_half): Likewise.
9858 (ix86_emit_i387_round): Likewise.
9859 (ix86_mangle_type): Likewise.
9860 (ix86_expand_round_sse4): Likewise.
9861 (expand_vec_perm_blend): Likewise.
9862 (canonicalize_vector_int_perm): Likewise.
9863 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
9864 (expand_vec_perm_1): Likewise.
9865 (expand_vec_perm_interleave3): Likewise.
9866 (expand_vec_perm_even_odd_pack): Likewise.
9867 (expand_vec_perm_even_odd_1): Likewise.
9868 (expand_vec_perm_broadcast_1): Likewise.
9869 (ix86_vectorize_vec_perm_const_ok): Likewise.
9870 (ix86_expand_vecop_qihi): Likewise.
9871 (ix86_expand_mul_widen_hilo): Likewise.
9872 (ix86_expand_sse2_abs): Likewise.
9873 (ix86_expand_pextr): Likewise.
9874 (ix86_expand_pinsr): Likewise.
9875 (ix86_preferred_simd_mode): Likewise.
9876 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
9877 * config/i386/sse.md (*andnot<mode>3): Likewise.
9878 (<mask_codefor><code><mode>3<mask_name>): Likewise.
9879 (*<code><mode>3): Likewise.
9880 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
9881 (ia64_expand_atomic_op): Likewise.
9882 (ia64_arg_type): Likewise.
9883 (ia64_mode_to_int): Likewise.
9884 (ia64_scalar_mode_supported_p): Likewise.
9885 (ia64_vector_mode_supported_p): Likewise.
9886 (expand_vec_perm_broadcast): Likewise.
9887 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
9888 (iq2000_function_arg_advance): Likewise.
9889 (iq2000_function_arg): Likewise.
9890 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
9891 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
9892 (m68k_libcall_value): Likewise.
9893 (m68k_function_value): Likewise.
9894 (sched_attr_op_type): Likewise.
9895 * config/mcore/mcore.c (mcore_output_move): Likewise.
9896 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
9898 (microblaze_function_arg): Likewise.
9899 * config/mips/mips.c (mips16_build_call_stub): Likewise.
9900 (mips_print_operand): Likewise.
9901 (mips_mode_ok_for_mov_fmt_p): Likewise.
9902 (mips_vector_mode_supported_p): Likewise.
9903 (mips_preferred_simd_mode): Likewise.
9904 (mips_expand_vpc_loongson_even_odd): Likewise.
9905 (mips_expand_vec_unpack): Likewise.
9906 (mips_expand_vi_broadcast): Likewise.
9907 (mips_expand_vector_init): Likewise.
9908 (mips_expand_vec_reduc): Likewise.
9909 (mips_expand_msa_cmp): Likewise.
9910 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
9911 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
9912 (cc_flags_for_mode): Likewise.
9913 * config/msp430/msp430.c (msp430_print_operand): Likewise.
9914 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
9915 (nds32_output_casesi_pc_relative): Likewise.
9916 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9917 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
9918 (nvptx_gen_unpack): Likewise.
9919 (nvptx_gen_pack): Likewise.
9920 (nvptx_gen_shuffle): Likewise.
9921 (nvptx_gen_wcast): Likewise.
9922 (nvptx_preferred_simd_mode): Likewise.
9923 * config/pa/pa.c (pa_secondary_reload): Likewise.
9924 * config/pa/predicates.md (base14_operand): Likewise.
9925 * config/powerpcspe/powerpcspe-c.c
9926 (altivec_resolve_overloaded_builtin): Likewise.
9927 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
9929 (rs6000_preferred_simd_mode): Likewise.
9930 (output_vec_const_move): Likewise.
9931 (rs6000_expand_vector_extract): Likewise.
9932 (rs6000_split_vec_extract_var): Likewise.
9933 (reg_offset_addressing_ok_p): Likewise.
9934 (rs6000_legitimate_offset_address_p): Likewise.
9935 (rs6000_legitimize_address): Likewise.
9936 (rs6000_emit_set_const): Likewise.
9937 (rs6000_const_vec): Likewise.
9938 (rs6000_emit_move): Likewise.
9939 (spe_build_register_parallel): Likewise.
9940 (rs6000_darwin64_record_arg_recurse): Likewise.
9941 (swap_selector_for_mode): Likewise.
9942 (spe_init_builtins): Likewise.
9943 (paired_init_builtins): Likewise.
9944 (altivec_init_builtins): Likewise.
9945 (do_load_for_compare): Likewise.
9946 (rs6000_generate_compare): Likewise.
9947 (rs6000_expand_float128_convert): Likewise.
9948 (emit_load_locked): Likewise.
9949 (emit_store_conditional): Likewise.
9950 (rs6000_output_function_epilogue): Likewise.
9951 (rs6000_handle_altivec_attribute): Likewise.
9952 (rs6000_function_value): Likewise.
9953 (emit_fusion_gpr_load): Likewise.
9954 (emit_fusion_p9_load): Likewise.
9955 (emit_fusion_p9_store): Likewise.
9956 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
9957 (fusion_gpr_mem_load): Likewise.
9958 (fusion_addis_mem_combo_load): Likewise.
9959 (fusion_addis_mem_combo_store): Likewise.
9960 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
9961 (fusion_gpr_mem_load): Likewise.
9962 (fusion_addis_mem_combo_load): Likewise.
9963 (fusion_addis_mem_combo_store): Likewise.
9964 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9966 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
9967 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
9968 (rs6000_preferred_simd_mode): Likewise.
9969 (output_vec_const_move): Likewise.
9970 (rs6000_expand_vector_extract): Likewise.
9971 (rs6000_split_vec_extract_var): Likewise.
9972 (reg_offset_addressing_ok_p): Likewise.
9973 (rs6000_legitimate_offset_address_p): Likewise.
9974 (rs6000_legitimize_address): Likewise.
9975 (rs6000_emit_set_const): Likewise.
9976 (rs6000_const_vec): Likewise.
9977 (rs6000_emit_move): Likewise.
9978 (rs6000_darwin64_record_arg_recurse): Likewise.
9979 (swap_selector_for_mode): Likewise.
9980 (paired_init_builtins): Likewise.
9981 (altivec_init_builtins): Likewise.
9982 (rs6000_expand_float128_convert): Likewise.
9983 (emit_load_locked): Likewise.
9984 (emit_store_conditional): Likewise.
9985 (rs6000_output_function_epilogue): Likewise.
9986 (rs6000_handle_altivec_attribute): Likewise.
9987 (rs6000_function_value): Likewise.
9988 (emit_fusion_gpr_load): Likewise.
9989 (emit_fusion_p9_load): Likewise.
9990 (emit_fusion_p9_store): Likewise.
9991 * config/rx/rx.c (rx_gen_move_template): Likewise.
9992 (flags_from_mode): Likewise.
9993 * config/s390/predicates.md (s390_alc_comparison): Likewise.
9994 (s390_slb_comparison): Likewise.
9995 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
9996 (s390_vector_mode_supported_p): Likewise.
9997 (s390_cc_modes_compatible): Likewise.
9998 (s390_match_ccmode_set): Likewise.
9999 (s390_canonicalize_comparison): Likewise.
10000 (s390_emit_compare_and_swap): Likewise.
10001 (s390_branch_condition_mask): Likewise.
10002 (s390_rtx_costs): Likewise.
10003 (s390_secondary_reload): Likewise.
10004 (__SECONDARY_RELOAD_CASE): Likewise.
10005 (s390_expand_cs): Likewise.
10006 (s390_preferred_simd_mode): Likewise.
10007 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
10008 * config/sh/sh.c (sh_print_operand): Likewise.
10009 (dump_table): Likewise.
10010 (sh_secondary_reload): Likewise.
10011 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10012 * config/sh/sh.md (casesi_worker_1): Likewise.
10013 (casesi_worker_2): Likewise.
10014 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
10015 (fcc_comparison_operator): Likewise.
10016 * config/sparc/sparc.c (sparc_expand_move): Likewise.
10017 (emit_soft_tfmode_cvt): Likewise.
10018 (sparc_preferred_simd_mode): Likewise.
10019 (output_cbranch): Likewise.
10020 (sparc_print_operand): Likewise.
10021 (sparc_expand_vec_perm_bmask): Likewise.
10022 (vector_init_bshuffle): Likewise.
10023 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10024 (spu_vector_mode_supported_p): Likewise.
10025 (spu_expand_insv): Likewise.
10026 (spu_emit_branch_or_set): Likewise.
10027 (spu_handle_vector_attribute): Likewise.
10028 (spu_builtin_splats): Likewise.
10029 (spu_builtin_extract): Likewise.
10030 (spu_builtin_promote): Likewise.
10031 (spu_expand_sign_extend): Likewise.
10032 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10033 (tilegx_simd_int): Likewise.
10034 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
10035 (tilepro_simd_int): Likewise.
10036 * config/v850/v850.c (const_double_split): Likewise.
10037 (v850_print_operand): Likewise.
10038 (ep_memory_offset): Likewise.
10039 * config/vax/vax.c (vax_rtx_costs): Likewise.
10040 (vax_output_int_move): Likewise.
10041 (vax_output_int_add): Likewise.
10042 (vax_output_int_subtract): Likewise.
10043 * config/visium/predicates.md (visium_branch_operator): Likewise.
10044 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
10045 (visium_print_operand_address): Likewise.
10046 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10047 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
10048 (xtensa_expand_conditional_branch): Likewise.
10049 (xtensa_copy_incoming_a7): Likewise.
10050 (xtensa_output_literal): Likewise.
10051 * dfp.c (decimal_real_maxval): Likewise.
10052 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
10054 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10055 Alan Hayward <alan.hayward@arm.com>
10056 David Sherwood <david.sherwood@arm.com>
10058 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
10059 (mode_nunits_inline): Likewise.
10060 (mode_inner_inline): Likewise.
10061 (mode_unit_size_inline): Likewise.
10062 (mode_unit_precision_inline): Likewise.
10063 (emit_insn_modes_h): Likewise. Also emit a #define of the
10065 (emit_mode_wider): Add an E_ prefix to mode names.
10066 (emit_mode_complex): Likewise.
10067 (emit_mode_inner): Likewise.
10068 (emit_mode_adjustments): Likewise.
10069 (emit_mode_int_n): Likewise.
10070 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
10071 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
10072 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
10073 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
10074 (CRC32_BUILTIN, ENTRY): Likewise.
10075 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
10076 (aarch64_pop_regs): Likewise.
10077 (aarch64_process_components): Likewise.
10078 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
10079 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
10080 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
10081 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
10082 * config/arm/arm.c (arm_init_libfuncs): Likewise.
10083 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
10085 * config/i386/i386-builtin.def (pcmpestr): Likewise.
10086 (pcmpistr): Likewise.
10087 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
10088 * config/mmix/mmix.c (mmix_output_condition): Likewise.
10089 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
10091 * config/rl78/rl78.c (mduc_regs): Likewise.
10092 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10093 (htm_expand_builtin): Likewise.
10094 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
10095 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
10096 * config/xtensa/xtensa.c (print_operand): Likewise.
10097 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
10098 (NUM_MODE_VECTOR_INT): Likewise.
10099 * genoutput.c (null_operand): Likewise.
10100 (output_operand_data): Likewise.
10101 * genrecog.c (print_parameter_value): Likewise.
10102 * lra.c (debug_operand_data): Likewise.
10104 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10106 * dwarf2out.c (scompare_loc_descriptor_wide)
10107 (scompare_loc_descriptor_narrow): New functions, split out from...
10108 (scompare_loc_descriptor): ...here.
10109 * expmed.c (emit_store_flag_int): New function, split out from...
10110 (emit_store_flag): ...here.
10112 2017-08-30 Richard Biener <rguenther@suse.de>
10114 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
10115 (dwarf2out_early_finish): Move setting of AT_pubnames from
10116 early debug output to early finish.
10118 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10120 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
10121 and -mdata-region to the assembler.
10123 2017-08-30 Richard Biener <rguenther@suse.de>
10125 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
10127 (gen_subprogram_die): Add DW_AT_object_pointer only early.
10128 (dwarf2out_early_global_decl): Only generate a DIE for the
10129 abstract origin if it doesn't already exist or is a declaration DIE.
10130 (resolve_addr): Do not add the linkage name twice when
10131 generating a stub DIE for the DW_TAG_GNU_call_site target.
10133 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10135 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10136 Use machine_mode rather than int for arg1_mode.
10138 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
10141 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
10142 that the second argument of the built-in functions to unpack
10143 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
10144 switch statement instead a lot of if statements.
10145 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
10146 Allow 64-bit values to be in Altivec registers as well as
10147 traditional floating point registers.
10148 (pack<mode>, FMOVE128_VSX iterator): Likewise.
10150 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
10152 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
10153 MAX_REGS_PER_ADDRESS == 1.
10155 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
10157 * config/i386/i386.opt (flag_fentry): Do not init to -1.
10158 * config/i386/i386.c (ix86_option_override_internal): Simplify
10159 setting of opts->x_flag_entry.
10161 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10162 Jakub Jelinek <jakub@redhat.com>
10163 Richard Biener <rguenther@suse.de>
10165 PR tree-optimization/81503
10166 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
10167 folded constant fits in the target type; reorder tests for clarity.
10169 2017-08-29 Martin Liska <mliska@suse.cz>
10171 * passes.def: Include pass_lower_switch.
10172 * stmt.c (dump_case_nodes): Remove and move to
10173 tree-switch-conversion.
10174 (case_values_threshold): Likewise.
10175 (expand_switch_as_decision_tree_p): Likewise.
10176 (emit_case_decision_tree): Likewise.
10177 (expand_case): Likewise.
10178 (balance_case_nodes): Likewise.
10179 (node_has_low_bound): Likewise.
10180 (node_has_high_bound): Likewise.
10181 (node_is_bounded): Likewise.
10182 (emit_case_nodes): Likewise.
10183 (struct simple_case_node): New struct.
10184 (add_case_node): Remove.
10185 (emit_case_dispatch_table): Use vector instead of case_list.
10186 (reset_out_edges_aux): Remove.
10187 (compute_cases_per_edge): Likewise.
10188 (expand_case): Build list of simple_case_node.
10189 (expand_sjlj_dispatch_table): Use it.
10190 * tree-switch-conversion.c (struct case_node): Moved from
10191 stmt.c and adjusted.
10192 (emit_case_nodes): Likewise.
10193 (node_has_low_bound): Likewise.
10194 (node_has_high_bound): Likewise.
10195 (node_is_bounded): Likewise.
10196 (case_values_threshold): Likewise.
10197 (reset_out_edges_aux): Likewise.
10198 (compute_cases_per_edge): Likewise.
10199 (add_case_node): Likewise.
10200 (dump_case_nodes): Likewise.
10201 (balance_case_nodes): Likewise.
10202 (expand_switch_as_decision_tree_p): Likewise.
10203 (emit_jump): Likewise.
10204 (emit_case_decision_tree): Likewise.
10205 (try_switch_expansion): Likewise.
10206 (do_jump_if_equal): Likewise.
10207 (emit_cmp_and_jump_insns): Likewise.
10208 (fix_phi_operands_for_edge): New function.
10209 (record_phi_operand_mapping): Likewise.
10210 (class pass_lower_switch): New pass.
10211 (pass_lower_switch::execute): New function.
10212 (make_pass_lower_switch): Likewise.
10213 (conditional_probability):
10214 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
10215 * tree-pass.h: Add make_pass_lower_switch.
10217 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10220 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
10223 2017-08-29 Richard Biener <rguenther@suse.de>
10225 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
10226 we do not add a DW_AT_inline attribute twice.
10227 (gen_subprogram_die): Remove code setting DW_AT_inline on
10228 DECL_ABSTRACT_P nodes.
10230 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
10232 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
10233 calls to internal functions.
10234 (gimplify_modify_expr): Likewise.
10235 * tree-call-cdce.c (use_internal_fn): Likewise.
10236 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
10237 (convert_to_divmod): Set the nothrow flag.
10238 * tree-if-conv.c (predicate_mem_writes): Likewise.
10239 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
10240 (vectorizable_call): Likewise.
10241 (vectorizable_store): Likewise.
10242 (vectorizable_load): Likewise.
10243 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
10244 (vect_recog_mask_conversion_pattern): Likewise.
10246 2017-08-29 Martin Liska <mliska@suse.cz>
10249 * gcc.c (driver_handle_option): Add new argument.
10250 * opts-common.c (handle_option): Pass
10251 target_option_override_hook.
10252 * opts-global.c (lang_handle_option): Add new option.
10253 (set_default_handlers): Add new argument.
10254 (decode_options): Likewise.
10255 * opts.c (target_handle_option): Likewise.
10256 (common_handle_option): Call target_option_override_hook.
10257 * opts.h (struct cl_option_handler_func): Add hook for
10258 target option override.
10259 (struct cl_option_handlers): Likewise.
10260 (set_default_handlers): Add new argument.
10261 (decode_options): Likewise.
10262 (common_handle_option): Likewise.
10263 (target_handle_option): Likewise.
10264 * toplev.c (toplev::main): Pass targetm.target_option.override
10267 2017-08-29 Richard Biener <rguenther@suse.de>
10268 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
10270 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
10271 life to the active subtree.
10273 2017-08-28 Jeff Law <law@redhat.com>
10275 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
10276 derive_equivalences.
10277 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
10278 Code moved into....
10279 (edge_info::derive_equivalences): New private member function
10281 * tree-ssa-dom.c (class edge_info): Changed from a struct
10282 to a class. Add ctor/dtor, methods and data members.
10283 (edge_info::edge_info): Renamed from allocate_edge_info.
10284 Initialize additional members.
10285 (edge_info::~edge_info): New.
10286 (free_dom_edge_info): Delete the edge info.
10287 (record_edge_info): Use new class & associated member functions.
10288 Tighten forms for testing for edge equivalences.
10289 (record_temporary_equivalences): Iterate over the simple
10290 equivalences rather than assuming there's only one per edge.
10291 (cprop_into_successor_phis): Iterate over the simple
10292 equivalences rather than assuming there's only one per edge.
10293 (optimize_stmt): Use operand_equal_p rather than pointer
10294 equality for mini-DSE code.
10296 2017-08-28 Nathan Sidwell <nathan@acm.org>
10298 * gcc.c (execute): Fold SIGPIPE handling into switch
10299 statement. Adjust internal error message.
10301 2017-08-28 Richard Biener <rguenther@suse.de>
10304 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
10305 Do nothing for removed DIEs.
10307 2017-08-28 Richard Biener <rguenther@suse.de>
10309 PR tree-optimization/81977
10310 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
10313 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
10316 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
10317 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
10318 using targetm.gen_mem_thread_fence.
10320 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
10323 * config/i386/i386.md (*<btsc><mode>): Change operand 2
10324 predicate to register_operand. Reorder operands.
10325 (*btr<mode>): Ditto.
10326 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
10327 (*btr<mode>_mask): Ditto.
10329 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
10331 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
10332 * config/rs6000/xmmintrin.h: New file.
10333 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
10335 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10338 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
10339 parameter and_insn and return it.
10340 (recombine_lvx_pattern): Insert a copy to ensure availability of
10341 the base register of the copied masking operation at the point of
10342 the instruction replacement.
10343 (recombine_stvx_pattern): Likewise.
10345 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
10347 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
10348 undocumented switches.
10349 (-mpower9-dform-vector): Likewise.
10350 (-mpower9-dform): Likewise.
10351 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
10352 comments to delete references to -mpower9-dform* switches.
10353 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
10354 Delete reference to -mpower9-dform* switches, test for
10355 -mpower9-vector instead.
10356 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
10357 (OTHER_P9_VECTOR_MASKS): Likewise.
10358 (POWERPC_MASKS): Likewise.
10359 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
10360 tests against -mpower9-dform* to -mpower9-vector. Delete code
10361 that checked for -mpower9-dform* consistancy with other options.
10362 Add test for -mpower9-misc to enable other power9 switches.
10363 (rs6000_init_hard_regno_mode_ok): Likewise.
10364 (rs6000_option_override_internal): Likewise.
10365 (rs6000_emit_prologue): Likewise.
10366 (rs6000_emit_epilogue): Likewise.
10367 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
10368 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
10369 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
10371 (emit_fusion_p9_store): Likewise.
10372 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
10373 resetting these macros if the assembler does not support ISA 3.0
10375 (TARGET_P9_DFORM_VECTOR): Likewise.
10376 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
10377 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
10379 2017-08-25 Alan Modra <amodra@gmail.com>
10381 PR rtl-optimization/81747
10382 * cse.c (cse_extended_basic_block): Don't attempt to record
10383 equivalences for degenerate conditional branches that branch
10384 to their fall-through.
10386 2017-08-24 Martin Sebor <msebor@redhat.com>
10388 PR middle-end/81908
10389 * gimple-fold.c (size_must_be_zero_p): New function.
10390 (gimple_fold_builtin_memory_op): Call it.
10392 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
10394 * config/rs6000/mm_malloc.h: New file.
10396 2017-08-24 Bin Cheng <bin.cheng@arm.com>
10398 PR tree-optimization/81913
10399 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
10400 analysis when either IVs in condition can wrap.
10402 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
10404 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
10405 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
10407 2017-08-24 Richard Biener <rguenther@suse.de>
10410 * targhooks.c (default_target_can_inline_p): Properly
10411 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
10412 is present and always compare.
10413 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
10414 infer -mfpmath=sse from TARGET_SSE_P.
10415 (ix86_can_inline_p): Properly use target_option_default_node when
10416 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
10418 2017-08-24 Richard Biener <rguenther@suse.de>
10421 * dwarf2out.c (output_die): Handle flag_generate_offload like
10423 (output_comp_unit): Likewise.
10424 (gen_array_type_die): Likewise.
10425 (dwarf2out_early_finish): Likewise.
10426 (note_variable_value_in_expr): Likewise.
10427 (dwarf2out_finish): Likewise. Adjust assert.
10428 * cgraphunit.c (symbol_table::compile): Move setting of
10429 flag_generate_offload earlier ...
10430 (symbol_table::finalize_compilation_unit): ... here, before
10431 early debug finalization.
10433 2017-08-24 Richard Biener <rguenther@suse.de>
10435 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
10436 and ipa-fnsummary.h.
10437 (ix86_can_inline_p): When ix86_fpmath flags do not match
10438 check whether the callee uses FP math at all.
10440 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
10442 PR middle-end/81931
10443 * tree-ssanames.c (get_nonzero_bits): Use element_precision
10444 instead of TYPE_PRECISION.
10446 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
10447 Alan Hayward <alan.hayward@arm.com>
10448 David Sherwood <david.sherwood@arm.com>
10450 * combine.c (make_extraction): Use subreg_offset_from_lsb.
10452 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
10454 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
10456 (ix86_frame::stack_realign_allocate): New field.
10457 (struct machine_frame_state): Modify comments.
10458 (machine_frame_state::sp_realigned_fp_end): New field.
10459 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
10460 layout calculation.
10461 (sp_valid_at): Add assertion to assure no attempt to access invalid
10462 offset of a realigned stack.
10463 (fp_valid_at): Likewise.
10464 (choose_baseaddr): Modify comments.
10465 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
10466 ix86_expand_prologue.
10467 (ix86_expand_prologue): Modify stack realignment and allocation.
10468 (ix86_expand_epilogue): Modify comments.
10469 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
10470 avx2_runtime, avx512f, and avx512f_runtime.
10472 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
10474 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
10475 (mstackrealign): Do not init to -1.
10476 * config/i386/i386.c (ix86_option_override_internal):
10477 Check opts_set, not opts when setting default value of
10478 opts->x_ix86_force_align_arg_pointer.
10480 2017-08-23 Richard Biener <rguenther@suse.de>
10482 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
10483 lang_hooks.decl_printable_name.
10484 * print-rtl-function.c (print_rtx_function): Likewise.
10485 * tree-pretty-print.c (dump_function_header): Likewise.
10487 2017-08-23 Richard Biener <rguenther@suse.de>
10490 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
10491 -g0 at compile-time.
10493 2017-08-23 Tamar Christina <tamar.christina@arm.com>
10495 PR middle-end/19706
10496 * doc/sourcebuild.texi (Other hardware attributes):
10499 2017-08-23 Tamar Christina <tamar.christina@arm.com>
10501 PR middle-end/19706
10502 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
10503 Fix single-use check.
10505 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
10507 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
10509 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
10511 * doc/install.texi: Modify to add more details on running selected
10514 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
10516 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
10517 is combined with -mabi=ms.
10518 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
10521 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10523 PR tree-optimization/81488
10524 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
10525 and cached_basis fields.
10526 (MAX_SPREAD): New constant.
10527 (alloc_cand_and_find_basis): Initialize new fields.
10528 (clear_visited): New function.
10529 (create_phi_basis_1): Rename from create_phi_basis, set visited
10530 and cached_basis fields.
10531 (create_phi_basis): New wrapper function.
10532 (phi_add_costs_1): Rename from phi_add_costs, add spread
10533 parameter, set visited field, short-circuit when limits reached.
10534 (phi_add_costs): New wrapper function.
10535 (record_phi_increments_1): Rename from record_phi_increments, set
10537 (record_phi_increments): New wrapper function.
10538 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
10539 (phi_incr_cost): New wrapper function.
10540 (all_phi_incrs_profitable_1): Rename from
10541 all_phi_incrs_profitable, set visited field.
10542 (all_phi_incrs_profitable): New wrapper function.
10544 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
10545 Alan Hayward <alan.hayward@arm.com>
10546 David Sherwood <david.sherwood@arm.com>
10548 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
10549 that takes the outer and inner modes.
10550 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
10551 comparison as the canonical test for a paradoxical subreg.
10552 * combine.c (simplify_set): Use paradoxical_subreg_p.
10553 (make_extraction): Likewise.
10554 (force_to_mode): Likewise.
10555 (rtx_equal_for_field_assignment_p): Likewise.
10556 (gen_lowpart_for_combine): Likewise.
10557 (simplify_comparison): Likewise.
10558 * cse.c (equiv_constant): Likewise.
10559 * expmed.c (store_bit_field_1): Likewise.
10560 * final.c (alter_subreg): Likewise.
10561 * fwprop.c (propagate_rtx): Likewise.
10562 (forward_propagate_subreg): Likewise.
10563 * ira-conflicts.c (ira_build_conflicts): Likewise.
10564 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
10565 * lra-constraints.c (curr_insn_transform): Likewise.
10566 (split_reg): Likewise.
10567 * lra-eliminations.c (move_plus_up): Likewise.
10568 (lra_eliminate_regs_1): Likewise.
10569 * recog.c (general_operand): Likewise.
10570 * ree.c (combine_reaching_defs): Likewise.
10571 * reload.c (push_reload): Likewise.
10572 (find_reloads): Likewise.
10573 * reload1.c (elimination_effects): Likewise.
10574 (compute_reload_subreg_offset): Likewise.
10575 (choose_reload_regs): Likewise.
10576 * rtlanal.c (subreg_lsb_1): Likewise.
10577 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10578 (simplify_subreg): Likewise.
10579 * var-tracking.c (track_loc_p): Likewise.
10580 * emit-rtl.c (byte_lowpart_offset): Likewise.
10581 (paradoxical_subreg_p): Delete out-of-line definition.
10583 2017-08-22 Jeff Law <law@redhat.com>
10585 PR tree-optimization/81741
10586 PR tree-optimization/71947
10587 * tree-ssa-dom.c: Include tree-inline.h.
10588 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
10589 equivalences if one is more expensive to compute than the other.
10590 * tree-ssa-scopedtables.h (class const_or_copies): Make
10591 record_const_or_copy_raw method private.
10592 (class avail_exprs_stack): New method simplify_binary_operation.
10593 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
10594 avail_exprs_stack::simplify_binary_operation as needed.
10595 (avail_exprs_stack::simplify_binary_operation): New function.
10597 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
10599 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
10600 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
10601 (DOT_SYMBOLS): Likewise.
10602 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
10603 (RELOCATABLE_NEEDS_FIXUP): Likewise.
10604 (RS6000_ABI_NAME): Likewise.
10605 (TARGET_CMODEL): Likewise.
10606 (TOC_SECTION_ASM_OP): Likewise.
10607 (SET_CMODEL): New macro.
10608 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
10610 2017-08-22 Richard Biener <rguenther@suse.de>
10612 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
10613 to allow for free-lang-data replacements similar to verify_type_variant.
10615 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
10616 Alan Hayward <alan.hayward@arm.com>
10617 David Sherwood <david.sherwood@arm.com>
10619 * config/aarch64/aarch64.md (casesi): Use DImode rather than
10620 VOIDmode for the LABEL_REF.
10622 2017-08-22 Richard Biener <rguenther@suse.de>
10624 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
10626 2017-08-22 Richard Biener <rguenther@suse.de>
10628 * common.opt (feliminate-dwarf2-dups): Ignore.
10629 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
10630 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
10631 same_die_p_wrap, compute_section_prefix,
10632 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
10633 (comdat_symbol_id, comdat_symbol_number): Likewise.
10634 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
10636 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
10637 (output_die): Mark unreachable path unreachable.
10638 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
10639 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
10640 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
10641 (dwarf2out_early_finish): Likewise.
10643 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
10645 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
10647 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
10650 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
10651 not VAR_P. Filter attribute warnings with OPT_Wattributes.
10652 (avr_attribute_table) <io, io_low, address>: Initialize
10653 .decl_required with true.
10655 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
10657 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
10658 undocumented debugging options.
10659 (-mvsx-scalar-double): Likewise.
10660 (-mallow-df-permute): Likewise.
10661 (-mvectorize-builtins): Likewise.
10662 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10663 (rs6000_builtin_vectorized_function): Likewise.
10664 (rs6000_builtin_md_vectorized_function): Likewise.
10665 (rs6000_opt_vars): Likewise.
10667 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
10670 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
10671 (*btrq_imm): Rename from *btrq.
10672 (*btcq_imm): Rename from *btcq.
10673 (btsc): New code attribute.
10674 (*<btsc><mode>): New insn pattern.
10675 (*btr<mode>): Ditto.
10676 (*<btsc><mode>_mask): New insn_and_split pattern.
10677 (*btr<mode>_mask): Ditto.
10679 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10680 Alan Hayward <alan.hayward@arm.com>
10681 David Sherwood <david.sherwood@arm.com>
10683 * function.c (pad_below): Simplify padding calculation.
10685 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10686 Alan Hayward <alan.hayward@arm.com>
10687 David Sherwood <david.sherwood@arm.com>
10689 * target.def (function_prologue): Remove frame size argument.
10690 (function_epilogue): Likewise.
10691 * doc/tm.texi: Regenerate.
10692 * final.c (final_start_function): Update call to function_prologue.
10693 (final_end_function): Update call to function_epilogue.
10694 (default_function_pro_epilogue): Remove frame size argument.
10695 * output.h (default_function_pro_epilogue): Likewise.
10696 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
10697 (arm_output_function_prologue): Likewise.
10698 * config/frv/frv.c (frv_function_prologue): Likewise.
10699 (frv_function_epilogue): Likewise.
10700 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
10701 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
10702 (ia64_output_function_epilogue): Likewise.
10703 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
10704 (m32r_output_function_epilogue): Likewise.
10705 * config/microblaze/microblaze.c (microblaze_function_prologue)
10706 (microblaze_function_epilogue): Likewise.
10707 * config/mips/mips.c (mips_output_function_prologue): Likewise.
10708 (mips_output_function_epilogue): Likewise.
10709 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
10710 (mmix_target_asm_function_epilogue): Likewise.
10711 * config/msp430/msp430.c (msp430_start_function): Likewise.
10712 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
10713 (nds32_asm_function_epilogue): Likewise.
10714 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
10715 * config/pa/pa.c (pa_output_function_prologue): Likewise.
10716 (pa_output_function_epilogue): Likewise.
10717 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
10718 (rs6000_output_function_epilogue): Likewise.
10719 * config/rl78/rl78.c (rl78_start_function): Likewise.
10720 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
10721 (rs6000_output_function_epilogue): Likewise.
10722 * config/rx/rx.c (rx_output_function_prologue): Likewise.
10723 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
10724 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
10725 (sparc_asm_function_epilogue): Likewise.
10727 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10729 * tree.h (type_has_mode_precision_p): New function.
10730 * convert.c (convert_to_integer_1): Use it.
10731 * expr.c (expand_expr_real_2): Likewise.
10732 (expand_expr_real_1): Likewise.
10733 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
10734 * match.pd: Likewise.
10735 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
10736 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
10737 * tree-tailcall.c (process_assignment): Likewise.
10738 * tree-vect-loop.c (vectorizable_reduction): Likewise.
10739 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
10740 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
10741 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10742 (vectorizable_assignment): Likewise.
10743 (vectorizable_shift): Likewise.
10744 (vectorizable_operation): Likewise.
10745 * tree-vrp.c (register_edge_assert_for_2): Likewise.
10747 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
10749 * match.pd: Add pow (C, x) simplification.
10751 2017-08-21 Richard Biener <rguenther@suse.de>
10753 PR tree-optimization/81900
10754 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
10755 for blocks with abnormal predecessors.
10756 (compute_antic): Do not set visited flag prematurely.
10758 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
10761 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
10763 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10765 * stor-layout.h (vector_type_mode): Move to...
10766 * tree.h (vector_type_mode): ...here.
10767 * stor-layout.c (vector_type_mode): Move to...
10768 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
10770 2017-08-21 Richard Biener <rguenther@suse.de>
10772 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
10773 register_external_die hooks.
10774 (debug_false_tree_charstarstar_uhwistar): Declare.
10775 (debug_nothing_tree_charstar_uhwi): Likewise.
10776 * debug.c (do_nothing_debug_hooks): Adjust.
10777 (debug_false_tree_charstarstar_uhwistar): New do nothing.
10778 (debug_nothing_tree_charstar_uhwi): Likewise.
10779 * dbxout.c (dbx_debug_hooks): Adjust.
10780 (xcoff_debug_hooks): Likewise.
10781 * sdbout.c (sdb_debug_hooks): Likewise.
10782 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
10783 * dwarf2out.c (macinfo_label_base): New global.
10784 (dwarf2out_register_external_die): New function for the
10785 register_external_die hook.
10786 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
10787 (dwarf2_debug_hooks): Use them.
10788 (dwarf2_lineno_debug_hooks): Adjust.
10789 (struct die_struct): Add with_offset flag.
10790 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
10791 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
10792 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
10793 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
10794 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
10795 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
10796 defining section names for the early LTO debug variants.
10797 (reset_indirect_string): New helper.
10798 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
10799 (print_dw_val): Add support for offsetted symbol references.
10800 (get_ultimate_context): Split out from is_cxx.
10801 (is_cxx): Use get_ultimate_context.
10802 (is_fortran): Add decl overload.
10803 (compute_comp_unit_symbol): Split out worker from
10804 compute_section_prefix.
10805 (compute_section_prefix): Call compute_comp_unit_symbol and
10806 set comdat_type_p here.
10807 (output_die): Skip DIE symbol output for the LTO added one.
10808 Handle DIE symbol references with offset.
10809 (output_comp_unit): Guard section name mangling properly.
10810 For LTO debug sections emit a symbol at the section beginning
10811 which we use to refer to its DIEs.
10812 (add_abstract_origin_attribute): For DIEs registered via
10813 dwarf2out_register_external_die directly refer to the early
10814 DIE rather than indirectly through the shadow one we created.
10815 Remove obsolete call to dwarf2out_abstract_function for
10816 non-function/block origins.
10817 (gen_array_type_die): When generating early LTO debug do
10818 not emit DW_AT_string_length.
10819 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
10820 late when in LTO. As suggested place a gcc_unreachable for
10821 the DECL_ABSTRACT_P case.
10822 (gen_subprogram_die): Avoid another specification DIE
10823 for early built declarations/definitions for the late LTO case.
10824 (gen_variable_die): Add type references for late duplicated VLA dies
10826 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
10827 we have the abstract instance already.
10828 (process_scope_var): Adjust decl DIE contexts in LTO which
10829 first puts them in limbo.
10830 (gen_decl_die): Do not generate type DIEs late apart from
10831 types for VLAs or for decls we do not yet have a DIE. Do not
10832 call dwarf2out_abstract_function late.
10833 (dwarf2out_early_global_decl): Make sure to create DIEs
10834 for abstract instances of a decl first.
10835 (dwarf2out_late_global_decl): Adjust comment.
10836 (output_macinfo_op): With multiple macro sections use
10837 macinfo_label_base to distinguish labels.
10838 (output_macinfo): Likewise. Update macinfo_label_base.
10839 Pass in the line info label.
10840 (note_variable_value_in_expr): When generating LTO resolve
10841 all variable values here by generating DIEs as needed.
10842 (init_sections_and_labels): Add early LTO debug flag parameter
10843 and generate different sections and names if set. Add generation
10844 counter for the labels so we can have multiple of them.
10845 (reset_dies): Helper to allow DIEs to be output multiple times.
10846 (dwarf2out_finish): When outputting DIEs to the fat part of an
10847 LTO object first reset DIEs.
10848 (dwarf2out_early_finish): Output early DIEs when generating LTO.
10849 (modified_type_die): Check for decl_ultimate_origin being self
10851 (gen_type_die_with_usage): Likewise.
10852 (gen_typedef_die): Allow decl_ultimate_origin being self.
10853 (set_decl_abstract_flags): Remove.
10854 (set_block_abstract_flags): Likewise.
10855 (dwarf2out_abstract_function): Treat the early generated DIEs
10856 as the abstract copy and only add DW_AT_inline and
10857 DW_AT_artificial here and call set_decl_origin_self.
10858 If the DIE has an abstract origin don't do anything.
10859 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
10860 if we have none yet (Go fails to build one, PR78628).
10861 (variably_modified_type_p): Prevent endless recursion for Ada
10862 cyclic pointer types.
10863 * lto-streamer-in.c: Include debug.h.
10864 (dref_queue): New global.
10865 (lto_read_tree_1): Stream in DIE references.
10866 (lto_input_tree): Register DIE references.
10867 (input_function): Stream DECL_DEBUG_ARGS.
10868 * lto-streamer-out.c: Include debug.h.
10869 (lto_write_tree_1): Output DIE references.
10870 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
10871 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
10872 (output_function): Stream DECL_DEBUG_ARGS.
10873 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
10874 Stream DECL_ABSTRACT_ORIGIN.
10875 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
10876 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
10877 DECL_CONTEXT for file-scope decls.
10878 * lto-streamer.h (struct dref_entry): Declare.
10879 (dref_queue): Likewise.
10880 * cfgexpand.c (pass_expand::execute): Do not call the
10881 outlining_inline_function hook here.
10882 * lto-wrapper.c (debug_obj): New global.
10883 (tool_cleanup): Unlink it if required.
10884 (debug_objcopy): New function.
10885 (run_gcc): Handle early debug sections in the IL files by
10886 extracting them to separate files, partially linkin them and
10887 feeding the result back as result to the linker.
10888 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
10889 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
10890 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
10891 sections into a separate segment.
10892 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
10894 (darwin_asm_dwarf_section): Likewise.
10895 (darwin_asm_output_dwarf_offset): Likewise.
10896 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
10898 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10899 Alan Hayward <alan.hayward@arm.com>
10900 David Sherwood <david.sherwood@arm.com>
10902 * read-md.h (md_reader::record_potential_iterator_use): Replace
10903 pointer argument with an rtx and an index.
10904 * read-rtl.c (iterator_group::apply_iterator): Likewise.
10905 (apply_mode_iterator): Likewise.
10906 (apply_code_iterator): Likewise.
10907 (apply_int_iterator): Likewise.
10908 (apply_subst_iterator): Likewise.
10909 (record_iterator_use): Likewise.
10910 (record_attribute_use): Likewise.
10911 (md_reader::record_potential_iterator_use): Likewise. Update calls
10912 to record_iterator_use and apply_iterator.
10913 (iterator_use): Replace ptr with x and index.
10914 (attribute_use): Likewise.
10915 (apply_attribute_uses): Update calls to apply_iterator.
10916 (apply_iterators): Likewise. Update initialization of iterator_use.
10917 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
10918 and record_potential_iterator_use.
10919 (rtx_reader::read_rtx_operand): Likewise.
10921 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10922 Alan Hayward <alan.hayward@arm.com>
10923 David Sherwood <david.sherwood@arm.com>
10925 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
10928 2017-08-21 Richard Biener <rguenther@suse.de>
10930 PR middle-end/81884
10931 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
10932 at struct end conservatively when comparing common bases.
10934 2017-08-21 Richard Biener <rguenther@suse.de>
10936 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
10937 (mem_ref_in_stmt): Remove.
10938 (determine_max_movement): Use ref index to get at the reference.
10939 (invariantness_dom_walker::before_dom_children): Deal with
10940 lim data already initialized.
10941 (gather_mem_refs_stmt): Initialize lim data and record ref index.
10943 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
10945 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
10946 (TARGET_ISA_ROUND): Ditto.
10947 (TARGET_ROUND): Ditto.
10948 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
10949 * config/i386/i386.md: Ditto.
10950 * config/i386/sse.md: Ditto.
10951 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
10952 with OPTION_MASK_ISA_SSE4_1.
10954 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
10957 * doc/extend.texi (x86 Built-in Functions): Correct the name of
10958 __builtin_ia32_lzcnt_u16.
10960 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
10963 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
10964 (rs6000_set_current_function): Rewrite function to use it.
10966 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
10969 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
10970 and TYPE_WARN_IF_NOT_ALIGN.
10971 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
10972 (handle_warn_if_not_align): New.
10973 (place_union_field): Call handle_warn_if_not_align.
10974 (place_field): Call handle_warn_if_not_align.
10975 Copy TYPE_WARN_IF_NOT_ALIGN.
10976 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
10977 (layout_type): Likewise.
10978 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
10980 (tree_decl_common): Add warn_if_not_align.
10981 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
10982 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
10983 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
10984 (DECL_WARN_IF_NOT_ALIGN): Likewise.
10985 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
10986 * doc/extend.texi: Document warn_if_not_aligned attribute.
10987 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
10989 2017-08-17 Martin Liska <mliska@suse.cz>
10992 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
10993 (get_data_dependence): Use it as pointer type.
10994 (distribute_loop): Likewise.
10996 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10998 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
10999 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
11000 (p8_vmrgow_<mode>_direct): New define_insn.
11001 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
11002 handle endianness for vmrgew and vmrgow permute patterns.
11004 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
11006 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
11007 * config/rs6000/rs6000-cpus.def: Remove comment.
11008 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
11009 (POWERPC_MASKS): Likewise.
11010 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
11011 use of TARGET_VSX_TIMODE.
11012 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
11013 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
11014 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
11015 (rs6000_option_override_internal): Remove dead code.
11016 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
11017 (rs6000_legitimize_reload_address): Likewise.
11018 (rs6000_legitimate_address_p): Likewise.
11019 (rs6000_opt_masks): Delete "vsx-timode".
11020 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
11021 from function comment.
11022 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
11023 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
11024 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
11026 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
11027 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
11028 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
11029 (VSX_TI): Likewise.
11031 (define_peephole2): Likewise.
11033 2017-08-17 Martin Sebor <msebor@redhat.com>
11036 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
11037 past the end of an array.
11038 (test_pp_format): Add test cases.
11040 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
11042 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
11043 missing ECF_NOTHROW flags.
11045 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
11048 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
11049 operands residing in integer registers.
11050 (*vsx_le_perm_load_<mode>): Likewise.
11051 (*vsx_le_perm_store_<mode>): Likewise.
11052 (define_peephole2): Add peepholes to optimize the above.
11054 2017-08-17 Marek Polacek <polacek@redhat.com>
11056 PR middle-end/81814
11057 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
11058 to mimic what shorten_compare did. Change the return type to bool.
11059 (fold_cond_expr_with_comparison): Update call to
11060 operand_equal_for_comparison_p.
11061 (fold_ternary_loc): Likewise.
11063 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
11065 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
11067 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
11069 2017-08-17 Richard Biener <rguenther@suse.de>
11071 * tree-ssa-structalias.c (solve_graph): When propagating
11072 to successors update the graphs succ edges and avoid duplicate work.
11074 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
11077 * config/i386/i386.c (ix86_option_override_internal): Save target
11078 specific options after ix86_stack_protector_guard_reg was changed.
11080 2017-08-17 Richard Biener <rguenther@suse.de>
11082 PR tree-optimization/81827
11083 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
11084 (new_var_info): Initialize it conservatively.
11085 (get_call_vi): Mark register vars.
11086 (new_scalar_tmp_constraint_exp): Likewise.
11087 (handle_rhs_call): Likewise.
11088 (handle_const_call): Likewise.
11089 (create_function_info_for): Likewise.
11090 (solve_constraints): Sort varinfos to separate register from
11091 non-register vars to pack points-to solution bitmaps during
11094 2017-08-17 Marek Polacek <polacek@redhat.com>
11096 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
11098 2017-08-17 Richard Biener <rguenther@suse.de>
11100 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
11101 to true when overflow is undefined and we saturated the result.
11103 2017-08-17 Alan Modra <amodra@gmail.com>
11106 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
11107 Don't use store multiple if only one reg needs saving.
11108 (interesting_frame_related_regno): New function.
11109 (rs6000_frame_related): Don't emit frame info for regs that
11111 (rs6000_emit_epilogue): Likewise.
11113 2017-08-16 Nathan Sidwell <nathan@acm.org>
11115 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
11116 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
11117 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
11118 * tree.c (free_lang_data_in_type): Use else-if chain. Always
11119 clear TYPE_LANG_1. Remove obsolete member-function stripping.
11120 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
11121 (verify_type): Adjust for TYPE_BINFO move.
11122 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
11123 process TYPE_BINFO directly.
11124 (hash_tree): Likewise.
11125 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
11127 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
11130 2017-08-16 David Malcolm <dmalcolm@redhat.com>
11132 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
11134 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
11137 * config/i386/i386.md (*anddi_1_btr): Change predicates of
11138 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
11139 Add ix86_binary_operator_ok to insn constraint.
11140 (*iordi_1_bts): Ditto.
11141 (*xordi_1_btc): Ditto.
11142 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
11143 Update corresponding peephole2 pattern.
11147 2017-08-16 Bin Cheng <bin.cheng@arm.com>
11149 PR tree-optimization/81832
11150 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
11151 copy loop header which has IFN_LOOP_DIST_ALIAS call.
11153 2017-08-16 Marek Polacek <polacek@redhat.com>
11156 * fold-const.c (fold_indirect_ref_1): Restore original behavior
11157 regarding size_zero_node.
11159 2017-08-16 Martin Liska <mliska@suse.cz>
11162 * config.gcc: Respect previously set extra_objs in case
11165 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
11167 PR tree-optimization/81835
11168 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
11169 the phi SSA_NAME. Check that the condition in a COND_EXPR does
11170 not depend on the phi.
11172 2017-08-16 Alan Modra <amodra@gmail.com>
11174 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
11177 2017-08-16 Alan Modra <amodra@gmail.com>
11179 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
11180 (save_reg_p): ..into this. Update all callers.
11181 (first_reg_to_save): Simplify.
11183 2017-08-16 Alan Modra <amodra@gmail.com>
11185 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
11188 2017-08-15 Joseph Myers <joseph@codesourcery.com>
11192 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
11193 constant count if that count is less than 32.
11195 2017-08-15 Nathan Sidwell <nathan@acm.org>
11197 * gcc.c (execute): Emit friendlier message if inferior is killed
11198 by an external cause.
11200 2017-08-15 Richard Biener <rguenther@suse.de>
11202 PR tree-optimization/81790
11203 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
11204 CONSTRUCTORs from simplifying and VN.
11206 2017-08-14 Martin Sebor <msebor@redhat.com>
11208 * builtin-attrs.def: Add comments.
11210 2017-08-14 Martin Sebor <msebor@redhat.com>
11213 * doc/extend.texi (attribute nonstring): Document new attribute.
11215 2017-08-14 Martin Sebor <msebor@redhat.com>
11218 * tree-diagnostic.c (default_tree_printer): Handle %G.
11219 * gimple-pretty-print.h (percent_G_format): Declare new function.
11220 * gimple-pretty-print.c (percent_G_format): Define.
11221 * tree-pretty-print.c (percent_K_format): Add argument.
11223 2017-08-14 Martin Sebor <msebor@redhat.com>
11225 PR translation/79998
11226 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
11227 Remove a stray space.
11229 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
11232 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
11233 (*iordi_1_bts): Ditto.
11234 (*xordi_1_btc): Ditto.
11236 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11239 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
11240 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11242 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
11243 quoted strings, and make more translator-friendly.
11244 (darwin_rs6000_override_options): Likewise.
11245 (rs6000_option_override_internal): Likewise.
11246 (rs6000_return_in_memory): Fix overlong line.
11247 (init_cmulative_args): Use quoted strings, and make more
11248 translator-friendly.
11249 (rs6000_pass_by_reference): Fix overlong line.
11250 (def_builtin): Use quoted strings.
11251 (altivec_expand_predicate_builtin): Use quoted strings, and make
11252 more translator-friendly.
11253 (htm_expand_builtin): Use quoted strings.
11254 (cpu_expand_builtin): Use quoted strings, and make more
11255 translator-friendly.
11256 (altivec_expand_builtin): Likewise.
11257 (paired_expand_predicate_builtin): Likewise.
11258 (rs6000_invalid_builtin): Likewise.
11259 (builtin_function_type): Use quoted strings.
11260 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
11261 more translator-friendly.
11262 (rs6000_trampoline_init): Likewise.
11263 (rs6000_handle_altivec_attribute): Likewise.
11264 (rs6000_inner_target_options): Use quoted strings.
11265 (rs6000_disable_incompatible_switches): Likewise.
11266 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
11267 strings, and make more translator-friendly.
11268 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
11270 2017-08-14 Bin Cheng <bin.cheng@arm.com>
11272 PR tree-optimization/81799
11273 * tree-loop-distribution.c (version_loop_by_alias_check): Force
11274 cond_expr to simple gimple operand.
11276 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
11278 PR middle-end/46932
11279 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
11281 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
11285 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
11286 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
11287 TARGET_GASISR_PROLOGUES.
11288 * config/avr/avr.c (avr_option_override): Same.
11289 (avr_pass_pre_proep::execute): Same.
11291 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
11294 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
11295 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
11297 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
11299 * config/i386/i386.md (*load_tp_<mode>): Redefine as
11300 define_insn_and_split. Split to a memory load from 0 in
11301 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
11302 using PTR mode iterator.
11303 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
11304 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
11305 (*add_tp_<mode>): Redefine as define_insn_and_split.
11306 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
11307 address space. Merge with *add_tp_x32 using PTR mode iterator.
11308 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
11309 Split to an add with a memory load from 0 in
11310 DEFAULT_TLS_SEG_REG address space.
11312 2017-08-12 Andrew Pinski <apinski@cavium.com>
11314 * config/aarch64/aarch64-option-extensions.def (rdma):
11315 Fix feature string to what Linux prints out in /proc/cpuinfo.
11317 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
11320 * dwarf2out.c (modified_type_die): For C typedef types that have
11321 an ultimate origin, process the ultimate origin instead of the
11323 (gen_typedef_die): Assert that input DECLs have no ultimate
11325 (gen_type_die_with_usage): For typedef variants that have an
11326 ultimate origin, just call gen_decl_die on the original DECL.
11327 (process_scope_var): Avoid creating DIEs for local typedefs and
11328 concrete static variables.
11330 2017-08-12 Alan Modra <amodra@gmail.com>
11334 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
11335 match gnu-user.h startfile.
11336 (ENDFILE_LINUX_SPEC): Similarly.
11338 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
11341 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
11343 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
11345 2017-08-11 Tamar Christina <tamar.christina@arm.com>
11346 * config/aarch64/aarch64.md (mov<mode>): Change.
11347 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
11348 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
11349 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
11351 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
11353 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
11354 for storage order barriers.
11356 2017-08-11 Martin Liska <mliska@suse.cz>
11359 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
11360 variables of void type.
11362 2017-08-11 Martin Liska <mliska@suse.cz>
11364 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
11365 TARGET_SUPPORTS_ALIASES.
11366 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
11367 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
11368 (optimize_weakref): Likewise.
11369 * symtab.c (symtab_node::noninterposable_alias): Likewise.
11370 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
11371 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
11373 2017-08-11 Martin Liska <mliska@suse.cz>
11376 * config/i386/i386.c (make_resolver_func): Do complete
11377 refactoring of the function.
11379 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
11382 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
11383 * config/i386/i386.c (ix86_stack_protect_guard): Use
11384 ix86_stack_protect_guard_symbol_str to generate varible declaration.
11385 * doc/invoke.texi (x86 Options): Document
11386 -mstack-protector-guard-symbol= option.
11388 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
11390 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
11391 * config/i386/i386.c (ix86_split_stack_guard): New function.
11392 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
11393 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
11394 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
11395 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
11396 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
11397 (split_stack_space_check): Call ix86_split_stack_guard.
11399 2017-08-10 Martin Sebor <msebor@redhat.com>
11401 * print-tree.c (print_node): Print location using the established
11403 Replace spaces with colons.
11404 (debug_raw, debug): Ditto.
11406 2017-08-10 Martin Sebor <msebor@redhat.com>
11409 * pretty-print.c (pp_format): Correct the handling of %s precision.
11411 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
11414 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
11416 (ix86_finalize_stack_frame_flags): This. Also clear
11417 frame_pointer_needed if -fno-omit-frame-pointer is used without
11419 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
11420 with ix86_finalize_stack_frame_flags.
11421 (ix86_expand_epilogue): Likewise.
11422 (ix86_expand_split_stack_prologue): Likewise.
11423 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
11425 2017-08-10 Martin Liska <mliska@suse.cz>
11428 * c-attribs.c (handle_target_attribute):
11429 Report warning for an empty string argument of target attribute.
11431 2017-08-09 Jakub Jelinek <jakub@redhat.com>
11434 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
11436 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
11437 or DECL_NONLOCAL labels.
11438 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
11439 or DECL_NONLOCAL labels here.
11441 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
11443 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
11444 to indicate when early gimple folding has been disabled.
11445 (rs6000_gimple_fold_builtin): Add debug content.
11446 (rs6000_invalid_builtin): Fix whitespace.
11447 (rs6000_expand_builtin): Fix whitespace.
11448 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
11450 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
11453 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
11454 SAVE_MULTIPLE if not all the registers that saves, should be saved.
11456 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
11458 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
11459 (qdf24xx): Likewise.
11460 * config/aarch64/aarch64.md: Include falkor.md.
11461 * config/aarch64/falkor.md: New.
11463 2017-08-09 Marek Polacek <polacek@redhat.com>
11466 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
11467 * diagnostic.c (emit_diagnostic): Add a comment.
11468 (emit_diagnostic_valist): New function.
11470 2017-08-09 Marek Polacek <polacek@redhat.com>
11473 * input.c (make_location): New overload.
11474 * input.h (make_location): Declare.
11476 2017-08-08 Alan Modra <amodra@gmail.com>
11477 H.J. Lu <hongjiu.lu@intel.com>
11480 * gcc.c (NO_PIE_SPEC): Delete.
11481 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
11483 (LINK_PIE_SPEC): ..to here.
11484 (LINK_COMMAND_SPEC): Support -no-pie.
11485 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
11486 chain of crtbegin*.o selection, update for PIE_SPEC changes and
11488 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
11489 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
11490 (ENDFILE_CRTEND_SPEC): Similarly.
11492 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
11495 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
11496 (mstack-protector-guard-offset=): Ditto.
11497 * config/i386/i386.c (ix86_option_override): Handle
11498 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
11500 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
11501 ix86_stack_protect_guard_offset variables.
11502 (TARGET_STACK_PROTECT_GUARD): Always define.
11503 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
11504 and -mstack-protector-guard-offset= options.
11506 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11508 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
11509 boundary case for the last candidate.
11511 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11513 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
11514 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
11516 2017-08-08 Tamar Christina <tamar.christina@arm.com>
11518 PR middle-end/19706
11519 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
11520 * config/aarch64/aarch64-builtins.c
11521 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
11522 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
11523 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
11525 2017-08-08 Tamar Christina <tamar.christina@arm.com>
11526 Andrew Pinski <pinskia@gmail.com>
11528 PR middle-end/19706
11529 * internal-fn.def (XORSIGN): New.
11530 * optabs.def (xorsign_optab): New.
11531 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
11532 (convert_expand_mult_copysign): New.
11533 (pass_optimize_widening_mul::execute): Call
11534 convert_expand_mult_copysign.
11536 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11538 PR tree-optimization/81354
11539 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
11540 Insert on edges rather than explicitly creating landing pads.
11541 (analyze_candidates_and_replace): Commit edge inserts.
11543 2017-08-08 Richard Biener <rguenther@suse.de>
11545 PR middle-end/81719
11546 * tree-ssa-loop-niter.c: Include tree-dfa.h.
11547 (expand_simple_operations): Also look through ADDR_EXPRs with
11548 MEM_REF bases treating them as POINTER_PLUS_EXPR.
11550 2017-08-08 Richard Biener <rguenther@suse.de>
11552 PR tree-optimization/81723
11553 * tree-vect-slp.c (struct bst_traits): New hash traits.
11554 (bst_fail): New global.
11555 (vect_build_slp_tree_2): New worker, split out from ...
11556 (vect_build_slp_tree): ... this now wrapping it with using
11557 bst_fail set to cache SLP tree build fails. Properly handle
11559 (vect_analyze_slp_instance): Allocate and free bst_fail.
11561 2017-08-08 Martin Liska <mliska@suse.cz>
11564 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
11565 LABEL_DECLs that can be from a different function.
11567 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11569 PR tree-optimization/81744
11570 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
11571 loop's number of iterations.
11573 2017-08-08 Martin Liska <mliska@suse.cz>
11575 * asan.c: Include header files.
11576 * attribs.c (build_decl_attribute_variant): New function moved
11578 (build_type_attribute_qual_variant): Likewise.
11579 (cmp_attrib_identifiers): Likewise.
11580 (simple_cst_list_equal): Likewise.
11581 (omp_declare_simd_clauses_equal): Likewise.
11582 (attribute_value_equal): Likewise.
11583 (comp_type_attributes): Likewise.
11584 (build_type_attribute_variant): Likewise.
11585 (lookup_ident_attribute): Likewise.
11586 (remove_attribute): Likewise.
11587 (merge_attributes): Likewise.
11588 (merge_type_attributes): Likewise.
11589 (merge_decl_attributes): Likewise.
11590 (merge_dllimport_decl_attributes): Likewise.
11591 (handle_dll_attribute): Likewise.
11592 (attribute_list_equal): Likewise.
11593 (attribute_list_contained): Likewise.
11594 * attribs.h (lookup_attribute): New function moved from tree.[ch].
11595 (lookup_attribute_by_prefix): Likewise.
11596 * bb-reorder.c: Include header files.
11597 * builtins.c: Likewise.
11598 * calls.c: Likewise.
11599 * cfgexpand.c: Likewise.
11600 * cgraph.c: Likewise.
11601 * cgraphunit.c: Likewise.
11602 * convert.c: Likewise.
11603 * dwarf2out.c: Likewise.
11604 * final.c: Likewise.
11605 * fold-const.c: Likewise.
11606 * function.c: Likewise.
11607 * gimple-expr.c: Likewise.
11608 * gimple-fold.c: Likewise.
11609 * gimple-pretty-print.c: Likewise.
11610 * gimple.c: Likewise.
11611 * gimplify.c: Likewise.
11612 * hsa-common.c: Likewise.
11613 * hsa-gen.c: Likewise.
11614 * internal-fn.c: Likewise.
11615 * ipa-chkp.c: Likewise.
11616 * ipa-cp.c: Likewise.
11617 * ipa-devirt.c: Likewise.
11618 * ipa-fnsummary.c: Likewise.
11619 * ipa-inline.c: Likewise.
11620 * ipa-visibility.c: Likewise.
11622 * lto-cgraph.c: Likewise.
11623 * omp-expand.c: Likewise.
11624 * omp-general.c: Likewise.
11625 * omp-low.c: Likewise.
11626 * omp-offload.c: Likewise.
11627 * omp-simd-clone.c: Likewise.
11628 * opts-global.c: Likewise.
11629 * passes.c: Likewise.
11630 * predict.c: Likewise.
11631 * sancov.c: Likewise.
11632 * sanopt.c: Likewise.
11633 * symtab.c: Likewise.
11634 * toplev.c: Likewise.
11635 * trans-mem.c: Likewise.
11636 * tree-chkp.c: Likewise.
11637 * tree-eh.c: Likewise.
11638 * tree-into-ssa.c: Likewise.
11639 * tree-object-size.c: Likewise.
11640 * tree-parloops.c: Likewise.
11641 * tree-profile.c: Likewise.
11642 * tree-ssa-ccp.c: Likewise.
11643 * tree-ssa-live.c: Likewise.
11644 * tree-ssa-loop.c: Likewise.
11645 * tree-ssa-sccvn.c: Likewise.
11646 * tree-ssa-structalias.c: Likewise.
11647 * tree-ssa.c: Likewise.
11648 * tree-streamer-in.c: Likewise.
11649 * tree-vectorizer.c: Likewise.
11650 * tree-vrp.c: Likewise.
11651 * tsan.c: Likewise.
11652 * ubsan.c: Likewise.
11653 * varasm.c: Likewise.
11654 * varpool.c: Likewise.
11655 * tree.c: Remove functions moved to attribs.[ch].
11656 * tree.h: Likewise.
11657 * config/aarch64/aarch64.c: Add attrs.h header file.
11658 * config/alpha/alpha.c: Likewise.
11659 * config/arc/arc.c: Likewise.
11660 * config/arm/arm.c: Likewise.
11661 * config/avr/avr.c: Likewise.
11662 * config/bfin/bfin.c: Likewise.
11663 * config/c6x/c6x.c: Likewise.
11664 * config/cr16/cr16.c: Likewise.
11665 * config/cris/cris.c: Likewise.
11666 * config/darwin.c: Likewise.
11667 * config/epiphany/epiphany.c: Likewise.
11668 * config/fr30/fr30.c: Likewise.
11669 * config/frv/frv.c: Likewise.
11670 * config/ft32/ft32.c: Likewise.
11671 * config/h8300/h8300.c: Likewise.
11672 * config/i386/winnt.c: Likewise.
11673 * config/ia64/ia64.c: Likewise.
11674 * config/iq2000/iq2000.c: Likewise.
11675 * config/lm32/lm32.c: Likewise.
11676 * config/m32c/m32c.c: Likewise.
11677 * config/m32r/m32r.c: Likewise.
11678 * config/m68k/m68k.c: Likewise.
11679 * config/mcore/mcore.c: Likewise.
11680 * config/microblaze/microblaze.c: Likewise.
11681 * config/mips/mips.c: Likewise.
11682 * config/mmix/mmix.c: Likewise.
11683 * config/mn10300/mn10300.c: Likewise.
11684 * config/moxie/moxie.c: Likewise.
11685 * config/msp430/msp430.c: Likewise.
11686 * config/nds32/nds32-isr.c: Likewise.
11687 * config/nds32/nds32.c: Likewise.
11688 * config/nios2/nios2.c: Likewise.
11689 * config/nvptx/nvptx.c: Likewise.
11690 * config/pa/pa.c: Likewise.
11691 * config/pdp11/pdp11.c: Likewise.
11692 * config/powerpcspe/powerpcspe.c: Likewise.
11693 * config/riscv/riscv.c: Likewise.
11694 * config/rl78/rl78.c: Likewise.
11695 * config/rx/rx.c: Likewise.
11696 * config/s390/s390.c: Likewise.
11697 * config/sh/sh.c: Likewise.
11698 * config/sol2.c: Likewise.
11699 * config/sparc/sparc.c: Likewise.
11700 * config/spu/spu.c: Likewise.
11701 * config/stormy16/stormy16.c: Likewise.
11702 * config/tilegx/tilegx.c: Likewise.
11703 * config/tilepro/tilepro.c: Likewise.
11704 * config/v850/v850.c: Likewise.
11705 * config/vax/vax.c: Likewise.
11706 * config/visium/visium.c: Likewise.
11707 * config/xtensa/xtensa.c: Likewise.
11709 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11712 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
11713 constraints since the -mupper-regs-* switches have been
11715 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
11716 into a vector from a double word element that was extracted from
11717 another vector, and eliminate extra XXPERMDI instructions.
11718 (vsx_concat_<mode>_2): Likewise.
11719 (vsx_concat_<mode>_3): Likewise.
11720 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
11721 concat to allow optimizing inserts from previous extracts.
11723 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
11725 * config/i386/i386.c (ix86_stack_protect_guard): Generate
11726 memory reference to a SSP offset in TLS address space.
11727 (ix86_print_operand) <case '@'>: Remove.
11728 (ix86_print_operand_punct_valid_p): Remove '@' code.
11729 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
11730 UNSPEC_SP_TLS_TEST.
11731 (stack_tls_protect_set_<mode>): Remove.
11732 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
11733 (stack_tls_protect_test_<mode>): Remove.
11734 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
11736 2017-08-07 Olivier Hainque <hainque@adacore.com>
11739 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
11741 2017-08-07 Douglas Rupp <rupp@adacore.com>
11743 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
11744 variable was referenced as multidir in command.
11746 2017-08-07 Jakub Jelinek <jakub@redhat.com>
11749 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
11752 2017-08-07 Martin Liska <mliska@suse.cz>
11754 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
11755 * config/rl78/rl78.c: Add include of attribs.h.
11756 * config/sh/sh.c: Likewise.
11757 * config/v850/v850.c: Likewise.
11759 2017-08-07 Tom de Vries <tom@codesourcery.com>
11761 PR middle-end/78266
11762 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
11764 2017-08-07 Martin Liska <mliska@suse.cz>
11766 * config/mips/mips.c: Include attribs.h.
11768 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
11771 * doc/invoke.texi: Document change in behvaior for -Ofast for
11774 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
11776 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
11778 (aarch64_pop_regs): Likewise.
11779 (aarch64_gen_load_pair): Likewise.
11780 (aarch64_save_callee_saves): Likewise.
11781 (aarch64_restore_callee_saves): Likewise.
11783 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11785 * config/i386/i386.c: Revert the last change.
11787 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11790 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
11792 (ix86_finalize_stack_frame_flags): This. Also clear
11793 frame_pointer_needed if -fno-omit-frame-pointer is used without
11795 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
11796 with ix86_finalize_stack_frame_flags.
11797 (ix86_expand_epilogue): Likewise.
11798 (ix86_expand_split_stack_prologue): Likewise.
11800 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11803 * config/i386/i386.c (get_builtin_code_for_version): Set priority
11804 to P_AES for Westmere.
11806 2017-08-07 Jonathan Yong <10walls@gmail.com>
11808 * config/i386/mingw.opt (fset-stack-executable): Removed.
11809 * config/i386/cygming.opt (fset-stack-executable): Moved
11811 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
11813 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
11815 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
11817 2017-08-07 Marek Polacek <polacek@redhat.com>
11819 PR middle-end/81737
11820 * fold-const.c (fold_indirect_ref_1): Check type_domain.
11822 2017-08-07 Martin Liska <mliska@suse.cz>
11824 * attribs.h (canonicalize_attr_name): New function.
11825 (cmp_attribs): Move from c-format.c and adjusted.
11826 (is_attribute_p): Moved from tree.h.
11827 * tree-inline.c: Add new includes.
11828 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
11829 (private_is_attribute_p): Remove.
11830 (private_lookup_attribute): Likewise.
11831 (private_lookup_attribute_by_prefix): Simplify.
11832 (remove_attribute): Use is_attribute_p.
11833 * tree.h: Remove removed declarations.
11835 2017-08-07 Jakub Jelinek <jakub@redhat.com>
11837 PR middle-end/81698
11838 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
11839 instead of computing it in the function. Formatting fix.
11840 (expand_case): Don't rely on default_edge being the first edge,
11841 clear it if removing it, pass default_edge to
11842 emit_case_dispatch_table.
11843 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
11846 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
11848 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
11849 insn in the function, emit NOP after the insn.
11851 2017-08-06 Tom de Vries <tom@codesourcery.com>
11853 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
11856 2017-08-06 Tom de Vries <tom@codesourcery.com>
11858 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
11861 2017-08-04 Yury Gribov <tetra2005@gmail.com>
11863 PR tree-optimization/57371
11864 * match.pd: New pattern.
11866 2017-08-04 Marek Polacek <polacek@redhat.com>
11868 PR middle-end/81695
11869 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
11870 perform the computation in offset_int.
11872 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11874 PR tree-optimization/81136
11875 * tree-vectorizer.h: Include tree-hash-traits.h.
11876 (vec_base_alignments): New typedef.
11877 (vec_info): Add a base_alignments field.
11878 (vect_record_base_alignments): Declare.
11879 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
11881 (DR_IS_CONDITIONAL_IN_STMT): New macro.
11882 (create_data_ref): Add an is_conditional_in_stmt argument.
11883 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
11884 the is_conditional_in_stmt field.
11885 (data_ref_loc): Add an is_conditional_in_stmt field.
11886 (get_references_in_stmt): Set the is_conditional_in_stmt field.
11887 (find_data_references_in_stmt): Update call to create_data_ref.
11888 (graphite_find_data_references_in_stmt): Likewise.
11889 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
11890 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
11891 (vect_record_base_alignment): New function.
11892 (vect_record_base_alignments): Likewise.
11893 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
11894 for nested statements even if we fail to compute a misalignment.
11895 Use pooled base alignments for unconditional references.
11896 (vect_find_same_alignment_drs): Compare base addresses instead
11898 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
11899 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
11901 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11903 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
11904 Add an explicit name for the enum. Use auto_vec for slp_instances
11905 and grouped_stores.
11906 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
11908 (_bb_vec_info): Add a constructor and destructor.
11909 (vinfo_for_stmt): Return NULL for uids of -1 as well.
11910 (destroy_loop_vec_info): Delete.
11911 (vect_destroy_datarefs): Likewise.
11912 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
11913 (vec_info::vec_info): New function.
11914 (vec_info::~vec_info): Likewise.
11915 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
11916 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
11917 destroy_loop_vec_info.
11918 * tree-vect-loop.c (new_loop_vec_info): Replace with...
11919 (_loop_vec_info::_loop_vec_info): ...this.
11920 (destroy_loop_vec_info): Replace with...
11921 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
11922 the stmt_vec_infos. Leave handling of vec_info information to its
11923 destructor. Remove explicit vector releases.
11924 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
11925 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
11926 * tree-vect-slp.c (new_bb_vec_info): Replace with...
11927 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
11928 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
11929 (destroy_bb_vec_info): Replace with...
11930 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
11931 information to its destructor.
11932 (vect_slp_analyze_bb_1): Use new and delete instead of
11933 new_bb_vec_info and destroy_bb_vec_info.
11934 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
11937 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11939 * tree-data-ref.h (subscript): Add access_fn field.
11940 (data_dependence_relation): Add could_be_independent_p.
11941 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
11942 (same_access_functions): Move to tree-data-ref.c.
11943 * tree-data-ref.c (ref_contains_union_access_p): New function.
11944 (access_fn_component_p): Likewise.
11945 (access_fn_components_comparable_p): Likewise.
11946 (dr_analyze_indices): Add a reference to access_fn_component_p.
11947 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
11949 (constant_access_functions): Likewise.
11950 (add_other_self_distances): Likewise.
11951 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
11952 (initialize_data_dependence_relation): Use XCNEW and remove
11953 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
11954 of access functions that have the same type. Allow the
11955 subsequence to end with different bases in some circumstances.
11956 Record the chosen access functions in SUB_ACCESS_FN.
11957 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
11958 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
11959 (subscript_dependence_tester_1): Likewise dra and drb.
11960 (build_classic_dist_vector): Update calls accordingly.
11961 (subscript_dependence_tester): Likewise.
11962 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
11963 DDR_COULD_BE_INDEPENDENT_P.
11964 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
11965 comp_alias_ddrs instead of may_alias_ddrs.
11966 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
11968 (vect_analyze_data_ref_dependence): Use it if
11969 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
11970 distance vectors if that fails.
11971 (dependence_distance_ge_vf): New function.
11972 (vect_prune_runtime_alias_test_list): Use it. Don't clear
11973 LOOP_VINFO_MAY_ALIAS_DDRS.
11975 2017-08-04 Richard Biener <rguenther@suse.de>
11977 PR middle-end/81705
11978 * fold-const.c (fold_binary_loc): Properly restrict
11979 minus_var0 && minus_var1 case when associating undefined overflow
11982 2017-08-04 Tom de Vries <tom@codesourcery.com>
11984 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
11986 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11988 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11989 Don't start diagnostic messages with a capital letter.
11990 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11992 (rs6000_invalid_builtin): Likewise.
11993 (rs6000_trampoline_init): Likewise.
11995 2017-08-03 Jakub Jelinek <jakub@redhat.com>
11998 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
11999 after setting changeable df flags.
12001 2017-08-03 Richard Biener <rguenther@suse.de>
12003 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
12004 up if the use is in USE - X.
12006 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
12008 * toplev.c (dumpfile.h): New include.
12009 (internal_error_reentered): New static function. Use it...
12010 (internal_error_function): ...here to handle reentered internal_error.
12012 2017-08-03 Richard Biener <rguenther@suse.de>
12014 PR middle-end/81148
12015 * fold-const.c (split_tree): Add minus_var and minus_con
12016 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
12017 here but always use minus_*.
12018 (associate_trees): Assert we never associate with MINUS_EXPR
12019 and NULL first operand. Do not recurse for PLUS_EXPR operands
12020 when associating as MINUS_EXPR either.
12021 (fold_binary_loc): Track minus_var and minus_con.
12023 2017-08-03 Tom de Vries <tom@codesourcery.com>
12026 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
12027 ACCEL_COMPILER, apply finish_options on
12028 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
12030 2017-08-03 Tom de Vries <tom@codesourcery.com>
12033 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
12034 function_entry_patch_area_size > 0.
12036 2017-08-03 Jakub Jelinek <jakub@redhat.com>
12039 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
12040 instead of 10??LU, perform unit multiplication in wide_int,
12041 don't change alloc_object_size_limit if the limit is larger
12044 PR tree-optimization/81655
12045 PR tree-optimization/81588
12046 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
12047 the case when ranges[i].low and high are 1 for unsigned type with
12050 PR middle-end/81052
12051 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
12052 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
12054 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12056 * tree-vrp.h: Add include guard.
12058 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
12061 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
12062 (ud2): New insn pattern.
12063 * config/i386/i386.c (ix86_expand_epilogue):
12064 For naked functions, generate ud2 instead of trap insn.
12066 2017-08-02 Marek Polacek <polacek@redhat.com>
12069 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
12071 2017-08-02 Tom de Vries <tom@codesourcery.com>
12072 Cesar Philippidis <cesar@codesourcery.com>
12074 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
12075 Add missing edge probabilities.
12077 2017-08-02 Tamar Christina <tamar.christina@arm.com>
12079 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
12080 Correct endianness.
12082 2017-08-02 Jakub Jelinek <jakub@redhat.com>
12084 PR middle-end/79499
12085 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
12086 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
12087 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
12088 prologue_seq sequences - if any.
12090 2017-08-02 Richard Biener <rguenther@suse.de>
12092 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
12093 via vectors if supported, integer extracts via punning if supported
12094 or otherwise vector extracts.
12096 2017-08-02 Richard Biener <rguenther@suse.de>
12098 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
12100 (bitmap_insert_into_set): ... this.
12102 2017-08-02 Richard Biener <rguenther@suse.de>
12104 PR tree-optimization/81633
12106 2015-08-17 Alan Hayward <alan.hayward@arm.com>
12108 PR tree-optimization/71752
12109 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
12111 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
12113 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
12114 (machine_function::call_ms2sysv_pad_out): Remove field.
12115 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
12116 (ix86_compute_frame_layout): Likewise.
12118 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
12121 * config/i386/i386.c (ix86_set_func_type): Disallow naked
12122 attribute with interrupt attribute.
12124 2017-08-01 Andrew Pinski <apinski@cavium.com>
12126 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
12127 BIT_INSERT_EXPR's operand 1
12128 to see if the types precision matches.
12130 2017-08-01 Martin Liska <mliska@suse.cz>
12132 PR middle-end/70140
12133 * builtins.c (expand_builtin_memcpy_args): Remove.
12134 (expand_builtin_memcpy): Call newly added function
12135 expand_builtin_memory_copy_args.
12136 (expand_builtin_memcpy_with_bounds): Likewise.
12137 (expand_builtin_mempcpy): Remove last argument.
12138 (expand_builtin_mempcpy_with_bounds): Likewise.
12139 (expand_builtin_memory_copy_args): New function created from
12140 expand_builtin_mempcpy_args with small modifications.
12141 (expand_builtin_mempcpy_args): Remove.
12142 (expand_builtin_stpcpy): Remove unused argument.
12143 (expand_builtin): Likewise.
12144 (expand_builtin_with_bounds): Likewise.
12146 2017-08-01 Martin Liska <mliska@suse.cz>
12149 Make mempcpy more optimal (PR middle-end/70140).
12151 2017-08-01 Jakub Jelinek <jakub@redhat.com>
12154 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12155 __builtin_vec_cmpne verify both arguments are compatible vectors
12156 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
12157 verify arg1_type is a pointer or array type. For __builtin_vec_st,
12158 move computation of aligned to after checking the argument types.
12162 * config/rs6000/vsx.md (vextract_fp_from_shorth,
12163 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
12166 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
12167 Doug Rupp <rupp@adacore.com>
12168 Olivier Hainque <hainque@adacore.com>
12170 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
12171 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
12173 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
12174 for TARGET_OS_CPP_BUILTIN.
12175 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
12176 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
12178 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
12179 passing required abi options to the assembler for EABI configurations.
12180 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
12181 of .text.hot and .text.unlikely sections for kernel modules when
12182 using ARM style exceptions.
12183 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
12184 options. Add EXTRA_CC1_SPEC.
12185 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
12187 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
12189 (ARM_TARGET2_DWARF_FORMAT): Define.
12190 * config/arm/t-vxworks: Adjust multilib control to removal of the
12191 Diab command line options.
12193 2017-08-01 Martin Liska <mliska@suse.cz>
12195 PR gcov-profile/81561
12196 * gcov.c (unblock): Make unblocking safe as we need to preserve
12197 index correspondence of blocks and block_lists.
12199 2017-08-01 Richard Biener <rguenther@suse.de>
12201 PR tree-optimization/81181
12202 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
12203 (compute_antic): ... end of iteration here.
12205 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
12207 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
12208 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
12209 (ftree-slp-vectorize): Likewise.
12210 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
12211 can no longer be set independent of flag_tree_loop_vectorize.
12212 * omp-general.c (emp_max_vf): Likewise.
12213 * opts.c (enable_fdo_optimizations): Remove references to
12214 flag_tree_vectorize, these are now implicit.
12215 (common_handle_option): Remove handling for OPT_ftree_vectorize,
12216 and leave it for the options machinery.
12218 2017-08-01 Martin Liska <mliska@suse.cz>
12220 PR middle-end/70140
12221 * builtins.c (expand_builtin_memcpy_args): Remove.
12222 (expand_builtin_memcpy): Call newly added function
12223 expand_builtin_memory_copy_args.
12224 (expand_builtin_memcpy_with_bounds): Likewise.
12225 (expand_builtin_mempcpy): Remove last argument.
12226 (expand_builtin_mempcpy_with_bounds): Likewise.
12227 (expand_builtin_memory_copy_args): New function created from
12228 expand_builtin_mempcpy_args with small modifications.
12229 (expand_builtin_mempcpy_args): Remove.
12230 (expand_builtin_stpcpy): Remove unused argument.
12231 (expand_builtin): Likewise.
12232 (expand_builtin_with_bounds): Likewise.
12234 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
12237 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
12238 print "ds:" only for immediates in generic address space.
12240 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
12243 * config/i386/i386.c (ix86_funciton_naked): New prototype.
12244 (ix86_function_ok_for_sibcall): Return false for naked functions.
12246 2017-08-01 Richard Biener <rguenther@suse.de>
12248 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
12249 (compute_antic): Seed worklist with exit block predecessors.
12250 * cfganal.c (dfs_find_deadend): For a cycle return the source
12251 of the edge closing it.
12253 2017-08-01 Tamar Christina <tamar.christina@arm.com>
12255 * config/aarch64/aarch64.c
12256 (aarch64_can_const_movi_rtx_p): Move 0 check.
12258 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12260 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
12261 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
12263 * match.pd: Ditto in address comparison pattern.
12265 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12267 PR tree-optimization/81627
12268 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
12269 closed ssa form for store-store chain.
12271 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12273 PR tree-optimization/81620
12274 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
12275 for store-store chain.
12277 2017-08-01 Jakub Jelinek <jakub@redhat.com>
12279 PR tree-optimization/81588
12280 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
12281 ranges[i].in_p, invert comparison code ccode. For >/>=,
12282 swap rhs1 and rhs2 and comparison code unconditionally,
12283 for </<= don't do that. Don't swap rhs1/rhs2 again if
12284 ranges[i].in_p, instead invert comparison code ccode if
12285 opcode or oe->rank is BIT_IOR_EXPR.
12288 * optabs.def (vec_extract_optab, vec_init_optab): Change from
12289 a direct optab to conversion optab.
12290 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
12291 with GET_MODE_INNER as last argument instead of optab_handler.
12292 * expmed.c (extract_bit_field_1): Likewise. Use vector from
12293 vector extraction if possible and optab is available.
12294 * expr.c (store_constructor): Use convert_optab_handler instead
12295 of optab_handler. Use vector initialization from smaller
12296 vectors if possible and optab is available.
12297 * tree-vect-stmts.c (vectorizable_load): Likewise.
12298 * doc/md.texi (vec_extract, vec_init): Document that the optabs
12299 now have two modes.
12300 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
12301 of vec_init from half-sized vectors with the same element mode.
12302 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
12303 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
12304 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
12305 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
12306 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
12307 after mode in gen_vec_extract* calls.
12308 (vec_extract<mode>): Renamed to ...
12309 (vec_extract<mode><ssescalarmodelower>): ... this.
12310 (vec_extract<mode><ssehalfvecmodelower>): New expander.
12311 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
12312 element mode after mode in gen_vec_init* calls.
12313 (VEC_INIT_HALF_MODE): New mode iterator.
12314 (vec_init<mode>): Renamed to ...
12315 (vec_init<mode><ssescalarmodelower>): ... this.
12316 (vec_init<mode><ssehalfvecmodelower>): New expander.
12317 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
12318 (vec_extractv2sfsf): ... this.
12319 (vec_initv2sf): Renamed to ...
12320 (vec_initv2sfsf): ... this.
12321 (vec_extractv2si): Renamed to ...
12322 (vec_extractv2sisi): ... this.
12323 (vec_initv2si): Renamed to ...
12324 (vec_initv2sisi): ... this.
12325 (vec_extractv4hi): Renamed to ...
12326 (vec_extractv4hihi): ... this.
12327 (vec_initv4hi): Renamed to ...
12328 (vec_initv4hihi): ... this.
12329 (vec_extractv8qi): Renamed to ...
12330 (vec_extractv8qiqi): ... this.
12331 (vec_initv8qi): Renamed to ...
12332 (vec_initv8qiqi): ... this.
12333 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
12334 (vec_init<mode>): Renamed to ...
12335 (vec_init<mode><VEC_base_l>): ... this.
12336 (vec_extract<mode>): Renamed to ...
12337 (vec_extract<mode><VEC_base_l>): ... this.
12338 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
12339 (vec_initv2sfsf): ... this.
12340 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
12341 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
12342 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
12343 element mode after mode in gen_vec_init* calls.
12344 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
12345 (vec_init<mode><Vel>): ... this.
12346 (vec_extract<mode>): Renamed to ...
12347 (vec_extract<mode><Vel>): ... this.
12348 * config/aarch64/iterators.md (Vel): New mode attribute.
12349 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
12350 Add element mode after mode in gen_vec_extract* calls.
12351 * config/s390/vector.md (non_vec_l): New mode attribute.
12352 (vec_extract<mode>): Renamed to ...
12353 (vec_extract<mode><non_vec_l>): ... this.
12354 (vec_init<mode>): Renamed to ...
12355 (vec_init<mode><non_vec_l>): ... this.
12356 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
12357 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
12359 * config/arm/iterators.md (V_elem_l): New mode attribute.
12360 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
12361 (vec_extract<mode><V_elem_l>): ... this.
12362 (vec_extractv2di): Renamed to ...
12363 (vec_extractv2didi): ... this.
12364 (vec_init<mode>): Renamed to ...
12365 (vec_init<mode><V_elem_l>): ... this.
12366 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
12367 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
12368 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
12369 Add element mode after gen_vec_extract* calls.
12370 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
12371 (vec_init<mode><unitmode>): ... this.
12372 (vec_extract<mode>): Renamed to ...
12373 (vec_extract<mode><unitmode>): ... this.
12374 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
12375 (vec_init<mode><unitmode>): ... this.
12376 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
12377 (vec_initv2sfsf): ... this.
12378 (vec_extractv2sf): Renamed to ...
12379 (vec_extractv2sfsf): ... this.
12380 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
12381 Add element mode after gen_vec_extract* calls.
12382 * config/mips/mips.md (unitmode): New mode iterator.
12383 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
12384 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
12385 * config/spu/spu.md (inner_l): New mode attribute.
12386 (vec_init<mode>): Renamed to ...
12387 (vec_init<mode><inner_l>): ... this.
12388 (vec_extract<mode>): Renamed to ...
12389 (vec_extract<mode><inner_l>): ... this.
12390 * config/sparc/sparc.md (veltmode): New mode iterator.
12391 (vec_init<VMALL:mode>): Renamed to ...
12392 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
12393 * config/ia64/vect.md (vec_initv2si): Renamed to ...
12394 (vec_initv2sisi): ... this.
12395 (vec_initv2sf): Renamed to ...
12396 (vec_initv2sfsf): ... this.
12397 (vec_extractv2sf): Renamed to ...
12398 (vec_extractv2sfsf): ... this.
12399 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
12400 (vec_init<mode>): Renamed to ...
12401 (vec_init<mode><VEC_base_l>): ... this.
12402 (vec_extract<mode>): Renamed to ...
12403 (vec_extract<mode><VEC_base_l>): ... this.
12404 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
12405 (vec_initv2sfsf): ... this.
12406 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
12407 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
12408 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
12409 gen_vec_init* calls.
12411 2017-08-01 Richard Biener <rguenther@suse.de>
12413 PR tree-optimization/81297
12414 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
12415 TREE_OVERFLOW from INTEGER_CSTs.
12417 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
12419 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
12421 2017-07-31 Carl Love <cel@us.ibm.com>
12423 * config/rs6000/rs6000-c: Add support for built-in functions
12424 vector signed char vec_xl_be (signed long long, signed char *);
12425 vector unsigned char vec_xl_be (signed long long, unsigned char *);
12426 vector signed int vec_xl_be (signed long long, signed int *);
12427 vector unsigned int vec_xl_be (signed long long, unsigned int *);
12428 vector signed long long vec_xl_be (signed long long, signed long long *);
12429 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
12430 vector signed short vec_xl_be (signed long long, signed short *);
12431 vector unsigned short vec_xl_be (signed long long, unsigned short *);
12432 vector double vec_xl_be (signed long long, double *);
12433 vector float vec_xl_be (signed long long, float *);
12434 * config/rs6000/altivec.h (vec_xl_be): Add #define.
12435 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
12436 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
12438 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
12439 (altivec_expand_builtin): Add switch statement to call
12440 altivec_expand_xl_be for each builtin.
12441 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
12442 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
12443 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
12444 __builtin_vsx_le_be_v16qi.
12445 * doc/extend.texi: Update the built-in documentation file for the
12446 new built-in functions.
12448 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
12451 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
12453 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
12455 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12457 * config.gcc: Add z14.
12458 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
12459 CPU model numbers for z13s and z14.
12460 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
12462 * config/s390/s390-opts.h (enum processor_type): Rename
12463 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
12464 * config/s390/s390.c (processor_table): Add field for CPU name to
12465 be passed to Binutils.
12466 (s390_asm_output_machine_for_arch): Use the new field in
12467 processor_table for Binutils.
12468 (s390_expand_builtin): Replace arch12 with z14.
12469 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
12470 (s390_get_sched_attrmask): Likewise.
12471 (s390_get_unit_mask): Likewise.
12472 * config/s390/s390.opt: Add z14 to processor_type enum.
12474 2017-07-31 Martin Jambor <mjambor@suse.cz>
12477 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
12478 regardless of optimization level.
12480 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
12481 Martin Liska <mliska@suse.cz>
12483 * predict.def: Remove old comment and adjust probability.
12484 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
12485 PREDICT statements.
12487 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
12490 * config/i386/i386.c (ix86_function_naked): New function.
12491 (ix86_can_use_return_insn_p): Return false for naked functions.
12492 (ix86_expand_prologue): Skip prologue for naked functions.
12493 (ix86_expand_epilogue): Skip epilogue for naked functions
12494 and emit trap instruction.
12495 (ix86_warn_func_return): New function.
12496 (ix86_attribute_table): Add "naked" attribute specification.
12497 (TARGET_WARN_FUNC_RETURN): Define.
12498 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
12500 2017-07-31 Martin Liska <mliska@suse.cz>
12502 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
12503 (dump_gimple_bb_header): Always dump BB info.
12504 (pp_cfg_jump): Do not append info about BB when dumping a jump.
12506 2017-07-31 Martin Liska <mliska@suse.cz>
12509 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
12510 also with current_function_decl non-null equality.
12512 2017-07-31 Jakub Jelinek <jakub@redhat.com>
12515 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
12516 change type to the element type, instead add eltype variable and
12517 use it where we are interested in the element type.
12519 PR tree-optimization/81603
12520 * ipa-polymorphic-call.c
12521 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
12522 offset arithmetic in offset_int, bail out if the resulting bit offset
12523 doesn't fit into shwi.
12525 2017-07-31 Martin Liska <mliska@suse.cz>
12527 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
12528 (gimplify_save_expr): Fix comment.
12530 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
12533 * config/i386/i386.c (ix86_function_arg): Update arguments for
12535 (ix86_compute_frame_layout): Set the initial stack offset to
12536 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
12537 INCOMING_FRAME_SP_OFFSET.
12538 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
12539 stack before exception handler returns.
12540 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
12541 the 'ERROR_CODE' for exception handler.
12543 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
12545 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
12546 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
12547 (ASM_OUTPUT_REG_POP): Ditto.
12548 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
12549 instead of asm_fprintf to output pure string.
12551 2017-07-29 Jakub Jelinek <jakub@redhat.com>
12553 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
12554 to imported_module_or_decl hook.
12555 (debug_nothing_tree_tree_tree_bool): Remove.
12556 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
12557 * debug.c (do_nothing_debug_hooks): Use
12558 debug_nothing_tree_tree_tree_bool_bool instead of
12559 debug_nothing_tree_tree_tree_bool.
12560 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12561 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
12562 * sdbout.c (sdb_debug_hooks): Likewise.
12563 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
12564 (gen_namespace_die): Add DW_AT_export_symbols attribute if
12566 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
12567 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
12568 attribute, don't add anything.
12570 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12572 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
12573 (fold_build2_stat_loc): Likewise.
12574 (fold_build3_stat_loc): Likewise.
12575 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
12576 (fold_build1_loc): Remove macro.
12577 (fold_build2_loc): Likewise.
12578 (fold_build3_loc): Likewise.
12580 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12582 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
12583 (gimple_build_debug_bind_source_stat): Likewise.
12584 * gimple.h (gimple_build_debug_bind): Remove macro.
12585 (gimple_build_debug_bind_source): Likewise.
12587 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12589 * bitmap.c (bitmap_alloc): Adjust.
12590 (bitmap_gc_alloc): Likewise.
12591 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
12593 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12595 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
12596 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
12597 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
12598 (bitmap_gc_alloc_stat): Likewise.
12599 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
12601 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12603 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
12604 * rtl.h (shallow_copy_rtx): Remove macro.
12606 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12608 * emit-rtl.c (gen_raw_REG): Adjust.
12609 * gengenrtl.c (gendef): Likewise.
12610 * rtl.c (rtx_alloc_stat): Remove _stat from name.
12611 * rtl.h (rtx_alloc): Remove macro.
12613 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12615 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
12616 (build_tree_list_stat): Likewise.
12617 * tree.h (build_tree_list): Remove macro.
12618 (build_tree_list_vec): Likewise.
12620 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12622 * tree.c (make_vector_stat): Remove _stat from name.
12623 (build_vector_stat): Likewise.
12624 * tree.h (make_vector_stat): Remove macro.
12625 (build_vector_stat): Likewise.
12627 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12629 * tree.h (build_var_debug_value): Remove prototype.
12631 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12633 * tree.c (tree_cons_stat): Remove _stat from name.
12634 * tree.h (tree_cons): Remove macro.
12636 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12638 * tree.c (build_vl_exp_stat): Remove _stat from name.
12639 * tree.h (build_vl_exp): Remove macro.
12641 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12643 * tree.c (build_decl_stat): Remove _stat from name.
12644 * tree.h (build_decl): Remove macro.
12646 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12648 * gimple.c (gimple_build_with_ops_stat): Adjust.
12649 (gimple_alloc_stat): Remove _stat from name.
12650 * gimple.h (gimple_alloc): Remove macro.
12652 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12654 * tree.c (make_tree_vec_stat): Remove _stat from name.
12655 (grow_tree_vec_stat): Likewise.
12656 * tree.h (make_tree_vec_stat): Adjust prototype.
12657 (grow_tree_vec_stat): Likewise.
12658 (make_tree_vec): Remove macro.
12659 (grow_tree_vec): Likewise.
12661 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12663 * fold-const.c (fold_build1_stat_loc): Adjust.
12664 (fold_build2_stat_loc): Likewise.
12665 (fold_build3_stat_loc): Likewise.
12666 * tree.c (build0_stat): Remove _stat from name.
12667 (build1_stat): Likewise.
12668 (build2_stat): Likewise.
12669 (build3_stat): Likewise.
12670 (build4_stat): Likewise.
12671 (build5_stat): Likewise.
12672 * tree.h (build1_loc): Remove macro, and rename _stat function
12674 (build2_loc): Likewise.
12675 (build3_loc): Likewise.
12676 (build4_loc): Likewise.
12677 (build5_loc): Likewise.
12679 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12681 * tree.c (make_int_cst_stat): Remove _stat from name.
12682 * tree.h (make_int_cst_stat): Adjust prototype.
12683 (make_int_cst): Remove macro.
12685 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12687 * tree.c (make_tre_binfo_stat): Remove _stat from name.
12688 * tree.h (make_tree_binfo_stat): Adjust prototype.
12689 (make_tree_binfo): Remove.
12691 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12693 * tree.c (copy_node_stat): Rename to copy_node.
12694 (build_distinct_type_copy): Adjust.
12695 * tree.h (copy_node_stat): Adjust prototype.
12696 (copy_node): Remove macro.
12698 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12700 * tree.c (make_node_stat): rename to make_node.
12701 (build_tree_list_stat): Adjust.
12702 (build0_stat): Likewise.
12703 (build2_stat): Likewise.
12704 (build3_stat): Likewise.
12705 (build4_stat): Likewise.
12706 (build5_stat): Likewise.
12707 (build_decl_stat): Likewise.
12708 * tree.h (make_node_stat): Adjust prototype.
12709 (make_node): remove macro.
12711 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
12713 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
12714 (PPC_FEATURE2_SCV): Likewise.
12715 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
12717 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12719 * config/aarch64/aarch64.c
12720 (aarch64_internal_mov_immediate): Add new special pattern.
12721 * config/aarch64/aarch64.md (*movdi_aarch64):
12722 Add reg/32bit const mov case.
12724 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12725 Richard Sandiford <richard.sandiford@linaro.org>
12727 * config/aarch64/aarch64.md (mov<mode>): Generalize.
12728 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
12729 Add integer and movi cases.
12730 (movi-split-hf-df-sf split, fp16): New.
12731 (enabled): Added TARGET_FP_F16INST.
12732 * config/aarch64/iterators.md (GPF_HF): New.
12733 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
12735 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12737 * config/aarch64/aarch64.c
12738 (aarch64_simd_container_mode): Add prototype.
12739 (aarch64_expand_mov_immediate): Add HI support.
12740 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
12741 (aarch64_can_const_movi_rtx_p): New.
12742 (aarch64_preferred_reload_class):
12743 Remove restrictions of using FP registers for certain SIMD operations.
12744 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
12745 (aarch64_valid_floating_const): Add integer move validation.
12746 (aarch64_simd_imm_scalar_p): Remove.
12747 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
12748 (aarch64_legitimate_constant_p): Expand list of supported cases.
12749 * config/aarch64/aarch64-protos.h
12750 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
12751 (aarch64_reinterpret_float_as_int): New.
12752 (aarch64_simd_imm_scalar_p): Remove.
12753 * config/aarch64/constraints.md (Uvi): New.
12754 (Dd): Split into Ds and new Dd.
12755 * config/aarch64/aarch64.md (*movsi_aarch64):
12757 (*movdi_aarch64): Add SIMD mov case.
12759 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12761 * tree-predcom.c: (struct chain): Handle store-store chain in which
12762 stores for elimination only store loop invariant values.
12763 (execute_pred_commoning_chain): Ditto.
12764 (prepare_initializers_chain_store_elim): Ditto.
12765 (prepare_finalizers): Ditto.
12766 (is_inv_store_elimination_chain): New function.
12767 (initialize_root_vars_store_elim_1): New function.
12769 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12771 * tree-predcom.c: Revise general description of the pass.
12772 (enum chain_type): New enum type for store elimination.
12773 (struct chain): New field supporting store elimination.
12774 (struct component): Ditto.
12775 (dump_chain): Dump store-stores chain.
12776 (release_chain): Release resources.
12777 (split_data_refs_to_components): Compute and create component
12778 contains only stores for elimination.
12779 (get_chain_last_ref_at): New function.
12780 (make_invariant_chain): Initialization.
12781 (make_rooted_chain): Specify chain type in parameter and record it.
12782 (add_looparound_copies): Skip for store-stores chain.
12783 (determine_roots_comp): Compute type of chain and pass it to
12785 (initialize_root_vars_store_elim_2): New function.
12786 (finalize_eliminated_stores): New function.
12787 (remove_stmt): Handle store for elimination.
12788 (execute_pred_commoning_chain): Execute predictive commoning on
12789 store-store chains.
12790 (determine_unroll_factor): Skip unroll for store-stores chain.
12791 (prepare_initializers_chain_store_elim): New function.
12792 (prepare_initializers_chain): Hanlde store-store chain.
12793 (prepare_finalizers_chain, prepare_finalizers): New function.
12794 (tree_predictive_commoning_loop): Return integer value indicating
12795 if loop is unrolled or lcssa form is corrupted.
12796 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
12798 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12800 * tree-predcom.c (initialize_root): Delete.
12801 (execute_pred_commoning_chain): Initialize root vars and replace
12802 reference of non-combined chain directly, rather than call above
12805 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12807 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
12808 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
12810 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12812 * tree-predcom.c (struct chain): New field init_seq.
12813 (release_chain): Release init_seq.
12814 (prepare_initializers_chain): Record intialization stmts in above
12816 (insert_init_seqs): New function.
12817 (tree_predictive_commoning_loop): Call insert_init_seqs.
12819 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12821 * tree-predcom.c (determine_roots_comp): Skip trivial components.
12823 2017-07-28 Richard Biener <rguenther@suse.de>
12825 * match.pd: Remove superfluous :c.
12826 * genmatch.c (simplify::id): Add member.
12827 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
12829 (current_id): New global.
12830 (dt_node::parent): Move from ...
12831 (dt_operand::parent): ... here. Add for_id member.
12832 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
12833 (decision_tree::find_node): Relax order requirement when
12834 merging DT_TRUE nodes to ones inbetween the current simplify
12835 and the one we try to merge with. Add diagnostic whenever
12836 we need to enforce pattern order by not merging.
12837 (decision_tree::insert): Set current_id.
12838 (decision_tree::print_node): Dump parent node and for_id.
12839 (parser::last_id): Add member.
12840 (parser::push_simplify): Assign unique id.
12841 (parser::parser): Initialize last_id.
12843 2017-07-28 Martin Liska <mliska@suse.cz>
12846 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
12847 gimple_build_debug_bind.
12849 2017-07-28 Richard Biener <rguenther@suse.de>
12851 PR tree-optimization/81502
12852 * match.pd: Add pattern combining BIT_INSERT_EXPR with
12854 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
12856 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
12857 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
12858 for BIT_FIELD_REF args.
12859 * fold-const.c (make_bit_field_ref): Likewise.
12860 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
12862 2017-07-28 Jakub Jelinek <jakub@redhat.com>
12865 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
12866 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
12867 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
12868 Or it into SANITIZER_UNDEFINED.
12869 * ubsan.c: Include gimple-fold.h and varasm.h.
12870 (ubsan_expand_ptr_ifn): New function.
12871 (instrument_pointer_overflow): New function.
12872 (maybe_instrument_pointer_overflow): New function.
12873 (instrument_object_size): Formatting fix.
12874 (pass_ubsan::execute): Call instrument_pointer_overflow
12875 and maybe_instrument_pointer_overflow.
12876 * internal-fn.c (expand_UBSAN_PTR): New function.
12877 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
12878 * sanitizer.def (__ubsan_handle_pointer_overflow,
12879 __ubsan_handle_pointer_overflow_abort): New builtins.
12880 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
12881 * internal-fn.def (UBSAN_PTR): New internal function.
12882 * opts.c (sanitizer_opts): Add pointer-overflow.
12883 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
12884 * fold-const.c (build_range_check): Compute pointer range check in
12885 integral type if pointer arithmetics would be needed. Formatting
12888 2017-07-28 Martin Liska <mliska@suse.cz>
12891 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
12892 parameters that are of a variable-length.
12894 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
12896 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
12898 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
12899 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
12900 (CRT_CALL_STATIC_FUNCTION): Likewise.
12901 (ASM_DEFAULT_SPEC): New define.
12902 (ASM_SPEC32): Likewise.
12903 (ASM_SPEC64): Likewise.
12904 (ASM_SPEC_COMMON): Likewise.
12905 (ASM_SPEC): Likewise.
12906 (INVALID_64BIT): Likewise.
12907 (LINK_OS_DEFAULT_SPEC): Likewise.
12908 (LINK_OS_SPEC32): Likewise.
12909 (LINK_OS_SPEC64): Likewise.
12910 (POWERPC_LINUX): Likewise.
12911 (PTRDIFF_TYPE): Likewise.
12912 (RESTORE_FP_PREFIX): Likewise.
12913 (RESTORE_FP_SUFFIX): Likewise.
12914 (SAVE_FP_PREFIX): Likewise.
12915 (SAVE_FP_SUFFIX): Likewise.
12916 (SIZE_TYPE): Likewise.
12917 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
12918 (TARGET_64BIT): Likewise.
12919 (TARGET_64BIT): Likewise.
12920 (TARGET_AIX): Likewise.
12921 (WCHAR_TYPE_SIZE): Likewise.
12922 (WCHAR_TYPE): Undefine.
12923 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
12924 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
12925 (CPP_OS_RTEMS_SPEC): Delete.
12926 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
12927 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
12929 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
12931 2017-07-28 Jakub Jelinek <jakub@redhat.com>
12933 PR tree-optimization/81578
12934 * tree-parloops.c (build_new_reduction): Bail out if
12935 reduction_code isn't one of the standard OpenMP reductions.
12936 Move the details printing after that decision.
12938 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12940 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
12941 related to reload_in_progress.
12942 (splat_input_operand): Likewise.
12943 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
12945 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
12947 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
12948 (TARGET_INSTANTIATE_DECLS): Likewise.
12949 (legitimate_indexed_address_p): Delete reload_in_progress code.
12950 (rs6000_debug_legitimate_address_p): Likewise.
12951 (rs6000_eliminate_indexed_memrefs): Likewise.
12952 (rs6000_emit_le_vsx_store): Likewise.
12953 (rs6000_emit_move_si_sf_subreg): Likewise.
12954 (rs6000_emit_move): Likewise.
12955 (register_to_reg_type): Likewise.
12956 (rs6000_pre_atomic_barrier): Likewise.
12957 (rs6000_machopic_legitimize_pic_address): Likewise.
12958 (rs6000_allocate_stack_temp): Likewise.
12959 (rs6000_address_for_fpconvert): Likewise.
12960 (rs6000_address_for_altivec): Likewise.
12961 (rs6000_secondary_memory_needed_rtx): Delete function.
12962 (rs6000_check_sdmode): Likewise.
12963 (rs6000_alloc_sdmode_stack_slot): Likewise.
12964 (rs6000_instantiate_decls): Likewise.
12965 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
12966 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
12967 Delete reload_in_progress.
12968 (*vec_reload_and_plus_<mptrsize>): Likewise.
12969 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
12970 (vsx_div_v2di): Likewise.
12971 (vsx_udiv_v2di): Likewise.
12973 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12975 * config/rs6000/rs6000.opt (mlra): Replace with stub.
12976 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
12977 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
12978 (rs6000_debug_reg_global): Delete print of LRA status.
12979 (rs6000_option_override_internal): Delete dead LRA related code.
12980 (rs6000_lra_p): Delete function.
12981 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
12983 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
12985 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
12986 * config/riscv/rtems.h: New file.
12988 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12989 Sudakshina Das <sudi.das@arm.com>
12991 * config/aarch64/aarch64.md
12992 (define_split for and<mode>3nr_compare): Move
12993 non aarch64_logical_operand to a register.
12994 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
12995 register immediate operand to a register.
12996 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
12998 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
13000 PR middle-end/81564
13001 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
13003 2017-07-27 Richard Biener <rguenther@suse.de>
13005 PR tree-optimization/81573
13006 PR tree-optimization/81494
13007 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
13008 multi defuse cycle case.
13010 2017-07-27 Richard Biener <rguenther@suse.de>
13012 PR tree-optimization/81571
13013 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
13016 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
13018 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
13019 earlier and only if MASK_FPU is set. Adjust formatting.
13021 2017-07-27 Martin Liska <mliska@suse.cz>
13023 * opt-functions.awk: Add validation of value of Init.
13024 * optc-gen.awk: Pass new argument.
13026 2017-07-27 Martin Liska <mliska@suse.cz>
13028 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
13029 Fix wrong condition.
13031 2017-07-27 Martin Liska <mliska@suse.cz>
13033 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
13034 BBs and edges seen by autoFDO.
13036 2017-07-27 Richard Biener <rguenther@suse.de>
13038 PR tree-optimization/81502
13039 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
13040 with incompatible but same sized type.
13041 (execute_update_addresses_taken): Likewise.
13043 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
13045 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
13046 flag_tree_loop_vectorize rather than flag_tree_vectorize.
13048 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13051 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
13052 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
13053 Change s_operand to memory_operand.
13055 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
13057 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
13058 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
13059 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
13060 Emit instructions rather than returning an expression. Handle TFmode
13061 and KFmode by casting to TImode.
13062 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
13063 (rs6000_emit_le_vsx_store): Likewise.
13064 * config/rs6000/vsx.md (VSX_TI): New iterator.
13065 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
13066 (*vsx_le_undo_permute_<mode>): Likewise.
13067 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
13068 emit the split sequence.
13069 (*vsx_le_perm_store_<mode>): Likewise.
13071 2017-07-27 Jakub Jelinek <jakub@redhat.com>
13073 PR tree-optimization/81555
13074 PR tree-optimization/81556
13075 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
13076 if true, force CHANGED for the recursive invocation.
13077 (reassociate_bb): Remember original length of ops array, pass
13078 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
13080 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
13081 attributes for noipa attribute. For naked attribute use
13082 lookup_attribute first before lookup_attribute_spec.
13083 * final.c (rest_of_handle_final): Disable IPA RA for functions with
13085 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
13086 for functions with noipa attribute.
13087 (cgraph_externally_visible_p): Return true for functions with noipa
13089 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
13090 for functions with noipa attribute.
13091 * doc/extend.texi: Document noipa function attribute.
13092 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
13093 also for functions with noipa attribute.
13094 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
13096 2017-07-26 Andrew Pinski <apinski@cavium.com>
13098 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
13099 vec_unalign_load_cost and vec_unalign_store_cost.
13101 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13103 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
13104 -mvsx-small-integer option.
13105 (ISA_3_0_MASKS_IEEE): Likewise.
13106 (OTHER_VSX_VECTOR_MASKS): Likewise.
13107 (POWERPC_MASKS): Likewise.
13108 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
13109 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
13110 code, only testing for DImode being allowed in non-VSX floating
13112 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
13113 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
13115 (rs6000_option_override_internal): Delete -mvsx-small-integer.
13116 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
13117 TARGET_P8_VECTOR test.
13118 (rs6000_secondary_reload_simple_move): Likewise.
13119 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
13120 since TARGET_P9_VECTOR was already tested.
13121 (rs6000_opt_masks): Remove -mvsx-small-integer.
13122 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
13123 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13125 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
13126 test for TARGET_VEXTRACTUB was used, and that uses
13128 (p9 extract splitter): Likewise.
13129 (vsx_extract_<mode>_di_p9): Likewise.
13130 (vsx_extract_<mode>_store_p9): Likewise.
13131 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
13132 for TARGET_P9_VECTOR was used. Delete code that is now dead with
13133 the elimination of TARGET_VSX_SMALL_INTEGER.
13134 (vsx_extract_<mode>_p8): Likewise.
13135 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
13136 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
13137 (vsx_set_<mode>_p9): Likewise.
13138 (vsx_set_v4sf_p9): Likewise.
13139 (vsx_set_v4sf_p9_zero): Likewise.
13140 (vsx_insert_extract_v4sf_p9): Likewise.
13141 (vsx_insert_extract_v4sf_p9_2): Likewise.
13142 * config/rs6000/rs6000.md (sign extend splitter): Change
13143 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
13144 (floatsi<mode>2_lfiwax_mem): Likewise.
13145 (floatunssi<mode>2_lfiwzx_mem): Likewise.
13146 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
13147 since a test for TARGET_P9_VECTOR was used.
13148 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13149 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
13150 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13151 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
13152 TARGET_P8_VECTOR test.
13153 (fix_trunc<mode>si2_stfiwx): Likewise.
13154 (fix_trunc<mode>si2_internal): Likewise.
13155 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
13156 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13158 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
13159 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
13160 TARGET_P8_VECTOR test.
13161 (fixuns_trunc<mode>si2_stfiwx): Likewise.
13162 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
13163 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13165 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
13166 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
13167 since a test for TARGET_P9_VECTOR was used.
13168 (splitter for loading small constants): Likewise.
13170 2017-07-26 Andrew Pinski <apinski@cavium.com>
13172 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
13175 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
13178 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
13179 (fp_valid_at): Likewise.
13181 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
13183 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
13184 (qdf24xx_addrcost_table): Likewise.
13185 (cortexa57_tunings): Update to use generic_branch_cost.
13186 (cortexa72_tunings): Likewise.
13187 (cortexa73_tunings): Likewise.
13188 (qdf24xx_tunings): Likewise.
13190 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
13192 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
13193 (thunderx2t99_branch_cost): Likewise.
13194 (cortexa35_tunings): Update to use generic_branch_cost.
13195 (cortexa53_tunings): Likewise.
13196 (cortexa57_tunings): Likewise.
13197 (cortexa72_tunings): Likewise.
13198 (cortexa73_tunings): Likewise.
13199 (thunderx2t99_tunings): Likewise.
13201 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
13203 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
13204 (sparc_option_override): Honour MASK_FSMULD.
13205 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
13206 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
13207 * config/sparc/sparc.opt (mfsmuld): New option.
13208 * doc/invoke.texi (mfsmuld): Document option.
13210 2017-07-26 Marek Polacek <polacek@redhat.com>
13212 PR middle-end/70992
13213 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
13215 2017-07-26 Richard Biener <rguenther@suse.de>
13217 * gimple-match-head.c (do_valueize): Return OP if valueize
13219 (get_def): New helper to get at the def stmt of a SSA name
13220 if valueize allows.
13221 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
13222 do_valueize to get at the def stmt.
13223 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
13225 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
13227 PR middle-end/46932
13228 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
13230 2017-07-26 Martin Liska <mliska@suse.cz>
13233 * function.c (expand_function_start): Make expansion of
13234 nonlocal_goto_save_area after parm_birth_insn.
13236 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
13238 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
13239 from all CPU target flags enable members.
13241 2017-07-26 Richard Biener <rguenther@suse.de>
13243 * genmatch.c (dt_simplify::gen): Make iterator vars const.
13244 (decision_tree::gen): Make 'type' const.
13245 (write_predicate): Likewise.
13247 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
13249 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
13250 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
13251 (rs6000_option_override_internal): Likewise.
13252 (rs6000_expand_vector_set): Likewise.
13253 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
13254 (TARGET_UPPER_REGS_SF): Likewise.
13255 (TARGET_UPPER_REGS_DI): Likewise.
13256 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
13257 (TARGET_DIRECT_MOVE_64BIT): Likewise.
13258 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
13259 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13260 (Splitters for DI constants in Altivec registers): Likewise.
13261 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
13262 (vsx_set_v4sf_p9): Likewise.
13263 (vsx_set_v4sf_p9_zero): Likewise.
13264 (vsx_insert_extract_v4sf_p9): Likewise.
13265 (vsx_insert_extract_v4sf_p9_2): Likewise.
13267 2017-07-25 Carl Love <cel@us.ibm.com>
13269 * doc/extend.texi: Update the built-in documentation file for the
13270 existing built-in functions
13271 vector signed char vec_cnttz (vector signed char);
13272 vector unsigned char vec_cnttz (vector unsigned char);
13273 vector signed short vec_cnttz (vector signed short);
13274 vector unsigned short vec_cnttz (vector unsigned short);
13275 vector signed int vec_cnttz (vector signed int);
13276 vector unsigned int vec_cnttz (vector unsigned int);
13277 vector signed long long vec_cnttz (vector signed long long);
13278 vector unsigned long long vec_cnttz (vector unsigned long long);
13280 2017-07-25 Andrew Pinski <apinski@cavium.com>
13282 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
13283 accesses where the use is for the first operand of a BIT_INSERT.
13285 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
13288 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
13289 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
13291 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
13293 * config/i386/gstabs.h: Delete.
13294 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
13296 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
13298 * config/i386/i386.c (ix86_decompose_address): Do not check for
13299 register RTX when looking at index_reg or base_reg.
13300 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
13302 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
13304 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
13305 to update EH info here.
13307 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
13309 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
13311 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
13313 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
13315 2017-07-25 Torsten Duwe <duwe@suse.de>
13317 * common.opt: Introduce -fpatchable-function-entry
13318 command line option, and its variables function_entry_patch_area_size
13319 and function_entry_patch_area_start.
13320 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
13321 including a two-value parser.
13322 * target.def (print_patchable_function_entry): New target hook.
13323 * targhooks.h (default_print_patchable_function_entry): New function.
13324 * targhooks.c (default_print_patchable_function_entry): Likewise.
13325 * toplev.c (process_options): Switch off IPA-RA if
13326 patchable function entries are being generated.
13327 * varasm.c (assemble_start_function): Look at the
13328 patchable-function-entry command line switch and current
13329 function attributes and maybe generate NOP instructions by
13330 calling the print_patchable_function_entry hook.
13331 * doc/extend.texi: Document patchable_function_entry attribute.
13332 * doc/invoke.texi: Document -fpatchable_function_entry
13333 command line option.
13334 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
13336 * doc/tm.texi: Re-generate.
13338 2017-07-25 Jakub Jelinek <jakub@redhat.com>
13341 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
13342 TARGET_AVX512DQ rather than TARGET_AVX512BW.
13344 2017-07-25 Tamar Christina <tamar.christina@arm.com>
13346 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
13348 2017-07-25 Richard Biener <rguenther@suse.de>
13350 PR tree-optimization/81455
13351 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
13352 not walk in cycles when looking for guards.
13354 2017-07-25 Richard Biener <rguenther@suse.de>
13356 PR tree-optimization/81529
13357 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
13358 when optimizing backedge uses.
13360 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
13362 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
13364 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
13365 to dl_section_ref. On AIX, append an expression to subtract
13366 the size of the section length to dl_section_ref.
13368 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
13370 * configure.ac: If any of the config.* scripts fail, exit 1.
13371 * configure: Regenerate.
13373 2017-07-25 Richard Biener <rguenther@suse.de>
13375 PR middle-end/81546
13376 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
13377 of immediate uses, be more verbose on errors.
13379 2017-07-25 Richard Biener <rguenther@suse.de>
13381 PR tree-optimization/81510
13382 * tree-vect-loop.c (vect_is_simple_reduction): When the
13383 reduction stmt is not inside the loop bail out.
13385 2017-07-25 Richard Biener <rguenther@suse.de>
13387 PR tree-optimization/81303
13388 * tree-vect-loop-manip.c (vect_loop_versioning): Build
13389 profitability check against LOOP_VINFO_NITERSM1.
13391 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
13393 * domwalk.c (cmp_bb_postorder): Simplify.
13394 (sort_bbs_postorder): New function. Use it...
13395 (dom_walker::walk): ...here to optimize common cases.
13397 2017-07-25 Martin Liska <mliska@suse.cz>
13400 * ipa-visibility.c (function_and_variable_visibility): Make the
13401 redirection just on target that supports aliasing.
13402 Fix GNU coding style.
13404 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
13407 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
13409 * config/arm/rtems.h: Likewise.
13410 * config/bfin/rtems.h: Likewise.
13411 * config/i386/rtemself.h: Likewise.
13412 * config/lm32/rtems.h: Likewise.
13413 * config/m32c/rtems.h: Likewise.
13414 * config/m68k/rtemself.h: Likewise.
13415 * config/microblaze/rtems.h: Likewise.
13416 * config/mips/rtems.h: Likewise.
13417 * config/moxie/rtems.h: Likewise.
13418 * config/nios2/rtems.h: Likewise.
13419 * config/powerpcspe/rtems.h: Likewise.
13420 * config/rs6000/rtems.h: Likewise.
13421 * config/rtems.h: Likewise.
13422 * config/sh/rtems.h: Likewise.
13423 * config/sh/rtemself.h: Likewise.
13424 * config/sparc/rtemself.h: Likewise.
13426 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
13429 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
13430 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
13431 * tree-ssa-structalias.c (alias_get_name): Same.
13433 2017-07-25 Bin Cheng <bin.cheng@arm.com>
13436 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
13437 instructions if no du chain is found.
13439 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
13441 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
13443 2017-07-25 Richard Biener <rguenther@suse.de>
13445 PR middle-end/81505
13446 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
13449 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
13451 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
13452 upper-regs options.
13453 (ISA_2_7_MASKS_SERVER): Likewise.
13454 (ISA_3_0_MASKS_IEEE): Likewise.
13455 (OTHER_P8_VECTOR_MASKS): Likewise.
13456 (OTHER_VSX_VECTOR_MASKS): Likewise.
13457 (POWERPC_MASKS): Likewise.
13458 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
13459 duplicate list of options.
13460 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
13461 explicit -mupper-regs options.
13462 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
13463 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
13464 alias for -mupper-regs-df.
13465 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13466 (rs6000_init_hard_regno_mode_ok): Likewise.
13467 (rs6000_option_override_internal): Likewise.
13468 (rs6000_opt_masks): Likewise.
13469 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
13470 options in terms of whether -mvsx or -mpower8-vector was used.
13471 (TARGET_UPPER_REGS_DI): Likewise.
13472 (TARGET_UPPER_REGS_SF): Likewise.
13473 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
13474 -mupper-regs-* options.
13476 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
13478 * passes.c (emergency_dump_function): Print some empty lines and a
13479 header before the RTL dump.
13481 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
13483 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
13485 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
13488 * config/aarch64/aarch64.c (aarch64_classify_symbol):
13489 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
13491 2017-07-24 Carl Love <cel@us.ibm.com>
13493 * config/rs6000/rs6000-c.c: Add support for built-in functions
13494 vector float vec_extract_fp32_from_shorth (vector unsigned short);
13495 vector float vec_extract_fp32_from_shortl (vector unsigned short);
13496 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
13497 vec_extract_fp_from_shortl): Add defines for the two builtins.
13498 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
13499 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
13501 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
13502 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
13503 * doc/extend.texi: Update the built-in documentation file for the
13504 new built-in function.
13506 2017-07-24 Jakub Jelinek <jakub@redhat.com>
13509 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
13511 * doc/generic.texi: Likewise.
13512 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
13513 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
13515 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
13517 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
13518 (aarch64_mls_elt_merge<mode>): Likewise.
13520 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
13522 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
13523 having __cxa_atexit.
13525 2017-07-23 Michael Collison <michael.collison@arm.com>
13527 * config/arm/arm.c (arm_option_override): Deprecate
13528 use of -mstructure-size-boundary.
13529 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
13530 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
13532 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13535 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
13536 Reduce cost estimate for direct moves.
13538 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
13541 * config/i386/i386.c (ix86_option_override_internal): Disable
13542 BMI, BMI2 and TBM instructions for -m16.
13544 2017-07-21 Carl Love <cel@us.ibm.com>
13546 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13547 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13548 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13549 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13550 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13551 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13552 VMULOSW): New enum "unspec" values.
13553 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13554 altivec_vmulosw): New patterns.
13555 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13556 VMULOSW): Add definitions.
13558 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
13560 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
13561 (qdf24xx): Likewise.
13562 * config/aarch64/aarch64-options-extensions.def (rdma); New.
13563 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
13564 (AARCH64_FL_V8_1): Renumber.
13565 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
13566 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
13567 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
13568 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
13569 rdma to feature modifiers list.
13571 2017-07-21 Yury Gribov <tetra2005@gmail.com>
13573 PR middle-end/56727
13574 * ipa-visibility (function_and_variable_visibility): Convert
13575 recursive PLT call to direct call if appropriate.
13577 2017-07-21 Andrew Pinski <apinski@cavium.com>
13579 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
13580 operand 1 to see if the types precision matches.
13581 * fold-const.c (operand_equal_p): Likewise.
13583 2017-07-21 Richard Biener <rguenther@suse.de>
13585 PR tree-optimization/81303
13586 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
13587 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
13588 (vect_peeling_hash_get_lowest_cost): Adjust.
13589 (vect_enhance_data_refs_alignment): Likewise. Use
13590 vect_get_peeling_costs_all_drs to compute the penalty for no
13591 peeling to match up costs.
13593 2017-07-21 Richard Biener <rguenther@suse.de>
13595 PR tree-optimization/81500
13596 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
13597 we didn't identify a reduction path.
13599 2017-07-21 Tom de Vries <tom@codesourcery.com>
13600 Cesar Philippidis <cesar@codesourcery.com>
13602 PR gcov-profile/81442
13603 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
13606 2017-07-21 Tom de Vries <tom@codesourcery.com>
13609 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
13611 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
13612 nvptx_override_options_after_change.
13614 2017-07-21 Ulrich Drepper <drepper@redhat.com>
13616 * dwarf2out.c (output_file_names): Avoid double testing for
13617 dwarf_version >= 5.
13619 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
13621 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
13623 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
13625 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
13627 (try_crossjump_to_edge): Do not punt on partitioned functions.
13629 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
13631 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
13632 Put all BBs reachable only via paths crossing cold region to cold
13634 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
13636 2016-07-21 Richard Biener <rguenther@suse.de>
13638 PR tree-optimization/81303
13639 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
13640 into account prologue and epilogue iterations when raising
13641 min_profitable_iters to sth at least covering one vector iteration.
13643 2017-07-21 Tamar Christina <tamar.christina@arm.com>
13645 * config/arm/arm.c (arm_test_cpu_arch_dat):
13648 2017-07-20 Nathan Sidwell <nathan@acm.org>
13650 Remove TYPE_METHODS.
13651 * tree.h (TYPE_METHODS): Delete.
13652 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
13653 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
13654 (dbxout_type_methods): Scan TYPE_FIELDS.
13655 (dbxout_type): Don't check TYPE_METHODS here.
13656 * function.c (use_register_for_decl): Always ignore register for
13657 class types when not optimizing.
13658 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
13659 * tree.c (free_lang_data_in_type): Stitch out member functions and
13660 templates from TYPE_FIELDS.
13661 (build_distinct_type_copy, verify_type_variant,
13662 verify_type): Member fns are on TYPE_FIELDS.
13663 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
13664 * tree-pretty-print.c (dump_generic_node): Likewise.
13666 2017-07-20 Jakub Jelinek <jakub@redhat.com>
13669 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
13670 V2TImode and V4TImode.
13671 (ix86_expand_vector_extract): Likewise.
13672 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
13673 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
13674 (ssescalarmode): Handle V4TImode and V2TImode.
13675 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
13676 (*vec_extractv2ti, *vec_extractv4ti): New insns.
13677 (VEXTRACTI128_MODE): New mode iterator.
13678 (splitter for *vec_extractv?ti first element): New.
13679 (VEC_INIT_MODE): New mode iterator.
13680 (vec_init<mode>): Consolidate 3 expanders into one using
13681 VEC_INIT_MODE mode iterator.
13683 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
13685 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
13686 non_spilled_static_chain_regno_p.
13688 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
13690 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
13692 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
13694 * bb-reorder.c (connect_traces): Allow copying of blocks within
13697 2017-07-20 Richard Biener <rguenther@suse.de>
13699 * gimple.h (gimple_phi_result): Add gphi * overload.
13700 (gimple_phi_result_ptr): Likewise.
13701 (gimple_phi_arg): Likewise. Adjust index assert to only
13702 allow actual argument accesses rather than all slots available
13704 (gimple_phi_arg_def): Add gphi * overload.
13705 * tree-phinodes.c (make_phi_node): Initialize only actual
13707 (resize_phi_node): Clear memory not covered by old node,
13708 do not initialize excess argument slots.
13709 (reserve_phi_args_for_new_edge): Initialize new argument slot
13712 2017-07-20 Bin Cheng <bin.cheng@arm.com>
13714 PR tree-optimization/81388
13716 2016-07-21 Bin Cheng <bin.cheng@arm.com>
13718 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
13719 by removing computation of may_be_zero.
13721 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13722 Tom de Vries <tom@codesourcery.com>
13724 PR middle-end/81030
13725 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
13726 when gimple level profile disagrees with what RTL expander did.
13728 2017-07-20 Richard Biener <rguenther@suse.de>
13730 PR tree-optimization/61171
13731 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
13732 (vect_analyze_stmt): Add slp instance parameter.
13733 (vectorizable_reduction): Likewise.
13734 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
13735 (vect_is_simple_reduction): Deal with chains not detected
13736 as SLP reduction chain, specifically not properly associated
13737 chains containing a mix of plus/minus.
13738 (get_reduction_op): Remove.
13739 (get_initial_defs_for_reduction): Simplify, pass in whether
13740 this is a reduction chain, pass in the SLP node for the PHIs.
13741 (vect_create_epilog_for_reduction): Get the SLP instance as
13743 (vectorizable_reduction): Get the SLP instance as arg.
13744 During analysis remember the SLP node with the PHIs in the
13745 instance. Simplify getting at the vectorized reduction PHIs.
13746 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
13747 through SLP instance.
13748 (vect_slp_analyze_operations): Likewise.
13749 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
13750 (vect_transform_stmt): Likewise.
13752 2017-07-20 Tom de Vries <tom@codesourcery.com>
13754 PR tree-optimization/81489
13755 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
13756 read of phi arg location to before loop that modifies phi.
13758 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13760 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
13763 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13765 PR middle-end/81331
13766 * except.c (execute): Fix ordering issue.
13768 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
13770 PR rtl-optimization/81423
13771 * combine.c (make_compound_operation_int): Don't try to optimize
13772 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
13774 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
13776 PR rtl-optimization/81423
13777 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
13778 with a constant that is -1 in the truncated to mode.
13780 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13782 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
13783 (determine_unlikely_bbs): ... here.
13784 * predict.h (propagate_unlikely_bbs_forward): Declare.
13785 * cfgexpand.c (pass_expand::execute): Use it.
13786 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
13788 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
13789 propagate_unlikely_bbs_forward.
13791 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13793 PR middle-end/81331
13794 * except.c (maybe_add_nop_after_section_switch): New function.
13797 2017-07-19 Tom de Vries <tom@codesourcery.com>
13799 * gimple.h (gimple_phi_set_arg): Make assert more strict.
13801 2017-07-19 Tom de Vries <tom@codesourcery.com>
13803 * gimple.h (gimple_phi_arg): Make assert more strict.
13805 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
13807 * config.gcc (powerpc*-*-*): Add mmintrin.h.
13808 * config/rs6000/mmintrin.h: New file.
13809 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
13811 2017-07-19 Jakub Jelinek <jakub@redhat.com>
13813 PR tree-optimization/81346
13814 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
13816 2017-07-19 Tom de Vries <tom@codesourcery.com>
13818 * config/nvptx/nvptx.md (VECIM): Add V2DI.
13820 2017-07-19 Tom de Vries <tom@codesourcery.com>
13822 * config/nvptx/nvptx-modes.def: Add V2DImode.
13823 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
13824 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
13825 (nvptx_output_mov_insn): Handle lack of mov.b128.
13826 (nvptx_print_operand): Handle 'H' and 'L' codes.
13827 (nvptx_vector_mode_supported): Allow V2DImode.
13828 (nvptx_preferred_simd_mode): New function.
13829 (nvptx_data_alignment): New function.
13830 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
13831 nvptx_preferred_simd_mode.
13832 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
13834 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
13836 2017-07-19 Tom de Vries <tom@codesourcery.com>
13838 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
13839 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
13840 (nvptx_vector_mode_supported): New function. Allow V2SImode.
13841 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
13842 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
13843 (mov<VECIM>_insn): New define_insn.
13844 (define_expand "mov<VECIM>): New define_expand.
13846 2017-07-19 Tom de Vries <tom@codesourcery.com>
13848 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
13850 2017-07-19 Jakub Jelinek <jakub@redhat.com>
13852 PR tree-optimization/81346
13853 * fold-const.h (fold_div_compare, range_check_type): Declare.
13854 * fold-const.c (range_check_type): New function.
13855 (build_range_check): Use range_check_type.
13856 (fold_div_compare): No longer static, rewritten into
13857 a match.pd helper function.
13858 (fold_comparison): Don't call fold_div_compare here.
13859 * match.pd (X / C1 op C2): New optimization using fold_div_compare
13860 as helper function.
13862 2017-07-19 Nathan Sidwell <nathan@acm.org>
13864 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
13865 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
13866 * tree.c (find_decls_types_r, verify_type): Use
13867 TYPE_{MIN,MAX}_VALUE_RAW.
13868 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
13869 (hash_tree): Likewise.
13870 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13872 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13875 2017-07-18 Tom de Vries <tom@codesourcery.com>
13877 PR middle-end/81464
13878 * omp-expand.c (expand_omp_for_static_chunk): Handle
13879 equal-argument loop exit phi.
13881 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
13884 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
13885 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
13886 operand 2 predicate.
13887 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
13888 operand 2 predicate.
13889 (ror,rol -> rorx splitters): Use const_int_operand as
13890 operand 2 predicate.
13892 2017-06-18 Richard Biener <rguenther@suse.de>
13894 PR tree-optimization/81410
13895 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
13896 the gap in the ! slp_perm SLP case after each group.
13898 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13900 PR middle-end/81463
13901 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
13904 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13906 PR middle-end/81462
13907 * predict.c (set_even_probabilities): Cleanup; do not affect
13908 probabilities that are already known.
13909 (combine_predictions_for_bb): Call even when count is set.
13911 2017-07-18 Nathan Sidwell <nathan@acm.org>
13913 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
13916 2017-07-18 Bin Cheng <bin.cheng@arm.com>
13919 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
13920 optimization for loop niter analysis.
13922 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
13925 * config/avr/avr.c (avr_optimize_casesi): Don't use
13926 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
13928 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
13930 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
13931 body_cost_vec from _vect_peel_extended_info.
13932 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
13933 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
13936 2017-07-18 Bin Cheng <bin.cheng@arm.com>
13938 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
13940 2017-07-18 Richard Biener <rguenther@suse.de>
13942 PR tree-optimization/80620
13943 PR tree-optimization/81403
13944 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
13945 info when re-using a VN table entry.
13947 2017-07-18 Richard Biener <rguenther@suse.de>
13949 PR tree-optimization/81418
13950 * tree-vect-loop.c (vectorizable_reduction): Properly compute
13951 vectype_in. Verify that with lane-reducing reduction operations
13952 we have a single def-use cycle.
13954 2017-07-17 Carl Love <cel@us.ibm.com>
13956 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
13958 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13959 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13960 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13961 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13962 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13963 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13964 VMULOSW): New enum "unspec" values.
13965 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13966 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
13967 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13968 altivec_vmulosw): New patterns.
13969 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13970 VMULOSW): Add definitions.
13971 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
13973 * config/alpha/alpha.c: Include predict.h.
13975 2017-07-17 Yury Gribov <tetra2005@gmail.com>
13977 * tree-vrp.c (compare_assert_loc): Fix comparison function
13978 to return predictable results.
13980 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13982 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
13984 (subdi3): Likewise.
13985 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
13986 * doc/invoke.texi (mexpand-adddi): Update text.
13988 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13990 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
13991 that also clobbers the CC register. The old expand code is moved
13993 (*arc_clzsi2): ... here.
13994 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
13995 the CC register. The old expand code is moved to ...
13996 (arc_ctzsi2): ... here.
13998 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14000 * config/arc/arc.opt (mindexed-loads): Use initial value
14001 TARGET_INDEXED_LOADS_DEFAULT.
14002 (mauto-modify-reg): Use initial value
14003 TARGET_AUTO_MODIFY_REG_DEFAULT.
14004 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
14005 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
14006 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
14007 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
14009 2017-07-17 Martin Liska <mliska@suse.cz>
14012 * opts.c (finish_options): Do not allow -fgnu-tm
14013 w/ -fsanitize={kernel-,}address. Say sorry.
14015 2017-07-17 Bin Cheng <bin.cheng@arm.com>
14018 * tree-loop-distribution.c (classify_partition): Only assert on
14019 numer of iterations.
14020 (merge_dep_scc_partitions): Delete prameter. Update function call.
14021 (distribute_loop): Remove code handling loop with unknown niters.
14022 (pass_loop_distribution::execute): Skip loop with unknown niters.
14024 2017-07-17 Bin Cheng <bin.cheng@arm.com>
14027 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
14028 function sort_partitions_by_post_order.
14030 2017-07-17 Bin Cheng <bin.cheng@arm.com>
14032 PR tree-optimization/81374
14033 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
14034 the max index of basic blocks, rather than number of basic blocks.
14036 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14038 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
14040 (arc_legitimate_pic_operand_p): Likewise.
14041 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
14043 (arc_needs_pcl_p): Likewise.
14044 (arc_legitimate_pc_offset_p): Likewise.
14045 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
14046 function is also used in constrains.md.
14047 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
14048 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
14049 PLUS. Only return true/false in known cases, otherwise assert.
14050 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
14051 is already called in arc_legitimate_constant_p.
14052 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
14054 (LEGITIMATE_PIC_OPERAND_P): Use
14055 arc_raw_symbolic_reference_mentioned_p function.
14056 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
14061 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14062 Andrew Burgess <andrew.burgess@embecosm.com>
14064 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
14065 (arc_return_address_register): New function.
14066 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
14067 (arc_handle_fndecl_attribute): Add naked attribute.
14068 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14069 (TARGET_WARN_FUNC_RETURN): Likewise.
14070 (arc_allocate_stack_slots_for_args): New function.
14071 (arc_warn_func_return): Likewise.
14072 (machine_function): Change type fn_type.
14073 (arc_compute_function_type): Consider new naked function type,
14074 change function return type.
14075 (arc_must_save_register): Adapt to handle new
14076 arc_compute_function_type's return type.
14077 (arc_expand_prologue): Likewise.
14078 (arc_expand_epilogue): Likewise.
14079 (arc_return_address_regs): Delete.
14080 (arc_return_address_register): New function.
14081 (arc_epilogue_uses): Use above function.
14082 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
14083 (arc_function_type): Change encoding, add naked type.
14084 (ARC_INTERRUPT_P): Change to handle the new encoding.
14085 (ARC_FAST_INTERRUPT_P): Likewise.
14086 (ARC_NORMAL_P): Define.
14087 (ARC_NAKED_P): Likewise.
14088 (arc_compute_function_type): Delete prototype.
14089 * config/arc/arc.md (in_ret_delay_slot): Use
14090 arc_return_address_register function.
14091 (simple_return): Likewise.
14092 (p_return_i): Likewise.
14094 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14096 PR tree-optimization/81428
14097 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
14098 can't be built for those types.
14100 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
14102 Remove stuff dead since r239246.
14104 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
14105 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
14106 (avr_inform_devices): Remove dead stuff.
14108 2017-07-17 Tamar Christina <tamar.christina@arm.com>
14110 * config/arm/arm_neon.h: Fix softp typo.
14112 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14114 PR tree-optimization/81365
14115 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
14116 aggregate moves onto bb predecessor edges, make sure there are no
14117 loads that could alias the lhs in between the start of bb and the
14120 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
14123 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
14124 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
14125 [LSHIFTRT, outer_code = TRUNCATE]: Same.
14127 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14129 PR tree-optimization/81396
14130 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
14131 (init_symbolic_number): Initialize it to 1.
14132 (perform_symbolic_merge): Add n_ops from both operands into the new
14134 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
14135 without base_addr as useless if they need more than one operation.
14136 (bswap_replace): Handle !bswap case for NULL base_addr.
14138 2017-07-17 Tom de Vries <tom@codesourcery.com>
14141 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
14144 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
14146 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
14147 conditional builtin define __FIX_LEON3FT_B2BST.
14149 2017-07-17 Daniel Cederman <cederman@gaisler.com>
14151 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
14152 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
14155 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
14157 PR rtl-optimization/81424
14158 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
14159 to remove potential trapping from operands if -fnon-call-exceptions.
14161 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14163 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
14164 profile_proability for scalling.
14165 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
14167 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14169 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
14171 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14173 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
14174 fixpoint arithmetics.
14176 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14178 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
14179 fixpoint arithmetics.
14181 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14183 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
14184 fixpoint arithmetics.
14186 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14188 * profile-count.h (profile_probability::from_reg_br_prob_note,
14189 profile_probability::to_reg_br_prob_note): New functions.
14190 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
14191 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
14192 * predict.c (probability_reliable_p): Update.
14193 (edge_probability_reliable_p): Update.
14194 (br_prob_note_reliable_p): Update.
14195 (invert_br_probabilities): Update.
14196 (add_reg_br_prob_note): New function.
14197 (combine_predictions_for_insn): Update.
14198 * asan.c (asan_clear_shadow): Update.
14199 * cfgbuild.c (compute_outgoing_frequencies): Update.
14200 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
14201 (update_br_prob_note): Update.
14202 (rtl_verify_edges): Update.
14203 (purge_dead_edges): Update.
14204 (fixup_reorder_chain): Update.
14205 * emit-rtl.c (try_split): Update.
14206 * ifcvt.c (cond_exec_process_insns): Update.
14207 (cond_exec_process_if_block): Update.
14208 (dead_or_predicable): Update.
14209 * internal-fn.c (expand_addsub_overflow): Update.
14210 (expand_neg_overflow): Update.
14211 (expand_mul_overflow): Update.
14212 * loop-doloop.c (doloop_modify): Update.
14213 * loop-unroll.c (compare_and_jump_seq): Update.
14214 * optabs.c (emit_cmp_and_jump_insn_1): Update.
14215 * predict.h: Update.
14216 * reorg.c (mostly_true_jump): Update.
14218 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
14219 * config/alpha/alpha.c (emit_unlikely_jump): Update.
14220 * config/arc/arc.c: (emit_unlikely_jump): Update.
14221 * config/arm/arm.c: (emit_unlikely_jump): Update.
14222 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
14223 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
14224 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
14225 (ix86_print_operand): Update.
14226 (ix86_split_fp_branch): Update.
14227 (predict_jump): Update.
14228 * config/ia64/ia64.c (ia64_print_operand): Update.
14229 * config/mmix/mmix.c (mmix_print_operand): Update.
14230 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
14231 (rs6000_expand_split_stack_prologue): Update.
14232 * config/rs6000/rs6000.c: Update.
14233 * config/s390/s390.c (s390_expand_vec_strlen): Update.
14234 (s390_expand_vec_movstr): Update.
14235 (s390_expand_cs_tdsi): Update.
14236 (s390_expand_split_stack_prologue): Update.
14237 * config/sh/sh.c (sh_print_operand): Update.
14238 (expand_cbranchsi4): Update.
14239 (expand_cbranchdi4): Update.
14240 * config/sparc/sparc.c (output_v9branch): Update.
14241 * config/spu/spu.c (get_branch_target): Update.
14242 (ea_load_store_inline): Update.
14243 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
14244 * config/tilepro/tilepro.c: Update.
14246 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
14248 * gimplify.c (mostly_copy_tree_r): Revert latest change.
14249 (gimplify_save_expr): Likewise.
14251 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14253 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
14255 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14257 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
14259 * timevar.def (TV_IPA_FNSUMMARY): Define.
14261 2017-07-16 Daniel Cederman <cederman@gaisler.com>
14263 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
14264 to back store errata sensitive sequence from being generated.
14265 (sqrtdf2_fix): Likewise.
14267 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14269 * tree-ssa-threadupdate.c (compute_path_counts,
14270 update_joiner_offpath_counts): Use profile_probability.
14272 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
14275 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14277 * config/arm/arm-c.c (arm_cpu_builtins): Define
14278 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
14280 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
14282 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14283 array entries to represent __ieee128 versions of the
14284 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
14285 scalar_extract_sig, and scalar_insert_exp built-in functions.
14286 (altivec_resolve_overloaded_builtin): Add special case handling
14287 for the __builtin_scalar_insert_exp function, as represented by
14288 the P9V_BUILTIN_VEC_VSIEDP constant.
14289 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
14290 exponent support for __ieee128 argument.
14291 (VSESQP): Add scalar extract signature support for __ieee128
14293 (VSTDCNQP): Add scalar test negative support for __ieee128
14295 (VSIEQP): Add scalar insert exponent support for __int128 argument
14296 with __ieee128 result.
14297 (VSIEQPF): Add scalar insert exponent support for __ieee128
14298 argument with __ieee128 result.
14299 (VSTDCQP): Add scalar test data class support for __ieee128
14301 (VSTDCNQP): Add overload support for scalar test negative with
14302 __ieee128 argument.
14303 (VSTDCQP): Add overload support for scalar test data class
14304 __ieee128 argument.
14305 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
14306 UNSPEC_VSX_SXSIGDP.
14307 (UNSPEC_VSX_SIEXPQP): New constant.
14308 (xsxexpqp): New insn for VSX scalar extract exponent quad
14310 (xsxsigqp): New insn for VSX scalar extract significand quad
14312 (xsiexpqpf): New insn for VSX scalar insert exponent quad
14313 precision with floating point argument.
14314 (xststdcqp): New expand for VSX scalar test data class quad
14316 (xststdcnegqp): New expand for VSX scalar test negative quad
14318 (xststdcqp): New insn to match expansions for VSX scalar test data
14319 class quad precision and VSX scalar test negative quad precision.
14320 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
14321 special case operand checking to enforce that second operand of
14322 VSX scalar test data class with quad precision argument is a 7-bit
14324 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
14325 prototypes and descriptions of __ieee128 versions of
14326 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
14327 scalar_test_data_class, and scalar_test_neg built-in functions.
14329 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14331 PR tree-optimization/81162
14332 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
14333 replace a negate with an add.
14335 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
14337 * doc/invoke.texi (arm/-mcpu): Document +crypto.
14339 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14341 * config/arm/arm-c.c (arm_cpu_builtins): Define
14342 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
14344 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14346 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
14347 (armv8-r): Set ARM Cortex-R52 as default CPU.
14348 * config/arm/arm-tables.opt: Regenerate.
14349 * config/arm/arm-tune.md: Regenerate.
14350 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
14352 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
14353 extension for -mcpu=cortex-r52.
14355 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14357 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
14358 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
14359 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
14360 (fp-armv8): Define it as FP_ARMv8 only.
14361 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
14362 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
14364 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
14365 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
14366 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
14367 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
14368 than TARGET_FPU_ARMV8.
14369 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
14370 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
14371 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
14373 * config/arm/neon.md (neon_vrint): Likewise.
14374 (neon_vcvt): Likewise.
14375 (neon_<fmaxmin_op><mode>): Likewise.
14376 (<fmaxmin><mode>3): Likewise.
14377 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
14378 * config/arm/predicates.md (arm_cond_move_operator): Check against
14379 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
14381 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
14383 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
14384 to top of function.
14386 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14388 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
14389 loop in comment with memset.
14391 2017-07-14 Martin Liska <mliska@suse.cz>
14393 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
14394 * dwarf2out.c (is_java): Remove the function.
14395 (output_pubname): Remove usage of the function.
14396 (lower_bound_default): Remove usage of DW_LANG_Java.
14397 (gen_compile_unit_die): Likewise.
14398 * gcc.c: Remove compiler defaults for .java and .zip files.
14399 * gimple-expr.c (remove_suffix): Change as there's no longer
14400 extension than 4-letter one.
14401 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
14402 (gimplify_save_expr): Likewise.
14403 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
14404 as it's possible even for other languages than Java.
14405 * langhooks.h (struct lang_hooks): Remove Java from a comment.
14406 * lto-opts.c (lto_write_options): Remove reference to Java.
14407 * opts.c (strip_off_ending): Update file extension handling.
14408 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
14409 * tree-eh.c (lower_resx): Likewise.
14410 * tree.c (free_lang_data_in_type): Remove dead code.
14411 (find_decls_types_r): Likewise.
14412 (build_common_builtin_nodes): Remove Java from a comment.
14413 (verify_type): Remove dead code.
14414 * varasm.c (assemble_external): Remove Java from a comment.
14416 2017-07-14 Martin Liska <mliska@suse.cz>
14418 * opts.c (finish_options): Add quotes.
14419 (common_handle_option): Likewise.
14421 2017-07-14 Martin Liska <mliska@suse.cz>
14423 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
14424 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
14425 Remove N_SO_PASCAL.
14426 * dwarf2out.c (lower_bound_default): Do not handle
14428 (gen_compile_unit_die): Likewise.
14429 * gcc.c: Remove default extension binding for GNU Pascal.
14430 * stmt.c: Remove Pascal language from a comment.
14431 * xcoffout.c: Likewise.
14433 2017-07-13 David Malcolm <dmalcolm@redhat.com>
14436 * diagnostic-show-locus.c (fixit_cmp): New function.
14437 (layout::layout): Sort m_fixit_hints.
14438 (column_range::column_range): Assert that the values are valid.
14439 (struct char_span): New struct.
14440 (correction::overwrite): New method.
14441 (struct source_line): New struct.
14442 (line_corrections::add_hint): Add assertions. Reimplement memcpy
14443 calls in terms of classes source_line and char_span, and
14444 correction::overwrite.
14445 (selftest::test_overlapped_fixit_printing_2): New function.
14446 (selftest::diagnostic_show_locus_c_tests): Call it.
14448 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
14450 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
14451 early if there is no lhs.
14453 2017-07-13 Martin Liska <mliska@suse.cz>
14455 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
14456 (gen_reference_type_die): Likewise.
14457 * stor-layout.c: Remove Pascal-related comment.
14459 2017-07-13 Martin Liska <mliska@suse.cz>
14461 * opts.c (finish_options): Add quotes to error messages.
14462 (parse_sanitizer_options): Likewise.
14464 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
14466 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
14468 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
14470 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
14472 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
14474 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
14476 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
14478 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
14481 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
14482 provides the hardware capability bits, define the macro
14483 __BUILTIN_CPU_SUPPORTS__.
14484 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
14485 if GLIBC does not provide the hardware capability bits. Add a
14486 gcc_unreachable call if the built-in cpu function is neither
14487 __builtin_cpu_is nor __builtin_cpu_supports.
14488 (rs6000_get_function_versions_dispatcher): Change the warning
14489 that an old GLIBC is used which does not export the capability
14490 bits to be an error.
14491 * doc/extend.texi (target_clones attribute): Document the
14492 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
14493 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
14494 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
14495 the macros defined by GCC if the newer GLIBC is available.
14497 2017-07-12 Jeff Law <law@redhat.com>
14499 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
14500 remaining includes slightly.
14501 * config/riscv/riscv-builtins.c: Include profile-count.h.
14503 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
14506 * config/avr/avr.c (avr_set_current_function): In diagnostic
14507 messages: Quote keywords and (parts of) identifiers.
14508 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
14511 2017-07-12 Carl Love <cel@us.ibm.com>
14513 * config/rs6000/rs6000-c.c: Add support for built-in functions
14514 vector bool char vec_revb (vector bool char);
14515 vector bool short vec_revb (vector short char);
14516 vector bool int vec_revb (vector bool int);
14517 vector bool long long vec_revb (vector bool long long);
14518 * doc/extend.texi: Update the built-in documentation file for the
14519 new built-in functions.
14521 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14523 * config/s390/s390.md: Remove movcc splitter.
14525 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14527 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
14528 load/store on condition.
14530 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
14533 * config/avr/avr.c (avr_encode_section_info)
14534 [progmem && !TREE_READONLY]: Error if progmem object needs
14537 2017-07-11 Michael Collison <michael.collison@arm.com>
14539 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
14542 2017-07-11 Carl Love <cel@us.ibm.com>
14544 * config/rs6000/rs6000-c.c: Add support for builtins
14545 vector unsigned int vec_parity_lsbb (vector signed int);
14546 vector unsigned int vec_parity_lsbb (vector unsigned int);
14547 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
14548 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
14549 vector unsigned long long vec_parity_lsbb (vector signed long long);
14550 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
14551 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
14552 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
14553 * doc/extend.texi: Update the built-in documentation file for the
14554 new built-in functions.
14556 2017-07-11 David Malcolm <dmalcolm@redhat.com>
14558 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
14559 (layout::m_primary_loc): New field.
14560 (layout::layout): Initialize new field. Move location filtering
14561 logic from here to...
14562 (layout::maybe_add_location_range): ...this new method. Add
14563 support for filtering to just the lines already specified by other
14565 (layout::will_show_line_p): New method.
14566 (gcc_rich_location::add_location_if_nearby): New method.
14567 (selftest::test_add_location_if_nearby): New test function.
14568 (selftest::diagnostic_show_locus_c_tests): Call it.
14569 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
14572 2017-07-11 Tom de Vries <tom@codesourcery.com>
14574 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
14575 (bb_first_real_insn): New function.
14576 (nvptx_single): Add extra initialization of broadcasted condition
14579 2017-07-11 Nathan Sidwell <nathan@acm.org>
14581 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
14583 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
14585 * doc/extend.texi (AVR Function Attributes): Remove weblink to
14586 Binutils doc as TEXI will mess them up.
14587 * doc/invoke.texi (AVR Options): Same here.
14589 2017-07-11 Daniel Cederman <cederman@gaisler.com>
14591 * config/sparc/sparc.opt (mfix-ut700): New option.
14592 (mfix-gr712rc): Likewise.
14593 (sparc_fix_b2bst): New variable.
14594 * doc/invoke.texi (SPARC options): Document them.
14595 (ARM options): Fix warnings.
14596 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
14597 instructions to prevent sequences that can trigger the store-store
14598 errata for certain LEON3FT processors.
14599 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
14600 (sparc_option_override): Set sparc_fix_b2bst appropriately.
14601 * config/sparc/sparc.md (fix_b2bst): New attribute.
14602 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
14604 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
14607 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
14609 (*rsqrtsf2_sse): Ditto.
14611 (div<mode>3): Macroize insn from divdf3 and divsf3
14612 using MODEF mode iterator.
14614 2017-07-10 Martin Sebor <msebor@redhat.com>
14616 PR tree-optimization/80397
14617 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
14618 instead of testing for equality to INTEGER_TYPE.
14620 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
14622 * config.gcc: Remove uclibc from arc target spec.
14624 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
14626 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
14628 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14631 * lto-wrapper.c (remove_option): New function.
14632 (merge_and_complain): Merge PIC/PIE options more realistically.
14634 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
14636 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
14640 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
14641 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
14642 * config.in: Regenerate.
14643 * configure: Regenerate.
14644 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
14645 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
14646 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
14647 (TARGET_GASISR_PROLOGUES): ...target mask.
14648 * common/config/avr/avr-common.c
14649 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
14650 Set -mgas-isr-prologues.
14651 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
14652 INSERT_PASS_BEFORE for it.
14653 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
14654 * config/avr/avr.c (avr_option_override)
14655 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
14656 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
14657 (avr_attribute_table) <no_gccisr>: Add new function attribute.
14658 (avr_set_current_function) <is_no_gccisr>: Init machine field.
14659 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
14661 (make_avr_pass_pre_proep): New function.
14662 (emit_push_sfr) <treg>: Add argument to function and use it
14663 instead of TMP_REG.
14664 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
14665 and set machine->gasisr.yes.
14666 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
14667 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
14668 __gcc_isr.n_pushed to .L__stack_usage.
14669 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
14670 (avr_asm_final_postscan_insn): ...this new static function.
14671 * config/avr/avr.h (machine_function)
14672 <is_no_gccisr, use_L__stack_usage>: New fields.
14673 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
14674 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
14675 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
14676 (gasisr, *gasisr): New expander and insn.
14677 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
14678 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
14679 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
14681 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
14683 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
14686 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
14688 Move jump-tables out of .text again.
14691 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
14692 (ASM_OUTPUT_ADDR_VEC): New function.
14693 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
14694 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
14695 INSN_ADDRESSes as asm comment.
14696 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
14697 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
14698 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
14699 * config/avr/avr.md (*tablejump): Adjust comment.
14700 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
14701 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
14703 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
14704 (avr_output_addr_vec): New proto.
14705 (avr_log_t) <insn_addresses>: New field.
14707 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
14710 * config/i386/i386.c (ix86_function_arg_advance): Set
14711 outgoing_args_on_stack to true if there are outgoing arguments
14713 (ix86_function_arg): Likewise.
14714 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
14715 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
14716 * config/i386/i386.h (machine_function): Add
14717 outgoing_args_on_stack.
14719 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
14721 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
14722 supporting pthreds.
14723 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
14725 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
14727 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
14728 (REAL_H): Remove $(MACHMODE_H).
14729 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
14731 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
14732 $(MACHMODE_H) and double-int.h.
14733 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
14735 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
14738 2017-07-07 Andrew Pinski <apinski@cavium.com>
14740 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
14741 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
14743 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14745 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
14746 Add warning if GCC was not configured to link against a GLIBC that
14747 exports the hardware capability bits.
14748 (make_resolver_func): Make resolver function private and not a
14749 COMDAT function. Create the name with clone_function_name instead
14750 of make_unique_name.
14753 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
14754 correct operand in doing the split.
14756 2017-07-07 Carl Love <cel@us.ibm.com>
14758 * config/rs6000/rs6000-c: Add support for built-in function
14759 vector unsigned short vec_pack_to_short_fp32 (vector float,
14761 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
14762 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
14763 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
14764 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
14765 (convert_4f32_8i16): Add define_expand.
14766 * doc/extend.texi: Update the built-in documentation file for the
14767 new built-in function.
14769 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14771 * config/sparc/m8.md: New file.
14772 * config/sparc/sparc.md: Include m8.md.
14774 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14776 * config/sparc/sparc.opt: New option -mvis4b.
14777 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
14778 (sparc_option_override): Handle VIS4B.
14779 (enum sparc_builtins): Define
14780 SPARC_BUILTIN_DICTUNPACK{8,16,32},
14781 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
14782 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
14783 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
14784 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
14785 (check_constant_argument): New function.
14786 (sparc_vis_init_builtins): Define builtins
14787 __builtin_vis_dictunpack{8,16,32},
14788 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
14789 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
14790 __builtin_vis_fpcmpde{8,16,32}shl and
14791 __builtin_vis_fpcmpur{8,16,32}shl.
14792 (sparc_expand_builtin): Check that the constant operands to
14793 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
14794 constant and in range.
14795 * config/sparc/sparc-c.c (sparc_target_macros): Handle
14797 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
14798 (SPARC_IMM5_P): Likewise.
14799 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
14800 (enabled): Handle vis4b.
14801 (UNSPEC_DICTUNPACK): New unspec.
14802 (UNSPEC_FPCMPSHL): Likewise.
14803 (UNSPEC_FPUCMPSHL): Likewise.
14804 (UNSPEC_FPCMPDESHL): Likewise.
14805 (UNSPEC_FPCMPURSHL): Likewise.
14806 (cpu_feature): New CPU feature `vis4b'.
14807 (dictunpack{8,16,32}): New insns.
14808 (FPCSMODE): New mode iterator.
14809 (fpcscond): New code iterator.
14810 (fpcsucond): Likewise.
14811 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
14812 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
14813 (fpcmpde{8,16,32}{si,di}shl): Likewise.
14814 (fpcmpur{8,16,32}{si,di}shl): Likewise.
14815 * config/sparc/constraints.md: Define constraints `q' for unsigned
14816 2-bit integer constants and `t' for unsigned 5-bit integer
14818 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
14820 (imm5_operand_dictunpack16): Likewise.
14821 (imm5_operand_dictunpack32): Likewise.
14822 (imm2_operand): Likewise.
14823 * doc/invoke.texi (SPARC Options): Document -mvis4b.
14824 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
14825 ditunpack* and fpcmp*shl builtins.
14827 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14829 * config.gcc: Handle m8 in --with-{cpu,tune} options.
14830 * config.in: Add HAVE_AS_SPARC6 define.
14831 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
14833 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
14835 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
14836 (CPP_CPU_SPEC): Handle m8.
14837 (ASM_CPU_SPEC): Likewise.
14838 * config/sparc/sparc-opts.h (enum processor_type): Add
14840 * config/sparc/sparc.c (m8_costs): New struct.
14841 (sparc_option_override): Handle TARGET_CPU_m8.
14842 (sparc32_initialize_trampoline): Likewise.
14843 (sparc64_initialize_trampoline): Likewise.
14844 (sparc_issue_rate): Likewise.
14845 (sparc_register_move_cost): Likewise.
14846 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
14847 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
14848 (ASM_CPU64_DEFAULT_SPEC): Likewise.
14849 (CPP_CPU_SPEC): Handle M8.
14850 (ASM_CPU_SPEC): Likewise.
14851 (AS_M8_FLAG): Define.
14852 * config/sparc/sparc.md: Add m8 to the cpu attribute.
14853 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
14854 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
14856 * configure: Regenerate.
14857 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
14860 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14862 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
14864 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
14865 ("*movdi_insn_sp32"): Do not set v3pipe.
14866 ("*movsi_insn"): Likewise.
14867 ("*movdi_insn_sp64"): Likewise.
14868 ("*movsf_insn"): Likewise.
14869 ("*movdf_insn_sp32"): Likewise.
14870 ("*movdf_insn_sp64"): Likewise.
14871 ("*zero_extendsidi2_insn_sp64"): Likewise.
14872 ("*sign_extendsidi2_insn"): Likewise.
14873 ("*mov<VM32:mode>_insn"): Likewise.
14874 ("*mov<VM64:mode>_insn_sp64"): Likewise.
14875 ("*mov<VM64:mode>_insn_sp32"): Likewise.
14876 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
14877 ("<vlop:code><VL:mode>3"): Likewise.
14878 ("*not_<vlop:code><VL:mode>3"): Likewise.
14879 ("*nand<VL:mode>_vis"): Likewise.
14880 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
14881 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
14882 ("one_cmpl<VL:mode>2"): Likewise.
14883 ("faligndata<VM64:mode>_vis"): Likewise.
14884 ("alignaddrsi_vis"): Likewise.
14885 ("alignaddrdi_vis"): Likweise.
14886 ("alignaddrlsi_vis"): Likewise.
14887 ("alignaddrldi_vis"): Likewise.
14888 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
14889 ("bmaskdi_vis"): Likewise.
14890 ("bmasksi_vis"): Likewise.
14891 ("bshuffle<VM64:mode>_vis"): Likewise.
14892 ("cmask8<P:mode>_vis"): Likewise.
14893 ("cmask16<P:mode>_vis"): Likewise.
14894 ("cmask32<P:mode>_vis"): Likewise.
14895 ("pdistn<P:mode>_vis"): Likewise.
14896 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
14898 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14900 * config/sparc/sparc.md ("subtype"): New insn attribute.
14901 ("*wrgsr_sp64"): Set insn subtype.
14902 ("*rdgsr_sp64"): Likewise.
14903 ("alignaddrsi_vis"): Likewise.
14904 ("alignaddrdi_vis"): Likewise.
14905 ("alignaddrlsi_vis"): Likewise.
14906 ("alignaddrldi_vis"): Likewise.
14907 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
14908 ("fexpand_vis"): Likewise.
14909 ("fpmerge_vis"): Likewise.
14910 ("faligndata<VM64:mode>_vis"): Likewise.
14911 ("bshuffle<VM64:mode>_vis"): Likewise.
14912 ("cmask8<P:mode>_vis"): Likewise.
14913 ("cmask16<P:mode>_vis"): Likewise.
14914 ("cmask32<P:mode>_vis"): Likewise.
14915 ("fchksm16_vis"): Likewise.
14916 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
14917 ("fmean16_vis"): Likewise.
14918 ("fp<plusminus_insn>64_vis"): Likewise.
14919 ("<plusminus_insn>v8qi3"): Likewise.
14920 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
14921 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
14922 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
14923 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
14924 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
14925 ("*movqi_insn"): Likewise.
14926 ("*movhi_insn"): Likewise.
14927 ("*movsi_insn"): Likewise.
14928 ("movsi_pic_gotdata_op"): Likewise.
14929 ("*movdi_insn_sp32"): Likewise.
14930 ("*movdi_insn_sp64"): Likewise.
14931 ("movdi_pic_gotdata_op"): Likewise.
14932 ("*movsf_insn"): Likewise.
14933 ("*movdf_insn_sp32"): Likewise.
14934 ("*movdf_insn_sp64"): Likewise.
14935 ("*zero_extendhisi2_insn"): Likewise.
14936 ("*zero_extendqihi2_insn"): Likewise.
14937 ("*zero_extendqisi2_insn"): Likewise.
14938 ("*zero_extendqidi2_insn"): Likewise.
14939 ("*zero_extendhidi2_insn"): Likewise.
14940 ("*zero_extendsidi2_insn_sp64"): Likewise.
14941 ("ldfsr"): Likewise.
14942 ("prefetch_64"): Likewise.
14943 ("prefetch_32"): Likewise.
14944 ("tie_ld32"): Likewise.
14945 ("tie_ld64"): Likewise.
14946 ("*tldo_ldub_sp32"): Likewise.
14947 ("*tldo_ldub1_sp32"): Likewise.
14948 ("*tldo_ldub2_sp32"): Likewise.
14949 ("*tldo_ldub_sp64"): Likewise.
14950 ("*tldo_ldub1_sp64"): Likewise.
14951 ("*tldo_ldub2_sp64"): Likewise.
14952 ("*tldo_ldub3_sp64"): Likewise.
14953 ("*tldo_lduh_sp32"): Likewise.
14954 ("*tldo_lduh1_sp32"): Likewise.
14955 ("*tldo_lduh_sp64"): Likewise.
14956 ("*tldo_lduh1_sp64"): Likewise.
14957 ("*tldo_lduh2_sp64"): Likewise.
14958 ("*tldo_lduw_sp32"): Likewise.
14959 ("*tldo_lduw_sp64"): Likewise.
14960 ("*tldo_lduw1_sp64"): Likewise.
14961 ("*tldo_ldx_sp64"): Likewise.
14962 ("*mov<VM32:mode>_insn"): Likewise.
14963 ("*mov<VM64:mode>_insn_sp64"): Likewise.
14964 ("*mov<VM64:mode>_insn_sp32"): Likewise.
14966 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14968 * config/sparc/sparc.md ("type"): New insn type viscmp.
14969 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
14971 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
14972 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
14973 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
14974 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
14976 ("n7_vis_logical_11cycle"): Likewise.
14977 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
14978 * config/sparc/niagara2.md ("niag3_vis": Likewise.
14979 * config/sparc/niagara.md ("niag_vis"): Likewise.
14980 * config/sparc/ultra3.md ("us3_fga"): Likewise.
14981 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
14983 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14985 * config/sparc/sparc.md: New instruction type `bmask'.
14986 (bmaskdi_vis): Use the `bmask' type.
14987 (bmasksi_vis): Likewise.
14988 * config/sparc/ultra3.md (us3_array): Likewise.
14989 * config/sparc/niagara7.md (n7_array): Likewise.
14990 * config/sparc/niagara4.md (n4_array): Likewise.
14991 * config/sparc/niagara2.md (niag2_vis): Likewise.
14992 (niag3_vis): Likewise.
14993 * config/sparc/niagara.md (niag_vis): Likewise.
14995 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14997 * ipa-comdats.c: Remove optimize check from gate.
14998 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
14999 for functions not optimized.
15000 (ipa_fn_summary_read): Skip optimize check.
15001 (ipa_fn_summary_write): Likewise.
15002 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
15004 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
15006 (can_inline_edge_p): Check flag_pcc_struct_return for match.
15007 (check_callers): Give up on caller which is not optimized.
15008 (inline_small_functions): Likewise.
15009 (ipa_inline): Do not give up when not optimizing.
15010 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
15011 away unoptimizes cdtors.
15012 (whole_program_function_and_variable_visibility): Do
15013 ipa_discover_readonly_nonaddressable_vars in LTO mode.
15014 * ipa.c (process_references): Do not check optimize.
15015 (symbol_table::remove_unreachable_nodes): Update optimize check.
15016 (set_writeonly_bit): Update optimize check.
15017 (pass_ipa_cdtor_merge::gate): Do not check optimize.
15018 (pass_ipa_single_use::gate): Remove.
15020 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15022 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
15023 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
15024 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
15025 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
15026 permute_load, permute_store, adjust_extract, adjust_splat,
15027 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
15028 replace_swap_with_copy, dump_swap_insn_table,
15029 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
15030 recombine_lvx_pattern, recombine_stvx_pattern,
15031 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
15032 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
15033 to file rs6000-p8swap.c.
15034 * config/rs6000/rs6000-p8swap.c: New file.
15035 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
15036 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
15037 and rs6000*-*-* targets.
15039 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15041 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
15043 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15045 * lto-wrapper.c (merge_and_complain): Do not merge
15046 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
15047 fsigned_zeros, ftrapping_math, fwrapv.
15048 (append_compiler_options): Do not track these options.
15049 (append_linker_options): Likewie
15051 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15053 * cgraphunit.c (cgraph_node::finalize_function): When
15054 !flag_toplevel_reorde set no_reorder flag.
15055 (varpool_node::finalize_decl): Likewise.
15056 (symbol_table::compile): Drop no toplevel reorder path.
15058 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15060 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
15061 edges; zero probability is not better than uninitialized.
15063 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
15065 * asan.h (asan_sanitize_allocas_p): Declare.
15066 * asan.c (asan_sanitize_allocas_p): New function.
15067 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
15068 (handle_builtin_alloca): Likewise.
15069 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
15070 if !asan_sanitize_allocas_p.
15071 * params.def (asan-instrument-allocas): Add new option.
15072 * params.h (ASAN_PROTECT_ALLOCAS): Define.
15073 * opts.c (common_handle_option): Disable allocas sanitization for
15076 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
15078 * asan.c: Include gimple-fold.h.
15079 (get_last_alloca_addr): New function.
15080 (handle_builtin_stackrestore): Likewise.
15081 (handle_builtin_alloca): Likewise.
15082 (asan_emit_allocas_unpoison): Likewise.
15083 (get_mem_refs_of_builtin_call): Add new parameter, remove const
15084 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
15085 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
15086 (instrument_builtin_call): Pass gimple iterator to
15087 get_mem_refs_of_builtin_call.
15088 (last_alloca_addr): New global.
15089 * asan.h (asan_emit_allocas_unpoison): Declare.
15090 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
15091 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
15092 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
15093 if function calls alloca.
15094 * gimple-fold.c (replace_call_with_value): Remove static keyword.
15095 * gimple-fold.h (replace_call_with_value): Declare.
15096 * internal-fn.c: Include asan.h.
15097 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
15098 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
15100 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15102 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
15103 (C_SELFTEST_FLAGS): New.
15104 (CPP_SELFTEST_FLAGS): New.
15105 (SELFTEST_DEPS): New, from deps of s-selftest.
15106 (C_SELFTEST_DEPS): New, from deps of s-selftest.
15107 (CPP_SELFTEST_DEPS): New.
15108 (selftest): Add dependency on s-selftest-c++.
15109 (s-selftest): Rename to...
15110 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
15111 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
15112 than SELFTEST_FLAGS.
15113 (selftest-gdb): Rename to...
15114 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
15116 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
15117 (selftest-valgrind): Rename to...
15118 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
15120 (selftest-valgrind): Reintroduce as an alias for
15121 selftest-c-valgrind.
15122 (s-selftest-c++): New.
15123 (selftest-c++-gdb): New.
15124 (selftest-c++-valgrind): New.
15126 2017-07-06 Olivier Hainque <hainque@adacore.com>
15128 * gcc.c (process_command): When deciding if undefined variables
15129 should be ignored when processing specs, accept "gcc -v" as well.
15131 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15133 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
15134 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
15136 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
15138 * config/arm/arm-cpus.in (armv8-r): Add new entry.
15139 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
15140 * config/arm/arm-tables.opt: Regenerate.
15141 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
15143 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
15145 2017-07-06 Carl Love <cel@us.ibm.com>
15147 * ChangeLog: Clean up from mid air collision
15149 2017-07-06 Carl Love <cel@us.ibm.com>
15151 * config/rs6000/rs6000-c.c: Add support for built-in functions
15152 vector signed int vec_subc (vector signed int, vector signed int);
15153 vector signed __int128 vec_subc (vector signed __int128,
15154 vector signed __int128);
15155 vector unsigned __int128 vec_subc (vector unsigned __int128,
15156 vector unsigned __int128);
15157 vector signed int vec_sube (vector signed int, vector signed int,
15158 vector signed int);
15159 vector unsigned int vec_sube (vector unsigned int,
15160 vector unsigned int,
15161 vector unsigned int);
15162 vector signed __int128 vec_sube (vector signed __int128,
15163 vector signed __int128,
15164 vector signed__int128);
15165 vector unsigned __int128 vec_sube (vector unsigned __int128,
15166 vector unsigned __int128,
15167 vector unsigned __int128);
15168 vector signed int vec_subec (vector signed int, vector signed int,
15169 vector signed int);
15170 vector unsigned int vec_subec (vector unsigned int,
15171 vector unsigned int,
15172 vector unsigned int);
15173 vector signed __int128 vec_subec (vector signed __int128,
15174 vector signed __int128,
15175 vector signed__int128);
15176 vector unsigned __int128 vec_subec (vector unsigned __int128,
15177 vector unsigned __int128,
15178 vector unsigned __int128);
15179 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
15180 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
15181 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
15182 BU_ALTIVEC_OVERLOAD_X definitions.
15183 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
15184 * doc/extend.texi: Update the built-in documentation file for the new
15185 built-in functions.
15187 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15190 * diagnostic-show-locus.c (layout::layout): Use start and finish
15191 spelling location for the start and finish of each range.
15192 * genmatch.c (linemap_client_expand_location_to_spelling_point):
15193 Add unused aspect param.
15194 * input.c (expand_location_1): Add "aspect" param, and use it
15195 to access the correct part of the location.
15196 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
15198 (expand_location_to_spelling_point): Likewise.
15199 (linemap_client_expand_location_to_spelling_point): Add "aspect"
15200 param, and pass it to expand_location_1.
15202 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
15204 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
15205 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
15206 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
15207 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
15208 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
15209 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
15210 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
15211 _mm_maskz_getmant_ss): New intrinsics.
15212 (__builtin_ia32_getexpss128_mask): Changed to ...
15213 __builtin_ia32_getexpss128_round ... this.
15214 (__builtin_ia32_getexpsd128_mask): Changed to ...
15215 __builtin_ia32_getexpsd128_round ... this.
15216 * config/i386/i386-builtin-types.def
15217 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
15218 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
15219 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
15220 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
15221 __builtin_ia32_getmantss_mask_round): New builtins.
15222 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
15223 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
15224 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
15225 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
15226 * config/i386/sse.md
15227 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
15228 avx512f_sgetexp<mode><mask_scalar_name>
15229 <round_saeonly_scalar_name> ... this.
15230 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
15231 %0, %1, %2<round_saeonly_op3>}): Changed to ...
15232 vgetexp<ssescalarmodesuffix>
15233 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15234 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
15235 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
15236 avx512f_vgetmant<mode><mask_scalar_name>
15237 <round_saeonly_scalar_name> ... this.
15238 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
15239 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
15240 vgetmant<ssescalarmodesuffix>
15241 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
15242 %0<mask_scalar_operand4>, %1, %2
15243 <round_saeonly_scalar_mask_op4>, %3} ... this.
15244 * config/i386/subst.md (mask_scalar_operand4,
15245 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
15246 round_saeonly_scalar_nimm_predicate): New subst attributes.
15248 2017-07-06 Julia Koval <julia.koval@intel.com>
15250 * config/i386/i386.c (ix86_erase_embedded_rounding):
15251 Remove code for old rounding pattern.
15253 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
15255 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
15257 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
15259 * doc/sourcebuild.texi (Test Directives, Variants of
15260 dg-require-support): Add documentation for dg-require-stack-check.
15262 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
15264 * config/i386/subst.md (mask_scalar, round_scalar,
15265 round_saeonly_scalar): New meta-templates.
15266 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
15267 round_scalar_mask_operand3, round_scalar_mask_op3,
15268 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
15269 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
15270 round_saeonly_scalar_constraint,
15271 round_saeonly_scalar_prefix): New subst attribute.
15272 * config/i386/sse.md
15273 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
15274 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
15275 <round_scalar_name> ... this.
15276 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
15277 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
15278 <round_scalar_name> ... this.
15279 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
15280 <sse>_vm<code><mode>3<mask_scalar_name>
15281 <round_saeonly_scalar_name> ... this.
15282 (v<plusminus_mnemonic><ssescalarmodesuffix>
15283 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
15284 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
15285 v<plusminus_mnemonic><ssescalarmodesuffix>
15286 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15287 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
15288 (v<multdiv_mnemonic><ssescalarmodesuffix>
15289 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
15290 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
15291 v<multdiv_mnemonic><ssescalarmodesuffix>
15292 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15293 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
15294 (v<maxmin_float><ssescalarmodesuffix>
15295 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
15296 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
15297 v<maxmin_float><ssescalarmodesuffix>
15298 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15299 %0<mask_scalar_operand3>, %1, %<iptr>2
15300 <round_saeonly_scalar_mask_op3>} ... this.
15302 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
15304 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
15305 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
15307 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
15308 Alan Hayward <alan.hayward@arm.com>
15309 David Sherwood <david.sherwood@arm.com>
15311 * combine.c (simplify_if_then_else): Remove "enum" before
15313 * compare-elim.c (can_eliminate_compare): Likewise.
15314 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
15316 (aarch64_lookup_simd_builtin_type): Likewise.
15317 (aarch64_simd_builtin_type): Likewise.
15318 (aarch64_init_simd_builtin_types): Likewise.
15319 (aarch64_simd_expand_args): Likewise.
15320 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
15322 (aarch64_reverse_mask): Likewise.
15323 (aarch64_simd_emit_reg_reg_move): Likewise.
15324 (aarch64_gen_adjusted_ldpstp): Likewise.
15325 (aarch64_ccmp_mode_to_code): Likewise.
15326 (aarch64_operands_ok_for_ldpstp): Likewise.
15327 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15328 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
15330 (aarch64_min_divisions_for_recip_mul): Likewise.
15331 (aarch64_reassociation_width): Likewise.
15332 (aarch64_get_condition_code_1): Likewise.
15333 (aarch64_simd_emit_reg_reg_move): Likewise.
15334 (aarch64_simd_attr_length_rglist): Likewise.
15335 (aarch64_reverse_mask): Likewise.
15336 (aarch64_operands_ok_for_ldpstp): Likewise.
15337 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15338 (aarch64_gen_adjusted_ldpstp): Likewise.
15339 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
15341 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
15342 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
15343 (arm_lookup_simd_builtin_type): Likewise.
15344 (arm_simd_builtin_type): Likewise.
15345 (arm_init_simd_builtin_types): Likewise.
15346 (arm_expand_builtin_args): Likewise.
15347 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
15348 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
15349 (ft32_setup_incoming_varargs): Likewise.
15350 (ft32_function_arg): Likewise.
15351 (ft32_function_arg_advance): Likewise.
15352 (ft32_pass_by_reference): Likewise.
15353 (ft32_arg_partial_bytes): Likewise.
15354 (ft32_valid_pointer_mode): Likewise.
15355 (ft32_addr_space_pointer_mode): Likewise.
15356 (ft32_addr_space_legitimate_address_p): Likewise.
15357 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
15359 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
15360 (ix86_emit_outlined_ms2sysv_restore): Likewise.
15361 (iamcu_alignment): Likewise.
15362 (canonicalize_vector_int_perm): Likewise.
15363 (ix86_noce_conversion_profitable_p): Likewise.
15364 (ix86_mpx_bound_mode): Likewise.
15365 (ix86_operands_ok_for_move_multiple): Likewise.
15366 * config/microblaze/microblaze-protos.h
15367 (microblaze_expand_conditional_branch_reg): Likewise.
15368 * config/microblaze/microblaze.c
15369 (microblaze_expand_conditional_branch_reg): Likewise.
15370 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
15372 (rs6000_reassociation_width): Likewise.
15373 (rs6000_invalid_binary_op): Likewise.
15374 (fusion_p9_p): Likewise.
15375 (emit_fusion_p9_load): Likewise.
15376 (emit_fusion_p9_store): Likewise.
15377 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
15379 (riscv_hard_regno_mode_ok_p): Likewise.
15380 (riscv_address_insns): Likewise.
15381 (riscv_split_symbol): Likewise.
15382 (riscv_legitimize_move): Likewise.
15383 (riscv_function_value): Likewise.
15384 (riscv_hard_regno_nregs): Likewise.
15385 (riscv_expand_builtin): Likewise.
15386 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
15387 (riscv_build_integer): Likewise.
15388 (riscv_split_integer): Likewise.
15389 (riscv_legitimate_constant_p): Likewise.
15390 (riscv_cannot_force_const_mem): Likewise.
15391 (riscv_regno_mode_ok_for_base_p): Likewise.
15392 (riscv_valid_base_register_p): Likewise.
15393 (riscv_valid_offset_p): Likewise.
15394 (riscv_valid_lo_sum_p): Likewise.
15395 (riscv_classify_address): Likewise.
15396 (riscv_legitimate_address_p): Likewise.
15397 (riscv_address_insns): Likewise.
15398 (riscv_load_store_insns): Likewise.
15399 (riscv_force_binary): Likewise.
15400 (riscv_split_symbol): Likewise.
15401 (riscv_force_address): Likewise.
15402 (riscv_legitimize_address): Likewise.
15403 (riscv_move_integer): Likewise.
15404 (riscv_legitimize_const_move): Likewise.
15405 (riscv_legitimize_move): Likewise.
15406 (riscv_address_cost): Likewise.
15407 (riscv_subword): Likewise.
15408 (riscv_output_move): Likewise.
15409 (riscv_canonicalize_int_order_test): Likewise.
15410 (riscv_emit_int_order_test): Likewise.
15411 (riscv_function_arg_boundary): Likewise.
15412 (riscv_pass_mode_in_fpr_p): Likewise.
15413 (riscv_pass_fpr_single): Likewise.
15414 (riscv_pass_fpr_pair): Likewise.
15415 (riscv_get_arg_info): Likewise.
15416 (riscv_function_arg): Likewise.
15417 (riscv_function_arg_advance): Likewise.
15418 (riscv_arg_partial_bytes): Likewise.
15419 (riscv_function_value): Likewise.
15420 (riscv_pass_by_reference): Likewise.
15421 (riscv_setup_incoming_varargs): Likewise.
15422 (riscv_print_operand): Likewise.
15423 (riscv_elf_select_rtx_section): Likewise.
15424 (riscv_save_restore_reg): Likewise.
15425 (riscv_for_each_saved_reg): Likewise.
15426 (riscv_register_move_cost): Likewise.
15427 (riscv_hard_regno_mode_ok_p): Likewise.
15428 (riscv_hard_regno_nregs): Likewise.
15429 (riscv_class_max_nregs): Likewise.
15430 (riscv_memory_move_cost): Likewise.
15431 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
15432 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
15433 (rl78_addr_space_address_mode): Likewise.
15434 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15436 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15437 (rs6000_reassociation_width): Likewise.
15438 (rs6000_invalid_binary_op): Likewise.
15439 (fusion_p9_p): Likewise.
15440 (emit_fusion_p9_load): Likewise.
15441 (emit_fusion_p9_store): Likewise.
15442 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
15443 (ok_for_simple_move_operands): Likewise.
15444 (ok_for_simple_move_strict_operands): Likewise.
15445 (ok_for_simple_arith_logic_operands): Likewise.
15446 (visium_legitimize_reload_address): Likewise.
15447 (visium_select_cc_mode): Likewise.
15448 (output_cbranch): Likewise.
15449 (visium_split_double_move): Likewise.
15450 (visium_expand_copysign): Likewise.
15451 (visium_expand_int_cstore): Likewise.
15452 (visium_expand_fp_cstore): Likewise.
15453 * config/visium/visium.c (visium_pass_by_reference): Likewise.
15454 (visium_function_arg): Likewise.
15455 (visium_function_arg_advance): Likewise.
15456 (visium_libcall_value): Likewise.
15457 (visium_setup_incoming_varargs): Likewise.
15458 (visium_legitimate_constant_p): Likewise.
15459 (visium_legitimate_address_p): Likewise.
15460 (visium_legitimize_address): Likewise.
15461 (visium_secondary_reload): Likewise.
15462 (visium_register_move_cost): Likewise.
15463 (visium_memory_move_cost): Likewise.
15464 (prepare_move_operands): Likewise.
15465 (ok_for_simple_move_operands): Likewise.
15466 (ok_for_simple_move_strict_operands): Likewise.
15467 (ok_for_simple_arith_logic_operands): Likewise.
15468 (visium_function_value_1): Likewise.
15469 (rtx_ok_for_offset_p): Likewise.
15470 (visium_legitimize_reload_address): Likewise.
15471 (visium_split_double_move): Likewise.
15472 (visium_expand_copysign): Likewise.
15473 (visium_expand_int_cstore): Likewise.
15474 (visium_expand_fp_cstore): Likewise.
15475 (visium_split_cstore): Likewise.
15476 (visium_select_cc_mode): Likewise.
15477 (visium_split_cbranch): Likewise.
15478 (output_cbranch): Likewise.
15479 (visium_print_operand_address): Likewise.
15480 * expmed.c (flip_storage_order): Likewise.
15481 * expmed.h (emit_cstore): Likewise.
15482 (flip_storage_order): Likewise.
15483 * genrecog.c (validate_pattern): Likewise.
15484 * hsa-gen.c (gen_hsa_addr): Likewise.
15485 * internal-fn.c (expand_arith_overflow): Likewise.
15486 * ira-color.c (allocno_copy_cost_saving): Likewise.
15487 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15488 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
15489 (process_invariant_for_inheritance): Likewise.
15490 * lra-eliminations.c (move_plus_up): Likewise.
15491 * omp-low.c (lower_oacc_reductions): Likewise.
15492 * simplify-rtx.c (simplify_subreg): Likewise.
15493 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
15494 (TARGET_CHKP_BOUND_MODE): Likewise..
15495 * targhooks.c (default_chkp_bound_mode): Likewise.
15496 (default_setup_incoming_vararg_bounds): Likewise.
15497 * targhooks.h (default_chkp_bound_mode): Likewise.
15498 (default_setup_incoming_vararg_bounds): Likewise.
15499 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
15500 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
15501 (have_whole_vector_shift): Likewise.
15502 * tree-vect-stmts.c (vectorizable_load): Likewise.
15503 * doc/tm.texi: Regenerate.
15505 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
15507 Graceful degrade if Binutils PR21472 is not available.
15510 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
15511 .rodata in flash test fails.
15512 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
15513 * confgure: Regenerate.
15514 * config.in: Regenerate.
15515 * config/avr/avr.c (avr_asm_named_section)
15516 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
15517 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
15518 (avr_asm_init_sections): Same.
15520 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15522 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
15523 (fma<VH:mode>4_intrinsic): Likewise.
15524 (*fmsub<VCVTF:mode>4): Likewise.
15525 (*fmsub<VH:mode>4_intrinsic): Likewise.
15527 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
15530 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
15531 Don't depend on "optimize > 0".
15532 (out_movhi_r_mr, out_movqi_mr_r): Same.
15533 (out_movhi_mr_r, out_movqi_r_mr): Same.
15534 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
15535 io_address_operand on "optimize > 0".
15537 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15539 * tree-loop-distribution.c: Add general explanantion on the pass.
15540 (generate_loops_for_partition): Mark distributed loop.
15541 (pg_add_dependence_edges): New parameter. Handle alias data
15542 dependence specially and record it in the parameter if asked.
15543 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
15544 (init_partition_graph_vertices, add_partition_graph_edge): New.
15545 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
15546 (free_partition_graph_vdata, build_partition_graph): New.
15547 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
15548 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
15549 (data_ref_segment_size, latch_dominated_by_data_ref): New.
15550 (compute_alias_check_pairs, version_loop_by_alias_check): New.
15551 (version_for_distribution_p, finalize_partitions): New.
15552 (distribute_loop): Handle alias data dependence specially. Factor
15553 out loop fusion code as functions and call these functions.
15555 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15557 * tree-loop-distribution.c (classify_partition): New parameter and
15558 better handle reduction statement.
15559 (rdg_build_partitions): Revise comment.
15560 (distribute_loop): Compute statements in all partitions and pass it
15561 to classify_partition.
15563 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15565 * tree-loop-distribution.c (enum partition_type): New.
15566 (struct partition): New field type.
15567 (partition_merge_into): Add parameter. Update partition type.
15568 (data_dep_in_cycle_p, update_type_for_merge): New functions.
15569 (build_rdg_partition_for_vertex): Compute partition type.
15570 (rdg_build_partitions): Dump partition type.
15571 (distribute_loop): Update calls to partition_merge_into.
15573 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15575 * tree-loop-distribution.c (struct ddr_hasher): New.
15576 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
15578 (classify_partition): Call get_data_dependence.
15579 (pg_add_dependence_edges): Ditto.
15580 (distribute_loop): Release data dependence hash table.
15582 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15584 * tree-loop-distribution.c (ref_base_address): Delete.
15585 (similar_memory_accesses): Rename ...
15586 (share_memory_accesses): ... to this. Check if partitions access
15587 the same memory reference.
15588 (distribute_loop): Call share_memory_accesses.
15590 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15592 * tree-loop-distribution.c (struct partition): New field recording
15593 its data reference.
15594 (partition_alloc, partition_free): Init and release data refs.
15595 (partition_merge_into): Merge data refs.
15596 (build_rdg_partition_for_vertex): Collect data refs for partition.
15597 (pg_add_dependence_edges): Change parameters from vector to bitmap.
15599 (distribute_loop): Remve data refs from vertice data of partition
15602 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15604 * tree-loop-distribution.c (params.h): Include header file.
15605 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
15606 (datarefs_vec): New global var.
15607 (create_rdg_vertices): Use datarefs_vec directly.
15608 (free_rdg): Don't free data references.
15609 (build_rdg): Update use. Don't free data references.
15610 (distribute_loop): Compute global variable for data references.
15611 Bail out if there are too many data references.
15613 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15615 * tree-loop-distribution.c (loop_nest): New global var.
15616 (build_rdg): Use loop directly, rather than loop nest.
15617 (pg_add_dependence_edges): Remove loop nest parameter. Use global
15619 (distribute_loop): Compute global variable loop nest. Update use.
15621 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15623 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
15624 (partition_merge_into): New parameter. Dump reason for fusion.
15625 (distribute_loop): Update use of partition_merge_into.
15627 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15629 * tree-loop-distribution.c (bb_top_order_index): New.
15630 (bb_top_order_index_size, bb_top_order_cmp): New.
15631 (stmts_from_loop): Use topological order.
15632 (pass_loop_distribution::execute): Compute and release topological
15633 order for basic blocks.
15635 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15637 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
15640 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15642 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
15643 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
15644 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
15645 * internal-fn.def (LOOP_DIST_ALIAS): New.
15646 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
15647 (fold_loop_internal_call): ... this.
15648 (vect_loop_dist_alias_call): New function.
15649 (set_uid_loop_bbs): Call fold_loop_internal_call.
15650 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
15653 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
15656 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
15657 Require dead FLAGS_REG at the beginning of a peephole.
15659 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
15662 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
15663 arguments in the call to __builtin_ia32_sbb_u32.
15664 (_subborrow_u64): Swap _X and _Y arguments in the call to
15665 __builtin_ia32_sbb_u64.
15667 2017-07-04 Jakub Jelinek <jakub@redhat.com>
15670 * tree-vrp.c (compare_assert_loc): Turn into a function template
15671 with stable template parameter. Only test if a->e is NULL,
15672 !a->e == !b->e has been verified already. Use e == NULL or
15673 e != NULL instead of e or ! e tests. If stable is true, don't use
15674 iterative_hash_expr, on the other side allow a or b or both NULL
15675 and sort the NULLs last.
15676 (process_assert_insertions): Sort using compare_assert_loc<false>
15677 instead of compare_assert_loc, later sort using
15678 compare_assert_loc<true> before calling process_assert_insertions_for
15679 in a loop. Use break instead of continue once seen NULL pointer.
15681 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15683 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
15684 Cortex-R7 and Cortex-R8 processors.
15686 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15688 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
15689 uninitialized while src is not.
15691 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
15693 * common/config/arm/arm-common.c: Adjust include path for
15695 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
15696 (arm-cpu.h): Create in build directory. Adjust dependency rules.
15697 (arm-cpu-data.h): Likewise.
15698 (arm-cpu-cdata.h): Likewise.
15699 * config/arm/arm-cpu.h: Delete.
15700 * config/arm/arm-cpu-cdata.h: Delete.
15701 * config/arm/arm-cpu-data.h: Delete.
15703 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
15705 * config/arm/arm-cpus.in (cortex-a55): New.
15706 (cortex-a75): Likewise.
15707 (cortex-a75.cortex-a55): Likewise.
15708 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
15710 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
15711 * config/arm/arm-cpu-cdata.h: Regenerate.
15712 * config/arm/arm-cpu-data.h: Regenerate.
15713 * config/arm/arm-cpu.h: Regenerate.
15714 * config/arm/arm-tables.opt: Regenerate.
15715 * config/arm/arm-tune.md: Regenerate.
15717 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15719 * haifa-sched.c (sched_create_recovery_edges): Update profile.
15721 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15723 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
15726 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
15728 PR tree-optimization/81292
15729 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
15730 full_string_p, also call adjust_related_strinfos if the adjustment
15731 is simple, otherwise invalidate related strinfos.
15733 2017-07-04 Martin Liska <mliska@suse.cz>
15736 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
15737 newly created variable as DECL_IGNORED_P.
15739 2017-07-04 Martin Liska <mliska@suse.cz>
15742 * ipa-inline.c (inline_small_functions):
15743 Use xstrdup_for_dump.
15745 2017-07-04 Tom de Vries <tom@codesourcery.com>
15747 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
15749 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
15752 * config/darwin.c (darwin_function_switched_text_sections):
15755 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
15757 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
15759 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
15761 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
15763 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15765 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
15766 min_profitable_iters, and th as inclusive lower bounds.
15767 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
15768 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
15769 for min_profitable_iters and min_profitable_estimate.
15770 (vect_transform_loop): Treat th as an inclusive lower bound.
15771 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
15773 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
15776 * config/darwin.c (darwin_function_switched_text_sections):
15777 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
15778 in two pieces, and suppress the use of buf.
15780 2017-07-03 Nathan Sidwell <nathan@acm.org>
15782 * hash-table.h (hash_table_mod1): Fix indentation.
15784 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15786 PR middle-end/81290
15787 * predict.c (force_edge_cold): Be more careful about propagation
15789 * profile-count.h (profile_probability::guessed,
15790 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
15792 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
15794 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
15796 * doc/invoke.texi (rcpc architecture extension): Document it.
15798 2017-07-03 Richard Biener <rguenther@suse.de>
15800 PR tree-optimization/60510
15801 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
15802 the scalar reduction PHI and use it.
15803 (vectorizable_reduction): Properly guard the single_defuse_cycle
15804 path for non-SLP reduction chains where we cannot use it.
15805 Rework reduc_def/index and vector type deduction. Rework
15806 vector operand gathering during reduction op code-gen.
15807 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
15808 chains dissolve the chain and leave it to non-SLP reduction
15811 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15813 * tree-data-ref.h (dr_alignment): Declare.
15814 * tree-data-ref.c (dr_alignment): New function.
15815 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
15816 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
15818 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
15820 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15822 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
15823 and base_misalignment fields.
15824 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
15825 * tree-data-ref.c: Include builtins.h.
15826 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
15827 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
15828 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
15829 * tree-vect-data-refs.c: Include tree-cfg.h.
15830 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
15831 fields instead of calculating an alignment here.
15832 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
15833 innermost_loop_behavior fields.
15835 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15837 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
15839 (DR_STEP_ALIGNMENT): New macro.
15840 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
15841 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
15842 (create_data_ref): Print it.
15843 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
15844 to tell whether the step preserves vector (mis)alignment.
15845 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
15846 Move the check for an integer step and generalise to all INTEGER_CST.
15847 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
15848 Print the outer step alignment.
15850 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15852 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
15853 with offset_alignment.
15854 (DR_ALIGNED_TO): Delete.
15855 (DR_OFFSET_ALIGNMENT): New macro.
15856 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
15857 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
15858 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
15859 (create_data_ref): Likewise.
15860 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
15861 (vect_analyze_data_refs): Likewise.
15862 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
15863 creating dummy innermost behavior.
15865 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15867 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
15868 with a "innermost_loop_behavior *" and refeence tree.
15869 * tree-data-ref.c (dr_analyze_innermost): Likewise.
15870 (create_data_ref): Update call accordingly.
15871 * tree-predcom.c (find_looparound_phi): Likewise.
15873 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15875 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
15876 fields with dr_wrt_vec_loop.
15877 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
15878 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
15879 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
15880 (vect_dr_behavior): New function.
15881 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
15882 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
15883 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
15884 track whether the step preserves the misalignment.
15885 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
15886 Use vect_dr_behavior.
15887 (vect_setup_realignment): Update call accordingly.
15888 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
15889 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
15890 call to vect_create_addr_base_for_vector_ref.
15891 (vect_create_cond_for_align_checks): Likewise.
15892 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
15893 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
15894 (vect_recog_mask_conversion_pattern): Likewise.
15895 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
15896 (new_stmt_vec_info): Remove redundant zeroing.
15898 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
15900 * common/config/arm/arm-common.c (arm_be8_option): New function.
15901 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
15902 (ISA_ARMv6): Add isa_bit_be8.
15903 * config/arm/arm.h (arm_be8_option): Add prototype.
15904 (BE8_SPEC_FUNCTION): New define.
15905 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
15906 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
15907 (mlittle-endian): Similarly.
15908 (mbe8, mbe32): New options.
15909 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
15910 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
15912 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15914 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
15916 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15918 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
15919 (cleanup_tree_cfg_bb): Use it.
15920 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
15922 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
15924 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15927 * loop-doloop.c (add_test): Update profile.
15929 2017-07-03 Martin Liska <mliska@suse.cz>
15932 * sanopt.c (rewrite_usage_of_param): New function.
15933 (sanitize_rewrite_addressable_params): Likewise.
15934 (pass_sanopt::execute): Call rewrite_usage_of_param.
15936 2017-07-03 Richard Biener <rguenther@suse.de>
15938 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
15939 back to using VIEW_CONVERT_EXPR.
15941 2017-07-03 Martin Liska <mliska@suse.cz>
15944 * doc/extend.texi: Document when a resolver function is
15945 generated for target_clones.
15947 2017-07-03 Martin Liska <mliska@suse.cz>
15949 * asan.c (asan_emit_stack_protection): Unpoison just red zones
15950 and shadow memory of auto variables which are subject of
15951 use-after-scope sanitization.
15952 (asan_expand_mark_ifn): Add do set only when is_poison.
15954 2016-07-03 Richard Biener <rguenther@suse.de>
15956 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
15958 (vect_force_simple_reduction): Record reduction def -> phi mapping.
15959 (vectorizable_reduction): Perform reduction PHI creation when
15960 visiting a reduction PHI and adjust and simplify code generation
15961 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
15962 (vect_transform_loop): Visit reduction PHIs.
15963 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
15964 defs into the SLP tree.
15965 (vect_build_slp_tree): Reduction defs terminate the recursion.
15966 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
15968 (vect_get_vec_defs_for_stmt_copy): Export.
15969 (vect_get_vec_defs): Likewise.
15970 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
15972 (vect_get_vec_defs_for_stmt_copy): Declare.
15973 (vect_get_vec_defs): Likewise.
15975 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15977 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
15978 parameter with a "loop" parameter and use it instead of the
15979 loop containing DR_STMT. Don't check simple_iv when doing
15980 BB analysis. Describe the two analysis modes in the comment.
15982 2017-07-03 Tom de Vries <tom@codesourcery.com>
15984 PR tree-optimization/69468
15985 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
15986 (find_same_succ_bb): Handle ignore_edge_flags.
15988 2017-07-03 Tom de Vries <tom@codesourcery.com>
15990 PR tree-optimization/81192
15991 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
15993 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
15995 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
15997 2017-07-03 Tom de Vries <tom@codesourcery.com>
15999 PR tree-optimization/81192
16000 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
16001 BB_SAME_SUCC (bb) == NULL.
16003 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16005 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
16008 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16010 * dumpfile.c: Include profile-count.h
16011 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
16013 (insert_cond_bb): Update profile.
16014 * tree-cfg.h (insert_cond_bb): Update prototype.
16015 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
16016 * tree-dump.c: Do not include tree-cfg.
16018 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16020 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
16022 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16024 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
16027 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16029 * tree-complex.c (expand_complex_div_wide): update profile.
16031 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16032 Alan Hayward <alan.hayward@arm.com>
16033 David Sherwood <david.sherwood@arm.com>
16035 * Makefile.in (MACHMODE_H): Remove insn-modes.h
16036 (CORETYPES_H): New define.
16037 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
16038 (insn-modes-inline.h, s-modes-inline-h): New rules.
16039 (generated_files): Add insn-modes-inline.h.
16040 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
16041 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
16042 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
16043 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
16044 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
16045 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
16046 (build/gencodes.o, build/genconditions.o): Likewise.
16047 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
16048 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
16049 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
16050 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
16051 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
16052 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
16053 * coretypes.h: Include everything up to real.h for generators.
16054 Include insn-modes.h first. Include wide-int-print.h after
16055 wide-int.h. Include insn-modes-inline.h and then machmode.h.
16056 * machmode.h: Don't include insn-modes.h here.
16057 * function-tests.c: Remove includes of signop.h, machmode.h,
16058 double-int.h and wide-int.h.
16060 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
16062 * optc-save-gen.awk: Likewise.
16063 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
16064 * godump.c: Remove include of wide-int-print.h.
16065 * pretty-print.h: Likewise.
16066 * wide-int-print.cc: Likewise.
16067 * wide-int.cc: Likewise.
16068 * hash-map-tests.c: Remove include of signop.h.
16069 * hash-set-tests.c: Likewise.
16070 * rtl-tests.c: Likewise.
16071 * mkconfig.sh: Remove include of machmode.h.
16072 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
16074 (emit_insn_modes_inline_h): ...this new function. Emit the code
16075 into an insn-modes-inline.h header file, adding appropriate
16076 include guards and end comments.
16077 (emit_insn_modes_c_header): Remove include of machmode.h.
16078 (emit_min_insn_modes_c_header): Include coretypes.h rather than
16080 (main): Handle -i flag and call emit_insn_modes_inline_h when
16083 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16085 * tree-ssa-strlen.c (strinfo): Rename the length field to
16086 nonzero_chars. Add a full_string_p field.
16087 (compare_nonzero_chars, zero_length_string_p): New functions.
16088 (get_addr_stridx): Add an offset_out parameter.
16089 Use compare_nonzero_chars.
16090 (get_stridx): Update accordingly. Use compare_nonzero_chars.
16091 (new_strinfo): Update after above changes to strinfo.
16092 (set_endptr_and_length): Set full_string_p.
16093 (get_string_length): Update after above changes to strinfo.
16094 (unshare_strinfo): Update call to new_strinfo.
16095 (maybe_invalidate): Likewise.
16096 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
16097 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
16098 as a uhwi instead of an shwi. Update after above changes to
16099 strinfo and new_strinfo.
16100 (zero_length_string): Assert that chainsi contains full strings.
16101 Use zero_length_string_p. Update call to new_strinfo.
16102 (adjust_related_strinfos): Update after above changes to strinfo.
16103 Copy full_string_p from origsi.
16104 (adjust_last_stmt): Use zero_length_string_p.
16105 (handle_builtin_strlen): Update after above changes to strinfo and
16106 new_strinfo. Install the lhs as the string length if the previous
16107 entry didn't describe a full string.
16108 (handle_builtin_strchr): Update after above changes to strinfo
16110 (handle_builtin_strcpy): Likewise.
16111 (handle_builtin_strcat): Likewise.
16112 (handle_builtin_malloc): Likewise.
16113 (handle_pointer_plus): Likewise.
16114 (handle_builtin_memcpy): Likewise. Track nonzero characters
16115 that aren't necessarily followed by a nul terminator.
16116 (handle_char_store): Likewise.
16118 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16120 PR tree-optimization/80769
16121 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
16122 for malloc and calloc. Document the new invariant that all related
16123 strinfos have delayed lengths or none do.
16124 (verify_related_strinfos): Move earlier in file.
16125 (set_endptr_and_length): New function, split out from...
16126 (get_string_length): ...here. Also set the lengths of related
16128 (zero_length_string): Assert that chainsi has known (rather than
16130 (adjust_related_strinfos): Likewise.
16132 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16134 PR tree-optimization/81136
16135 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
16136 assert that two references with the same misalignment have the same
16137 compile-time misalignment if those compile-time misalignments
16140 2017-07-01 Andi Kleen <ak@linux.intel.com>
16142 * print-tree.c (print_node): Print all attributes.
16144 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16146 * cfg.c (scale_bbs_frequencies): New function.
16147 * cfg.h (scale_bbs_frequencies): Declare it.
16148 * cfgloopanal.c (single_likely_exit): Cleanup.
16149 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
16151 (scale_loop_profile): Likewise.
16152 (loop_version): Likewise.
16153 (create_empty_loop_on_edge): Update.
16154 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
16155 scale_loop_frequencies, scale_loop_profile, loopify,
16156 loop_version): Update prototypes.
16157 * modulo-sched.c (sms_schedule): Update.
16158 * predict.c (unlikely_executed_edge_p): Also check probability.
16159 (probably_never_executed_edge_p): Fix typo.
16160 * tree-if-conv.c (version_loop_for_if_conversion): Update.
16161 * tree-parloops.c (gen_parallel_loop): Update.
16162 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
16163 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16164 * tree-ssa-loop-split.c (split_loop): Update.
16165 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
16166 * tree-vect-loop-manip.c (vect_do_peeling): Update.
16167 (vect_loop_versioning): Update.
16168 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16170 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16172 * trans-mem.c (split_bb_make_tm_edge): Update profile.
16174 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16176 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
16177 to keep profile consistent.
16179 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16181 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
16182 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
16183 * profile-count.h (max_safe_multiplier): Make unsigned.
16184 (profile_count::guessed_zero): New.
16186 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16188 * bb-reorder.c (fix_up_crossing_landing_pad,
16189 fix_crossing_conditional_branches): Use make_single_succ_edge
16190 to keep profile consistent.
16192 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16194 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
16197 2017-07-01 Jakub Jelinek <jakub@redhat.com>
16200 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
16201 the right scopes, make sure cond_jump isn't preserved between multiple
16202 iterations. Search for fallthru edge whenever there are 3+ edges and
16203 use find_fallthru_edge for it.
16205 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16207 Patch by Alexander Monakov <amonakov@ispras.ru>
16208 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
16209 probabilities consistently.
16211 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16213 * pa.c (pa_expand_compare_and_swap_loop): Update call of
16214 emit_cmp_and_jump_insns.
16216 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16219 * tree-inline.c (expand_call_inline): Combine profile statuses.
16221 2017-06-30 Andrew Pinski <apinski@cavium.com>
16223 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
16224 fold_stmt returned true.
16226 2017-06-30 Nathan Sidwell <nathan@acm.org>
16228 * ggc.h (empty_string): Delete.
16229 * cfgexpand.c (expand_asm_stmt): Use plain "".
16230 * optabs.c (expand_asm_memory_barrier): Likewise.
16231 * stringpool.c (empty_string): Delete.
16232 (digit_vector, digit_string): Delete.
16233 (ggc_alloc_string): Use plain "", don't optimize single digit
16234 strings. Use ggc_alloc_atomic.
16236 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
16238 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
16239 comparison set and one other set, use the cost of the non-comparison
16242 2017-06-30 Nathan Sidwell <nathan@acm.org>
16244 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
16247 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
16249 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
16250 loops. Remove now unneeded calls to gimple_switch_set_label() that
16251 just set removed labels to NULL_TREE.
16253 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
16255 * tree-ssanames.c (set_range_info_raw): Abstract from ...
16256 (set_range_info): ...here. Only call set_range_info_raw if domain
16258 (set_nonzero_bits): Call set_range_info_raw.
16259 * tree-ssanames.h (set_range_info_raw): New.
16261 2017-06-30 Jakub Jelinek <jakub@redhat.com>
16264 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
16265 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
16266 of nonimmediate_operand and <store_mask_constraint> instead of m
16267 for the input operand. For V8FI iterator, always split if input
16268 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
16269 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
16270 <store_mask_predicate> instead of register_operand and
16271 <store_mask_constraint> instead of v for the input operand. Make
16272 sure both operands aren't MEMs for if not <mask_applied>.
16274 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
16276 * lto-wrapper.c (copy_file) Close both file descriptors before
16279 2017-06-30 Martin Liska <mliska@suse.cz>
16282 * multiple_target.c (create_dispatcher_calls): Make ifunc
16283 also for function that don't have calls or are not referenced.
16285 2017-06-30 Richard Biener <rguenther@suse.de>
16287 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
16288 analyze the first scalar stmt. Move vector type computation
16289 for the BB case here from ...
16290 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
16291 live operation processing in the SLP case properly.
16293 2017-06-30 Richard Biener <rguenther@suse.de>
16295 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
16297 2017-06-30 Martin Liska <mliska@suse.cz>
16300 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
16301 before BUILT_IN_UNWIND_RESUME when ASAN is used.
16303 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
16305 * doc/invoke.texi (AArch64): Add missing options and remove redundant
16308 2017-06-30 Richard Biener <rguenther@suse.de>
16310 PR tree-optimization/81249
16311 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
16312 condition reduction result to original scalar type.
16314 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16316 * profile-count.h (enum profile_quality): Fix typos and whitespace
16319 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16321 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
16322 type for branch probabilities.
16324 2017-06-29 Julian Brown <julian@codesourcery.com>
16325 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16327 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
16328 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
16329 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
16330 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
16332 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16334 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
16335 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
16336 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
16337 CC usage from generic code to here.
16338 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
16339 CC usage into the target macros.
16341 2017-06-29 Maya Rashish <coypu@sdf.org>
16343 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
16346 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16348 * arm/arm-builtins.c: Include profile-count.h
16349 * except.c (sjlj_emit_function_enter): Use
16350 profile_probability::unlikely.
16352 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16354 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
16355 and tocrel_offset be pointer args rather than implicitly using
16357 (legitimate_constant_pool_address_p, rs6000_emit_move,
16358 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
16359 tocrel_offset and use in toc_relative_expr_p call.
16360 (print_operand, print_operand_address): Use static tocrel_base_oac
16361 and tocrel_offset_oac.
16362 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
16365 2017-06-29 Maya Rashish <coypu@sdf.org>
16367 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
16369 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
16371 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
16372 objects, take into account only the alignment of 'op0' and 'mode1' if
16375 2017-06-29 Steve Ellcey <sellcey@cavium.com>
16377 * ccmp.c (ccmp_tree_comparison_p): New function.
16378 (ccmp_candidate_p): Update to use above function.
16379 (get_compare_parts): New function.
16380 (expand_ccmp_next): Update to use new functions.
16381 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
16383 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
16384 take mode as argument.
16385 * ccmp.h (expand_ccmp_expr): Add mode as argument.
16386 * expr.c (expand_expr_real_1): Pass mode as argument.
16388 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
16390 * combine.c (combine_instructions): Print insns to dump_file, together
16393 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16395 * asan.c (asan_emit_stack_protection): Update.
16396 (create_cond_insert_point): Update.
16397 * auto-profile.c (afdo_propagate_circuit): Update.
16398 * basic-block.h (struct edge_def): Turn probability to
16399 profile_probability.
16400 (EDGE_FREQUENCY): Update.
16401 * bb-reorder.c (find_traces_1_round): Update.
16402 (better_edge_p): Update.
16403 (sanitize_hot_paths): Update.
16404 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
16405 (make_single_succ_edge): Update.
16406 (check_bb_profile): Update.
16407 (dump_edge_info): Update.
16408 (update_bb_profile_for_threading): Update.
16409 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
16410 probabilitycount to 0.
16411 * cfgbuild.c (compute_outgoing_frequencies): Update.
16412 * cfgcleanup.c (try_forward_edges): Update.
16413 (outgoing_edges_match): Update.
16414 (try_crossjump_to_edge): Update.
16415 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
16416 (expand_gimple_tailcall): Update.
16417 (construct_init_block): Use make_single_succ_edge.
16418 (construct_exit_block): Use make_single_succ_edge.
16419 * cfghooks.c (verify_flow_info): Update.
16420 (redirect_edge_succ_nodup): Update.
16421 (split_edge): Update.
16422 (account_profile_record): Update.
16423 * cfgloopanal.c (single_likely_exit): Update.
16424 * cfgloopmanip.c (scale_loop_profile): Update.
16425 (set_zero_probability): Remove.
16426 (duplicate_loop_to_header_edge): Update.
16427 * cfgloopmanip.h (loop_version): Update prototype.
16428 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
16429 (force_nonfallthru_and_redirect): Update.
16430 (update_br_prob_note): Update.
16431 (rtl_verify_edges): Update.
16432 (purge_dead_edges): Update.
16433 (rtl_lv_add_condition_to_bb): Update.
16434 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
16435 * cgraphunit.c (init_lowered_empty_function): Update.
16436 (cgraph_node::expand_thunk): Update.
16437 * cilk-common.c: Include profile-count.h
16438 * dojump.c (inv): Remove.
16439 (jumpifnot): Update.
16440 (jumpifnot_1): Update.
16441 (do_jump_1): Update.
16443 (do_jump_by_parts_greater_rtx): Update.
16444 (do_compare_rtx_and_jump): Update.
16445 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
16446 do_jump_1. do_compare_rtx_and_jump): Update prototype.
16447 * dwarf2cfi.c: Include profile-count.h
16448 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
16449 (sjlj_emit_dispatch_table): Likewise.
16450 * explow.c: Include profile-count.h
16451 * expmed.c (emit_store_flag_force): Update.
16452 (do_cmp_and_jump): Update.
16453 * expr.c (compare_by_pieces_d::generate): Update.
16454 (compare_by_pieces_d::finish_mode): Update.
16455 (emit_block_move_via_loop): Update.
16456 (store_expr_with_bounds): Update.
16457 (store_constructor): Update.
16458 (expand_expr_real_2): Update.
16459 (expand_expr_real_1): Update.
16460 * expr.h (try_casesi, try_tablejump): Update prototypes.
16461 * gimple-pretty-print.c (dump_probability): Update.
16462 (dump_profile): New.
16463 (dump_gimple_label): Update.
16464 (dump_gimple_bb_header): Update.
16465 * graph.c (draw_cfg_node_succ_edges): Update.
16466 * hsa-gen.c (convert_switch_statements): Update.
16467 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
16468 (find_if_case_1): Update.
16469 (find_if_case_2): Update.
16470 * internal-fn.c (expand_arith_overflow_result_store): Update.
16471 (expand_addsub_overflow): Update.
16472 (expand_neg_overflow): Update.
16473 (expand_mul_overflow): Update.
16474 (expand_vector_ubsan_overflow): Update.
16475 * ipa-cp.c (good_cloning_opportunity_p): Update.
16476 * ipa-split.c (split_function): Use make_single_succ_edge.
16477 * ipa-utils.c (ipa_merge_profiles): Update.
16478 * loop-doloop.c (add_test): Update.
16479 (doloop_modify): Update.
16480 * loop-unroll.c (compare_and_jump_seq): Update.
16481 (unroll_loop_runtime_iterations): Update.
16482 * lra-constraints.c (lra_inheritance): Update.
16483 * lto-streamer-in.c (input_cfg): Update.
16484 * lto-streamer-out.c (output_cfg): Update.
16485 * mcf.c (adjust_cfg_counts): Update.
16486 * modulo-sched.c (sms_schedule): Update.
16487 * omp-expand.c (expand_omp_for_init_counts): Update.
16488 (extract_omp_for_update_vars): Update.
16489 (expand_omp_ordered_sink): Update.
16490 (expand_omp_for_ordered_loops): Update.
16491 (expand_omp_for_generic): Update.
16492 (expand_omp_for_static_nochunk): Update.
16493 (expand_omp_for_static_chunk): Update.
16494 (expand_cilk_for): Update.
16495 (expand_omp_simd): Update.
16496 (expand_omp_taskloop_for_outer): Update.
16497 (expand_omp_taskloop_for_inner): Update.
16498 * omp-simd-clone.c (simd_clone_adjust): Update.
16499 * optabs.c (expand_doubleword_shift): Update.
16500 (expand_abs): Update.
16501 (emit_cmp_and_jump_insn_1): Update.
16502 (expand_compare_and_swap_loop): Update.
16503 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
16504 * predict.c (predictable_edge_p): Update.
16505 (edge_probability_reliable_p): Update.
16506 (set_even_probabilities): Update.
16507 (combine_predictions_for_insn): Update.
16508 (combine_predictions_for_bb): Update.
16509 (propagate_freq): Update.
16510 (estimate_bb_frequencies): Update.
16511 (force_edge_cold): Update.
16512 * profile-count.c (profile_count::dump): Add missing space into dump.
16513 (profile_count::debug): Add newline.
16514 (profile_count::differs_from_p): Explicitly convert to unsigned.
16515 (profile_count::stream_in): Update.
16516 (profile_probability::dump): New member function.
16517 (profile_probability::debug): New member function.
16518 (profile_probability::differs_from_p): New member function.
16519 (profile_probability::differs_lot_from_p): New member function.
16520 (profile_probability::stream_in): New member function.
16521 (profile_probability::stream_out): New member function.
16522 * profile-count.h (profile_count_quality): Rename to ...
16523 (profile_quality): ... this one.
16524 (profile_probability): New.
16525 (profile_count): Update.
16526 * profile.c (compute_branch_probabilities): Update.
16527 * recog.c (peep2_attempt): Update.
16528 * sched-ebb.c (schedule_ebbs): Update.
16529 * sched-rgn.c (find_single_block_region): Update.
16530 (compute_dom_prob_ps): Update.
16531 (schedule_region): Update.
16532 * sel-sched-ir.c (compute_succs_info): Update.
16533 * stmt.c (struct case_node): Update.
16534 (do_jump_if_equal): Update.
16535 (get_outgoing_edge_probs): Update.
16536 (conditional_probability): Update.
16537 (emit_case_dispatch_table): Update.
16538 (expand_case): Update.
16539 (expand_sjlj_dispatch_table): Update.
16540 (emit_case_nodes): Update.
16541 * targhooks.c: Update.
16542 * tracer.c (better_p): Update.
16543 (find_best_successor): Update.
16544 * trans-mem.c (expand_transaction): Update.
16545 * tree-call-cdce.c: Update.
16546 * tree-cfg.c (gimple_split_edge): Upate.
16547 (move_sese_region_to_fn): Upate.
16548 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
16549 * tree-eh.c (lower_resx): Upate.
16550 (cleanup_empty_eh_move_lp): Upate.
16551 * tree-if-conv.c (version_loop_for_if_conversion): Update.
16552 * tree-inline.c (copy_edges_for_bb): Update.
16553 (copy_cfg_body): Update.
16554 * tree-parloops.c (gen_parallel_loop): Update.
16555 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
16556 (gimple_gen_time_profiler): Update.
16557 * tree-ssa-dce.c (remove_dead_stmt): Update.
16558 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
16559 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
16560 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
16561 (unloop_loops): Update.
16562 (try_peel_loop): Update.
16563 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16564 * tree-ssa-loop-split.c (connect_loops): Update.
16565 (split_loop): Update.
16566 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
16567 (hoist_guard): Update.
16568 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
16569 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
16570 (value_replacement): Update.
16571 * tree-ssa-reassoc.c (branch_fixup): Update.
16572 * tree-ssa-tail-merge.c (replace_block_by): Update.
16573 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
16574 (create_edge_and_update_destination_phis): Update.
16575 (compute_path_counts): Update.
16576 (recompute_probabilities): Update.
16577 (update_joiner_offpath_counts): Update.
16578 (freqs_to_counts_path): Update.
16579 (duplicate_thread_path): Update.
16580 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
16581 (struct switch_conv_info): Update.
16582 (gen_inbound_check): Update.
16583 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
16584 (vect_do_peeling): Update.
16585 (vect_loop_versioning): Update.
16586 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16587 (optimize_mask_stores): Update.
16588 * ubsan.c (ubsan_expand_null_ifn): Update.
16589 * value-prof.c (gimple_divmod_fixed_value): Update.
16590 (gimple_divmod_fixed_value_transform): Update.
16591 (gimple_mod_pow2): Update.
16592 (gimple_mod_pow2_value_transform): Update.
16593 (gimple_mod_subtract): Update.
16594 (gimple_mod_subtract_transform): Update.
16595 (gimple_ic): Update.
16596 (gimple_stringop_fixed_value): Update.
16597 (gimple_stringops_transform): Update.
16598 * value-prof.h: Update.
16600 2017-06-29 Carl Love <cel@us.ibm.com>
16602 * config/rs6000/rs6000-c.c: Add support for built-in functions
16603 vector signed int vec_signed (vector float);
16604 vector signed long long vec_signed (vector double);
16605 vector signed int vec_signed2 (vector double, vector double);
16606 vector signed int vec_signede (vector double);
16607 vector signed int vec_signedo (vector double);
16608 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
16609 instruction generator.
16610 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
16611 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
16612 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
16614 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
16615 vunsignede_v2df): Add define_expands.
16616 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
16617 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
16618 VEC_UNSIGNEDO): Add definitions.
16619 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
16620 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
16621 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
16622 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
16623 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
16624 * config/rs6000/altivec.h (vec_signed, vec_signed2,
16625 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
16626 vec_unsignede, vec_unsignedo): Add builtin defines.
16627 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
16629 * doc/extend.texi: Update the built-in documentation file for the
16630 new built-in functions.
16632 2017-06-29 Richard Biener <rguenther@suse.de>
16634 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
16635 reduction chains to LOOP_VINFO_REDUCTIONS.
16636 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
16637 SLP reductions after processing reduction chains.
16639 2017-06-29 Nathan Sidwell <nathan@acm.org>
16641 * builtins.c (fold_builtin_FUNCTION): Use
16642 lang_hooks.decl_printable_name.
16644 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
16646 PR middle-end/81194
16647 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
16648 with only one label.
16649 * stmt.c (expand_case): Assert NCASES is greater than one.
16651 2017-06-29 Richard Biener <rguenther@suse.de>
16653 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
16655 (group_case_labels): Likewise.
16656 (find_taken_edge): Push sanity checking on val to workers...
16657 (find_taken_edge_cond_expr): ... here
16658 (find_taken_edge_switch_expr): ... and here, handle cases
16659 with just a default label.
16660 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
16661 (group_case_labels): Likewise.
16662 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
16663 group_case_labels does anything cleanup the CFG again.
16665 2017-06-29 Bin Cheng <bin.cheng@arm.com>
16667 PR tree-optimization/81196
16668 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
16669 exit condition comparing two IVs.
16671 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
16673 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
16674 profile to the dummy entry at the end of the list of architectures.
16675 * config/arm/arm-cpu-cdata.h: Regenerated.
16677 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16678 Michael Collison <michael.collison@arm.com>
16681 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
16683 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
16684 (*aarch64_reg_<mode>3_minus_mask): New pattern.
16685 (*aarch64_<optab>_reg_di3_mask2): New pattern.
16686 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
16687 of shift when the shift amount is masked with constant equal to
16688 the size of the mode.
16689 * config/aarch64/predicates.md (subreg_lowpart_operator): New
16692 2017-06-29 Martin Liska <mliska@suse.cz>
16694 * config/i386/i386.opt: Change range from [1,5] to [0,5].
16696 2017-06-29 Yury Gribov <tetra2005@gmail.com>
16699 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
16702 (edge_growth_cache_entry::edge_growth_cache_entry): New
16704 (reset_edge_growth_cache): Update to use constructor.
16706 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16708 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
16709 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
16710 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
16712 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
16714 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
16715 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
16717 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
16719 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
16720 (*-linux-uclibc*): Add t-uclibc tmake_file.
16721 * config/t-musl: New.
16722 * config/t-uclibc: New.
16724 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
16726 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
16728 (gen_comm_data): Emit architectural setting of arch_prof.
16729 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
16731 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
16732 (armv8-m.base, armv8-m.main): Likewise.
16733 * arm-protos.h (arm_build_target): Add profile field.
16734 (arch_option): Likewise.
16735 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
16737 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
16738 arm_active_target.profile.
16740 2017-06-28 Richard Biener <rguenther@suse.de>
16742 PR middle-end/81227
16743 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
16744 TYPE_OVERFLOW_WRAPS.
16745 * match.pd (negate_expr_p): Likewise.
16746 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
16747 fold_build2, not fold_binary.
16749 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
16751 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
16752 Convert memory address to Pmode.
16753 (aarch64_print_operand): Assert MEM operands are always Pmode.
16755 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
16758 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
16759 Remove redundant if.
16760 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
16761 * config/arm/aarch-common-protos.h
16762 (aarch_forward_to_shift_is_not_shifted_re): Remove.
16763 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
16765 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
16768 * multiple_target.c (create_dispatcher_calls): Set the default
16769 clone to be static, not public.
16771 2017-06-28 Richard Biener <rguenther@suse.de>
16773 * tree-vect-loop.c (vectorizable_reduction): Move special
16774 cond reduction IV var creation ...
16775 (vect_create_epilog_for_reduction): ... here. Remove induction_index
16776 parameter. Use STMT_VINFO_VECTYPE.
16777 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
16780 2017-06-28 Martin Liska <mliska@suse.cz>
16783 * ipa-visibility.c (non_local_p): Handle visibility.
16785 2017-06-28 Martin Liska <mliska@suse.cz>
16788 * common.opt: Add IntegerRange to various options.
16789 * opt-functions.awk (integer_range_info): New function.
16790 * optc-gen.awk: Add integer_range_info to cl_options struct.
16791 * opts-common.c (decode_cmdline_option): Handle
16792 CL_ERR_INT_RANGE_ARG.
16793 (cmdline_handle_error): Likewise.
16794 * opts.c (print_filtered_help): Show valid interval in
16795 when --help is provided.
16796 * opts.h (struct cl_option): Add range_min and range_max fields.
16797 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
16799 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
16801 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
16802 (x * C EQ/NE y * C): New transformation.
16804 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
16806 * genmultilib (combination_space): Accept '+' in option names.
16808 2017-06-28 Martin Liska <mliska@suse.cz>
16811 * asan.c (instrument_derefs): Bail out inner references
16812 that are hard register variables.
16814 2017-06-28 Jakub Jelinek <jakub@redhat.com>
16817 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
16818 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
16820 2017-06-28 Richard Biener <rguenther@suse.de>
16822 * tree-vectorizer.h (vect_get_vec_defs): Remove.
16823 (vect_get_slp_defs): Adjust.
16824 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
16826 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
16828 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
16829 get_initial_defs_for_reduction instead of vect_get_vec_defs.
16830 (vectorizable_reduction): Adjust.
16831 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
16833 (vect_get_slp_defs): Likewise.
16834 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
16835 (vectorizable_bswap): Adjust.
16836 (vectorizable_call): Likewise.
16837 (vectorizable_conversion): Likewise.
16838 (vectorizable_assignment): Likewise.
16839 (vectorizable_shift): Likewise.
16840 (vectorizable_operation): Likewise.
16841 (vectorizable_store): Likewise.
16842 (vectorizable_condition): Likewise.
16843 (vectorizable_comparison): Likewise.
16845 2017-06-28 Michael Collison <michael.collison@arm.com>
16848 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
16850 (arm_gen_movmemqi): Removed unused variable 'i'.
16851 Convert 'for' loop into 'while' loop.
16852 (arm_expand_prologue): Remove last unnecessary set of insn.
16853 (thumb_pop): Remove unused variable 'pushed_words'.
16854 (thumb_exit): Remove last unnecessary set of regs_to_pop.
16856 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16858 * config/s390/predicates.md: Use s390_rel_address_ok_p.
16859 * config/s390/s390-protos.h: Add prototype of
16860 s390_rel_address_ok_p.
16861 * config/s390/s390.c (s390_got_symbol): New function.
16862 (s390_rel_address_ok_p): New function.
16863 (legitimize_pic_address): Use s390_rel_address_ok_p.
16864 (s390_load_got): Use s390_got_symbol.
16865 (s390_option_override): Issue error if
16866 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
16867 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
16869 * config/s390/s390.opt: New option mpic-data-is-text-relative.
16871 2017-06-27 Andrew Pinski <apinski@cavium.com>
16873 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
16874 (X * copysign (1.0, X)): New pattern.
16875 (X * copysign (1.0, -X)): New pattern.
16876 (copysign (-1.0, CST)): New pattern.
16878 2017-06-27 Joseph Myers <joseph@codesourcery.com>
16880 * genmultilib (combination_space): Remove variable.
16881 Validate reuse rules against regular expression for any sequence
16882 of multilib options in any order.
16884 2017-06-27 Michael Collison <michael.collison@arm.com>
16886 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
16887 call aarch64_split_simd_combine.
16888 * (aarch64_combine_internal<mode>): Delete pattern.
16889 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
16890 Allow register and subreg operands.
16892 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16894 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
16895 specific need, just fallback on defaults.
16896 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
16898 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16899 Olivier Hainque <hainque@adacore.com>
16901 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
16903 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
16904 targets. Pick a default if no particular attempt applied.
16905 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
16908 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16910 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
16911 (x86_64-wrs-vxworks7): Likewise.
16913 2017-06-27 Marek Polacek <polacek@redhat.com>
16916 * ubsan.c (instrument_null): Check get_base_address's result for null.
16918 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
16920 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
16922 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
16924 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
16925 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
16926 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
16927 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
16928 New function types.
16929 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
16930 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
16931 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
16932 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
16933 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
16934 BUILT_IN_FEUPDATEENV): New builtins.
16935 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
16936 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
16937 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
16938 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
16940 (builtin_structptr_types): Adjust size.
16941 * tree.c (builtin_structptr_types): Add four entries.
16943 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16944 Olivier Hainque <hainque@adacore.com>
16946 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
16947 (TLS_SYM): New local macro, forcing reference to __tls__ on
16948 link command lines for VxWorks 7 RTPs, triggering initialization
16950 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
16951 OS features TLS support, true for RTPs on VxWorks 7.
16952 * config/vxworks.c (vxworks_override_options): Setup emutls
16955 2017-06-27 Jakub Jelinek <jakub@redhat.com>
16957 * predict.c (test_prediction_value_range): Use -1U instead of -1
16958 to avoid narrowing conversion warning.
16959 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
16960 to avoid narrowing conversion warning.
16961 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
16963 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
16965 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16967 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
16968 64bit configurations.
16969 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
16970 (SIZE_TYPE): Likewise.
16971 * config/vxworks.c (vxworks_emutls_var_fields): Use
16972 long_unsigned_type_node instead of unsigned_type_node as the offset
16973 field type, which is "pointer" mode in emutls.c.
16975 2017-06-27 Jakub Jelinek <jakub@redhat.com>
16978 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
16980 PR middle-end/81207
16981 * gimple-fold.c (replace_call_with_call_and_fold): Handle
16982 gimple_vuse copying separately from gimple_vdef copying.
16984 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16986 * value-prof.c (free_hist): Remove call to memset and the enclosing if
16989 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
16990 Olivier Hainque <hainque@adacore.com>
16992 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
16993 for all vxworks7 targets.
16994 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
16995 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
16996 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
16997 variations for VX6/VX7 and 32/64bits later on in ...
16998 (VXWORKS_LIB_SPEC): Leverage new macros.
16999 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
17000 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
17002 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
17004 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
17005 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
17007 2017-06-26 Carl Love <cel@us.ibm.com>
17009 * config/rs6000/rs6000-c.c: Add support for built-in functions
17010 vector bool char vec_reve (vector bool char);
17011 vector signed char vec_reve (vector signed char);
17012 vector unsigned char vec_reve (vector unsigned char);
17013 vector bool int vec_reve (vector bool int);
17014 vector signed int vec_reve (vector signed int);
17015 vector unsigned int vec_reve (vector unsigned int);
17016 vector bool long long vec_reve (vector bool long long);
17017 vector signed long long vec_reve (vector signed long long);
17018 vector unsigned long long vec_reve (vector unsigned long long);
17019 vector bool short vec_reve (vector bool short);
17020 vector signed short vec_reve (vector signed short);
17021 vector double vec_reve (vector double);
17022 vector float vec_reve (vector float);
17023 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
17024 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
17025 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
17026 (altivec_vreve): New pattern.
17027 * config/rs6000/altivec.h (vec_reve): New define.
17028 * doc/extend.texi (vec_rev): Update the built-in documentation file
17029 for the new built-in functions.
17031 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17033 PR tree-optimization/71815
17034 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
17036 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
17038 (slsr_process_phi): Likewise.
17039 (replace_uncond_cands_and_profitable_phis): Don't replace a
17040 multiply candidate with a stride of 1 (copy or cast).
17041 (phi_incr_cost): Call uses_consumed_by_stmt rather than
17043 (lowest_cost_path): Likewise.
17044 (total_savings): Likewise.
17046 2017-06-26 Richard Biener <rguenther@suse.de>
17049 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
17050 Use def_builtin_pure for all gather builtins.
17052 2017-06-26 Richard Biener <rguenther@suse.de>
17054 PR tree-optimization/81203
17055 * tree-tailcall.c (find_tail_calls): Do not move stmts into
17056 non-dominating BBs.
17058 2017-06-26 Marek Polacek <polacek@redhat.com>
17061 * doc/invoke.texi: Document -Wmultistatement-macros.
17063 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
17065 * doc/sourcebuild.texi (ARM-specific attributes): Document new
17066 arm_neon_ok_no_float_abi effective target.
17068 2017-06-26 Richard Biener <rguenther@suse.de>
17070 PR tree-optimization/80928
17071 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
17072 (copy_bbs): Set BB_DUPLICATED flag early.
17073 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
17075 (execute_on_shrinking_pred): Likewise.
17076 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
17077 BB_DUPLICATED blocks.
17078 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
17079 iterate over all PHIs considering removal of *gsi.
17081 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
17083 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
17086 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17088 * config/rs6000/rs6000-string.c: (expand_block_clear,
17089 do_load_for_compare, select_block_compare_mode,
17090 compute_current_alignment, expand_block_compare,
17091 expand_strncmp_align_check, expand_strn_compare,
17092 expand_block_move, rs6000_output_load_multiple)
17093 Move functions related to string/block move/compare
17094 to a separate file.
17095 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
17096 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
17097 for this function which is now used in two files.
17098 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
17099 * config.gcc: Add rs6000-string.o to extra_objs for
17100 targets powerpc*-*-* and rs6000*-*-*.
17102 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
17105 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
17106 32-bit, since indexed is not valid for DImode.
17107 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
17108 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
17109 (define_peephole2 for Altivec d-form load): Add 32-bit support.
17110 (define_peephole2 for Altivec d-form store): Likewise.
17113 * multiple_target.c (create_dispatcher_calls): Only create the
17114 dispatcher call if the function is the default clone of a
17115 versioned function.
17117 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
17119 PR middle-end/80902
17120 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
17121 a call, force the call to not be a tail call.
17123 2017-06-23 Jeff Law <law@redhat.com>
17125 * doc/contrib.texi: Add entry for Steven Pemberton's work on
17128 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
17130 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
17131 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
17132 handling for early expansion of vector shifts (sl,sr,sra,rl).
17133 (builtin_function_type): Add vector shift right instructions
17134 to the unsigned argument list.
17136 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
17138 rtl-optimizatoin/79286
17139 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
17140 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
17141 trap. PIC register plus a const unspec without offset can never trap.
17143 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
17145 * tree.h (builtin_structptr_type): New type.
17146 (builtin_structptr_types): Declare new array.
17147 * tree.c (builtin_structptr_types): New array.
17148 (free_lang_data, build_common_tree_nodes): Use it.
17150 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
17153 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
17156 2017-06-22 Matt Turner <mattst88@gmail.com>
17158 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
17159 Lake models to skylake case. Assume skylake for unknown
17160 models with clflushopt.
17162 2017-06-22 Jeff Law <law@redhat.com>
17164 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
17165 frame sizes that do not satisfy aarch64_uimm12_shift.
17167 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
17169 * profile-count.h (apply_probability,
17170 apply_scale, probability_in): Fix checks for zero.
17172 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17174 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
17175 * doc/cppdiropts.texi (-I @var{dir}): Document it.
17177 2016-06-22 Richard Biener <rguenther@suse.de>
17179 * tree-vect-loop.c (vect_model_reduction_cost): Handle
17180 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
17181 REDUC_MAX_EXPR support.
17182 (vectorizable_reduction): Likewise.
17183 (vect_create_epilog_for_reduction): Likewise.
17185 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
17187 * match.pd (A / (1 << B) -> A >> B): New.
17188 * generic-match-head.c: Include optabs-tree.h.
17189 * gimple-match-head.c: Likewise.
17190 * optabs-tree.h (target_supports_op_p): New.
17191 * optabs-tree.c (target_supports_op_p): New.
17193 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17195 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
17196 $gcc_cv_ld --help output.
17197 (gcc_cv_ld_demangle): Likewise.
17198 (gcc_cv_ld_eh_frame_hdr): Likewise.
17199 (gcc_cv_ld_pie): Likewise.
17200 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
17201 (gcc_cv_ld_buildid): Likewise.
17202 (gcc_cv_ld_sysroot): Likewise.
17203 (ld_bndplt_support): Likewise.
17204 (ld_pushpopstate_support): Likewise.
17205 * configure: Regenerate.
17206 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
17208 2017-06-21 Jakub Jelinek <jakub@redhat.com>
17211 * config/i386/sse.md (round<mode>2): Renumber match_dup and
17212 operands indexes to avoid gap between operands and match_dups.
17214 2017-06-21 Andrew Pinski <apinski@cavium.com>
17216 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
17217 Increment Arith_shift and Arith_shift_reg by 1.
17218 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
17220 * config/aarch64/aarch64.c (thunderx_tunings): Enable
17221 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
17222 (aarch64_strip_extend): Add new argument and test for it.
17223 (aarch64_cheap_mult_shift_p): New function.
17224 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
17225 add a cost if it is true.
17226 Update calls to aarch64_strip_extend.
17227 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
17229 2017-06-21 Andrew Pinski <apinski@cavium.com>
17231 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
17233 (thunderxt88): Likewise.
17234 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
17235 (thunderx_prefetch_tune): New variable.
17236 (thunderx2t99_prefetch_tune): Update for the correct values.
17237 (thunderxt88_tunings): New variable.
17238 (thunderx_tunings): Use thunderx_prefetch_tune instead of
17239 generic_prefetch_tune.
17240 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
17242 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17244 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
17245 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
17246 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
17247 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
17248 (aarch64_atomic_cas<mode>, GPI): Likewise.
17250 2017-06-21 Martin Liska <mliska@suse.cz>
17252 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
17253 statements on cold and hot labels.
17254 * predict.c (tree_estimate_probability_bb): Remove the
17255 prediction from this place.
17257 2017-06-21 Martin Liska <mliska@suse.cz>
17259 PR tree-optimization/79489
17260 * gimplify.c (maybe_add_early_return_predict_stmt): New
17262 (gimplify_return_expr): Call the function.
17263 * predict.c (tree_estimate_probability_bb): Remove handling
17265 * predict.def: Update comment about early return predictor.
17266 * gimple-predict.h (is_gimple_predict): New function.
17267 * predict.def: Change default value of early return to 66.
17268 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
17270 * passes.def: Put pass_strip_predict_hints to the beginning of
17273 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
17275 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
17276 FUNCTION_DECL declarations.
17277 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
17279 (dwaf2out_decl): Likewise.
17280 * godump.c (go_early_global_decl): Skip call to the real debug hook
17281 for FUNCTION_DECL declarations.
17282 * passes.c (rest_of_decl_compilation): Skip call to the
17283 early_global_decl debug hook for FUNCTION_DECL declarations, unless
17284 -fdump-go-spec is passed.
17286 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
17288 * config/i386/i386.c (struct builtin_isa): New field pure_p.
17289 Reorder for compactness.
17290 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
17291 (def_builtin_pure, def_builtin_pure2): New functions.
17292 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
17294 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
17296 * match.pd (nop_convert): New predicate.
17297 ((A +- CST1) +- CST2): Allow some NOP conversions.
17299 2017-06-21 Jakub Jelinek <jakub@redhat.com>
17302 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
17303 with ctors/dtors if GOVD_SHARED is set.
17305 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
17307 * config/aarch64/aarch64.md (movti_aarch64):
17308 Emit mov rather than orr.
17309 (movtf_aarch64): Likewise.
17310 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
17311 Emit mov rather than orr.
17313 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
17315 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
17316 Swap alternatives, make integer dup more expensive.
17318 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
17320 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
17321 Return true for non-tls symbols.
17323 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
17325 * config/aarch64/aarch64-cores.def (cortex-a55): New.
17326 (cortex-a75): Likewise.
17327 (cortex-a75.cortex-a55): Likewise.
17328 * config/aarch64/aarch64-tune.md: Regenerate.
17329 * doc/invoke.texi (-mtune): Document new values for -mtune.
17331 2017-06-21 Tom de Vries <tom@codesourcery.com>
17333 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
17334 stack_size feature.
17335 (Effective-Target Keywords, Other attributes): Suggest using
17336 dg-add-options stack_size feature to get stack limit in stack_size
17337 effective target documentation.
17339 2017-06-21 Julian Brown <julian@codesourcery.com>
17340 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17342 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
17343 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
17344 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
17346 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
17347 attribute type list for neon_multiply.
17348 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
17349 attribute type list for neon_multiply.
17350 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
17351 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
17352 attribute type list for neon_multiply.
17353 * config/arm/types.md (crypto_pmull): Add.
17354 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
17355 attribute type list.
17357 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
17359 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
17362 2017-06-20 Jakub Jelinek <jakub@redhat.com>
17364 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
17365 to make sure not to dereference a NULL cost_classes_ptr pointer.
17367 2017-06-20 Carl Love <cel@us.ibm.com>
17369 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17370 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17371 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17372 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17373 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17374 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17375 VMULOSW): New enum "unspec" values.
17376 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17377 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17378 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17379 altivec_vmulosw): New patterns.
17380 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17381 VMULOSW): Add definitions.
17383 2017-06-20 Julia Koval <julia.koval@intel.com>
17385 * config/i386/i386.c: Fix rounding expand for new pattern.
17386 * config/i386/subst.md: Fix pattern (parallel -> unspec).
17388 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17390 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
17391 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
17393 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17395 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
17398 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17400 * config/aarch64/aarch64-cores.def: Rearrange to sort by
17401 architecture, then by implementer ID.
17402 * config/aarch64/aarch64-tune.md: Regenerate.
17404 2017-06-20 Richard Biener <rguenther@suse.de>
17406 PR middle-end/81097
17407 * fold-const.c (split_tree): Fold to type before negating.
17409 2017-06-20 David Malcolm <dmalcolm@redhat.com>
17411 * diagnostic-show-locus.c
17412 (selftest::test_fixit_deletion_affecting_newline): New function.
17413 (selftest::diagnostic_show_locus_c_tests): Call it.
17415 2017-06-20 Andreas Schwab <schwab@suse.de>
17418 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
17421 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17423 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
17424 __ARM_FEATURE_COPROC according to support.
17426 2017-06-20 Jakub Jelinek <jakub@redhat.com>
17428 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
17429 Rewritten to avoid overflow for > 32-bit pointers.
17432 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
17433 by removing enum keyword.
17434 (ubsan_type_descriptor): Likewise. Formatting fix.
17437 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
17438 splitter): Require TARGET_SSE2 in the condition.
17440 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
17443 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
17444 for doing vector set of SFmode on ISA 3.0.
17445 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
17446 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
17448 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
17449 SFmode value into a V4SF variable that was extracted from another
17450 V4SF variable without converting the element to double precision
17451 and back to single precision vector format.
17452 (vsx_insert_extract_v4sf_p9_2): Likewise.
17454 2017-06-19 Jakub Jelinek <jakub@redhat.com>
17456 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
17457 in UWHI to avoid undefined overflow.
17460 * ubsan.h (enum ubsan_encode_value_phase): New.
17461 (ubsan_encode_value): Change second argument to
17462 enum ubsan_encode_value_phase with default value of
17463 UBSAN_ENCODE_VALUE_GENERIC.
17464 * ubsan.c (ubsan_encode_value): Change second argument to
17465 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
17466 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
17467 create_tmp_var_raw instead of create_tmp_var and use a
17469 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
17470 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
17471 ubsan_encode_value callers.
17474 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
17475 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
17476 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
17478 2017-06-19 Richard Biener <rguenther@suse.de>
17480 PR middle-end/81118
17481 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
17482 estimates if we changed anything.
17484 2017-06-19 Richard Biener <rguenther@suse.de>
17486 PR tree-optimization/80887
17487 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
17488 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
17489 simplified lookups, then reset mprts_hook.
17490 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
17492 (try_to_simplify): Likewise.
17494 2017-06-19 Martin Liska <mliska@suse.cz>
17497 * gimplify.c (gimplify_switch_expr):
17498 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
17500 2017-06-19 Martin Liska <mliska@suse.cz>
17502 * doc/install.texi: Document that PGO runs in 4 stages.
17504 2017-06-19 Martin Liska <mliska@suse.cz>
17507 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
17508 to dispatcher function name.
17509 * multiple_target.c (replace_function_decl): New function.
17510 (create_dispatcher_calls): Redirect both edges and references.
17512 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
17514 * profile-count.c (profile_count::dump): Dump quality.
17515 (profile_count::differs_from_p): Update for unsigned val.
17516 * profile-count.h (profile_count_quality): New enum.
17517 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
17519 2017-06-19 Richard Biener <rguenther@suse.de>
17521 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
17522 struct function as arg.
17523 (estimate_numbers_of_iterations): Export overload with loop arg.
17524 (free_numbers_of_iterations_estimates_loop): Use an overload of
17525 free_numbers_of_iterations_estimates instead.
17526 * tree-cfg.c (remove_bb): Adjust.
17527 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
17528 * tree-parloops.c (gen_parallel_loop): Likewise.
17529 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
17531 (tree_unroll_loops_completely): Likewise.
17532 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
17533 Use an overload instead and export.
17534 (estimated_loop_iterations): Adjust.
17535 (max_loop_iterations): Likewise.
17536 (likely_max_loop_iterations): Likewise.
17537 (estimate_numbers_of_iterations): Take struct function as arg
17539 (loop_exits_before_overflow): Adjust.
17540 (free_numbers_of_iterations_estimates_loop): Use an overload.
17541 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
17542 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
17544 2017-06-19 Richard Biener <rguenther@suse.de>
17547 * ipa-prop.c (find_constructor_constant_at_offset): Handle
17548 RANGE_EXPR conservatively.
17550 2017-06-16 Carl Love <cel@us.ibm.com>
17552 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17553 definitions for vec_float, vec_float2, vec_floato,
17554 vec_floate built-ins.
17555 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
17556 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
17558 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
17559 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
17560 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
17561 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
17562 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
17563 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
17564 vec_floato): Add builtin defines.
17565 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
17566 Update the built-in documentation file for the new built-in
17569 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17571 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
17572 (mthumb): Mark as the negative of -marm.
17574 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17576 * doc/invoke.texi (ARM Options, -mcpu): Document supported
17578 (ARM Options, -mtune): Document that this accepts the same
17579 extension options as -mcpu.
17580 (ARM Options, -mfpu): Document addition of -mfpu=auto.
17582 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17584 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
17585 permitted extensions.
17587 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17589 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
17590 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
17591 (armv8-m.main): Add option +nodsp.
17592 * config/arm/arm-cpu-cdata.h: Regenerated.
17594 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17596 * config/arm/t-fuchsia: New file.
17597 * config.gcc (arm*-*-fuchsia*): Use it.
17599 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17601 * config/arm/t-symbian: Rewrite for new option infrastructure.
17603 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17605 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
17606 (MULTILIB_REQUIRED): Likewise.
17608 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17610 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
17611 (MULTILIB_RESUE): Likewise.
17612 (MULTILIB_MATCHES): Likewise.
17613 (MULTLIB_REQUIRED): Likewise.
17615 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17617 * config/arm/t-rtems: Rewrite for new option framework.
17619 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17621 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
17622 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
17623 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
17624 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
17625 * config/arm/t-multilib: ... here.
17626 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
17627 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
17628 armv7-a and armv8*-a when A-profile libraries have not been built.
17629 * config/arm/t-rmprofile: Rewrite.
17631 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17633 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
17634 with a backslash. Remove the backslash after substituting unescaped
17636 * doc/fragments.texi (MULTILIB_REUSE): Document it.
17638 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17640 * config.gcc: (arm*-*-*): When building a-profile libraries, force
17641 the driver to pass through the default setting of -mfloat-abi.
17642 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
17644 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
17645 (all_feat_combs): New rule.
17646 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
17648 * config/arm/t-aprofile: Rewrite.
17650 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17652 * config/arm/arm.h (FPUTYPE_AUTO): Define.
17653 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
17654 fpu is not specified by the user/command-line.
17655 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
17656 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
17657 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
17658 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
17659 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
17660 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
17662 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17664 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
17665 * config/arm/t-arm-elf: Rewritten.
17667 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17669 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
17670 have some floating-point instructions.
17671 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
17672 (TARGET_MAYBE_HARD_FLOAT): New macro.
17673 * config/arm/arm-builtins.c (arm_init_builtins): Use
17674 TARGET_MAYBE_HARD_FLOAT.
17675 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
17677 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17679 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
17680 (configargs.h): Include it.
17681 (arm_print_hint_for_fpu_option): New function.
17682 (arm_parse_fpu_option): New function.
17683 (candidate_extension): New class.
17684 (arm_canon_for_multilib): New function.
17685 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
17686 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
17687 (ARCH_CANONICAL_SPECS): New macro.
17688 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
17690 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17692 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
17693 are set after handling multilib fragments. Set target_cpu_default2
17696 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17698 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
17700 (arm*-*-*): Set target_cpu_default2 to a quoted string.
17701 * config/arm/parsecpu.awk (check_cpu): Validate any extension
17703 (check_arch): Likewise.
17704 * config/arm/arm.c (arm_configure_build_target): Handle
17705 TARGET_CPU_DEFAULT being a string constant. Scan any feature
17706 options in the default.
17708 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17710 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
17711 when an option is an alias of another.
17712 * config/arm/parsecpu.awk (optalias): New parser token.
17713 (gen_comm_data): Mark non-alias options as such. Emit entries
17714 for extension aliases.
17715 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
17716 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
17717 (armv6kz, armv6zk, armv6t2): Likewise.
17718 (armv7): Make vfpv3-d16 an alias.
17719 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
17721 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
17722 Sort in canonical order.
17723 (armv8-a): Sort in canonical order.
17724 (armv8.1-a, armv8.2-a): Likewise.
17725 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
17727 (cortex-a9): Sort in canonical order.
17728 * config/arm/arm.c (selftests.h): Include it.
17729 (arm_test_cpu_arch_data): New function.
17730 (arm_run_self_tests): New function.
17731 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
17732 (targetm): Move declaration to the end of the file.
17733 * arm-cpu-cdata.h: Regenerated.
17735 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17737 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
17738 call to target_mode_check describing the type of option passed.
17739 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
17740 (arm_target_thumb_only): Use arm_parse_arch_option_name or
17741 arm_parse_cpu_option_name to match parameters against list of
17743 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
17744 arm_arch_core_flags data structure.
17745 * config/arm/arm-cpu_cdata.h: Regenerated.
17747 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17749 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
17751 (arm_print_hint_for_cpu_option): Likewise.
17752 (arm_print_hint_for_arch_option): Likewise.
17753 (arm_parse_cpu_option_name): Likewise.
17754 (arm_parse_arch_option_name): Likewise.
17755 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
17756 of entries in the all_fpus list.
17757 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
17758 (arm_parse_cpu_option_name): Declare.
17759 (arm_parse_arch_option_name): Declare.
17760 (arm_parse_option_features): Declare.
17761 (arm_intialize_isa): Declare.
17762 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
17764 (gen_comm_data): ... here. Make definitions non-static.
17765 * config/arm/arm-cpu-data.h: Regenerated.
17766 * config/arm/arm-cpu-cdata.h: Regenerated.
17768 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17770 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
17771 (cpu_arch_extension): New structure.
17772 (cpu_arch_option, arch_option, cpu_option): New structures.
17773 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
17774 architecture types.
17775 (gen_data): Generate new format data tables.
17776 * config/arm/arm.c (cpu_tune): New structure.
17777 (cpu_option, processors): Delete.
17778 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
17779 (arm_print_hint_for_cpu_option): ... this and ...
17780 (arm_print_hint_for_arch_option): ... this.
17781 (arm_parse_arch_cpu_name): Delete. Replace with ...
17782 (arm_parse_cpu_option_name): ... this and ...
17783 (arm_parse_arch_option_name): ... this.
17784 (arm_unrecognized_feature): Change type of target parameter to
17786 (arm_parse_arch_cpu_features): Delete. Replace with ...
17787 (arm_parse_option_features): ... this.
17788 (arm_configure_build_target): Rework to use new configuration data
17790 (arm_print_tune_info): Rework for new configuration data tables.
17791 * config/arm/arm-cpu-data.h: Regenerated.
17792 * config/arm/arm-cpu.h: Regenerated.
17794 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17796 * Makefile.in (OBJS): Move sbitmap.o from here ...
17797 (OBJS-libcommon): ... to here.
17799 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17801 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
17802 (ISA_ALL_CRYPTO): New macro.
17803 (ISA_ALL_SIMD): New macro
17804 (ISA_ALL_FP): New macro.
17805 * config/arm/arm.c (fpu_bitlist): Update initializer.
17806 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
17808 (arm9e): Add fpu. Add option for nofp
17809 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
17810 (arm926ej-s, arm1026ej-s): Likewise.
17811 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
17812 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
17813 neon-fp16, neon-vfpv4, nofp and nosimd.
17814 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
17815 (cortex-a8): Add fpu. Add option for nofp.
17816 (cortex-a9): Add fpu. Add options for nosimd and nofp.
17817 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
17818 (cortex-r4f): Add fpu.
17819 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
17820 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
17822 (cortex-r8): Likewise.
17823 (cortex-m4): Add fpu. Add option for nofp.
17824 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
17825 (cortex-a17.cortex-a7): Likewise.
17826 (cortex-a32): Add fpu. Add options for crypto and nofp.
17827 (cortex-a35, cortex-a53): Likewise.
17828 (cortex-a57): Add fpu. Add option for crypto.
17829 (cortex-a72, cortex-a73): Likewise.
17830 (exynos-m1): Likewise.
17831 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
17832 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
17833 (cortex-m33): Add fpu. Add option for nofp.
17834 * config/arm/arm-cpu-cdata.h: Regenerated
17835 * config/arm/arm-cpu-data.h: Regenerated.
17837 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17839 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
17840 (armv5te, armv5tej): Likewise.
17841 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
17842 (armv7): Add options fp and vfpv3-d16.
17843 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
17844 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
17846 (armv7ve): Likewise.
17847 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
17848 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
17849 (armv8-a): Add nocrypto option.
17850 (armv8.1-a, armv8.2-a): Likewise.
17851 (armv8-m.main): add options fp, fp.dp and nofp.
17853 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17855 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
17857 (armv8-a+crc): Delete.
17858 (armv8.1-a): Add options simd, crypto and nofp.
17859 (armv8.2-a): Add options fp16, simd, crypto and nofp.
17860 (armv8.2-a+fp16): Delete.
17861 (armv8-m.main): Add option dsp.
17862 (armv8-m.main+dsp): Delete.
17863 (cortex-a8): Add fpu. Add nofp option.
17864 (cortex-a9): Add fpu. Add nofp and nosimd options.
17865 * config/arm/parsecpu.awk (gen_data): Generate option tables and
17866 link to main cpu and architecture data structures.
17867 (gen_comm_data): Only put isa attributes from the main architecture
17869 (option): New statement for architecture and CPU entries.
17870 * arm.c (struct cpu_option): New structure.
17871 (struct processors): Add entry for options.
17872 (arm_unrecognized_feature): New function.
17873 (arm_parse_arch_cpu_name): Ignore any characters after the first
17875 (arm_parse_arch_cpu_feature): New function.
17876 (arm_configure_build_target): Separate out any CPU and architecture
17877 features and parse separately. Don't error out if -mfpu=auto is
17878 used with only an architecture string.
17879 (arm_print_asm_arch_directives): New function.
17880 (arm_file_start): Call it.
17881 * config/arm/arm-cpu-cdata.h: Regenerated.
17882 * config/arm/arm-cpu-data.h: Likewise.
17883 * config/arm/arm-tables.opt: Likewise.
17885 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17887 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
17888 assembler when it is not -mfpu=auto.
17890 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17892 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
17893 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
17894 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
17895 (ASM_CPU_SPEC): Rewrite.
17896 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
17897 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
17898 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
17899 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
17900 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
17901 copied string is NUL-terminated. Also strip any characters prefixed
17903 (arm_rewrite_selected_arch): New function.
17904 (arm_rewrite_march): New function.
17906 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17908 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
17909 (x_arm_cpu_string, x_arm_tune_string): Likewise.
17910 (march, mcpu, mtune): Convert to string-based options.
17911 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
17912 (arm_parse_arch_cpu_name): New function.
17913 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
17914 identify selected architecture or CPU.
17915 (arm_option_save): New function.
17916 (TARGET_OPTION_SAVE): Redefine.
17917 (arm_option_restore): Restore string options.
17918 (arm_option_print): Print string options.
17920 2017-06-16 Martin Sebor <msebor@redhat.com>
17922 PR tree-optimization/80933
17923 PR tree-optimization/80934
17924 * builtins.c (fold_builtin_3): Do not handle bcmp here.
17925 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
17926 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
17927 (gimple_fold_builtin): Call them.
17929 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17931 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
17932 as unlikely; update profile.
17934 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17936 * predict.c (force_edge_cold): Handle declaring edges impossible
17939 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17941 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
17943 (try_unroll_loop_completely): Fix reporting.
17945 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17947 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
17949 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
17952 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
17953 if given a non-constant argument for an intrinsic which requires a
17956 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17958 * profile.c (compare_freqs): New function.
17959 (branch_prob): Sort edge list.
17960 (find_spanning_tree): Assume that the list is priority sorted.
17962 2017-06-16 Richard Biener <rguenther@suse.de>
17964 PR tree-optimization/81090
17965 * passes.def (pass_record_bounds): Remove.
17966 * tree-pass.h (make_pass_record_bounds): Likewise.
17967 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
17968 make_pass_record_bounds): Likewise.
17969 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
17970 not free niter estimates at the beginning but at the end.
17971 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
17973 2017-06-16 Richard Biener <rguenther@suse.de>
17975 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
17976 initializer to workaround ICE in host GCC 4.8.
17978 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17980 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
17982 (clone_inlined_nodes): Update.
17984 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17986 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
17987 prefetch settings, and enable prefetching by default at -O3.
17989 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17991 * config/aarch64/aarch64.c (aarch64_override_options_internal):
17992 Set flag_prefetch_loop_arrays according to tuning data.
17994 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17996 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
17997 New tune structure.
17998 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
17999 [Unrelated to main purpose of the patch] Place the pointer field last
18000 to enable type checking errors when tune structure are wrongly merged.
18001 * config/aarch64/aarch64.c (generic_prefetch_tune,)
18002 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
18003 (thunderx2t99_prefetch_tune): New tune constants.
18004 (tune_params *_tunings): Update all tunings (no functional change).
18005 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
18006 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
18007 from tunings structures.
18009 2017-06-16 Jakub Jelinek <jakub@redhat.com>
18012 * ubsan.c (instrument_null): Add T argument, use it instead
18013 of computing it based on IS_LHS.
18014 (instrument_object_size): Likewise.
18015 (pass_ubsan::execute): Adjust instrument_null and
18016 instrument_object_size callers to pass gimple_get_lhs or
18017 gimple_assign_rhs1 result to it. Use instrument_null instead of
18018 calling get_base_address and instrument_mem_ref. Handle
18019 aggregate call arguments for object-size sanitization.
18021 2017-06-16 Yury Gribov <tetra2005@gmail.com>
18023 PR tree-optimization/81089
18024 * tree-vrp.c (is_masked_range_test): Validate operands of
18027 2017-06-15 Martin Sebor <msebor@redhat.com>
18030 * dumpfile.c (dump_register): Avoid calling memset to initialize
18031 a class with a default ctor.
18032 * gcc.c (struct compiler): Remove const qualification.
18033 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
18034 * hash-table.h: Ditto.
18035 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
18037 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
18038 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
18040 * params.h (struct param_info): Make struct members non-const.
18041 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
18042 with default initialization.
18043 * vec.h (vec_copy_construct, vec_default_construct): New helper
18045 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
18046 with vec_copy_construct.
18047 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
18048 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
18049 * doc/invoke.texi (-Wclass-memaccess): Document.
18051 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18053 * emit-rtl.h (is_leaf): Update comment about local
18054 register allocator.
18056 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18059 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
18060 for a variable to have a section before checking if the section has a
18062 Set section to.persistent if persistent attribute is set.
18063 Warn if .persistent attribute is used on an automatic variable.
18065 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
18067 PR rtl-optimization/80474
18068 * reorg.c (update_block): Do not ignore instructions in a delay slot.
18070 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
18072 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
18075 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
18077 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
18078 (casesi): Emit bounds checking as RTL.
18079 (casesi_internal_mips16_<mode>): Remove bounds checking.
18081 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
18083 * config/xtensa/xtensa.c (xtensa_option_override): Append
18084 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
18085 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
18086 xtensa_doloop_hooks): Define unconditionally.
18087 (xtensa_reorg_loops): Only call reorg_loops in the presence of
18089 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
18090 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
18091 for it in xtensa_option_override.
18092 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
18093 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
18095 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
18097 * doc/cppopts.texi: Document '-' special value to -MF.
18099 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
18101 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
18102 (cortex_a53_fconst): Likewise.
18103 (cortex_a53_fpmul): Likewise.
18104 (cortex_a53_f_load_64): Likewise.
18105 (cortex_a53_f_load_many): Likewise.
18106 (cortex_a53_advsimd_alu): Likewise.
18107 (cortex_a53_advsimd_alu_q): Likewise.
18108 (cortex_a53_advsimd_mul): Likewise.
18109 (cortex_a53_advsimd_mul_q): Likewise.
18110 (fpmac bypass): Add new bypass for fpmac-fpmac case.
18111 Add missing fmul, r2f_cvt and fconst cases.
18113 2017-06-14 Richard Biener <rguenther@suse.de>
18115 PR middle-end/81088
18116 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
18118 (fold_binary_loc): When associating do not treat pre-existing
18119 TREE_OVERFLOW on literal constants as a reason to allow
18120 TREE_OVERFLOW on associated literal constants.
18122 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18124 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
18125 (MASK_FEATURES): New macro.
18126 * config/sparc/sparc.c (sparc_option_override): Remove the special
18127 handling of -mfpu and generalize it to all MASK_FEATURES switches.
18129 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18131 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
18132 a division of 0 if non-call exceptions are enabled.
18134 2017-06-14 Andrew Pinski <apinski@cavium.com>
18135 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18138 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
18139 Improve vector initialization code gen for only variable case.
18141 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18143 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
18145 2017-06-14 Richard Biener <rguenther@suse.de>
18147 PR tree-optimization/81083
18148 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
18151 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18153 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
18154 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
18155 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
18156 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
18157 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
18158 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
18160 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18162 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
18163 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
18165 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18167 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
18169 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18171 * config/rs6000/t-rtems: Don't handle SPE.
18173 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18175 * config/rs6000/t-linux: Don't handle SPE.
18177 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18179 * config/rs6000/eabispe.h: Delete file.
18181 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18183 * config/rs6000/t-spe: Delete file.
18185 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18187 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
18188 (rs6000_legitimate_offset_address_p): Return false for anything in
18189 V2SImode or V2SFmode.
18191 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18193 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
18194 except V2SF and V2SI. Rearrange the vector modes, and add comments.
18195 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
18197 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
18198 (rs6000_legitimate_offset_address_p): Ditto.
18199 (rs6000_emit_move): Ditto.
18200 (rs6000_init_builtins): Remove V4HI_type_node.
18202 2017-06-13 Martin Liska <mliska@suse.cz>
18205 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
18206 (gate_asan): Likewise.
18207 * asan.h (asan_no_sanitize_address_p): Remove the function.
18208 (sanitize_flags_p): New function.
18209 * builtins.def: Fix coding style.
18210 * common.opt: Use renamed enum value.
18211 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
18212 * doc/extend.texi: Document no_sanitize attribute.
18213 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
18214 to SANITIZE_UNDEFINED_NONDEFAULT.
18215 * gcc.c (sanitize_spec_function): Use the renamed enum value.
18216 * gimple-fold.c (optimize_atomic_compare_exchange_p):
18217 Use sanitize_flags_p.
18218 * gimplify.c (gimplify_function_tree): Likewise.
18219 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
18220 * opts.c (parse_no_sanitize_attribute): New function.
18221 (common_handle_option): Use renamed enum value.
18222 * opts.h (parse_no_sanitize_attribute): Declare.
18223 * tree.c (sanitize_flags_p): New function.
18224 * tree.h: Declared here.
18225 * tsan.c: Use sanitize_flags_p.
18226 * ubsan.c (ubsan_expand_null_ifn): Likewise.
18227 (instrument_mem_ref): Likewise.
18228 (instrument_bool_enum_load): Likewise.
18229 (do_ubsan_in_current_function): Remove the function.
18230 (pass_ubsan::execute): Use sanitize_flags_p.
18231 * ubsan.h: Remove do_ubsan_in_current_function
18232 * tree-cfg.c (print_no_sanitize_attr_value): New function.
18233 (dump_function_to_file): Use it here.
18235 2017-06-13 Martin Jambor <mjambor@suse.cz>
18237 PR tree-optimization/80803
18238 PR tree-optimization/81063
18239 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
18240 (propagate_subaccesses_across_link): Enqueue subtree whenever
18241 necessary instead of relying on the caller.
18243 2017-06-13 Martin Jambor <mjambor@suse.cz>
18245 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
18246 that have a first_link.
18247 (sort_and_splice_var_accesses): Do not check first_link before
18249 (subtree_mark_written_and_enqueue): Likewise.
18250 (propagate_all_subaccesses): Likewise and do not stop at first
18251 parent with a first_link.
18253 2017-06-13 Martin Jambor <mjambor@suse.cz>
18255 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
18258 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18260 * match.pd: New pattern.
18262 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18264 * tree-vrp.c (is_masked_range_test): New function.
18265 (register_edge_assert_for): Determine ranges for
18268 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18270 PR tree-optimization/67328
18271 * fold-const.c (maskable_range_p): New function.
18272 (build_range_check): Generate bittests if possible.
18274 2017-06-13 Martin Liska <mliska@suse.cz>
18276 * gimple-pretty-print.c (dump_probability): Add new argument.
18277 (dump_edge_probability): Dump both probability and count.
18278 (dump_gimple_label): Likewise.
18279 (dump_gimple_bb_header): Likewise.
18281 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
18284 * config/avr/avr-devices.c: Fix indentation.
18285 * config/avr/gen-avr-mmcu-specs.c: Dito.
18287 2017-06-13 Richard Biener <rguenther@suse.de>
18289 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
18290 instead get vector type from stmt_info.
18291 (vectorizable_reduction): Adjust. Remove dead code.
18293 2017-06-13 Richard Biener <rguenther@suse.de>
18295 PR middle-end/81065
18296 * fold-const.c (extract_muldiv_1): Remove bogus distribution
18297 case of C * (x * C2 + C3).
18298 (fold_addr_of_array_ref_difference): Properly fold index difference.
18300 2017-06-12 David S. Miller <davem@davemloft.net>
18303 * config/sparc/sparc.md (return expander): Emit frame blockage if
18304 function uses alloca.
18306 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
18308 * combine.c (make_field_assignment): Check len rather than the mode
18309 precision when calling force_to_mode.
18311 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
18313 Support multilibs and devices that see flash in RAM address range.
18316 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
18317 (avr_mcu_t) <flash_pm_offset>: New field.
18318 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
18319 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
18320 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
18321 (AVR_TINY_PM_OFFSET): Remove macro.
18322 * config/avr/avr.opt (-mshort-calls): New option.
18323 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18324 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
18325 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
18326 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
18327 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
18328 instead of avr_arch->have_jmp_call.
18329 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
18330 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
18331 avr_arch->flash_pm_offset to define.
18332 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
18333 new field flash_pm_offset. Add entry for avrxmega3.
18334 (avr_texinfo): Add entry for avrxmega3.
18335 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
18336 attiny212, attiny214,
18337 attiny412, attiny414, attiny416, attiny417,
18338 attiny814, attiny816, attiny817,
18339 attiny1614, attiny1616, attiny1617,
18340 attiny3214, attiny3216, attiny3217.
18341 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
18342 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
18343 (avr_print_operand_address) [AVR_TINY]: Same.
18344 (avr_asm_init_sections) <readonly_data_section>: Only patch
18345 callback if avr_arch->flash_pm_offset = 0.
18346 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
18347 for rodata if avr_arch->flash_pm_offset != 0.
18348 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
18349 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
18350 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
18351 (m_options): Append opt_rcall.
18352 (m_dirnames): Append dir_rcall.
18353 * config/avr/t-multilib: Regenerate.
18355 * configure.ac [target=avr]: Check whether avrxmega3 default
18356 linker description file works as needed.
18357 * configure: Regenerate.
18358 * doc/avr-mmcu.texi: Regenerate.
18359 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
18360 <__AVR_ARCH__>: Document avrxmega3 and 103.
18361 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
18362 <__AVR_SHORT_CALLS__>: Document it.
18363 <__AVR_PM_BASE_ADDRESS__>: Document it.
18364 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
18365 (AVR Variable Attributes) <progmem>: Document this is
18366 not needed for avrxmega3.
18367 (AVR Named Address Spaces) <__flash>: Dito.
18369 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
18371 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
18373 2017-06-12 Doug Rupp <rupp@adacore.com>
18375 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
18376 Append vxworks-stdint.h to the tm_file list.
18377 * config/vxworks-stdint.h: New file.
18379 2017-06-12 Martin Liska <mliska@suse.cz>
18381 PR tree-optimization/81041
18382 * tree-profile.c (gimple_gen_ic_func_profiler):
18383 Create an extra BB in profile-generate
18384 (gimple_gen_time_profiler): Likewise.
18386 2017-06-12 Jakub Jelinek <jakub@redhat.com>
18388 PR tree-optimization/81003
18389 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
18390 (update_range_test): Use it instead of force_gimple_operand_gsi.
18392 2017-06-12 Richard Biener <rguenther@suse.de>
18394 PR tree-optimization/81053
18395 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
18396 with backedge value not defined in loop. Simplify def stmt
18399 2017-06-11 Tom de Vries <tom@codesourcery.com>
18402 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
18404 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
18405 nvptx_cannot_force_const_mem.
18407 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18409 * opts.c (finish_options): Move test for flag_split_stack after
18410 it has been initialized.
18412 2017-06-11 Jason Merrill <jason@redhat.com>
18414 * tree.h (id_equal): New.
18415 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
18416 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
18417 instead of strcmp of IDENTIFIER_POINTER.
18419 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18421 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
18422 (mark_all_inlined_calls_cdtor): Fix formating.
18423 (inline_transform): Rescale profile before inlining.
18425 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18427 * cgraph.h (cgraph_edge::clone): Update prototype.
18428 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
18429 (cgraph_node::create_clone): Update.
18430 (cgraph_node::create_version_clone): Update.
18431 * tree-inline.c (copy_bb): Update.
18432 (expand_call_inline): Update.
18434 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
18436 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
18437 factored out from ...
18438 (rs6000_emit_prologue): ... here.
18440 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
18442 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
18443 factored out from ...
18444 (rs6000_emit_prologue): ... here.
18446 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18448 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
18450 (handle_missing_profiles): Fix computation of tp_first_run.
18451 (counts_to_freqs): Do not touch freqs when count is 0.
18453 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18455 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
18458 2017-06-10 Tom de Vries <tom@codesourcery.com>
18460 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
18461 attributes): Document signal effective target.
18463 2017-06-10 Tom de Vries <tom@codesourcery.com>
18465 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
18466 Document effective target stack_size.
18468 2017-06-09 David Malcolm <dmalcolm@redhat.com>
18470 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
18471 to the edit_context if they can be auto-applied.
18473 2017-06-9 Ian Lance Taylor <iant@golang.org>
18475 * opts.c (finish_options): If -fsplit-stack, disable implicit
18476 -forder-blocks-and-partition.
18477 * doc/invoke.texi (Optimize Options): Document that when using
18478 -fsplit-stack -forder-blocks-and-partition is not implicitly
18481 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18483 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
18484 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
18485 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
18486 * builtins.def (abort, trap, unreachable): Declare cold.
18487 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
18488 * tree-core.h (ECF_COLD): New.
18489 * tree.c (set_call_expr_flags): Handle ECF_COLD.
18490 (build_common_builtin_nodes): Mark unreachable and abort as cold.
18492 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18494 * predict.c (unlikely_executed_stmt_p): Cleanup.
18496 2017-06-09 Richard Biener <rguenther@suse.de>
18498 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
18499 model if the ref is always written to.
18501 2017-06-09 Tamar Christina <tamar.christina@arm.com>
18503 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
18505 2017-06-09 Tamar Christina <tamar.christina@arm.com>
18507 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
18510 2017-06-09 Tom de Vries <tom@codesourcery.com>
18513 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
18514 "target cannot support label values" when encountering LABEL_REF.
18516 2017-06-09 Martin Liska <mliska@suse.cz>
18518 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
18519 (gimple_gen_ic_func_profiler): Emit direct comparison
18520 of __gcov_indirect_call_callee with NULL.
18521 (gimple_gen_time_profiler): Change probability from
18522 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
18524 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18526 * profile.c (edge_gcov_counts): Turn to pointer.
18527 (compute_branch_probabilities, compute_branch_probabilities): Update.
18528 (branch_prob): Do not clear edge_gcov_count.
18529 * profile.h (edge_gcov_counts): Turn to pointer.
18530 (edge_gcov_count): Update.
18532 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18534 * gimple.h (gimple_check_failed): Mark cold.
18536 2017-06-09 Richard Biener <rguenther@suse.de>
18538 PR tree-optimization/66623
18539 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
18540 refactor check_reduction into two parts, properly computing
18541 whether we have to check reduction validity for outer loop
18544 2017-06-09 Richard Biener <rguenther@suse.de>
18546 PR tree-optimization/79483
18547 * graphite-scop-detection.c (order): New global.
18548 (get_order): Compute bb to order mapping that satisfies code
18549 generation constraints.
18550 (cmp_pbbs): New helper.
18551 (build_scops): Start domwalk at entry block, sort generated
18554 2017-06-09 Richard Biener <rguenther@suse.de>
18556 PR middle-end/81007
18557 * ipa-polymorphic-call.c
18558 (ipa_polymorphic_call_context::restrict_to_inner_class):
18559 Skip FIELD_DECLs with error_mark_node type.
18560 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
18563 2017-06-09 Martin Liska <mliska@suse.cz>
18565 * predict.c (struct branch_predictor): New struct.
18566 (test_prediction_value_range): New test.
18567 (predict_c_tests): New function.
18568 * selftest-run-tests.c (selftest::run_tests): Run the function.
18569 * selftest.h: Declare new tests.
18571 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
18574 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
18575 gen_add3_insn did not fail.
18576 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
18577 r0, construct that number in a temporary reg and add that reg to r0.
18578 If asked to put the result in r0 as well, fail.
18580 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
18582 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18583 for early expansion of vec_eqv.
18585 2017-06-08 Jakub Jelinek <jakub@redhat.com>
18587 PR middle-end/81005
18588 * ubsan.c (instrument_null): Avoid pointless code temporary.
18589 (pass_ubsan::execute): Instrument aggregate arguments of calls.
18591 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
18595 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
18598 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
18601 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18603 * predict.c (unlikely_executed_edge_p): Move ahead.
18604 (probably_never_executed_edge_p): Use it.
18606 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
18608 PR middle-end/79988
18609 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
18610 gimple_call_builtin_p call.
18612 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18614 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
18615 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
18616 rtl_check_failed_type2, rtl_check_failed_code1,
18617 rtl_check_failed_code2, rtl_check_failed_code_mode,
18618 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
18619 rtvec_check_failed_bounds, rtl_check_failed_flag,
18620 _fatal_insn_not_found, _fatal_insn): Likewise.
18621 * tree.h (tree_contains_struct_check_failed,
18622 tree_check_failed, tree_not_check_failed,
18623 tree_class_check_failed, tree_range_check_failed,
18624 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
18625 tree_vec_elt_check_failed, phi_node_elt_check_failed,
18626 tree_operand_check_failed, omp_clause_check_failed,
18627 omp_clause_operand_check_failed, omp_clause_range_check_failed):
18630 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18632 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
18633 flag_branch_probabilities.
18634 * ipa-inline.c (edge_badness): Likewise.
18635 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
18636 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
18637 * predict.c (maybe_hot_frequency_p): Likewise.
18638 (probably_never_executed): Likewise.
18639 * sched-ebb.c (schedule_ebbs): Likewise.
18640 * sched-rgn.c (find_single_block_region): Likewise.
18641 * tracer.c (tail_duplicate): Likewise.
18643 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18645 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
18646 longer requires x_flag_profile_use.
18648 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18650 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
18651 instead of flag_reorder_blocks_and_partition.
18652 * dbxout.c (dbxout_function_end): Likewise.
18653 * dwarf2out.c (gen_subprogram_die): Likewise.
18654 * haifa-sched.c (sched_create_recovery_edges): Likewise.
18655 * hw-doloop.c (reorg_loops): Likewise.
18656 * varasm.c (assemble_start_function,
18657 assemble_end_function): Likewise.
18658 (decide_function_section): Do not check for
18659 flag_reorder_blocks_and_partition.
18661 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
18663 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
18665 (chkp_get_hard_register_fake_addr_expr): Ditto.
18666 (chkp_build_addr_expr): Add check for hard reg case.
18667 (chkp_parse_array_and_component_ref): Ditto.
18668 (chkp_find_bounds_1): Ditto.
18669 (chkp_process_stmt): Don't generate bounds store for
18672 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18674 * predict.c (maybe_hot_bb_p): Do not check profile status.
18675 (maybe_hot_edge_p): Likewise.
18676 (probably_never_executed): Check for zero counts even if profile
18678 (unlikely_executed_edge_p): New function.
18679 (unlikely_executed_stmt_p): New function.
18680 (unlikely_executed_bb_p): New function.
18681 (set_even_probabilities): Use unlikely predicates.
18682 (combine_predictions_for_bb): Likewise.
18683 (predict_paths_for_bb): Likewise.
18684 (predict_paths_leading_to_edge): Likewise.
18685 (determine_unlikely_bbs): New function.
18686 (estimate_bb_frequencies): Use it.
18687 (compute_function_frequency): Use zero counts even if profile is
18689 * profile-count.h: Fix typo.
18691 2017-08-08 Julia Koval <julia.koval@intel.com>
18693 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
18694 _mm512_mask_cvtsepi16_storeu_epi8,
18695 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
18696 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
18697 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
18698 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
18699 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
18700 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
18701 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
18702 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
18703 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
18704 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
18705 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
18706 __builtin_ia32_pmovuswb256mem_mask,
18707 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
18708 __builtin_ia32_pmovwb512mem_mask): New builtins.
18710 2017-08-08 Julia Koval <julia.koval@intel.com>
18712 PR target/73350,80862
18713 * config/i386/subst.md (round): Fix round pattern.
18714 * config/i386/i386.c (ix86_erase_embedded_rounding):
18715 Fix erasing rounding for the fixed pattern.
18717 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18719 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
18721 2017-06-08 Martin Liska <mliska@suse.cz>
18723 PR gcov-profile/80911
18724 * gcov.c (block_info::block_info): New constructor.
18726 2017-06-07 Carl Love <cel@us.ibm.com>
18728 * config/rs6000/rs6000-c: The return type of the following
18729 built-in functions was implemented as int not long long. Fix sign
18730 of return value for the unsigned version of vec_mulo and vec_mule.
18731 vector unsigned long long vec_bperm (vector unsigned long long,
18732 vector unsigned char)
18733 vector signed long long vec_mule (vector signed int,
18735 vector unsigned long long vec_mule (vector unsigned int,
18736 vector unsigned int)
18737 vector signed long long vec_mulo (vector signed int,
18739 vector unsigned long long vec_mulo (vector unsigned int,
18740 vector unsigned int)
18741 * doc/extend.texi: Fix the documentation for the built-in
18744 2017-06-07 Carl Love <cel@us.ibm.com>
18747 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
18750 2017-06-07 Carl Love <cel@us.ibm.com>
18752 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
18753 support, Generate doublehv for signed int/float for BE case only.
18755 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
18757 * doc/invoke.texi (mcx16): Rewrite.
18759 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18761 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
18762 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
18763 *mov<mode>_softfloat, and an anonymous splitter): Use
18764 nonimmediate_operand instead of rs6000_nonimmediate_operand.
18766 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18768 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
18770 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
18771 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
18772 (rs6000_debug_reg_global): Adjust.
18773 (rs6000_init_hard_regno_mode_ok): Adjust.
18774 (rs6000_dbx_register_number): Adjust.
18775 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
18776 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
18777 Remove SPE_ACC and SPEFSCR.
18778 (REG_ALLOC_ORDER): Ditto.
18779 (FRAME_POINTER_REGNUM): Change to 111.
18780 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
18781 (REG_CLASS_NAMES): Ditto.
18782 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
18783 (REGISTER_NAMES): Ditto.
18784 (ADDITIONAL_REG_NAMES): Ditto.
18785 (rs6000_reg_names): Ditto.
18786 * config/rs6000/rs6000.md: Renumber some register number
18789 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18791 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
18793 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
18794 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
18796 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
18797 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
18798 Delete the SPE high registers.
18799 (REG_ALLOC_ORDER): Ditto.
18800 (enum reg_class): Remove SPE_HIGH_REGS.
18801 (REG_CLASS_NAMES): Ditto.
18802 (REG_CLASS_CONTENTS): Delete the SPE high registers.
18803 (REGISTER_NAMES): Ditto.
18804 (rs6000_reg_names): Ditto.
18805 * doc/tm.texi.in: Remove SPE as example.
18806 * doc/tm.texi: Regenerate.
18808 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18810 * config/rs6000/8540.md (ppc8540_brinc): Delete.
18811 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
18812 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
18813 * config/rs6000/rs6000.md (type): Remove "brinc".
18815 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18817 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
18818 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
18819 * config/rs6000/linuxspe.h: Delete file.
18820 * config/rs6000/rs6000.md: Don't include spe.md.
18821 * config/rs6000/spe.h: Delete file.
18822 * config/rs6000/spe.md: Delete file.
18823 * config/rs6000/t-rs6000: Remove spe.md.
18825 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18827 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
18828 (reg_or_none500mem_operand): Delete.
18829 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
18830 instead of reg_or_none500mem_operand.
18832 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18834 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
18835 handling of SPE flags.
18836 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
18838 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18840 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
18842 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
18843 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
18844 paired_divv2sf3): Similar.
18845 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
18846 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
18847 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
18849 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
18850 Rename the paired_* instruction patterns.
18851 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
18853 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
18854 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
18855 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
18856 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
18857 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
18858 PAIRED_VECTOR_MODE.
18859 (struct machine_function): Delete field spe_insn_chain_scanned_p.
18860 (spe_func_has_64bit_regs_p): Delete.
18861 (spe_expand_predicate_builtin): Delete.
18862 (spe_expand_evsel_builtin): Delete.
18863 (TARGET_DWARF_REGISTER_SPAN): Do not define.
18864 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
18865 (invalid_e500_subreg): Delete.
18866 (rs6000_legitimize_address): Always force_reg op2 as well, for
18867 paired single memory accesses.
18868 (rs6000_member_type_forces_blk): Delete.
18869 (rs6000_spe_function_arg): Delete.
18870 (rs6000_expand_unop_builtin): Delete SPE handling.
18871 (rs6000_expand_binop_builtin): Ditto.
18872 (spe_expand_stv_builtin): Delete.
18873 (bdesc_2arg_spe): Delete.
18874 (spe_expand_builtin): Delete.
18875 (spe_expand_predicate_builtin): Delete.
18876 (spe_expand_evsel_builtin): Delete.
18877 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
18878 (spe_init_builtins): Delete.
18879 (spe_func_has_64bit_regs_p): Delete.
18880 (savres_routine_name): Delete "info" parameter. Adjust callers.
18881 (rs6000_emit_stack_reset): Ditto.
18882 (rs6000_dwarf_register_span): Delete.
18883 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
18884 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
18885 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
18887 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
18889 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
18890 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
18891 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
18892 mulv2sf3, divv2sf3): Delete expanders.
18894 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18896 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
18898 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18900 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
18901 * config/rs6000/rs6000.c: Ditto.
18903 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18905 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
18906 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
18907 comparison_operator.
18909 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18911 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
18912 * config/rs6000/rs6000.opt: Ditto.
18913 * config/rs6000/t-rtems: Ditto.
18915 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18917 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
18918 TARGET_E500_SINGLE by 0, simplify.
18919 * config/rs6000/rs6000.c: Ditto.
18920 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
18921 (spe_build_register_parallel): Delete.
18922 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
18923 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
18924 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
18925 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
18926 (E500_CONVERT): Delete.
18927 * config/rs6000/spe.md: Remove many patterns and all define_constants.
18929 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18931 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
18932 * config/rs6000/dfp.md: Ditto.
18933 (negdd2, *negdd2_fpr): Merge.
18934 (absdd2, *absdd2_fpr): Merge.
18935 (negtd2, *negtd2_fpr): Merge.
18936 (abstd2, *abstd2_fpr): Merge.
18937 * config/rs6000/e500.h: Delete file.
18938 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
18939 TARGET_FPRS by 1 and simplify.
18940 * config/rs6000/rs6000-c.c: Ditto.
18941 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
18942 TARGET_DF_SPE by 0.
18943 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
18945 * config/rs6000/rs6000.md: Ditto.
18946 (floatdidf2, *floatdidf2_fpr): Merge.
18947 (move_from_CR_gt_bit): Delete.
18948 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
18949 (E500_CR_IOR_COMPARE): Delete.
18950 (All patterns that require !TARGET_FPRS): Delete.
18951 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
18953 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18955 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
18957 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18959 * graphds.c (add_edge): Intitialize edge's attached data.
18960 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
18961 pointer parameter. Call pointed function on each edge during
18962 graph traversing. Skip traversing the edge when the function
18964 (graphds_dfs, graphds_scc): Ditto.
18965 (for_each_edge): New parameter. Pass the new parameter to callback
18967 * graphds.h (skip_edge_callback): New function pointer type.
18968 (graphds_dfs, graphds_scc): New function pointer parameter.
18969 (graphds_edge_callback, for_each_edge): New parameter.
18971 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18973 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
18974 out code checking if runtime alias check is possible to below ...
18975 Call the new function.
18976 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
18977 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
18979 2017-06-07 Marek Polacek <polacek@redhat.com>
18982 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
18983 TYPE_OVERFLOW_WRAPS check.
18985 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18987 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
18988 if versioning is required.
18989 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
18990 peeling with the check for versioning.
18992 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18994 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
18995 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
18996 Set true to new parameter if new ssa variable is defined.
18997 (vect_gen_vector_loop_niters): Refactor. Set range information
18998 for the new vector loop bound variable.
18999 (vect_do_peeling): Ditto.
19001 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19003 * tree-affine.c (ssa.h): Include header file.
19004 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
19005 has wrapping overflow behavior.
19007 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19009 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
19011 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19013 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
19014 (tree_to_aff_combination): ... here.
19016 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19018 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
19019 reg_pressure model function.
19020 (ivopts_global_cost_for_size): Delete.
19021 (determine_set_costs, iv_ca_recount_cost): Call new model function
19022 ivopts_estimate_reg_pressure.
19024 2017-06-07 Tamar Christina <tamar.christina@arm.com>
19026 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
19027 expensive than udiv. Remove floating point cases from mod.
19029 2017-06-07 Tamar Christina <tamar.christina@arm.com>
19031 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
19032 Increase idiv cost.
19034 2017-06-07 Tamar Christina <tamar.christina@arm.com>
19036 * config/aarch64/aarch64.md
19037 (copysignsf3): Fix mask generation.
19039 2017-06-07 Jakub Jelinek <jakub@redhat.com>
19041 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
19043 (class dump_manager): Add register_dumps method.
19044 * dumpfile.c: Include langhooks.h.
19045 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
19046 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
19047 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
19048 (dump_manager::dump_register): Start with 512 entries instead of 32.
19049 (dump_manager::register_dumps): New method.
19050 * toplev.c (general_init): Instead of invoking register_dumps
19051 langhook, invoke register_dumps method on the dump manager.
19052 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
19055 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
19057 * doc/md.texi: Clarify the restrictions on a define_insn condition.
19058 Say that # requires an associated define_split to exist, and that
19059 the define_split must be suitable for use after register allocation.
19061 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19063 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
19064 (compute_outgoing_frequencies): Also initialize zero counts.
19065 (find_many_sub_basic_blocks): Do not produce uninitialized profile
19066 around loops; preserve more of profile when nothing changes.
19068 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
19070 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
19072 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
19073 * config/arm/arm-cpu-cdata.h: Regenerate.
19074 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
19075 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
19076 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
19077 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
19078 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
19080 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
19082 * config/arm/t-rmprofile: Likewise.
19083 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
19085 2017-06-06 David S. Miller <davem@davemloft.net>
19088 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
19089 blockage if function uses alloca.
19091 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19093 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
19094 New "uid" fields to hold pretty-print IDs of group and ref.
19095 Memory references are now identified as <group_id>:<ref_id>
19096 instead of using [random] addresses.
19097 (dump_mem_details): Simplify, no functional change.
19098 (dump_mem_ref): Simplify and make output more concise.
19099 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
19100 (find_or_create_group): Initialize group uid.
19101 (record_ref): Initialize ref uid. Improve debug output.
19102 (prune_group_by_reuse, should_issue_prefetch_p,)
19103 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
19104 (mark_nontemporal_store, determine_loop_nest_reuse):
19105 Improve debug output.
19107 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19109 * dbgcnt.def (prefetch): New debug counter.
19110 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
19111 (schedule_prefetches): Stop issueing prefetches if debug counter
19114 2017-06-06 Tom de Vries <tom@codesourcery.com>
19116 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
19117 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
19119 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19121 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
19122 Use aarch64_reg_or_zero predicate for operand 4.
19123 (aarch64_compare_and_swap<mode> define_insn_and_split):
19124 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
19125 (aarch64_store_exclusive<mode>): Likewise for operand 2.
19127 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
19129 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
19130 (arm_compute_save_core_reg_mask): This.
19131 (thumb1_compute_save_reg_mask): Rename into ...
19132 (thumb1_compute_save_core_reg_mask): This.
19133 (arm_compute_save_reg0_reg12_mask): Adapt comment.
19134 (arm_compute_frame_layout): Likewise.
19136 2017-06-06 Richard Biener <rguenther@suse.de>
19138 PR tree-optimization/80974
19139 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
19140 keep or clear leaders SSA info.
19142 2017-06-06 Tom de Vries <tom@codesourcery.com>
19144 * config/nvptx/nvptx.c (split_mode_p): New function.
19145 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
19147 2017-06-06 Tom de Vries <tom@codesourcery.com>
19149 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
19151 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19154 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
19157 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19159 * shrink-wrap.c (handle_simple_exit): Update profile.
19160 (try_shrink_wrapping): Upate profile.
19162 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19164 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
19165 (tree_guess_outgoing_edge_probabilities): New.
19166 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
19167 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
19169 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19171 * ipa-split.c (split_function): Initialize return bb profile.
19173 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19175 * profile.c (compute_branch_probabilities): Also initialize
19176 EXIT_BLOCK profile.
19178 2017-06-06 Richard Biener <rguenther@suse.de>
19180 PR tree-optimization/80928
19181 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
19182 (vect_analyze_loop_operations): Properly guard analysis for
19184 (vect_transform_loop): Likewise.
19185 (vect_analyze_loop_2): Also reset SLP type on PHIs.
19186 (vect_model_induction_cost): Do not cost for pure SLP.
19187 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
19188 of induction in inner loop vectorization.
19189 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
19190 (vect_get_and_check_slp_defs): Handle vect_induction_def.
19191 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
19193 (vect_analyze_slp_cost_1): Cost induction.
19194 (vect_detect_hybrid_slp_stmts): Handle PHIs.
19195 (vect_get_slp_vect_defs): Likewise.
19196 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
19197 (vect_transform_stmt): Handle SLP reductions.
19198 * tree-vectorizer.h (vectorizable_induction): Adjust.
19200 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19202 * config/rs6000/rs6000.c (make_resolver_func): Update
19203 init_lowered_empty_function call.
19205 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
19207 * doc/invoke.texi: Document the -fprofile-abs-path option.
19208 * common.opt (fprofile-abs-path): New option.
19209 * gcov-io.h (gcov_write_filename): Declare.
19210 * gcov-io.c (gcov_write_filename): New function.
19211 * coverage.c (coverage_begin_function): Use gcov_write_filename.
19212 * profile.c (output_location): Likewise.
19214 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19216 * shring-wrap.c: Revert accidental commit.
19218 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
19220 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
19222 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19224 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
19226 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
19227 profile in callgraph edge.
19228 * profile-count.h (apply_probability): If THIS is 0, then result is 0
19229 (apply_scale): Likewise.
19230 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
19231 Also scale profile when inlining function with zero profile.
19232 (initialize_cfun): Update exit block profile even when it is zero.
19233 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
19234 when profile is read.
19236 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19238 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
19239 (CLONE_*): New constants to define the processors we can generate
19240 code for with the target_clone attribute.
19241 (rs6000_clone_map): New array to identify which clone processors
19242 the current program is running on.
19243 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
19244 target_clone attribute.
19245 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
19246 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
19247 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
19248 (cpu_expand_builtin): Add support for target_clone attribute.
19249 (rs6000_valid_attribute_p): Allow "default" attribute.
19250 (get_decl_name): New debug function to simplify printing the
19251 current function name in debugging statements.
19252 (rs6000_clone_priority): New functions to support the target_clone
19253 attribute, and be able to generate code to switch between ISA 2.05
19254 through ISA 3.0 (power6 through power9).
19255 (rs6000_compare_version_priority): Likewise.
19256 (rs6000_get_function_versions_dispatcher): Likewise.
19257 (make_resolver_func): Likewise.
19258 (add_condition_to_bb): Likewise.
19259 (dispatch_function_versions): Likewise.
19260 (rs6000_generate_version_dispatcher_body): Likewise.
19261 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
19262 (fusion_gpr_load_p): Fix a spacing issue.
19263 * doc/extend.texi (Common Function Attributes): Document that the
19264 PowerPC supports the target_clone attribute.
19266 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
19268 * config/arm/arm.h: explain F symbol found in description of ARM
19269 register allocation in its legend.
19271 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19273 * config/mips/frame-header-opt.c: Include profile-count.h.
19274 * config/riscv/riscv.c: Include profile-count.h
19276 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19278 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
19280 (sm_set_flag_if_changed): Add bbs field.
19281 (execute_sm_if_changed_flag_set): Pass BBS.
19282 (execute_sm): Update.
19284 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19286 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
19289 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19291 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
19292 (peephole2): New peephole2 to emit the above.
19293 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
19295 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19297 * config/aarch64/aarch64.c (define_peephole2 above
19298 *sub_<shift>_<mode>): New peephole.
19300 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19302 * config/i386/i386.c (make_resolver_func): Update.
19303 * Makefile.in: Add profile-count.h and profile-count.o
19304 * auto-profile.c (afdo_indirect_call): Update to new API.
19305 (afdo_set_bb_count): Update.
19306 (afdo_propagate_edge): Update.
19307 (afdo_propagate_circuit): Update.
19308 (afdo_calculate_branch_prob): Update.
19309 (afdo_annotate_cfg): Update.
19310 * basic-block.h: Include profile-count.h
19311 (struct edge_def): Turn count to profile_count.
19312 (struct basic_block_def): Likewie.
19313 (REG_BR_PROB_BASE): Move to profile-count.h
19314 (RDIV): Move to profile-count.h
19315 * bb-reorder.c (max_entry_count): Turn to profile_count.
19316 (find_traces): Update.
19317 (rotate_loop):Update.
19318 (connect_traces):Update.
19319 (sanitize_hot_paths):Update.
19320 * bt-load.c (migrate_btr_defs): Update.
19321 * cfg.c (RDIV): Remove.
19322 (init_flow): Use alloc_block.
19323 (alloc_block): Uninitialize count.
19324 (unchecked_make_edge): Uninitialize count.
19325 (check_bb_profile): Update.
19326 (dump_edge_info): Update.
19327 (dump_bb_info): Update.
19328 (update_bb_profile_for_threading): Update.
19329 (scale_bbs_frequencies_int): Update.
19330 (scale_bbs_frequencies_gcov_type): Update.
19331 (scale_bbs_frequencies_profile_count): New.
19332 * cfg.h (update_bb_profile_for_threading): Update.
19333 (scale_bbs_frequencies_profile_count): Declare.
19334 * cfgbuild.c (compute_outgoing_frequencies): Update.
19335 (find_many_sub_basic_blocks): Update.
19336 * cfgcleanup.c (try_forward_edges): Update.
19337 (try_crossjump_to_edge): Update.
19338 * cfgexpand.c (expand_gimple_tailcall): Update.
19339 (construct_exit_block): Update.
19340 * cfghooks.c (verify_flow_info): Update.
19341 (dump_bb_for_graph): Update.
19342 (split_edge): Update.
19343 (make_forwarder_block): Update.
19344 (duplicate_block): Update.
19345 (account_profile_record): Update.
19346 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
19347 (get_estimated_loop_iterations): Update.
19348 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
19349 (single_likely_exit): Update.
19350 * cfgloopmanip.c (scale_loop_profile): Update.
19352 (set_zero_probability): Update.
19353 (lv_adjust_loop_entry_edge): Update.
19354 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
19355 (purge_dead_edges): Update.
19356 (rtl_account_profile_record): Update.
19357 * cgraph.c (cgraph_node::create): Uninitialize count.
19358 (symbol_table::create_edge): Uninitialize count.
19359 (cgraph_update_edges_for_call_stmt_node): Update.
19360 (cgraph_edge::dump_edge_flags): Update.
19361 (cgraph_node::dump): Update.
19362 (cgraph_edge::maybe_hot_p): Update.
19363 * cgraph.h: Include profile-count.h
19364 (create_clone), create_edge, create_indirect_edge): Update.
19365 (cgraph_node): Turn count to profile_count.
19366 (cgraph_edge0: Likewise.
19367 (make_speculative, clone): Update.
19368 (create_edge): Update.
19369 (init_lowered_empty_function): Update.
19370 * cgraphclones.c (cgraph_edge::clone): Update.
19371 (duplicate_thunk_for_node): Update.
19372 (cgraph_node::create_clone): Update.
19373 * cgraphunit.c (cgraph_node::analyze): Update.
19374 (cgraph_node::expand_thunk): Update.
19375 * final.c (dump_basic_block_info): Update.
19376 * gimple-streamer-in.c (input_bb): Update.
19377 * gimple-streamer-out.c (output_bb): Update.
19378 * graphite.c (print_global_statistics): Update.
19379 (print_graphite_scop_statistics): Update.
19380 * hsa-brig.c: Include basic-block.h.
19381 * hsa-dump.c: Include basic-block.h.
19382 * hsa-gen.c (T sum_slice): Update.
19383 (convert_switch_statements):Update.
19384 * hsa-regalloc.c: Include basic-block.h.
19385 * ipa-chkp.c (chkp_produce_thunks): Update.
19386 * ipa-cp.c (struct caller_statistics): Update.
19387 (init_caller_stats): Update.
19388 (gather_caller_stats): Update.
19389 (ipcp_cloning_candidate_p): Update.
19390 (good_cloning_opportunity_p): Update.
19391 (get_info_about_necessary_edges): Update.
19392 (dump_profile_updates): Update.
19393 (update_profiling_info): Update.
19394 (update_specialized_profile): Update.
19395 (perhaps_add_new_callers): Update.
19396 (decide_about_value): Update.
19397 (ipa_cp_c_finalize): Update.
19398 * ipa-devirt.c (struct odr_type_warn_count): Update.
19399 (struct decl_warn_count): Update.
19400 (struct final_warning_record): Update.
19401 (possible_polymorphic_call_targets): Update.
19402 (ipa_devirt): Update.
19403 * ipa-fnsummary.c (redirect_to_unreachable): Update.
19404 * ipa-icf.c (sem_function::merge): Update.
19405 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
19406 * ipa-inline.c (compute_uninlined_call_time): Update.
19407 (compute_inlined_call_time): Update.
19408 (want_inline_small_function_p): Update.
19409 (want_inline_self_recursive_call_p): Update.
19410 (edge_badness): Update.
19411 (lookup_recursive_calls): Update.
19412 (recursive_inlining): Update.
19413 (inline_small_functions): Update.
19414 (dump_overall_stats): Update.
19415 (dump_inline_stats): Update.
19416 * ipa-profile.c (ipa_profile_generate_summary): Update.
19417 (ipa_propagate_frequency): Update.
19418 (ipa_profile): Update.
19419 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19420 * ipa-utils.c (ipa_merge_profiles): Update.
19421 * loop-doloop.c (doloop_modify): Update.
19422 * loop-unroll.c (report_unroll): Update.
19423 (unroll_loop_runtime_iterations): Update.
19424 * lto-cgraph.c (lto_output_edge): Update.
19425 (lto_output_node): Update.
19426 (input_node): Update.
19427 (input_edge): Update.
19428 (merge_profile_summaries): Update.
19429 * lto-streamer-in.c (input_cfg): Update.
19430 * lto-streamer-out.c (output_cfg): Update.
19431 * mcf.c (create_fixup_graph): Update.
19432 (adjust_cfg_counts): Update.
19433 (sum_edge_counts): Update.
19434 * modulo-sched.c (sms_schedule): Update.
19435 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
19436 * predict.c (maybe_hot_count_p): Update.
19437 (probably_never_executed): Update.
19438 (dump_prediction): Update.
19439 (combine_predictions_for_bb): Update.
19440 (propagate_freq): Update.
19441 (handle_missing_profiles): Update.
19442 (counts_to_freqs): Update.
19443 (rebuild_frequencies): Update.
19444 (force_edge_cold): Update.
19445 * predict.h: Include profile-count.h
19446 (maybe_hot_count_p, counts_to_freqs): UPdate.
19447 * print-rtl-function.c: Do not include cfg.h
19448 * print-rtl.c: Include basic-block.h
19449 * profile-count.c: New file.
19450 * profile-count.h: New file.
19451 * profile.c (is_edge_inconsistent): Update.
19452 (correct_negative_edge_counts): Update.
19453 (is_inconsistent): Update.
19454 (set_bb_counts): Update.
19455 (read_profile_edge_counts): Update.
19456 (compute_frequency_overlap): Update.
19457 (compute_branch_probabilities): Update; Initialize and deinitialize
19459 (branch_prob): Update.
19460 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
19461 (edge_gcov_count): New.
19462 (bb_gcov_count): New.
19463 * shrink-wrap.c (try_shrink_wrapping): Update.
19464 * tracer.c (better_p): Update.
19465 * trans-mem.c (expand_transaction): Update.
19466 (ipa_tm_insert_irr_call): Update.
19467 (ipa_tm_insert_gettmclone_call): Update.
19468 * tree-call-cdce.c: Update.
19469 * tree-cfg.c (gimple_duplicate_sese_region): Update.
19470 (gimple_duplicate_sese_tail): Update.
19471 (gimple_account_profile_record): Update.
19472 (execute_fixup_cfg): Update.
19473 * tree-inline.c (copy_bb): Update.
19474 (copy_edges_for_bb): Update.
19475 (initialize_cfun): Update.
19476 (freqs_to_counts): Update.
19477 (copy_cfg_body): Update.
19478 (expand_call_inline): Update.
19479 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19480 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
19481 (try_unroll_loop_completely): Update.
19482 (try_peel_loop): Update.
19483 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19484 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
19485 * tree-ssa-loop-split.c (connect_loops): Update.
19486 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
19487 * tree-ssa-reassoc.c (branch_fixup): Update.
19488 * tree-ssa-tail-merge.c (replace_block_by): Update.
19489 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
19490 (compute_path_counts): Update.
19491 (update_profile): Update.
19492 (recompute_probabilities): Update.
19493 (update_joiner_offpath_counts): Update.
19494 (estimated_freqs_path): Update.
19495 (freqs_to_counts_path): Update.
19496 (clear_counts_path): Update.
19497 (ssa_fix_duplicate_block_edges): Update.
19498 (duplicate_thread_path): Update.
19499 * tree-switch-conversion.c (case_bit_test_cmp): Update.
19500 (struct switch_conv_info): Update.
19501 * tree-tailcall.c (decrease_profile): Update.
19502 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19503 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19504 * value-prof.c (check_counter): Update.
19505 (gimple_divmod_fixed_value): Update.
19506 (gimple_mod_pow2): Update.
19507 (gimple_mod_subtract): Update.
19508 (gimple_ic_transform): Update.
19509 (gimple_stringop_fixed_value): Update.
19510 * value-prof.h (gimple_ic): Update.
19512 2017-06-02 Carl Love <cel@us.ibm.com>
19514 * config/rs6000/rs6000-c: Add support for built-in functions
19515 vector double vec_doublee (vector signed int);
19516 vector double vec_doublee (vector unsigned int);
19517 vector double vec_doublee (vector float);
19518 vector double vec_doubleh (vector signed int);
19519 vector double vec_doubleh (vector unsigned int);
19520 vector double vec_doubleh (vector float);
19521 vector double vec_doublel (vector signed int);
19522 vector double vec_doublel (vector unsigned int);
19523 vector double vec_doublel (vector float);
19524 vector double vec_doubleo (vector signed int);
19525 vector double vec_doubleo (vector unsigned int);
19526 vector double vec_doubleo (vector float);.
19527 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
19528 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
19530 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
19531 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
19532 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
19534 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
19535 vec_doublel, vec_doubleh.
19536 * doc/extend.texi: Update the built-in documentation file for the
19537 new built-in functions.
19539 2017-06-02 David Malcolm <dmalcolm@redhat.com>
19542 * ipa-inline-analysis.c (free_growth_caches): Set
19543 edge_removal_hook_holder to NULL after removing it.
19545 2017-06-02 Sudakshina Das <sudi.das@arm.com>
19547 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
19548 comparision with zero.
19550 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
19551 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19552 for early expansion of vec_min and vec_max builtins.
19553 (builtin_function_type): Add min/max unsigned variants to those
19554 identified as having unsigned arguments.
19556 2017-06-02 Olivier Hainque <hainque@adacore.com>
19558 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
19560 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19562 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
19563 Use VALL_F16 iterator rather than VALL.
19565 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19567 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
19568 Emit CBNZ inside loop when doing a strong exchange and comparing
19569 against zero. Generate the CC flags after the loop.
19571 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
19573 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
19574 (dl_section_ref): New.
19575 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
19576 On AIX, append an expression to subtract the size of the
19577 section length to dl_section_ref.
19579 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
19581 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19582 for early expansion of vector absolute builtins.
19584 2017-06-02 Richard Biener <rguenther@suse.de>
19586 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
19587 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
19589 2017-06-02 Richard Biener <rguenther@suse.de>
19591 PR tree-optimization/80948
19592 * tree-tailcall.c (find_tail_calls): Track stmts to move in
19593 stmt order as well.
19595 2017-06-02 Richard Biener <rguenther@suse.de>
19597 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
19599 * tree-vect-stmts.c (process_use): Do not mark backedge defs
19600 for inductions as relevant.
19602 2017-06-02 Richard Biener <rguenther@suse.de>
19604 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
19605 (vectorizable_induction): ... this. Remove dead code.
19607 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
19609 * builtins. (expand_builtin_alloca): Remove second parameter and
19610 infer its value from the first parameter instead.
19611 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
19613 2017-06-02 Jakub Jelinek <jakub@redhat.com>
19615 PR rtl-optimization/80903
19616 * loop-doloop.c (add_test): Unshare sequence.
19618 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
19620 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
19622 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
19624 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
19626 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
19627 xlogue_layout::get_instance, logue_layout::xlogue_layout,
19628 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
19629 (xlogue_layout::get_stub_rtx): Make static.
19630 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
19631 (xlogue_layout::compute_stub_managed_regs): Rename to...
19632 (xlogue_layout::count_stub_managed_regs): ...this.
19633 (xlogue_layout::is_stub_managed_reg): New function.
19634 (xlogue_layout::m_stub_names): Rename to...
19635 (xlogue_layout::s_stub_names): ...this, make static.
19636 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
19637 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
19638 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
19639 xlogue_layout::s_stub_names): Instantiate statics.
19640 (stub_managed_regs): Remove.
19641 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
19642 (disable_call_ms2sysv_xlogues): Rename to...
19643 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
19644 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
19646 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
19647 change after reload_completed.
19648 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
19650 (ix86_expand_prologue): Likewise.
19651 (ix86_expand_epilogue): Likewise.
19652 (ix86_expand_split_stack_prologue): Likewise.
19653 (ix86_compute_frame_layout): Remove frame parameter ...
19654 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
19655 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
19657 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
19658 (ix86_frame): Move from here ...
19659 * config/i386/i386.h (ix86_frame): ... to here.
19660 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
19661 complete ix86_frame data structure instead. Remove some_ld_name.
19663 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
19665 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
19666 symbols that hold a DECL_VALUE_EXPR.
19668 2017-06-01 Martin Jambor <mjambor@suse.cz>
19670 PR tree-optimization/80898
19671 * tree-sra.c (process_subtree_disqualification): Removed.
19672 (disqualify_candidate): Do not acll
19673 process_subtree_disqualification.
19674 (subtree_mark_written_and_enqueue): New function.
19675 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
19676 RHS has been disqualified and re-queue LHS if necessary. Apart
19677 from that, ignore disqualified RHS.
19679 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19681 * config/s390/s390.c (s390_emit_epilogue): Disable early return
19682 address fetch for z10 or later.
19684 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19686 * config/arc/arc.md (tst_movb): Add guard when splitting.
19688 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19690 * config/arc/arc.c (arc_can_eliminate): Test against
19691 arc_frame_pointer_needed.
19693 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19695 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
19696 to prevent store reordering.
19697 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
19698 (type): Add block type.
19699 (stack_tie): Define special instruction to be used in
19702 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19704 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
19705 constraint. It is not valid for the pattern.
19706 (noncommutative_binary_comparison): Likewise.
19708 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19710 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
19713 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19715 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
19716 be used by the reg-alloc.
19718 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19720 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
19721 reg-alloc when having mul64 or mul32x16 instructions.
19722 (mulsidi3): Likewise.
19723 (umulsidi3): Likewise.
19724 (mulsi32x16): New pattern.
19725 (mulsi64): Likewise.
19726 (mulsidi64): Likewise.
19727 (umulsidi64): Likewise.
19728 (MUL32x16_REG): Define.
19729 (mul64_600): Use MUL32x16_REG.
19730 (mac64_600): Likewise.
19731 (umul64_600): Likewise.
19732 (umac64_600): Likewise.
19734 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19736 * config/arc/arc.md (mulsi3_700): Make it commutative.
19738 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
19740 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
19742 (*sign_extendsidi2_insn): Likewise for movstosw.
19744 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
19746 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
19747 the type of the input discriminant value. Convert the
19748 discriminant value of signedness vary.
19750 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
19752 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
19753 Add to -Wall section.
19755 2017-06-01 Richard Biener <rguenther@suse.de>
19757 PR middle-end/66313
19758 * fold-const.c (fold_plusminus_mult_expr): If the factored
19759 factor may be zero use a wrapping type for the inner operation.
19760 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
19761 and handle moved defs.
19762 (process_assignment): Properly guard the unary op case. Return a
19763 tri-state indicating that moving the stmt before the call may allow
19764 to continue. Pass through to_move.
19765 (find_tail_calls): Handle moving unrelated defs before
19768 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
19771 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
19772 splitter result in the canonical way.
19774 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
19776 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
19777 also for 32bit target. Update insn attributes.
19778 (zero-extendsidi2 splitter): Allow all registers for operand 1.
19780 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
19782 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
19783 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
19784 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
19785 (_mm_maskz_min_ss): New intrinsics.
19787 2017-05-31 Martin Liska <mliska@suse.cz>
19789 * tree-vect-loop.c (vect_create_epilog_for_reduction):
19790 Change comment style to one we normally use.
19791 (vectorizable_reduction): Likewise.
19792 (vectorizable_induction): Likewise.
19793 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19794 (vectorizable_call): Likewise.
19795 (vectorizable_simd_clone_call): Likewise.
19796 (vectorizable_conversion): Likewise.
19797 (vectorizable_assignment): Likewise.
19798 (vectorizable_shift): Likewise.
19799 (vectorizable_operation): Likewise.
19800 (vectorizable_store): Likewise.
19801 (vectorizable_load): Likewise.
19802 * tree-vectorizer.h: Likewise.
19804 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
19806 * passes.c (emergency_dump_function): New.
19807 * tree-pass.h (emergency_dump_function): Declare.
19808 * plugin.c (plugins_internal_error_function): Remove.
19809 * plugin.h (plugins_internal_error_function): Remove declaration.
19810 * toplev.c (internal_error_function): New static function. Use it...
19811 (general_init): ...here.
19813 2017-05-31 Graham Markall <graham.markall@embecosm.com>
19815 * config/arc/arc.c (arc_print_operand): Handle constant operands.
19816 (arc_rtx_costs): Add costs for new patterns.
19817 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
19818 * config/arc/predicates.md: Add _1_2_3_operand predicate.
19820 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
19822 * tree-ssa-strlen.c (get_next_strinfo): New function.
19823 (get_stridx_plus_constant): Use it.
19824 (zero_length_string): Likewise.
19825 (adjust_related_strinfos): Likewise.
19826 (adjust_last_stmt): Likewise.
19828 2017-05-31 Richard Biener <rguenther@suse.de>
19831 * config/i386/i386.c (ix86_expand_builtin): Remove assert
19832 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
19834 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
19836 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
19837 loop_vinfo argument and use of dependence distance vectors.
19838 Check instead whether the two references differ only in their
19839 initial value and assume that they have the same alignment if the
19840 difference is a multiple of the vector alignment.
19841 (vect_analyze_data_refs_alignment): Update call accordingly.
19843 2017-05-31 Martin Liska <mliska@suse.cz>
19846 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
19848 2017-05-31 Bin Cheng <bin.cheng@arm.com>
19850 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
19851 (create_intersect_range_checks): Move from ...
19852 * tree-data-ref.c (create_intersect_range_checks_index)
19853 (create_intersect_range_checks): ... to here.
19854 (create_runtime_alias_checks): New function factored from ...
19855 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
19856 here. Call above function.
19857 * tree-data-ref.h (create_runtime_alias_checks): New function.
19859 2017-05-31 Bin Cheng <bin.cheng@arm.com>
19861 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
19862 segment length for dr_b and compute it in wide_int.
19864 2017-05-31 Richard Biener <rguenther@suse.de>
19866 PR tree-optimization/80906
19867 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
19868 and pass through iv_map.
19869 (copy_bb_and_scalar_dependences): Adjust.
19870 (translate_pending_phi_nodes): Likewise.
19871 (copy_loop_close_phi_args): Handle code-generating IVs instead
19874 2017-05-30 David Malcolm <dmalcolm@redhat.com>
19876 * diagnostic-color.c (color_dict): Add "type-diff".
19877 (parse_gcc_colors): Update comment.
19878 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
19879 -fdiagnostics-show-template-tree and -fno-elide-type.
19880 (GCC_COLORS): Add type-diff to example.
19882 (-fdiagnostics-show-template-tree): New.
19883 (-fno-elide-type): New.
19884 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
19885 the pp_format_decoder callback. Call any m_format_postprocessor's
19887 (pretty_printer::pretty_printer): Initialize
19888 m_format_postprocessor.
19889 (pretty_printer::~pretty_printer): Delete any
19890 m_format_postprocessor.
19891 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
19892 (class format_postprocessor): New class.
19893 (struct pretty_printer::format_decoder): Document the new parameters.
19894 (struct pretty_printer::m_format_postprocessor): New field.
19895 * tree-diagnostic.c (default_tree_printer): Update for new
19896 bool and const char ** params.
19897 * tree-diagnostic.h (default_tree_printer): Likewise.
19899 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
19901 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
19902 (lwa_operand): Delete rs6000_gen_cell_microcode test.
19903 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
19904 rs6000_gen_cell_microcode code.
19905 (rs6000_final_prescan_insn): Delete.
19906 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
19907 "warn-cell-microcode" entries.
19908 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
19909 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
19910 throughout. Change cc_reg_not_micro_cr0_operand to
19911 cc_reg_not_cr0_operand throughout.
19912 (*extendhi<mode>2_noload): Delete.
19913 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
19914 (mwarn-cell-microcode): Delete.
19915 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
19916 -mgen-cell-microcode and -mwarn-cell-microcode.
19918 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
19921 * config/i386/constraints.md (Yd): New constraint.
19923 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
19924 and (?Yd, r) alternatives. Update insn attributes.
19925 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
19926 and (?*Yd, r) alternatives. Update insn attributes.
19927 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
19929 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
19931 * gimplify.c (gimplify_modify_expr): Don't create a
19932 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
19935 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
19937 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
19939 2017-05-30 Richard Biener <rguenther@suse.de>
19941 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
19942 and reduc_def fields.
19943 (STMT_VINFO_REDUC_TYPE): New define.
19944 (STMT_VINFO_REDUC_DEF): Likewise.
19945 (vect_force_simple_reduction): Adjust prototype.
19946 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
19947 (vect_is_simple_reduction): Remove check_reduction argument.
19948 (vect_force_simple_reduction): Adjust and set
19949 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
19950 (vectorizable_reduction): Do not re-do reduction analysis
19951 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
19952 * tree-parloops.c (gather_scalar_reductions): Adjust.
19954 2017-05-30 Richard Biener <rguenther@suse.de>
19956 PR middle-end/80901
19957 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
19960 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
19962 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
19963 Introduce unknown_misalignment parameter and remove vf.
19964 (vect_peeling_hash_get_lowest_cost):
19965 Pass unknown_misalignment parameter.
19966 (vect_enhance_data_refs_alignment):
19967 Fix unsupportable data ref treatment.
19969 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19971 * tree-vect-data-refs.c (vect_get_data_access_cost):
19972 Workaround for SLP handling.
19973 (vect_enhance_data_refs_alignment):
19974 Compute costs for doing no peeling at all, compare to the best
19975 peeling costs so far and avoid peeling if cheaper.
19977 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19979 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
19980 Return peeling info and set costs to zero for unlimited cost
19982 (vect_enhance_data_refs_alignment): Also inspect all datarefs
19983 with unknown misalignment. Compute and costs for unknown
19984 misalignment, compare them to the costs for known misalignment
19985 and choose the cheapest for peeling.
19987 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19989 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
19990 (vect_get_peeling_costs_all_drs): Create function.
19991 (vect_peeling_hash_get_lowest_cost):
19992 Use vect_get_peeling_costs_all_drs.
19993 (vect_peeling_supportable): Create function.
19994 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
19996 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19998 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
19999 DR_HAS_NEGATIVE_STEP.
20000 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
20001 (vect_enhance_data_refs_alignment): Use.
20002 (vect_duplicate_ssa_name_ptr_info): Use.
20003 * tree-vectorizer.h (dr_misalignment): Use.
20004 (known_alignment_for_access_p): Use.
20006 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20009 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
20011 (has_section_name): New function.
20013 2017-05-30 Martin Liska <mliska@suse.cz>
20016 * auto-profile.c (get_function_decl_from_block): Fix
20019 2017-05-30 Richard Biener <rguenther@suse.de>
20021 PR middle-end/80876
20022 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
20024 2017-05-30 Martin Liska <mliska@suse.cz>
20026 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
20027 * dumpfile.h (struct dump_file_info): Remove ctors.
20029 2017-05-30 Martin Liska <mliska@suse.cz>
20031 * predict.def: Fix GNU coding style.
20033 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
20035 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
20036 Mark 'to' argument with ATTRIBUTE_UNUSED.
20038 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
20040 * config/xtensa/xtensa.c (xtensa_emit_call): Use
20041 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
20042 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
20045 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
20047 * doc/install.texi (Options specification): Restore entry of
20048 --enable-sjlj-exceptions.
20050 2017-05-27 Michael Eager <eager@eagercon.com>
20053 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
20055 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
20057 * config/microblaze/microblaze.h
20058 (FIXED_REGISTERS): Update in macro.
20059 (CALL_USED_REGISTERS): Update in macro.
20061 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
20063 * doc/install.texi: Add links to macOS binary distributions.
20065 2017-05-27 Jakub Jelinek <jakub@redhat.com>
20069 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20071 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
20073 2017-05-26 Martin Liska <mliska@suse.cz>
20075 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
20077 2017-05-26 Martin Liska <mliska@suse.cz>
20079 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
20080 always leading ';; '.
20081 (dump_bb_info): Likewise.
20082 (brief_dump_cfg): Likewise.
20083 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
20084 * dumpfile.c: Remove usage of TDF_VERBOSE.
20085 * dumpfile.h (enum dump_kind): Likewise.
20086 (dump_gimple_bb_header): Do not use TDF_COMMENT.
20087 * print-tree.c (debug_verbose): Remove.
20088 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
20089 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
20090 * tree-diagnostic.c (default_tree_printer): Replace
20091 TDF_DIAGNOSTIC with TDF_SLIM.
20093 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20095 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
20096 in parameter loop, rather than loop_vinfo.
20097 (create_intersect_range_checks): Ditto.
20098 (vect_create_cond_for_alias_checks): Update call to above functions.
20100 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20102 PR tree-optimization/80815
20103 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
20104 for merging runtime alias checks. Handle negative DR_STEPs.
20106 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20108 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
20110 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
20111 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
20112 out code pruning runtime alias checks.
20113 * tree-data-ref.c (prune_runtime_alias_test_list): New function
20114 factored out from above.
20115 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
20117 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
20119 (prune_runtime_alias_test_list): New decalaration.
20121 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20123 * tree-vect-data-refs.c (compare_tree): Rename and move ...
20124 * tree-data-ref.c (data_ref_compare_tree): ... to here.
20125 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
20126 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
20127 (operator==, comp_dr_with_seg_len_pair): Ditto.
20128 (vect_prune_runtime_alias_test_list): Ditto.
20130 2017-05-26 Martin Liska <mliska@suse.cz>
20133 * params.def: Bound partial-inlining-entry-probability param.
20135 2017-05-26 Marek Polacek <polacek@redhat.com>
20138 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
20141 2017-05-26 Richard Biener <rguenther@suse.de>
20143 PR tree-optimization/80842
20144 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
20147 2017-05-26 Richard Biener <rguenther@suse.de>
20149 PR tree-optimization/80844
20150 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
20152 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
20154 * doc/md.texi (Machine Constraints): Update x86 family
20155 machine constraints section to match 'config/i386/constraints.md'.
20157 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
20159 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
20161 2017-05-25 Nathan Sidwell <nathan@acm.org>
20163 * doc/invoke.texi (--enable-languages): Update documentation.
20165 2017-05-25 Martin Liska <mliska@suse.cz>
20167 * dumpfile.c: Add TDF_FOLDING.
20168 * dumpfile.h (enum dump_kind): Likewise.
20169 * genmatch.c (dt_simplify::gen_1): Use it.
20171 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20173 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
20175 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20177 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
20178 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
20180 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20182 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
20183 * match.pd (X == C): Rewrite it here.
20184 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
20185 with_certain_nonzero_bits2): New predicates.
20186 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
20188 2017-05-24 Nathan Sidwell <nathan@acm.org>
20190 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
20193 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
20196 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
20198 * config/powerpcspe: New port. Files are copied from the rs6000
20199 port, with "rs6000" in filenames replaced by "powerpcspe".
20201 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
20203 PR rtl-optimization/80754
20204 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
20206 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
20208 * config/sparc/sparc.md (length): Return the correct value for -mflat
20209 sibcalls to match output_sibcall.
20211 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
20215 * config/rs6000/rs6000.c (struct machine_function): Add new field
20217 (rs6000_get_separate_components): Init that field, use it.
20218 (rs6000_components_for_bb): Use the field.
20220 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
20222 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
20224 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
20226 PR middle-end/80823
20227 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
20229 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20232 * config/s390/s390.c (s390_check_qrst_address): Check incoming
20233 address against address_operand predicate.
20234 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
20236 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
20238 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
20239 parameters passed indirectly.
20241 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
20243 * config/i386/i386.md (*movdi_internal): Remove SSE4
20244 alternative 18 (?r, *v). Update insn attributes.
20245 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
20246 Update insn attributes.
20247 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
20248 Update insn attributes.
20249 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
20250 alternative 1 (r, v). Remove isa attribute.
20251 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
20252 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
20253 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
20255 2017-05-23 Tom de Vries <tom@codesourcery.com>
20257 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
20260 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20262 * cgraphunit.c (symbol_table::process_new_functions): Update.
20263 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
20264 (inline_generate_summary): Rename to ...
20265 (ipa_fn_summary_generate): ... this one.
20266 (inline_read_summary): Rename to ...
20267 (ipa_fn_summary_read): ... this one.
20268 (inline_write_summary): Rename to ...
20269 (ipa_fn_summary_write): ... this one.
20270 (inline_free_summary): Rename to ...
20271 (ipa_free_fn_summary): ... this one.
20272 (pass_data_local_fn_summary, pass_local_fn_summary,
20273 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
20274 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
20275 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
20276 make_pass_ipa_fn_summary): New.
20277 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
20278 inline_write_summary, inline_free_summary): Remove.
20279 (ipa_free_fn_summary) : New.
20280 * ipa-inline.c (ipa_inline): Update.
20281 (pass_ipa_inline): Do not generate summaries.
20282 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
20284 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
20285 and add pass_ipa_fn_summary.
20286 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
20288 (make_pass_inline_parameters): Remove.
20290 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
20292 * omp-low.c (struct omp_context): Remove "default_kind" member.
20295 * omp-offload.c (execute_oacc_device_lower): Remove the
20296 parallelism dimensions function attributes for unparallelized
20297 OpenACC kernels constructs.
20299 2017-05-23 Martin Liska <mliska@suse.cz>
20301 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
20303 (cgraph_edge::make_speculative): Likewise.
20304 (cgraph_edge::resolve_speculation): Likewise.
20305 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
20306 (cgraph_node::dump): Likewise.
20307 * cgraph.h: Likewise.
20308 * cgraphunit.c (analyze_functions): Likewise.
20309 (symbol_table::compile): Likewise.
20310 * ipa-cp.c (print_all_lattices): Likewise.
20311 (determine_versionability): Likewise.
20312 (initialize_node_lattices): Likewise.
20313 (ipcp_verify_propagated_values): Likewise.
20314 (estimate_local_effects): Likewise.
20315 (update_profiling_info): Likewise.
20316 (create_specialized_node): Likewise.
20317 (perhaps_add_new_callers): Likewise.
20318 (decide_about_value): Likewise.
20319 (decide_whether_version_node): Likewise.
20320 (identify_dead_nodes): Likewise.
20321 (ipcp_store_bits_results): Likewise.
20322 * ipa-devirt.c (dump_targets): Likewise.
20323 (ipa_devirt): Likewise.
20324 * ipa-icf.c (sem_item::dump): Likewise.
20325 (sem_function::equals): Likewise.
20326 (sem_variable::equals): Likewise.
20327 (sem_item_optimizer::read_section): Likewise.
20328 (sem_item_optimizer::execute): Likewise.
20329 (congruence_class::dump): Likewise.
20330 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
20331 (dump_inline_summary): Likewise.
20332 (estimate_node_size_and_time): Likewise.
20333 (inline_analyze_function): Likewise.
20334 * ipa-inline-transform.c (inline_call): Likewise.
20335 * ipa-inline.c (report_inline_failed_reason): Likewise.
20336 (want_early_inline_function_p): Likewise.
20337 (edge_badness): Likewise.
20338 (update_edge_key): Likewise.
20339 (inline_small_functions): Likewise.
20340 * ipa-profile.c (ipa_profile): Likewise.
20341 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
20342 (ipa_make_edge_direct_to_target): Likewise.
20343 (remove_described_reference): Likewise.
20344 (ipa_impossible_devirt_target): Likewise.
20345 (propagate_controlled_uses): Likewise.
20346 (ipa_print_node_params): Likewise.
20347 (ipcp_transform_function): Likewise.
20348 * ipa-pure-const.c (pure_const_read_summary): Likewise.
20349 (propagate_pure_const): Likewise.
20350 * ipa-reference.c (generate_summary): Likewise.
20351 (read_write_all_from_decl): Likewise.
20352 (propagate): Likewise.
20353 (ipa_reference_read_optimization_summary): Likewise.
20354 * ipa-utils.c (ipa_merge_profiles): Likewise.
20355 * ipa.c (walk_polymorphic_call_targets): Likewise.
20356 (symbol_table::remove_unreachable_nodes): Likewise.
20357 (ipa_single_use): Likewise.
20358 * passes.c (execute_todo): Likewise.
20359 * predict.c (drop_profile): Likewise.
20360 * symtab.c (symtab_node::get_dump_name): New function.
20361 (symtab_node::dump_name): Likewise.
20362 (symtab_node::dump_asm_name): Likewise.
20363 (symtab_node::dump_references): Likewise.
20364 (symtab_node::dump_referring): Likewise.
20365 (symtab_node::dump_base): Likewise.
20366 (symtab_node::debug_symtab): Likewise.
20367 * tree-sra.c (convert_callers_for_node): Likewise.
20368 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
20369 * value-prof.c (init_node_map): Likewise.
20371 2017-05-23 Martin Liska <mliska@suse.cz>
20373 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
20374 and symtab_node::debug_symtab to symbol_table::debug.
20375 * cgraphunit.c (analyze_functions): Use the renamed function.
20376 (symbol_table::compile): Likewise.
20377 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
20378 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
20379 * passes.c (execute_todo): Likewise.
20380 * symtab.c (symbol_table::dump): New function.
20381 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
20383 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20385 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
20386 that nonconst implies exec.
20388 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20390 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
20391 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
20392 (inline_edge_summary_vec): Turn into ...
20393 (ipa_call_summaries): ... this one.
20394 (redirect_to_unreachable, edge_set_predicate,
20395 evaluate_properties_for_edge, inline_summary_alloc,
20396 reset_ipa_call_summary, reset_inline_summary,
20397 inline_summary_t::duplicate): Update.
20398 (inline_edge_duplication_hook): Turn to ...
20399 (ipa_call_summary_t::duplicate): ... this one.
20400 (inline_edge_removal_hook): Turn to ...
20401 (ipa_call_summary_t::remove): ... this one.
20402 (dump_inline_edge_summary): Turn to ...
20403 (dump_ipa_call_summary): ... this one.
20404 (estimate_function_body_sizes): Update.
20405 (inline_update_callee_summaries): Update.
20406 (remap_edge_change_prob): Update.
20407 (remap_edge_summaries): Update.
20408 (inline_merge_summary): Update.
20409 (do_estimate_edge_time): Update.
20410 (inline_generate_summary): Update.
20411 (inline_read_section): Update.
20412 (inline_read_summary): Update.
20413 (inline_free_summary): Update.
20414 * ipa-inline.c (can_inline_edge_p): Update.
20415 (compute_inlined_call_time): Update.
20416 (want_inline_small_function_p): Update.
20417 (edge_badness): Update.
20418 (early_inliner): Update.
20419 * ipa-inline.h (inline_edge_summary): Turn to ...
20420 (ipa_call_summary): ... this one.
20421 (ipa_call_summary_t): New class.
20422 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
20423 (ipa_call_summaries): New.
20424 (inline_edge_summary): Remove.
20425 (estimate_edge_growth): Update.
20426 * ipa-profile.c (ipa_propagate_frequency_1): Update.
20427 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20428 * ipa-split.c (execute_split_functions): Update.
20429 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
20431 2017-05-23 Tom de Vries <tom@codesourcery.com>
20433 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
20434 attributes): Document rdrand effective target.
20436 2017-05-23 Tom de Vries <tom@codesourcery.com>
20438 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
20439 attributes): Sort alphabetically.
20441 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
20443 * config/avr/genmultilib.awk: Use gsub instead of gensub.
20445 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
20448 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
20449 V2DF/V2DI splat into two separate patterns, one that handles
20450 registers, and the other that only handles memory. Drop support
20451 for splatting from a GPR on ISA 2.07 and then splitting the
20452 splat into direct move and splat.
20453 (vsx_splat_<mode>_reg): Likewise.
20454 (vsx_splat_<mode>_mem): Likewise.
20456 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
20458 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
20460 2017-05-22 Jakub Jelinek <jakub@redhat.com>
20462 PR middle-end/80809
20463 * omp-low.c (finish_taskreg_remap): New function.
20464 (finish_taskreg_scan): If unit size of ctx->record_type
20465 is non-constant, unshare the size expression and replace
20466 decls in it with possible outer var refs.
20468 PR middle-end/80809
20469 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
20470 GOVD_SHARED rather than GOVD_PRIVATE with it.
20471 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
20472 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
20474 PR middle-end/80853
20475 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
20476 as last argument to build_outer_var_ref for pointer bases of array
20477 section reductions.
20479 2017-05-19 Martin Sebor <msebor@redhat.com>
20481 * print-tree.c (print_node): Print DECL_READ_P flag.
20483 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20485 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
20486 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
20487 * cgraph.c: Likewise.
20488 * cgraphunit.c: Likewise.
20489 * gengtype.c: Likewise.
20490 * ipa-cp.c: Likewise.
20491 * ipa-devirt.c: Likewise.
20492 * ipa-icf.c: Likewise.
20493 * ipa-predicate.c: Likewise.
20494 * ipa-profile.c: Likewise.
20495 * ipa-prop.c: Likewise.
20496 * ipa-split.c: Likewise.
20498 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
20499 edge_predicate_pool, dump_inline_hints,
20500 inline_summary::account_size_time, redirect_to_unreachable,
20501 edge_set_predicate, set_hint_predicate,
20502 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
20503 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
20504 inline_summary_t::remove, remap_hint_predicate_after_duplication,
20505 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
20506 ipa_call_summary_t::remove, initialize_growth_caches,
20507 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
20508 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
20509 mark_modified, unmodified_parm_1, unmodified_parm,
20510 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
20511 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
20512 compute_bb_predicates, will_be_nonconstant_expr_predicate,
20513 will_be_nonconstant_predicate, record_modified_bb_info,
20514 get_minimal_bb, record_modified, param_change_prob,
20515 phi_result_unknown_predicate, predicate_for_phi_result,
20516 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
20517 estimate_function_body_sizes, compute_inline_parameters,
20518 compute_inline_parameters_for_curren, pass_data_inline_parameters,
20519 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
20520 inline_update_callee_summaries, remap_edge_change_prob,
20521 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
20522 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
20523 inline_analyze_function, inline_summary_t::insert,
20524 inline_generate_summary, read_ipa_call_summary, inline_read_section,
20525 inline_read_summary, write_ipa_call_summary, inline_write_summary,
20526 inline_free_summary): Move to ipa-fnsummary.h
20527 (predicate_t): Remove.
20528 * ipa-fnsummary.c: New file.
20529 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
20530 (enum inline_hints_vals, inline_hints, agg_position_info,
20531 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
20532 inline_summaries, ipa_call_summary, ipa_call_summary_t,
20533 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
20534 dump_inline_summary, dump_inline_hints, inline_generate_summary,
20535 inline_read_summary, inline_write_summary, inline_free_summary,
20536 inline_analyze_function, initialize_inline_failed,
20537 inline_merge_summary, inline_update_overall_summary,
20538 compute_inline_parameters): Move to ipa-fnsummary.h
20539 * ipa-fnsummary.h: New file.
20540 * ipa-inline-transform.h: Include ipa-inline.h.
20541 * ipa-inline.c: LIkewise.
20543 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20545 * ipa-inline.c (edge_badness): Use inlined_time instead of
20546 inline_summaries->get.
20548 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20550 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
20552 2017-05-22 Nathan Sidwell <nathan@acm.org>
20554 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
20555 (fdump-lang): Document 'raw' option.
20556 * dumpfile.h (TDI_tu): Delete.
20557 * dumpfile.c (dump_files): Remove translation-unit.
20558 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
20560 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
20562 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
20563 command option from $(AWK) call.
20564 * config/avr/genmultilib.awk: Simplify and rewrite so that it
20565 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
20566 [FORMAT]: Remove handling of variable.
20567 * config/avr/t-multilib: Regenerate.
20569 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20571 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
20573 (dump_inline_summary): Do not print self_time.
20574 (estimate_function_body_sizes): Do not set self_time.
20575 (compute_inline_parameters): Likewise.
20576 (inline_read_section, inline_write_summary): Do not stream self_time.
20577 * ipa-inline.h (inline_summary): Drop self_time.
20579 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20581 * ipa-inline-analysis.c (account_size_time): Rename to ...
20582 (inline_summary::account_size_time): ... this one.
20583 (reset_ipa_call_summary): Turn to ...
20584 (ipa_call_summary::reset): ... this one.
20585 (reset_inline_summary): Turn to ...
20586 (inline_summary::reset): ... this one.
20587 (inline_summary_t::remove): Update.
20588 (inline_summary_t::duplicate): Update.
20589 (ipa_call_summary_t::remove): Update.
20590 (dump_inline_summary): Update.
20591 (estimate_function_body_sizes): Update.
20592 (compute_inline_parameters): Update.
20593 (estimate_node_size_and_time): Update.
20594 (inline_merge_summary): Update.
20595 (inline_update_overall_summary): Update.
20596 (inline_read_section): Update.
20597 (inline_write_summary): Update.
20598 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
20599 add account_size_time and reset member functions.
20600 (ipa_call_summary): Add reset function.
20601 * ipa-predicate.h (predicate::operator &): Constify.
20603 2017-05-22 Richard Biener <rguenther@suse.de>
20605 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
20607 2017-05-19 Jason Merrill <jason@redhat.com>
20609 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
20611 2017-05-19 Marek Polacek <polacek@redhat.com>
20614 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
20615 TYPE_OVERFLOW_WRAPS checks.
20617 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
20619 * tree-core.h (enum omp_clause_default_kind): Add
20620 "OMP_CLAUSE_DEFAULT_PRESENT".
20621 * tree-pretty-print.c (dump_omp_clause): Handle it.
20622 * gimplify.c (enum gimplify_omp_var_data): Add
20623 "GOVD_MAP_FORCE_PRESENT".
20624 (gimplify_adjust_omp_clauses_1): Map it to
20625 "GOMP_MAP_FORCE_PRESENT".
20626 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
20628 * gimplify.c (oacc_default_clause): Clarify.
20630 2017-05-19 Nathan Sidwell <nathan@acm.org>
20632 LANG_HOOK_REGISTER_DUMPS
20633 * toplev.c (general_init): Call register dump lang hook.
20634 * doc/invoke.texi: Document -fdump-lang option family.
20635 * dumpfile.c (dump_files): Remove class dump here.
20636 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
20637 * dumpfile.h (tree_dump_index): Remove TDI_class.
20638 * langhooks-def.h (lhd_register_dumps): Declare.
20639 (LANG_HOOKS_REGISTER_DUMPS): Define.
20640 (LANG_HOOKS_INITIALIZER): Add it.
20641 * langhooks.c (lhd_register_dumps): Define.
20642 * langhooks.h (struct lang_hooks): Add register_dumps.
20644 2017-05-19 Nathan Sidwell <nathan@acm.org>
20646 * context.h (context::set_passes): New.
20647 * context.c (context::context): Do not create pass manager.
20648 * toplev.c (general_init): Create pass manager here.
20650 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
20652 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
20653 use this splitter if two add or or instructions would also work for
20654 the constant we want to generate.
20656 2017-05-19 Richard Biener <rguenther@suse.de>
20659 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
20660 predicate evaluation.
20662 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
20664 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
20666 * ipa-inline.c (want_inline_small_function_p): Do not cast to
20669 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
20671 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
20672 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
20673 (inline_edge_summary_vec): Turn into ...
20674 (ipa_call_summaries): ... this one.
20675 (redirect_to_unreachable, edge_set_predicate,
20676 evaluate_properties_for_edge, inline_summary_alloc,
20677 reset_ipa_call_summary, reset_inline_summary,
20678 inline_summary_t::duplicate): Update.
20679 (inline_edge_duplication_hook): Turn to ...
20680 (ipa_call_summary_t::duplicate): ... this one.
20681 (inline_edge_removal_hook): Turn to ...
20682 (ipa_call_summary_t::remove): ... this one.
20683 (dump_inline_edge_summary): Turn to ...
20684 (dump_ipa_call_summary): ... this one.
20685 (estimate_function_body_sizes): Update.
20686 (inline_update_callee_summaries): Update.
20687 (remap_edge_change_prob): Update.
20688 (remap_edge_summaries): Update.
20689 (inline_merge_summary): Update.
20690 (do_estimate_edge_time): Update.
20691 (inline_generate_summary): Update.
20692 (inline_read_section): Update.
20693 (inline_read_summary): Update.
20694 (inline_free_summary): Update.
20695 * ipa-inline.c (can_inline_edge_p): Update.
20696 (compute_inlined_call_time): Update.
20697 (want_inline_small_function_p): Update.
20698 (edge_badness): Update.
20699 (early_inliner): Update.
20700 * ipa-inline.h (inline_edge_summary): Turn to ...
20701 (ipa_call_summary): ... this one.
20702 (ipa_call_summary_t): New class.
20703 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
20704 (ipa_call_summaries): New.
20705 (inline_edge_summary): Remove.
20706 (estimate_edge_growth): Update.
20707 * ipa-profile.c (ipa_propagate_frequency_1): Update.
20708 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20709 * ipa-split.c (execute_split_functions): Update.
20710 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
20712 2017-05-19 Richard Biener <rguenther@suse.de>
20714 PR middle-end/80764
20715 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
20717 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
20719 * config/rs6000/rs6000.c (struct machine_function): Add field
20720 fpr_is_wrapped_separately.
20721 (rs6000_get_separate_components): Use 64 components. Handle the
20722 new FPR components.
20723 (rs6000_components_for_bb): Handle the FPR components.
20724 (rs6000_emit_prologue_components): Handle the FPR components.
20725 (rs6000_emit_epilogue_components): Handle the FPR components.
20726 (rs6000_set_handled_components): Handle the FPR components.
20727 (rs6000_emit_prologue): Don't output prologue code for those FPRs
20728 that are already separately shrink-wrapped.
20729 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
20730 that are already separately shrink-wrapped.
20732 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
20735 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
20738 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
20739 (define_peephole2 for Altivec d-form load): Add peepholes to catch
20740 cases where the register allocator uses a move and an offsettable
20741 memory operation to/from a FPR register on ISA 2.06/2.07.
20742 (define_peephole2 for Altivec d-form store): Likewise.
20744 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
20747 * config/i386/mmx.md (*mov<mode>_internal): Enable
20748 alternatives 11, 12, 13 and 14 also for 32bit targets.
20749 Remove alternatives 15, 16, 17 and 18.
20750 * config/i386/sse.md (vec_concatv2di): Change
20751 alternative (!x, *y) to (x, ?!*Yn).
20753 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
20755 * dumpfile.h (enum dump_kind): Remove stray comma.
20757 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
20759 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
20760 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
20761 predicate::num_conditions
20762 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
20763 (CHANGED): turn into predicate::changed.
20764 (agg_position_info): Move to ipa-predicate.h
20765 (add_condition, predicate::add_clause, predicate::operator &=,
20766 predicate::or_with, predicate::evaluate, predicate::probability,
20767 dump_condition, dump_clause, predicate::dump,
20768 predicate::remap_after_duplication, predicate::remap_after_inlining,
20769 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
20770 (evaluate_conditions_for_known_args): Update.
20771 (set_cond_stmt_execution_predicate): Update.
20772 * ipa-inline.h: Include ipa-predicate.h
20773 (condition, inline_param_summary, conditions, agg_position_info,
20774 predicate): Move to ipa-predicate.h
20775 * ipa-predicate.c: New file.
20776 * ipa-predicate.h: New file.
20778 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
20780 * final.c (leaf_function_p): Check we are not in a sequence.
20782 2017-05-18 Martin Liska <mliska@suse.cz>
20784 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
20785 * dumpfile.c (dump_register): Use new enum dump_kind.
20786 (get_dump_file_name): Likewise.
20787 (dump_enable_all): Likewise.
20788 (dump_switch_p_1): Likewise.
20789 (enable_rtl_dump_file): Remove usage of TDF_RTL.
20790 * dumpfile.h (enum dump_kind): New enum type.
20791 (struct dump_file_info): Create constructor and
20792 format fields and comments.
20793 * passes.c (pass_manager::register_one_dump_file):
20795 * statistics.c (statistics_early_init): Likewise.
20796 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
20797 TDF_TREE with TDF_SLIM.
20798 (gather_memory_references_ref): Likewise.
20800 2017-05-18 Martin Liska <mliska@suse.cz>
20802 * vec.h (struct vnull): Use it.
20804 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
20806 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
20807 (true_predicate, false_predicate, true_predicate_p,
20808 false_predicate_p): Remove.
20809 (single_cond_predicate, not_inlined_predicate): Turn to member function
20811 (add_condition): Update.
20812 (add_clause): Turn to...
20813 (predicate::add_clause): ... this one; update; allow passing NULL
20815 (and_predicates): Turn to ...
20816 (predicate::operator &=): ... this one.
20817 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
20818 (or_predicates): Turn to ...
20819 (predicate::or_with): ... this one.
20820 (evaluate_predicate): Turn to ...
20821 (predicate::evaluate): ... this one.
20822 (predicate_probability): Turn to ...
20823 (predicate::probability): ... this one.
20824 (dump_condition): Update.
20825 (dump_predicate): Turn to ...
20826 (predicate::dump): ... this one.
20827 (account_size_time): Update.
20828 (edge_set_predicate): Update.
20829 (set_hint_predicate): UPdate.
20830 (evaluate_conditions_for_known_args): Update.
20831 (evaluate_properties_for_edge): Update.
20832 (remap_predicate_after_duplication): Turn to...
20833 (predicate::remap_after_duplication): ... this one.
20834 (remap_hint_predicate_after_duplication): Update.
20835 (inline_summary_t::duplicate): UPdate.
20836 (dump_inline_edge_summary): Update.
20837 (dump_inline_summary): Update.
20838 (set_cond_stmt_execution_predicate): Update.
20839 (set_switch_stmt_execution_predicate): Update.
20840 (compute_bb_predicates): Update.
20841 (will_be_nonconstant_expr_predicate): Update.
20842 (will_be_nonconstant_predicate): Update.
20843 (phi_result_unknown_predicate): Update.
20844 (predicate_for_phi_result): Update.
20845 (array_index_predicate): Update.
20846 (estimate_function_body_sizes): Update.
20847 (estimate_node_size_and_time): Update.
20848 (estimate_ipcp_clone_size_and_time): Update.
20849 (remap_predicate): Rename to ...
20850 (predicate::remap_after_inlining): ... this one.
20851 (remap_hint_predicate): Update.
20852 (inline_merge_summary): Update.
20853 (inline_update_overall_summary): Update.
20854 (estimate_size_after_inlining): Update.
20855 (read_predicate): Rename to ...
20856 (predicate::stream_in): ... this one.
20857 (read_inline_edge_summary): Update.
20858 (write_predicate): Rename to ...
20859 (predicate::stream_out): ... this one.
20860 (write_inline_edge_summary): Update.
20861 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
20862 (clause_t): Turn to uint32_t
20863 (predicate): Turn to class; implement constructor and operators
20865 (size_time_entry): Update.
20866 (inline_summary): Update.
20867 (inline_edge_summary): Update.
20869 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
20871 * fold-const.c (fold_binary_loc): Move transformation...
20872 * match.pd (C - X CMP X): ... here.
20874 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
20876 * config/sparc/sparc.c (sparc_option_override): Set function
20877 alignment for -mcpu=niagara7 to 64 to match the I$ line.
20878 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
20880 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
20882 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
20884 2017-05-18 Marek Polacek <polacek@redhat.com>
20887 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
20888 (pass_ubsan::execute): Call gimple_assign_single_p instead of
20889 gimple_assign_load_p.
20891 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
20893 PR middle-end/80692
20894 * real.c (do_compare): Give decimal_do_compare preference over
20895 comparing just the signs.
20897 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
20899 * doc/md.texi (Canonicalization of Instructions): Describe the
20900 canonical form of instructions that inherently set a condition
20903 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
20905 PR middle-end/80775
20906 * tree-cfg.c: Move deletion of unreachable case statements to after
20907 the merging of consecutive case labels.
20909 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
20911 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
20912 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
20913 restoring of callee-saved registers.
20915 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
20917 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
20918 * config/visium/visium.c (single_set_and_flags): Likewise.
20919 * config/visium/visium.md (Substitutions): Likewise.
20921 2017-05-17 Martin Liska <mliska@suse.cz>
20923 * cfg.c: Introduce dump_flags_t type and
20924 use it instead of int type.
20926 * cfghooks.c: Likewise.
20927 * cfghooks.h (struct cfg_hooks): Likewise.
20928 * cfgrtl.c: Likewise.
20929 * cfgrtl.h: Likewise.
20930 * cgraph.c (cgraph_node::get_body): Likewise.
20931 * coretypes.h: Likewise.
20932 * domwalk.c: Likewise.
20933 * domwalk.h: Likewise.
20934 * dumpfile.c (struct dump_option_value_info): Likewise.
20935 (dump_enable_all): Likewise.
20936 (dump_switch_p_1): Likewise.
20937 (opt_info_switch_p): Likewise.
20938 * dumpfile.h (enum tree_dump_index): Likewise.
20939 (struct dump_file_info): Likewise.
20940 * genemit.c: Likewise.
20941 * generic-match-head.c: Likewise.
20942 * gengtype.c (open_base_files): Likewise.
20943 * gimple-pretty-print.c: Likewise.
20944 * gimple-pretty-print.h: Likewise.
20945 * graph.c (print_graph_cfg): Likewise.
20946 * graphite-scop-detection.c (dot_all_sese): Likewise.
20947 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
20948 * loop-unroll.c (report_unroll): Likewise.
20949 * passes.c (pass_manager::register_one_dump_file): Likewise.
20950 * print-tree.c: Likewise.
20951 * statistics.c: Likewise.
20952 * tree-cfg.c: Likewise.
20953 * tree-cfg.h: Likewise.
20954 * tree-dfa.c: Likewise.
20955 * tree-dfa.h: Likewise.
20956 * tree-dump.c (dump_function): Likewise.
20957 * tree-dump.h (struct dump_info): Likewise.
20958 * tree-pretty-print.c: Likewise.
20959 * tree-pretty-print.h: Likewise.
20960 * tree-ssa-live.c: Likewise.
20961 * tree-ssa-live.h: Likewise.
20962 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
20963 * tree-vect-loop.c: Likewise.
20964 * tree-vect-slp.c: Likewise.
20966 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
20967 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20969 PR tree-optimization/80457
20970 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
20971 of all arguments to a statement as scalar_to_vec operations.
20972 (vectorizable_call): Adjust call to vect_model_simple_cost for
20974 (vectorizable_conversion): Likewise.
20975 (vectorizable_assignment): Likewise.
20976 (vectorizable_shift): Likewise.
20977 (vectorizable_operation): Likewise.
20978 (vectorizable_comparison): Likewise.
20979 (vect_is_simple_cond): Record the def types for operands.
20980 (vectorizable_condition): Likewise, call vect_model_simple_cost.
20981 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
20982 for statement argument count.
20984 2017-05-16 Carl Love <cel@us.ibm.com>
20986 * config/rs6000/rs6000-c: Add support for built-in functions
20987 vector unsigned long long vec_bperm (vector unsigned long long,
20988 vector unsigned char)
20989 vector signed long long vec_mule (vector signed int,
20991 vector unsigned long long vec_mule (vector unsigned int,
20992 vector unsigned int)
20993 vector signed long long vec_mulo (vector signed int,
20995 vector unsigned long long vec_mulo (vector unsigned int,
20996 vector unsigned int)
20997 vector signed char vec_sldw (vector signed char,
20998 vector signed char,
21000 vector unsigned char vec_sldw (vector unsigned char,
21001 vector unsigned char,
21003 vector signed short vec_sldw (vector signed short,
21004 vector signed short,
21006 vector unsigned short vec_sldw (vector unsigned short,
21007 vector unsigned short,
21009 vector signed int vec_sldw (vector signed int,
21012 vector unsigned int vec_sldw (vector unsigned int,
21013 vector unsigned int,
21015 vector signed long long vec_sldw (vector signed long long,
21016 vector signed long long,
21018 vector unsigned long long vec_sldw (vector unsigned long long,
21019 vector unsigned long long,
21021 * config/rs6000/rs6000-c: Add support for built-in functions
21022 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
21023 * config/rs6000/altivec.h: Add defintion for vec_sldw.
21024 * doc/extend.texi: Update the built-in documentation for the
21025 new built-in functions.
21027 2017-05-16 Marek Polacek <polacek@redhat.com>
21031 * tree.c (save_expr): Don't fold the expression.
21033 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
21035 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
21036 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
21037 and (?*y,m). Update insn attributes.
21039 2017-05-16 Martin Liska <mliska@suse.cz>
21041 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
21042 flags argument of print_gimple_stmt, print_gimple_expr,
21043 print_generic_stmt and print_generic_expr.
21044 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
21045 * coretypes.h: Likewise.
21046 * except.c (dump_eh_tree): Likewise.
21047 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
21048 * gimple-pretty-print.h: Likewise.
21049 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
21050 (backprop::push_to_worklist): Likewise.
21051 (backprop::pop_from_worklist): Likewise.
21052 (backprop::process_use): Likewise.
21053 (backprop::intersect_uses): Likewise.
21054 (note_replacement): Likewise.
21055 * gimple-ssa-store-merging.c
21056 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
21057 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
21058 (pass_store_merging::execute): Likewise.
21059 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
21060 (ssa_base_cand_dump_callback): Likewise.
21061 (dump_incr_vec): Likewise.
21062 (replace_refs): Likewise.
21063 (replace_mult_candidate): Likewise.
21064 (create_add_on_incoming_edge): Likewise.
21065 (create_phi_basis): Likewise.
21066 (insert_initializers): Likewise.
21067 (all_phi_incrs_profitable): Likewise.
21068 (introduce_cast_before_cand): Likewise.
21069 (replace_one_candidate): Likewise.
21070 * gimplify.c (gimplify_expr): Likewise.
21071 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
21072 (set_rename): Likewise.
21073 (rename_uses): Likewise.
21074 (copy_loop_phi_nodes): Likewise.
21075 (add_close_phis_to_merge_points): Likewise.
21076 (copy_loop_close_phi_args): Likewise.
21077 (copy_cond_phi_args): Likewise.
21078 (graphite_copy_stmts_from_block): Likewise.
21079 (translate_pending_phi_nodes): Likewise.
21080 * graphite-poly.c (print_pdr): Likewise.
21081 (dump_gbb_cases): Likewise.
21082 (dump_gbb_conditions): Likewise.
21083 (print_scop_params): Likewise.
21084 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
21085 (build_cross_bb_scalars_use): Likewise.
21086 (gather_bbs::before_dom_children): Likewise.
21087 * hsa-dump.c (dump_hsa_immed): Likewise.
21088 * ipa-cp.c (print_ipcp_constant_value): Likewise.
21089 (get_replacement_map): Likewise.
21090 * ipa-inline-analysis.c (dump_condition): Likewise.
21091 (estimate_function_body_sizes): Likewise.
21092 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
21093 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
21094 * ipa-prop.c (ipa_dump_param): Likewise.
21095 (ipa_print_node_jump_functions_for_edge): Likewise.
21096 (ipa_modify_call_arguments): Likewise.
21097 (ipa_modify_expr): Likewise.
21098 (ipa_dump_param_adjustments): Likewise.
21099 (ipa_dump_agg_replacement_values): Likewise.
21100 (ipcp_modif_dom_walker::before_dom_children): Likewise.
21101 * ipa-pure-const.c (check_stmt): Likewise.
21102 (pass_nothrow::execute): Likewise.
21103 * ipa-split.c (execute_split_functions): Likewise.
21104 * omp-offload.c (dump_oacc_loop_part): Likewise.
21105 (dump_oacc_loop): Likewise.
21106 * trans-mem.c (tm_log_emit): Likewise.
21107 (tm_memopt_accumulate_memops): Likewise.
21108 (dump_tm_memopt_set): Likewise.
21109 (dump_tm_memopt_transform): Likewise.
21110 * tree-cfg.c (gimple_verify_flow_info): Likewise.
21111 (print_loop): Likewise.
21112 * tree-chkp-opt.c (chkp_print_addr): Likewise.
21113 (chkp_gather_checks_info): Likewise.
21114 (chkp_get_check_result): Likewise.
21115 (chkp_remove_check_if_pass): Likewise.
21116 (chkp_use_outer_bounds_if_possible): Likewise.
21117 (chkp_reduce_bounds_lifetime): Likewise.
21118 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
21119 (chkp_mark_completed_bounds): Likewise.
21120 (chkp_register_incomplete_bounds): Likewise.
21121 (chkp_mark_invalid_bounds): Likewise.
21122 (chkp_maybe_copy_and_register_bounds): Likewise.
21123 (chkp_build_returned_bound): Likewise.
21124 (chkp_get_bound_for_parm): Likewise.
21125 (chkp_build_bndldx): Likewise.
21126 (chkp_get_bounds_by_definition): Likewise.
21127 (chkp_generate_extern_var_bounds): Likewise.
21128 (chkp_get_bounds_for_decl_addr): Likewise.
21129 * tree-chrec.c (chrec_apply): Likewise.
21130 * tree-data-ref.c (dump_data_reference): Likewise.
21131 (dump_subscript): Likewise.
21132 (dump_data_dependence_relation): Likewise.
21133 (analyze_overlapping_iterations): Likewise.
21134 * tree-inline.c (expand_call_inline): Likewise.
21135 (tree_function_versioning): Likewise.
21136 * tree-into-ssa.c (dump_defs_stack): Likewise.
21137 (dump_currdefs): Likewise.
21138 (dump_names_replaced_by): Likewise.
21139 (dump_update_ssa): Likewise.
21140 (update_ssa): Likewise.
21141 * tree-object-size.c (pass_object_sizes::execute): Likewise.
21142 * tree-parloops.c (build_new_reduction): Likewise.
21143 (try_create_reduction_list): Likewise.
21144 (ref_conflicts_with_region): Likewise.
21145 (oacc_entry_exit_ok_1): Likewise.
21146 (oacc_entry_exit_single_gang): Likewise.
21147 * tree-pretty-print.h: Likewise.
21148 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
21149 (get_scalar_evolution): Likewise.
21150 (add_to_evolution): Likewise.
21151 (get_loop_exit_condition): Likewise.
21152 (analyze_evolution_in_loop): Likewise.
21153 (analyze_initial_condition): Likewise.
21154 (analyze_scalar_evolution): Likewise.
21155 (instantiate_scev): Likewise.
21156 (number_of_latch_executions): Likewise.
21157 (gather_chrec_stats): Likewise.
21158 (final_value_replacement_loop): Likewise.
21159 (scev_const_prop): Likewise.
21160 * tree-sra.c (dump_access): Likewise.
21161 (disqualify_candidate): Likewise.
21162 (create_access): Likewise.
21163 (reject): Likewise.
21164 (maybe_add_sra_candidate): Likewise.
21165 (create_access_replacement): Likewise.
21166 (analyze_access_subtree): Likewise.
21167 (analyze_all_variable_accesses): Likewise.
21168 (sra_modify_assign): Likewise.
21169 (initialize_constant_pool_replacements): Likewise.
21170 (find_param_candidates): Likewise.
21171 (decide_one_param_reduction): Likewise.
21172 (replace_removed_params_ssa_names): Likewise.
21173 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
21174 * tree-ssa-copy.c (dump_copy_of): Likewise.
21175 (copy_prop_visit_cond_stmt): Likewise.
21176 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
21177 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
21178 (record_equivalences_from_stmt): Likewise.
21179 * tree-ssa-dse.c (compute_trims): Likewise.
21180 (delete_dead_call): Likewise.
21181 (delete_dead_assignment): Likewise.
21182 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
21183 (forward_propagate_into_cond): Likewise.
21184 (pass_forwprop::execute): Likewise.
21185 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
21186 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
21188 (move_computations_worker): Likewise.
21189 (execute_sm): Likewise.
21190 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
21191 (remove_exits_and_undefined_stmts): Likewise.
21192 (remove_redundant_iv_tests): Likewise.
21193 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
21194 (adjust_iv_update_pos): Likewise.
21195 * tree-ssa-math-opts.c (bswap_replace): Likewise.
21196 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
21197 (value_replacement): Likewise.
21198 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
21199 * tree-ssa-pre.c (print_pre_expr): Likewise.
21200 (get_representative_for): Likewise.
21201 (create_expression_by_pieces): Likewise.
21202 (insert_into_preds_of_block): Likewise.
21203 (eliminate_insert): Likewise.
21204 (eliminate_dom_walker::before_dom_children): Likewise.
21205 (eliminate): Likewise.
21206 (remove_dead_inserted_code): Likewise.
21207 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
21208 * tree-ssa-reassoc.c (get_rank): Likewise.
21209 (eliminate_duplicate_pair): Likewise.
21210 (eliminate_plus_minus_pair): Likewise.
21211 (eliminate_not_pairs): Likewise.
21212 (undistribute_ops_list): Likewise.
21213 (eliminate_redundant_comparison): Likewise.
21214 (update_range_test): Likewise.
21215 (optimize_range_tests_var_bound): Likewise.
21216 (optimize_vec_cond_expr): Likewise.
21217 (rewrite_expr_tree): Likewise.
21218 (rewrite_expr_tree_parallel): Likewise.
21219 (linearize_expr): Likewise.
21220 (break_up_subtract): Likewise.
21221 (linearize_expr_tree): Likewise.
21222 (attempt_builtin_powi): Likewise.
21223 (attempt_builtin_copysign): Likewise.
21224 (transform_stmt_to_copy): Likewise.
21225 (transform_stmt_to_multiply): Likewise.
21226 (dump_ops_vector): Likewise.
21227 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
21228 (print_scc): Likewise.
21229 (set_ssa_val_to): Likewise.
21230 (visit_reference_op_store): Likewise.
21231 (visit_use): Likewise.
21232 (sccvn_dom_walker::before_dom_children): Likewise.
21233 (run_scc_vn): Likewise.
21234 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
21236 (expr_hash_elt::print): Likewise.
21237 (const_and_copies::pop_to_marker): Likewise.
21238 (const_and_copies::record_const_or_copy_raw): Likewise.
21239 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
21240 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
21241 (dump_predicates): Likewise.
21242 (find_uninit_use): Likewise.
21243 (warn_uninitialized_phi): Likewise.
21244 (pass_late_warn_uninitialized::execute): Likewise.
21245 * tree-ssa.c (verify_vssa): Likewise.
21246 (verify_ssa): Likewise.
21247 (maybe_optimize_var): Likewise.
21248 * tree-vrp.c (dump_value_range): Likewise.
21249 (dump_all_value_ranges): Likewise.
21250 (dump_asserts_for): Likewise.
21251 (register_edge_assert_for_2): Likewise.
21252 (vrp_visit_cond_stmt): Likewise.
21253 (vrp_visit_switch_stmt): Likewise.
21254 (vrp_visit_stmt): Likewise.
21255 (vrp_visit_phi_node): Likewise.
21256 (simplify_cond_using_ranges_1): Likewise.
21257 (fold_predicate_in): Likewise.
21258 (evrp_dom_walker::before_dom_children): Likewise.
21259 (evrp_dom_walker::push_value_range): Likewise.
21260 (evrp_dom_walker::pop_value_range): Likewise.
21261 (execute_early_vrp): Likewise.
21263 2017-05-16 Richard Biener <rguenther@suse.de>
21265 * dwarf2out.c (loc_list_from_tree_1): Do not create
21266 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
21268 2017-05-16 Richard Biener <rguenther@suse.de>
21270 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
21272 (note_variable_value_in_expr): If we resolved the decl ref
21273 do not push to the stack.
21275 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
21277 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
21278 operations in fast-math mode.
21279 (vaddq_f16): Likewise.
21280 (vmul_f16): Likewise.
21281 (vmulq_f16): Likewise.
21282 (vsub_f16): Likewise.
21283 (vsubq_f16): Likewise.
21284 * config/arm/neon.md (add<mode>3): New.
21286 (fma:<VH:mode>3): New. Also remove outdated comment.
21289 2017-05-16 Martin Liska <mliska@suse.cz>
21293 * ipa-devirt.c (warn_types_mismatch): Fix typo.
21294 (odr_types_equivalent_p): Likewise.
21296 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
21298 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
21300 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
21303 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
21304 non-interunit SSE move alternatives with '?'.
21305 (zero-extendsidi peephole2): New peephole to skip intermediate
21306 general register in SSE zero-extend sequence.
21308 2017-05-15 Jeff Law <law@redhat.com>
21310 * reorg.c (relax_delay_slots): Create a new variable to hold
21311 the temporary target rather than clobbering TARGET_LABEL.
21313 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
21314 missing argument to extract_bit_field call.
21315 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
21317 2017-05-15 Martin Liska <mliska@suse.cz>
21320 * gcc.c (process_command): Do not allow empty argument of -o option.
21322 2017-05-15 Renlin Li <renlin.li@arm.com>
21324 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
21325 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
21326 * config/aarch64/constraints.md (Usf): Add long call check.
21327 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
21328 (call_value): Likewise.
21329 (sibcall): Likewise.
21330 (sibcall_value): Likewise.
21332 (call_value_insn): New.
21333 (sibcall_insn): Update rtx pattern.
21334 (sibcall_value_insn): Likewise.
21335 (call_internal): Remove.
21336 (call_value_internal): Likewise.
21337 (sibcall_internal): Likewise.
21338 (sibcall_value_internal): Likewise.
21339 (call_reg): Likewise.
21340 (call_symbol): Likewise.
21341 (call_value_reg): Likewise.
21342 (call_value_symbol): Likewise.
21344 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
21347 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
21349 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
21351 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
21352 compatible with CCGOCmode and with CCZmode.
21354 2017-05-14 Martin Sebor <msebor@redhat.com>
21356 PR middle-end/77671
21357 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
21358 (gimple_fold_builtin_snprintf): Same.
21359 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
21360 (gimple_fold_builtin_snprintf): Same.
21361 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
21362 of character types.
21363 (is_call_safe): New function.
21364 (try_substitute_return_value): Call it.
21365 (try_simplify_call): New function.
21366 (pass_sprintf_length::handle_gimple_call): Call it.
21368 2017-05-14 Martin Sebor <msebor@redhat.com>
21370 PR middle-end/80669
21371 * builtins.c (expand_builtin_stpncpy): Simplify.
21373 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
21375 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
21376 * config/i386/i386.h
21377 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
21378 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
21379 (struct machine_function): Add new members call_ms2sysv,
21380 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
21381 (struct machine_frame_state): New fields sp_realigned and
21382 sp_realigned_offset.
21383 * config/i386/i386.c
21384 (enum xlogue_stub): New enum.
21385 (enum xlogue_stub_sets): New enum.
21386 (class xlogue_layout): New class.
21387 (struct ix86_frame): New fields stack_realign_allocate_offset,
21388 stack_realign_offset and outlined_save_offset. Modify comments to
21389 detail stack layout when using out-of-line stubs.
21390 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
21391 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
21392 -mcall-ms2sysv-xlogues.
21393 (stub_managed_regs): New static variable.
21394 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
21395 registers managed by out-of-line stub.
21396 (disable_call_ms2sysv_xlogues): New function.
21397 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
21398 m->call_ms2sysv when appropriate and compute frame layout for
21400 (sp_valid_at, fp_valid_at): New inline functions.
21401 (choose_basereg): New function.
21402 (choose_baseaddr): Add align parameter, use choose_basereg and modify
21404 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
21405 Use align parameter of choose_baseaddr to generated aligned SSE movs
21407 (pro_epilogue_adjust_stack): Modify to track
21408 machine_frame_state::sp_realigned.
21409 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
21410 (ix86_nsaved_sseregs): Likewise.
21411 (ix86_emit_save_regs): Likewise.
21412 (ix86_emit_save_regs_using_mov): Likewise.
21413 (ix86_emit_save_sse_regs_using_mov): Likewise.
21414 (get_scratch_register_on_entry): Likewise.
21415 (gen_frame_set): New function.
21416 (gen_frame_load): Likewise.
21417 (gen_frame_store): Likewise.
21418 (emit_outlined_ms2sysv_save): Likewise.
21419 (emit_outlined_ms2sysv_restore): Likewise.
21420 (ix86_expand_prologue): Modify stack re-alignment code and call
21421 emit_outlined_ms2sysv_save when appropriate.
21422 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
21423 parameter rtx_insn *insn, which allows the function to be used to only
21424 generate the notes.
21425 (ix86_expand_epilogue): Modify validity checks of frame and stack
21426 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
21427 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
21428 * config/i386/predicates.md
21429 (save_multiple): New predicate.
21430 (restore_multiple): Likewise.
21431 * config/i386/sse.md
21432 (save_multiple<mode>): New pattern.
21433 (save_multiple_realign<mode>): Likewise.
21434 (restore_multiple<mode>): Likewise.
21435 (restore_multiple_and_return<mode>): Likewise.
21436 (restore_multiple_leave_return<mode>): Likewise.
21437 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
21439 2017-05-14 Julia Koval <julia.koval@intel.com>
21441 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
21442 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
21443 (__builtin_ia32_xsetbv): New builtins.
21444 * config/i386/i386.c (ix86_expand_special_args_builtin):
21446 (ix86_expand_builtin): Special expand for new intrinsics.
21447 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
21448 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
21449 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
21451 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21453 * cfganal.c (inverted_post_order_compute): Change argument type
21455 * cfganal.h (inverted_post_order_compute): Adjust prototype.
21456 * df-core.c (rest_of_handle_df_initialize): Adjust.
21457 (rest_of_handle_df_finish): Likewise.
21458 (df_analyze_1): Likewise.
21459 (df_analyze): Likewise.
21460 (loop_inverted_post_order_compute): Change argument to be a vec *.
21461 (df_analyze_loop): Adjust.
21462 (df_get_n_blocks): Likewise.
21463 (df_get_postorder): Likewise.
21464 * df.h (struct df_d): Change field to be a vec.
21465 * lcm.c (compute_laterin): Adjust.
21466 (compute_available): Likewise.
21467 * lra-lives.c (lra_create_live_ranges_1): Likewise.
21468 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
21469 * tree-ssa-pre.c (compute_antic): Likewise.
21471 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21473 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
21474 (depth_first_search::depth_first_search): Change structure init
21475 function to this constructor.
21476 (depth_first_search::add_bb): Rename function to this member.
21477 (depth_first_search::execute): Likewise.
21478 (flow_dfs_compute_reverse_finish): Adjust.
21480 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21482 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
21483 (longest_simple_path): Likewise.
21484 * shrink-wrap.c (spread_components): Likewise.
21485 (disqualify_problematic_components): Likewise.
21486 (emit_common_heads_for_components): Likewise.
21487 (emit_common_tails_for_components): Likewise.
21488 (insert_prologue_epilogue_for_components): Likewise.
21490 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21492 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
21495 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21497 * df-core.c (df_set_blocks): Start using auto_bitmap.
21498 (df_compact_blocks): Likewise.
21499 * df-problems.c (df_rd_confluence_n): Likewise.
21500 * df-scan.c (df_insn_rescan_all): Likewise.
21501 (df_process_deferred_rescans): Likewise.
21502 (df_update_entry_block_defs): Likewise.
21503 (df_update_exit_block_uses): Likewise.
21504 (df_entry_block_bitmap_verify): Likewise.
21505 (df_exit_block_bitmap_verify): Likewise.
21506 (df_scan_verify): Likewise.
21507 * lra-constraints.c (lra_constraints): Likewise.
21508 (undo_optional_reloads): Likewise.
21509 (lra_undo_inheritance): Likewise.
21510 * lra-remat.c (calculate_gen_cands): Likewise.
21511 (do_remat): Likewise.
21512 * lra-spills.c (assign_spill_hard_regs): Likewise.
21513 (spill_pseudos): Likewise.
21514 * tree-ssa-pre.c (bitmap_set_and): Likewise.
21515 (bitmap_set_subtract_values): Likewise.
21517 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21519 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
21520 management with auto_bitmap.
21521 (fix_inter_tick): Likewise.
21522 (fix_recovery_deps): Likewise.
21523 * ira.c (add_store_equivs): Likewise.
21524 (find_moveable_pseudos): Likewise.
21525 (split_live_ranges_for_shrink_wrap): Likewise.
21526 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
21527 (rtx_reuse_manager::seen_def_p): Likewise.
21528 (rtx_reuse_manager::set_seen_def): Likewise.
21529 * print-rtl.h (class rtx_reuse_manager): Likewise.
21531 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21533 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
21535 (migrate_btr_def): Likewise.
21536 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
21537 * df-core.c (loop_post_order_compute): Likewise.
21538 (loop_inverted_post_order_compute): Likewise.
21539 * hsa-common.h: Likewise.
21540 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
21541 * init-regs.c (initialize_uninitialized_regs): Likewise.
21542 * ipa-inline.c (resolve_noninline_speculation): Likewise.
21543 (inline_small_functions): Likewise.
21544 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
21545 * ira.c (combine_and_move_insns): Likewise.
21546 (build_insn_chain): Likewise.
21547 * loop-invariant.c (find_invariants): Likewise.
21548 * lower-subreg.c (propagate_pseudo_copies): Likewise.
21549 * predict.c (tree_predict_by_opcode): Likewise.
21550 (predict_paths_leading_to): Likewise.
21551 (predict_paths_leading_to_edge): Likewise.
21552 (estimate_loops_at_level): Likewise.
21553 (estimate_loops): Likewise.
21554 * shrink-wrap.c (try_shrink_wrapping): Likewise.
21555 (spread_components): Likewise.
21556 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
21557 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
21558 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
21559 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
21560 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
21561 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
21562 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
21563 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
21564 (mark_threaded_blocks): Likewise.
21565 (thread_through_all_blocks): Likewise.
21566 * tree-ssa.c (verify_ssa): Likewise.
21567 (execute_update_addresses_taken): Likewise.
21568 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
21570 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21572 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
21574 (post_order_compute): Likewise.
21575 (inverted_post_order_compute): Likewise.
21576 (pre_and_rev_post_order_compute_fn): Likewise.
21578 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21580 * genrecog.c (int_set::int_set): Explicitly construct our
21581 auto_vec base class.
21582 * vec.h (auto_vec::auto_vec): New constructor.
21584 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21586 * bitmap.h (class auto_bitmap): New constructor taking
21587 bitmap_obstack * argument.
21589 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21591 * bitmap.h (class auto_bitmap): Change type of m_bits to
21592 bitmap_head, and adjust ctor / dtor and member operators.
21594 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
21596 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
21597 when returned register mode doesn't match original mode.
21599 2017-05-12 Jeff Law <law@redhat.com>
21600 Jakub Jelinek <jakub@redhat.com>
21602 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
21603 we look for cc setter after the compare-elim changes.
21604 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
21605 within the vector to match what compare-elim now expects.
21606 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
21607 (xorsi3_flags, one_cmplsi2_flags): Likewise.
21609 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
21610 after the compare-elim changes.
21611 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
21612 the vector to match what compare-elim now expects.
21613 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
21614 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
21615 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
21616 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
21617 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
21619 * config/visium/visium.c (single_set_and_flags): Fix where
21620 we look for cc setter after the compare-elim changes.
21621 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
21622 with the vector to match what compare-elim now expects.
21623 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
21624 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
21625 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
21626 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
21627 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
21628 (neg<mode>2_insn_set_overflow): Likewise.
21630 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
21632 PR middle-end/79794
21633 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
21634 maybe_expand_insn call, set ops[0].target. If still set after call,
21635 set alt_rtl. Add extra arg to recursive calls.
21636 (extract_bit_field): Add alt_rtl argument. Pass to
21638 * expmed.h (extract_bit_field): Fix prototype.
21639 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
21640 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
21641 to extract_bit_field_calls.
21642 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
21643 Pass alt_rtl to extract_bit_field calls.
21644 * calls.c (store_unaligned_arguments_into_psuedos)
21645 load_register_parameters): Pass extra NULL to extract_bit_field calls.
21646 * optabs.c (maybe_legitimize_operand): Clear op->target when call
21648 * optabs.h (struct expand_operand): Add target bitfield.
21650 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
21652 * compare-elim.c (try_eliminate_compare): Canonicalize
21653 operation with embedded compare to
21654 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
21655 (set (reg) (operation)].
21657 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
21659 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
21662 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
21663 cost of adding a carry flag for ADC instruction.
21664 [case MINUS]: Ignore the cost of subtracting a carry flag
21665 for SBB instruction.
21667 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
21669 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
21671 * config/rs6000/bmiintrin.h: New file.
21672 * config/rs6000/bmi2intrin.h: New file.
21673 * config/rs6000/x86intrin.h: New file.
21675 2017-05-12 Jeff Law <law@redhat.com>
21677 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
21680 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
21682 PR middle-end/80707
21683 * tree-cfg.c: Remove cfg edges of unreachable case statements.
21685 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21687 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
21688 early expansion of vector divide builtins.
21689 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
21690 builtins identified as having unsigned arguments.
21692 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21694 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
21695 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
21696 expansion of vector logical operations (and, andc, or, xor,
21699 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21701 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
21702 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
21704 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21706 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
21707 early GIMPLE expansion of vector multiplies.
21709 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21711 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
21712 TARGET_HAVE_MOVT conditional.
21713 (movt splitter): Likewise.
21715 2017-05-12 Richard Biener <rguenther@suse.de>
21717 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
21718 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
21719 Fold all stmts not inplace.
21721 2017-05-12 Richard Biener <rguenther@suse.de>
21723 PR tree-optimization/80713
21724 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
21725 inserted_exprs bit for not removed stmts.
21727 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
21729 PR middle-end/69921
21730 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
21731 parallelized" attribute for parallelized OpenACC kernels.
21732 * omp-offload.c (execute_oacc_device_lower): Use it.
21734 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
21735 Set "oacc kernels" attribute.
21736 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
21737 parameter. Adjust all users.
21738 (oacc_fn_attrib_kernels_p): Remove function.
21739 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
21740 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
21741 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
21742 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
21743 assert "oacc kernels" attribute is set.
21745 2017-05-11 Carl Love <cel@us.ibm.com>
21747 * config/rs6000/rs6000-c: Add support for built-in functions
21748 vector unsigned char vec_popcnt (vector signed char)
21749 vector unsigned char vec_popcnt (vector unsigned char)
21750 vector unsigned short vec_popcnt (vector signed short)
21751 vector unsigned short vec_popcnt (vector unsigned short)
21752 vector unsigned int vec_popcnt (vector signed int)
21753 vector unsigned int vec_popcnt (vector unsigned int)
21754 vector unsigned long long vec_popcnt (vector signed long long)
21755 vector unsigned long long vec_popcnt (vector unsigned long long)
21756 vector signed long long vec_slo (vector signed long long,
21757 vector signed char)
21758 vector signed long long vec_slo (vector signed long long,
21759 vector unsigned char)
21760 vector unsigned long long vec_slo (vector unsigned long long,
21761 vector signed char)
21762 vector unsigned long long vec_slo (vector unsigned long long,
21763 vector unsigned char)
21764 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
21765 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
21766 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
21767 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
21768 * doc/extend.texi: Update the built-in documentation file for the
21769 new built-in functions.
21771 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
21773 * attribs.h (sorted_attr_string): Move machine independent
21774 functions for target clone support from the i386 port to common
21775 code. Rename ix86_function_versions to common_function_versions.
21776 Rename make_name to make_unique_name.
21777 (common_function_versions): Likewise.
21778 (make_unique_name): Likewise.
21779 (make_dispatcher_decl): Likewise.
21780 (is_function_default_version): Likewise.
21781 * attribs.c (attr_strcmp): Likewise.
21782 (sorted_attr_string): Likewise.
21783 (common_function_versions): Likewise.
21784 (make_unique_name): Likewise.
21785 (make_dispatcher_decl): Likewise.
21786 (is_function_default_version): Likewise.
21787 * config/i386/i386.c (attr_strcmp): Likewise.
21788 (sorted_attr_string): Likewise.
21789 (ix86_function_versions): Likewise.
21790 (make_name): Likewise.
21791 (make_dispatcher_decl): Likewise.
21792 (is_function_default_version): Likewise.
21793 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
21795 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21798 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
21799 Account for direct move costs for vec_construct of integer
21802 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
21805 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
21806 (UNSPEC_STX_ATOMIC): Ditto.
21807 (loaddi_via_sse): New insn.
21808 (storedi_via_sse): Ditto.
21809 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
21810 Update corresponding peephole2 patterns.
21811 (atomic_storedi_fpu): Ditto.
21813 2017-05-11 Julia Koval <julia.koval@intel.com>
21815 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
21816 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
21818 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
21819 (__builtin_ia32_rsqrt14ss_mask): New builtins.
21820 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
21822 2017-05-11 Nathan Sidwell <nathan@acm.org>
21824 * graphite-poly.c: Include dumpfile.h.
21826 * dumpfle.h (dump_function): Declare here ...
21827 * tree-dump.h (dump_function): ... not here.
21828 * dumpfile.c: #include tree-cfg.h.
21829 (dump_function): Move here from ...
21830 * tree-dump.c (dump_function): ... here.
21831 * gimplify.c: #include splay-tree.h, not tree-dump.h.
21832 * graphite-poly.c: Don't include tree-dump.h.
21833 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
21834 * print-tree.c: Likewise.
21835 * stor-layout.c: Likewise.
21836 * tree-nested.c: Likewise.
21838 * dumpfile.c (dump_start): Use TDF_FLAGS.
21839 (dump_enable_all): Fix TDF_KIND check thinko.
21841 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
21843 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21844 array entries to represent two legal parameterizations of the
21845 overloaded __builtin_cmpb function, as represented by the
21846 P6_OV_BUILTIN_CMPB constant.
21847 (altivec_resolve_overloaded_builtin): Add special case handling
21848 for the __builtin_cmpb function, as represented by the
21849 P6_OV_BUILTIN_CMPB constant.
21850 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
21851 (BU_P6_64BIT_2): New macro.
21852 (BU_P6_OVERLOAD_2): New macro
21853 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
21854 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
21855 (CMPB): Add overload support to represent both 32-bit and 64-bit
21856 compare-bytes function.
21857 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
21858 support for TARGET_CMPB.
21859 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
21860 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21861 documentation of the __builtin_cmpb overloaded built-in function.
21863 2017-05-11 Richard Biener <rguenther@suse.de>
21865 PR tree-optimization/80705
21866 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
21867 bases are not vectorizable.
21869 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21871 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
21872 when counting register pressure.
21874 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21876 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
21877 (iv_ca_more_deps): Renamed to ...
21878 (iv_ca_compare_deps): ... this.
21879 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
21881 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21883 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
21885 (determine_group_iv_costs): ... here.
21886 (find_inv_vars_cb): Record inv var if it's not recorded before.
21888 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21890 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
21891 (get_shiftadd_cost): Ditto.
21893 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21895 * tree-ssa-address.c: Include header file.
21896 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
21898 (add_to_parts): Refactor.
21899 (addr_to_parts): New parameter. Update use of move_hint_to_base.
21900 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
21903 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21905 PR tree-optimization/53090
21906 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
21908 (extract_cond_operands): Detect condition with IV on both sides
21909 and return COMP_IV_EXPR_2.
21910 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
21911 (rewrite_use_compare): Simplify by removing call to function
21912 extract_cond_operands.
21914 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21916 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
21917 (extract_cond_operands): Detect condition comparing against non-
21918 invariant bound and return appropriate enum value.
21919 (find_interesting_uses_cond): Update use of extract_cond_operands.
21920 Handle its return value accordingly.
21921 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
21923 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21925 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
21926 nonlinear iv_use computation in loop invariant sensitive way.
21928 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21930 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
21931 (find_iv_candidates): Call relate_compare_use_with_all_cands.
21933 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21935 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
21936 (dump_cand): Support iv_cand.inv_exprs.
21937 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
21939 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
21942 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21944 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
21946 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
21947 as local function. Include necessary header files.
21948 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
21950 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21952 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
21954 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21956 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
21957 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
21958 RSHIFT_EXPR and BIT_NOT_EXPR.
21960 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21962 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
21963 (adjust_setup_cost): New parameter supporting round up adjustment.
21964 (struct address_cost_data): Delete.
21965 (force_expr_to_var_cost): Don't bound cost with spill_cost.
21966 (split_address_cost, ptr_difference_cost): Delete.
21967 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
21968 (struct ainc_cost_data): New struct.
21969 (get_address_cost_ainc): New function.
21970 (get_address_cost, get_computation_cost): Reimplement.
21971 (determine_group_iv_cost_address): Record inv_expr for all uses of
21973 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
21974 (iv_ca_has_deps): Reimplemented to ...
21975 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
21977 (iv_ca_extend): Call iv_ca_more_deps.
21979 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21981 * tree-ssa-address.c (struct mem_address): Move to header file.
21982 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
21983 * tree-ssa-address.h (struct mem_address): Move from C file.
21984 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
21986 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21988 * tree-affine.h (aff_combination_type): New interface.
21989 (aff_combination_zero_p): Remove static.
21990 (aff_combination_const_p): New interface.
21991 (aff_combination_singleton_var_p): New interfaces.
21993 2017-05-11 Richard Biener <rguenther@suse.de>
21995 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
21996 Skip unreachable blocks and destinations.
21997 (eliminate): Move stmt removal and fixup ...
21998 (fini_eliminate): ... here. Skip inserted exprs.
21999 (pass_pre::execute): Move fini_pre after fini_eliminate.
22000 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
22001 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
22002 PRE to get rid of dead code that has invalid SSA form and
22003 split critical edges again.
22005 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22007 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22009 2017-05-11 Richard Biener <rguenther@suse.de>
22011 * passes.c (execute_function_todo): Verify loops if they are
22012 said to be up-to-date.
22013 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
22014 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
22016 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
22019 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
22020 handle calling assemble_external ourself.
22023 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
22024 modes with zero size. Enhance comment.
22026 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22028 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
22029 built-ins for vec_xl and vec_xst with short and char pointer
22032 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
22034 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
22035 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
22036 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
22037 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
22038 (_mm_maskz_min_round_ss): New intrinsics.
22039 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
22040 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
22041 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
22042 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
22043 (__builtin_ia32_minss_mask_round): New builtins.
22044 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22045 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22046 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
22048 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
22049 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
22051 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
22054 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
22056 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
22057 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
22058 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
22059 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
22060 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
22061 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
22062 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
22063 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22064 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
22065 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
22066 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
22067 (__builtin_ia32_mulss_mask_round): New builtins.
22068 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22069 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22070 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
22072 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
22073 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
22075 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
22078 2017-05-10 Julia Koval <julia.koval@intel.com>
22080 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
22081 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
22082 (_mm256_setr_m128i): New intrinsics.
22084 2017-05-10 Julia Koval <julia.koval@intel.com>
22086 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
22087 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
22088 (_mm_maskz_rcp14_ss): New intrinsics.
22089 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
22090 (__builtin_ia32_rcp14ss_mask): New builtins.
22091 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
22093 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
22095 PR tree-optimization/51513
22096 * tree-cfg.c (gimple_seq_unreachable_p): New function.
22097 (assert_unreachable_fallthru_edge_p): Use it.
22098 (group_case_labels_stmt): Likewise.
22099 * tree-cfg.h: Prototype it.
22100 * stmt.c: Include cfghooks.h and tree-cfg.h.
22101 (emit_case_dispatch_table) <gap_label>: New local variable.
22102 Use it to fill dispatch table gaps.
22103 Test for default_label before updating probabilities.
22104 (expand_case) <default_label>: Remove unneeded initialization.
22105 Test for unreachable default case statement and remove its edge.
22106 Set default_label accordingly.
22107 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
22109 2017-05-10 Carl Love <cel@us.ibm.com>
22111 * config/rs6000/rs6000-c: Add support for built-in functions
22112 vector signed char vec_neg (vector signed char)
22113 vector signed short int vec_neg (vector short int)
22114 vector signed int vec_neg (vector signed int)
22115 vector signed long long vec_neg (vector signed long long)
22116 vector float vec_neg (vector float)
22117 vector double vec_neg (vector double)
22118 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
22120 * config/rs6000/altivec.h: Add define for vec_neg
22121 * doc/extend.texi: Update the built-in documentation for the
22122 new built-in functions.
22124 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22126 PR tree-optimization/77644
22127 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
22129 2017-05-10 Nathan Sidwell <nathan@acm.org>
22131 * dumpfile.h (TDI_lang_all): New.
22132 (TDF_KIND): New. Renumber others
22133 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
22135 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
22137 (get_dump_file_name): Adjust suffix generation.
22138 (dump_enable_all): Use TDF_KIND.
22139 * doc/invoke.texi (-fdump-lang-all): Document.
22141 * dumpfile.h: Tabify.
22143 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
22146 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
22147 Move member access before delete.
22149 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
22151 * tree-inline.c (expand_call_inline): Split block at stmt
22154 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
22157 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
22158 are now unused after splitting mov{sf,sd}_hardfloat.
22159 (f32_lr2): Likewise.
22160 (f32_lm): Likewise.
22161 (f32_lm2): Likewise.
22162 (f32_li): Likewise.
22163 (f32_li2): Likewise.
22164 (f32_lv): Likewise.
22165 (f32_sr): Likewise.
22166 (f32_sr2): Likewise.
22167 (f32_sm): Likewise.
22168 (f32_sm2): Likewise.
22169 (f32_si): Likewise.
22170 (f32_si2): Likewise.
22171 (f32_sv): Likewise.
22172 (f32_dm): Likewise.
22173 (f32_vsx): Likewise.
22174 (f32_av): Likewise.
22175 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
22176 For movsf, order stores so the VSX stores occur before the GPR
22177 store which encourages the register allocator to use a traditional
22178 FPR instead of a GPR. For movsd, order the stores so that the GPR
22179 store comes before the VSX stores to allow the power6 to work.
22180 This is due to the power6 not having a 32-bit integer store
22181 instruction from a FPR.
22182 (movsf_hardfloat): Likewise.
22183 (movsd_hardfloat): Likewise.
22185 2017-05-09 Martin Sebor <msebor@redhat.com>
22187 PR translation/80280
22188 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
22191 PR translation/80280
22192 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
22193 data member added in r247778.
22194 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
22196 2017-05-09 Nathan Sidwell <nathan@acm.org>
22198 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
22200 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
22203 2017-05-09 Marek Polacek <polacek@redhat.com>
22205 * doc/invoke.texi: Fix typo.
22207 2017-05-09 Richard Biener <rguenther@suse.de>
22209 * tree-vrp.c (vrp_val_is_max): Adjust comment.
22210 (vrp_val_is_min): Likewise.
22211 (set_value_range_to_value): Likewise.
22212 (set_value_range_to_nonnegative): Likewise.
22213 (gimple_assign_nonzero_p): Likewise.
22214 (gimple_stmt_nonzero_p): Likewise.
22215 (vrp_int_const_binop): Likewise. Remove unreachable case.
22216 (adjust_range_with_scev): Adjust comments.
22217 (compare_range_with_value): Likewise.
22218 (extract_range_from_phi_node): Likewise.
22219 (test_for_singularity): Likewise.
22221 2017-05-09 Richard Biener <rguenther@suse.de>
22223 * tree-vrp.c (get_single_symbol): Add assert that we don't
22224 get overflowed constants as invariant part.
22225 (compare_values_warnv): Add comment before the TREE_NO_WARNING
22226 checks. Use wi::cmp instead of recursing for integer constants.
22227 (compare_values): Just ignore whether we assumed undefined
22228 overflow instead of failing the compare.
22229 (extract_range_for_var_from_comparison_expr): Add comment before the
22230 TREE_NO_WARNING sets.
22231 (test_for_singularity): Likewise.
22232 (extract_range_from_comparison): Do not disable optimization
22233 when we assumed undefined overflow.
22234 (extract_range_basic): Remove init of unused var.
22236 2017-05-09 Richard Biener <rguenther@suse.de>
22238 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
22239 (extract_range_from_multiplicative_op_1): Adjust.
22240 (extract_range_from_binary_expr_1): Use int_const_binop.
22242 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
22245 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
22246 rs6000_store_data_bypass_p in seven define_bypass directives and
22247 in several comments.
22248 * config/rs6000/rs6000-protos.h: Add prototype for
22249 rs6000_store_data_bypass_p function.
22250 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
22251 function implements slightly different (rs6000-specific) semantics
22252 than store_data_bypass_p, returning false rather than aborting
22253 with assertion error when arguments do not satisfy the
22254 requirements of store data bypass.
22255 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
22256 rs6000_store_data_bypass_p.
22258 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
22260 * config/xtensa/xtensa-protos.h
22261 (xtensa_initial_elimination_offset): New declaration.
22262 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22263 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
22264 macro definition, add case for FRAME_POINTER_REGNUM when
22265 FRAME_GROWS_DOWNWARD.
22266 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
22267 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
22268 xtensa_initial_elimination_offset.
22270 2017-05-08 Nathan Sidwell <nathan@acm.org>
22272 * doc/invoke.texi: Alphabetize -fdump options.
22274 2017-05-08 Martin Sebor <msebor@redhat.com>
22276 PR translation/80280
22277 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
22279 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
22281 * target.def (compute_frame_layout): New optional target hook.
22282 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
22283 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
22284 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
22286 * reload1.c (verify_initial_elim_offsets): Likewise.
22287 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
22288 (use_simple_return_p): Call arm_compute_frame_layout if needed.
22289 (arm_get_frame_offsets): Split up into this ...
22290 (arm_compute_frame_layout): ... and this function.
22292 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
22294 * config/aarch64/constraints.md (Usa): New constraint.
22295 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
22297 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
22299 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
22300 with_multilib_list after it has been checked.
22302 2017-05-08 Richard Biener <rguenther@suse.de>
22304 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
22305 (bitmap_set_subtract_values): Likewise.
22307 2017-05-08 Richard Biener <rguenther@suse.de>
22309 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
22310 (gimple_assign_nonzero): ... this and remove strict_overflow_p
22312 (gimple_stmt_nonzero_warnv_p): Rename to ...
22313 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
22315 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
22316 (extract_range_basic): Adjust, do not disable propagation on
22317 strict overflow sensitive simplification.
22318 (vrp_visit_cond_stmt): Likewise.
22320 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
22322 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
22323 body size unconditionally.
22325 2017-05-07 Jeff Law <law@redhat.com>
22328 2017-05-06 Jeff Law <law@redhat.com>
22329 PR tree-optimization/78496
22330 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
22333 PR tree-optimization/78496
22334 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
22335 (simplify_stmt_using_ranges): Call it.
22336 (vrp_dom_walker::before_dom_children): Extract equivalences
22337 from an ASSERT_EXPR with an equality comparison against a
22340 2017-05-06 Jeff Law <law@redhat.com>
22342 PR tree-optimization/78496
22343 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
22346 PR tree-optimization/78496
22347 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
22348 (simplify_stmt_using_ranges): Call it.
22349 (vrp_dom_walker::before_dom_children): Extract equivalences
22350 from an ASSERT_EXPR with an equality comparison against a
22353 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
22355 * lra-constraints.c (lra_copy_reg_equiv): New function.
22356 (split_reg): Use it to copy equivalence information from the
22357 original register to the spill register.
22359 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
22361 PR rtl-optimization/75964
22362 * simplify-rtx.c (simplify_const_relational_operation): Remove
22363 invalid handling of comparisons of integer ABS.
22365 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
22367 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
22368 initialize to zero.
22369 (init_regs): Remove declaration.
22370 (function_arg_advance_32): Initialize error_p as boolean variable.
22372 2017-05-05 Nathan Sidwell <nathan@acm.org>
22374 * store-motion.c (remove_reachable_equiv_notes): Reformat long
22375 lines. Use for (;;).
22377 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22379 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
22380 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
22381 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
22382 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
22383 VF=2 that require versioning.
22385 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22387 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
22390 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22392 * diagnostic.h (diagnostic_override_option_index): Convert from
22393 macro to inline function.
22395 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22397 * diagnostic.c (last_module_changed_p): New function.
22398 (set_last_module): New function.
22399 (diagnostic_report_current_module): Convert macro usage to
22400 the above functions.
22401 * diagnostic.h (diagnostic_context::last_module): Strengthen
22402 from const line_map * to const line_map_ordinary *.
22403 (diagnostic_last_module_changed): Delete macro.
22404 (diagnostic_set_last_module): Delete macro.
22406 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22408 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
22409 with diagnostic_report_diagnostic.
22410 (diagnostic_n_impl_richloc): Likewise.
22411 * diagnostic.h (report_diagnostic): Delete macro.
22412 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
22413 with diagnostic_report_diagnostic.
22414 * substring-locations.c (format_warning_va): Likewise.
22416 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22418 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
22419 save/restor of format_spec. Move option-printing code to...
22420 (print_option_information): ...this new function, and
22421 reimplement by simply printing to the pretty_printer,
22422 rather than appending to the format string.
22424 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22426 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
22427 handling logic into...
22428 (update_effective_level_from_pragmas): ...this new function.
22430 2017-05-04 Andrew Waterman <andrew@sifive.com>
22432 * config/riscv/riscv.opt (mstrict-align): New option.
22433 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
22434 (SLOW_UNALIGNED_ACCESS): Define.
22435 (riscv_slow_unaligned_access): Declare.
22436 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
22438 (riscv_slow_unaligned_access): New variable.
22439 (rocket_tune_info): Set slow_unaligned_access to true.
22440 (optimize_size_tune_info): Set slow_unaligned_access to false.
22441 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
22442 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
22443 (riscv_option_override): Set riscv_slow_unaligned_access.
22444 * doc/invoke.texi: Add -mstrict-align to RISC-V.
22446 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
22448 * config/riscv/riscv.md: Unify indentation.
22450 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22455 * config/rs6000/rs6000.md (u code attribute): Add FIX and
22457 (extendsi<mode>2): Add support for doing sign extension via
22458 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
22459 don't have ISA 3.0 instructions.
22460 (extendsi<mode>2 splitter): Likewise.
22461 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
22462 generate the normal insns since SImode can now go in vector
22463 registers. Disallow the special UNSPECs needed for previous
22464 machines to hide SImode being used. Add new insns
22465 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
22466 (fix_trunc<mode>si2_stfiwx): Likewise.
22467 (fix_trunc<mode>si2_internal): Likewise.
22468 (fixuns_trunc<mode>si2): Likewise.
22469 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22470 (fctiw<u>z_<mode>_smallint): Likewise.
22471 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
22472 of floating point to 32-bit integer from doing a direct move to
22473 the GPR registers to do a store.
22474 (fctiwz_<mode>): Break long line.
22476 2017-05-05 Bin Cheng <bin.cheng@arm.com>
22478 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
22479 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
22480 (addr_list, addr_offset_valid_p): New.
22481 (split_address_groups): Check offset validity with above function.
22482 (gt-tree-ssa-loop-ivopts.h): Include header file.
22484 2017-05-05 Nathan Sidwell <nathan@acm.org>
22486 * config.gcc (arm*-*-*): Add missing 'fi'.
22488 2017-05-05 Steve Ellcey <sellcey@cavium.com>
22490 * doc/invoke.texi (-fopt-info): Explicitly say order of options
22491 included in -fopt-info does not matter.
22492 * doc/optinfo.texi (-fopt-info): Fix description of default
22493 behavour. Explicitly say order of options included in -fopt-info
22496 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22498 * config.gcc: Allow combinations of aprofile and rmprofile values for
22499 --with-multilib-list.
22500 * config/arm/t-multilib: New file.
22501 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
22502 variables. Remove setting of ISA and floating-point ABI in
22503 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
22504 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
22505 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
22506 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
22508 * config/arm/t-rmprofile: Likewise except for the matches changes.
22509 * doc/install.texi (--with-multilib-list): Document the combination of
22510 aprofile and rmprofile values and warn about pitfalls in doing that.
22512 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
22514 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
22515 (movdi_aarch64): Likewise.
22517 2017-05-05 Jakub Jelinek <jakub@redhat.com>
22519 PR tree-optimization/80632
22520 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
22522 (build_arrays): Initialize it for virtual phis.
22523 (fix_phi_nodes): Use it for virtual phis.
22525 PR tree-optimization/80558
22526 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
22527 [x, y] op z into [x op, y op z] for op & or | if conditions
22530 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
22531 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22534 * config/arm/arm.md (use_literal_pool): Remove.
22535 (64-bit immediate split): No longer takes cost into consideration
22536 if arm_disable_literal_pool is enabled.
22537 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
22538 used when arm_disable_literal_pool is enabled.
22539 (arm_max_const_double_inline_cost): Remove use of
22540 arm_disable_literal_pool.
22541 (push_minipool_fix): Add assert.
22542 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
22543 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
22544 (no_literal_pool_sf_immediate): New.
22546 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22548 PR tree-optimization/80613
22549 * tree-ssa-dce.c (propagate_necessity): Remove cases for
22550 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
22552 2017-05-05 Richard Biener <rguenther@suse.de>
22554 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
22556 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
22558 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
22559 of this flag from insn conditions due to removal from r247495.
22561 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
22563 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
22565 (arm_early_store_addr_dep_ptr): Likewise.
22566 * config/arm/aarch-common-protos.h
22567 (arm_early_load_addr_dep_ptr): Add prototype.
22568 (arm_early_store_addr_dep_ptr): Likewise.
22569 * config/arm/cortex-a53.md: Add new bypasses.
22571 2017-05-05 Jakub Jelinek <jakub@redhat.com>
22573 * tree.c (next_type_uid): Change type to unsigned.
22574 (type_hash_canon): Decrement back next_type_uid if
22575 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
22576 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
22579 2017-05-04 Martin Sebor <msebor@redhat.com>
22581 * builtins.c: Fix a trivial typo in a comment.
22583 PR middle-end/79234
22584 * builtins.c (check_sizes): Adjust to handle reading past the end.
22585 Avoid printing excessive upper bound of ranges. Use %E to print
22586 tree nodes instead of converting them to %wu.
22587 (expand_builtin_memchr): New function.
22588 (compute_dest_size): Rename...
22589 (compute_objsize): ...to this.
22590 (expand_builtin_memcpy): Adjust.
22591 (expand_builtin_mempcpy): Adjust.
22592 (expand_builtin_strcat): Adjust.
22593 (expand_builtin_strcpy): Adjust.
22594 (check_strncat_sizes): Adjust.
22595 (expand_builtin_strncat): Adjust.
22596 (expand_builtin_strncpy): Adjust and simplify.
22597 (expand_builtin_memset): Adjust.
22598 (expand_builtin_bzero): Adjust.
22599 (expand_builtin_memcmp): Adjust.
22600 (expand_builtin): Handle memcmp.
22601 (maybe_emit_chk_warning): Check strncat just once.
22603 2017-05-04 Martin Sebor <msebor@redhat.com>
22605 PR preprocessor/79214
22606 PR middle-end/79222
22607 PR middle-end/79223
22608 * builtins.c (check_sizes): Add inlining context and issue
22609 warnings even when -Wno-system-headers is set.
22610 (check_strncat_sizes): Same.
22611 (expand_builtin_strncat): Same.
22612 (expand_builtin_memmove): New function.
22613 (expand_builtin_stpncpy): Same.
22614 (expand_builtin): Handle memmove and stpncpy.
22616 2017-05-04 Bin Cheng <bin.cheng@arm.com>
22618 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
22619 which is not used any more.
22621 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22623 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
22625 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22627 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
22628 (cortexa53_tunings): Likewise.
22629 (cortexa57_tunings): Likewise.
22630 (cortexa72_tunings): Likewise.
22631 (cortexa73_tunings): Likewise.
22633 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22635 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
22636 Set loop alignment to 8.
22638 2017-05-04 Martin Sebor <msebor@redhat.com>
22640 PR translation/80280
22641 * builtins.c (expand_builtin_object_size): Add missing quoting to
22642 %D and like directives.
22643 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
22644 (hsa_type_for_tree_type): Same.
22645 (verify_function_arguments): Same.
22646 * symtab.c (symbol_table::change_decl_assembler_name): Same.
22647 * varasm.c (get_section): Same.
22650 2017-05-04 Martin Sebor <msebor@redhat.com>
22652 PR translation/80280
22653 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
22655 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22657 * config/aarch64/aarch64.c (generic_addrcost_table):
22658 Change HI/TI mode setting.
22660 2017-05-04 Martin Jambor <mjambor@suse.cz>
22662 PR tree-optimization/80622
22663 * tree-sra.c (comes_initialized_p): New function.
22664 (build_accesses_from_assign): Only set write lazily when
22665 comes_initialized_p is false.
22666 (analyze_access_subtree): Use comes_initialized_p.
22667 (propagate_subaccesses_across_link): Assert !comes_initialized_p
22668 instead of testing for PARM_DECL.
22670 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22672 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
22673 constraint on operand 0 to allow more general addressing modes.
22674 Adjust output template.
22675 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
22677 * config/aarch64/aarch64-protos.h
22678 (aarch64_address_valid_for_prefetch_p): Declare prototype.
22679 * config/aarch64/constraints.md (Dp): New address constraint.
22680 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
22683 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
22685 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
22686 update use of estimate_ipcp_clone_size_and_time.
22687 (estimate_local_effects): Update use of
22688 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
22689 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
22690 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
22691 Return nonspecialized time.
22693 2017-05-04 Richard Biener <rguenther@suse.de>
22695 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
22696 for the last VUSE which def dominates the PHI. Directly call
22698 (get_continuation_for_phi_1): Remove.
22700 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
22702 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
22703 to explain the use of truncating division. Cap the number of
22704 iterations to the maximum given by nb_iterations_upper_bound,
22707 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22709 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
22710 * configure: Regenerate.
22711 * config.in: Regenerate.
22712 * config/i386/driver-mingw32.c: new file.
22713 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
22714 * config.host: Link driver-mingw32.o on MinGW host.
22715 * doc/install.texi: Document new --enable-mingw-wildcard configure
22718 2017-05-04 Marek Polacek <polacek@redhat.com>
22720 PR tree-optimization/80612
22721 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
22723 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22724 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
22726 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
22727 (movt splitter): Likewise.
22728 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
22729 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
22730 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
22731 block for Thumb-1 with MOVT.
22732 (thumb2_legitimate_address_p): Move code block ...
22733 (can_avoid_literal_pool_for_label_p): ... into this new function.
22734 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
22736 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
22737 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
22738 "M-profile targets with the MOVT instruction".
22740 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22742 * config/arm/arm-builtins.c (arm_init_builtins): Rename
22743 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
22744 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
22746 2017-05-04 Martin Liska <mliska@suse.cz>
22748 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
22749 variable cond_code.
22751 2017-05-04 Richard Biener <rguenther@suse.de>
22753 * tree.c (array_at_struct_end_p): Handle arrays at struct
22754 end with flexarrays more conservatively. Refactor and treat
22755 arrays of arrays or aggregates more strict. Fix
22756 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
22757 * tree.c (array_at_struct_end_p): Adjust prototype.
22758 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
22759 * gimple-fold.c (get_range_strlen): Likewise.
22760 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
22762 2017-05-04 Richard Biener <rguenther@suse.de>
22764 PR tree-optimization/31130
22765 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
22767 (supports_overflow_infinity): Likewise.
22768 (is_negative_overflow_infinity): Likewise.
22769 (is_positive_overflow_infinity): Likewise.
22770 (is_overflow_infinity): Likewise.
22771 (stmt_overflow_infinity): Likewise.
22772 (overflow_infinity_range_p): Likewise.
22773 (usable_range_p): Remove as always returning true.
22774 (make_overflow_infinity): Remove.
22775 (negative_overflow_infinity): Likewise.
22776 (positive_overflow_infinity): Likewise.
22777 (avoid_overflow_infinity): Likewise.
22778 (set_value_range): Adjust accordingly.
22779 (set_value_range_to_nonnegative): Likewise, remove now unused
22780 overflow_infinity arg.
22781 (vrp_operand_equal_p): Adjust.
22782 (update_value_range): Likewise.
22783 (range_int_cst_singleton_p): Likewise.
22784 (operand_less_p): Likewise.
22785 (compare_values_warnv): Likewise.
22786 (extract_range_for_var_from_comparison_expr): Likewise.
22787 (vrp_int_const_binop): Likewise.
22788 (zero_nonzero_bits_from_vr): Likewise.
22789 (extract_range_from_multiplicative_op_1): Likewise.
22790 (extract_range_from_binary_expr_1): Likewise.
22791 (extract_range_from_unary_expr): Likewise.
22792 (extract_range_from_comparison): Likewise.
22793 (extract_range_basic): Likewise.
22794 (adjust_range_with_scev): Likewise.
22795 (compare_ranges): Likewise.
22796 (compare_range_with_value): Likewise.
22797 (dump_value_range): Likewise.
22798 (test_for_singularity): Likewise, remove strict_overflow_p parameter
22800 (simplify_cond_using_ranges): Adjust.
22802 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22804 * brig-builtins.def: Added a builtin for class_f64.
22805 * builtin-types.def: Added a builtin type needed by class_f64.
22807 2017-05-03 Jason Merrill <jason@redhat.com>
22809 * timevar.def: Add TV_CONSTEXPR.
22811 2017-05-03 David Malcolm <dmalcolm@redhat.com>
22813 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
22815 2017-05-03 Martin Jambor <mjambor@suse.cz>
22817 * ipa-prop.c (ipa_update_after_lto_read): Removed.
22818 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
22819 * ipa-cp.c (ipcp_propagate_stage): Do not call
22820 ipa_update_after_lto_read.
22821 * ipa-inline.c (ipa_inline): Likewise.
22823 2017-05-03 Martin Jambor <mjambor@suse.cz>
22825 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
22826 tag. Added a default constructor and a destructor.
22827 (ipa_edge_args_sum_t): New class;
22828 (ipa_edge_args_sum): Declare.
22829 (ipa_edge_args_vector): Remove declaration.
22830 (IPA_EDGE_REF): Use ipa_edge_args_sum.
22831 (ipa_free_edge_args_substructures): Remove declaration.
22832 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
22833 (ipa_edge_args_info_available_for_edge_p): Likewise.
22834 * ipa-prop.c (ipa_edge_args_vector): Removed.
22835 (edge_removal_hook_holder): Likewise.
22836 (edge_duplication_hook_holder): Likewise.
22837 (ipa_edge_args_sum): New variable.
22838 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
22839 ipa_edge_args_vector.
22840 (ipa_free_edge_args_substructures): Likewise.
22841 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
22842 ipa_edge_args_vector.
22843 (ipa_edge_removal_hook): Turned into method
22844 ipa_edge_args_sum_t::remove.
22845 (ipa_edge_duplication_hook): Turned into method
22846 ipa_edge_args_sum_t::duplicate.
22847 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
22848 registering edge hooks.
22849 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
22850 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
22851 ipa_edge_args_sum instead of ipa_edge_args_vector.
22852 * ipa-profile.c (ipa_profile): Likewise.
22854 2017-05-03 Martin Jambor <mjambor@suse.cz>
22856 * symbol-summary.h (function_summary): New method exists.
22857 (function_summary::symtab_removal): Deallocate through release.
22858 (call_summary): New class.
22859 (gt_ggc_mx): New overload.
22860 (gt_pch_nx): Likewise.
22861 (gt_pch_nx): Likewise.
22863 2017-05-03 Jeff Law <law@redhat.com>
22865 PR tree-optimization/78496
22866 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
22867 from simplify_cond_using_ranges. Split off code to walk
22868 backwards through casts into ...
22869 (simplify_cond_using_ranges_2): New function.
22870 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
22871 (execute_vrp): After identifying jump threads, call
22872 simplify_cond_using_ranges_2.
22874 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
22877 * ipa-inline.h (inline_summary): Add ctor.
22878 (create_ggc): Do not use ggc_cleared_alloc.
22880 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
22881 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22883 * gcc.c (handle_braces): Support escaping in switch matching
22885 * doc/invoke.texi (Spec Files): Document it.
22886 Remove superfluous @code markup in items.
22888 2017-05-03 David Malcolm <dmalcolm@redhat.com>
22890 * diagnostic-show-locus.c (struct column_range): New struct.
22891 (get_affected_columns): New function.
22892 (get_printed_columns): New function.
22893 (struct correction): New struct.
22894 (correction::ensure_capacity): New function.
22895 (correction::ensure_terminated): New function.
22896 (struct line_corrections): New struct.
22897 (line_corrections::~line_corrections): New dtor.
22898 (line_corrections::add_hint): New function.
22899 (layout::print_trailing_fixits): Reimplement in terms of the new
22901 (selftest::test_overlapped_fixit_printing): New function.
22902 (selftest::diagnostic_show_locus_c_tests): Call it.
22904 2017-05-03 Nathan Sidwell <nathan@acm.org>
22906 Canonicalize canonical type hashing
22907 * tree.h (type_hash_canon_hash): Declare.
22908 * tree.c (type_hash_list, attribute_hash_list): Move into
22909 type_hash_canon_hash.
22910 (build_type_attribute_qual_variant): Break out hash code calc into
22911 type_hash_canon_hash.
22912 (type_hash_canon_hash): New. Generic type hash computation.
22913 (build_range_type_1, build_array_type_1, build_function_type,
22914 build_method_type_directly, build_offset_type, build_complex_type,
22915 make_vector_type): Call it.
22917 2017-05-03 Richard Biener <rguenther@suse.de>
22919 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
22920 When all DRs have unknown misaligned do not always peel
22921 when there is a store but apply the same costing model as if
22922 there were only loads.
22924 2017-05-03 Richard Biener <rguenther@suse.de>
22927 PR tree-optimization/80492
22928 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
22929 compare_base_decls returning dont-know properly.
22931 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
22933 * config/arm/iterators.md (CCSI): New mode iterator.
22934 (arch): New mode attribute.
22935 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
22936 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
22937 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
22938 code iterator for success result mode.
22939 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
22940 the corresponding new insn generators.
22942 2017-05-03 Bin Cheng <bin.cheng@arm.com>
22945 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22946 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22948 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
22950 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
22951 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
22952 (DDR_A): Wrap DDR argument in brackets.
22953 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
22954 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
22955 (DDR_REVERSED_P): Likewise.
22957 2017-05-03 Jakub Jelinek <jakub@redhat.com>
22959 PR tree-optimization/79472
22960 * tree-switch-conversion.c (struct switch_conv_info): Add
22961 contiguous_range and default_case_nonstandard fields.
22962 (collect_switch_conv_info): Compute contiguous_range and
22963 default_case_nonstandard fields, don't clear final_bb if
22964 contiguous_range and only the default case doesn't have the required
22966 (check_all_empty_except_final): Set default_case_nonstandard instead
22967 of failing if contiguous_range and the default case doesn't have empty
22969 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
22970 and only the default case doesn't have the required constants. Skip
22972 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
22973 if default_case_nonstandard.
22974 (build_constructors): Build constant 1 just once. Assert that default
22975 values aren't inserted in between cases if contiguous_range. Skip
22977 (build_arrays): Skip virtual phis.
22978 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
22979 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
22980 Handle virtual phis.
22981 (gen_inbound_check): Handle default_case_nonstandard case.
22982 (process_switch): Adjust check_final_bb caller. Call
22983 gather_default_values with the first non-default case instead of
22984 default case if default_case_nonstandard.
22986 2017-05-02 Nathan Sidwell <nathan@acm.org>
22988 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
22989 check. Fix formatting.
22991 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
22993 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
22994 errors when comparing specialized and unspecialized times.
22996 2017-05-02 David Malcolm <dmalcolm@redhat.com>
22998 * diagnostic-show-locus.c
22999 (layout::should_print_annotation_line_p): Make private.
23000 (layout::print_annotation_line): Make private.
23001 (layout::annotation_line_showed_range_p): Make private.
23002 (layout::show_ruler): Make private.
23003 (layout::print_source_line): Make private. Pass in line and
23004 line_width, rather than calling location_get_source_line. Drop
23006 (layout::print_leading_fixits): New method.
23007 (layout::print_any_fixits): Rename to...
23008 (layout::print_trailing_fixits): ...this, and make private.
23009 Don't print newline fixits.
23010 (diagnostic_show_locus): Move logic for printing one row into...
23011 (layout::print_line): ...this new function. Move the
23012 location_get_source_line call and error-handling from
23013 print_source_line to here. Call print_leading_fixits, and rename
23014 print_any_fixits to print_trailing_fixits.
23015 (selftest::test_fixit_insert_containing_newline): Update now that
23016 newlines are partially supported.
23017 (selftest::test_fixit_insert_containing_newline_2): New test.
23018 (selftest::test_fixit_replace_containing_newline): Update comments.
23019 (selftest::diagnostic_show_locus_c_tests): Call the new test.
23020 * edit-context.c (class added_line): New class.
23021 (class edited_line): Describe newline handling in comment.
23022 (edited_line::actually_edited_p): New method.
23023 (edited_line::print_content): Delete redundant decl.
23024 (edited_line::m_predecessors): New field.
23025 (edited_file::print_content): Call edited_line::print_content.
23026 (edited_file::print_diff): Update to support newlines.
23027 (edited_file::print_diff_hunk): Likewise.
23028 (edited_file::print_run_of_changed_lines): New function.
23029 (edited_file::print_diff_line): Convert to...
23030 (print_diff_line): ...this.
23031 (edited_file::get_effective_line_count): New function.
23032 (edited_line::edited_line): Initialize new field m_predecessors.
23033 (edited_line::~edited_line): Clean up m_predecessors.
23034 (edited_line::apply_fixit): Handle newlines.
23035 (edited_line::get_effective_line_count): New function.
23036 (edited_line::print_content): New function.
23037 (edited_line::print_diff_lines): New function.
23038 (selftest::test_applying_fixits_insert_containing_newline): New
23040 (selftest::test_applying_fixits_replace_containing_newline): New
23042 (selftest::insert_line): New function.
23043 (selftest::test_applying_fixits_multiple_lines): Add example of
23045 (selftest::edit_context_c_tests): Call the new tests.
23047 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23049 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
23050 parameter cand. Update dump information.
23051 (get_computation_cost): Update uses.
23053 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23055 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
23056 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
23057 (get_computation_at, rewrite_use_address): Update use of
23058 get_computation_aff.
23060 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23062 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
23063 (get_computation): Delete.
23064 (get_computation_cost): Implement like get_computation_cost_at.
23065 Use get_computation_at.
23066 (get_computation_cost_at): Delete.
23067 (rewrite_use_nonlinear_expr): Use get_computation_at.
23068 (rewrite_use_compare, remove_unused_ivs): Ditto.
23070 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23072 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
23074 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23076 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
23077 (ivopts_global_cost_for_size): Rename parameter and update uses.
23078 (iv_ca_recount_cost): Update uses.
23079 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
23080 candidates seperately in n_invs and n_cands.
23081 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
23083 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23085 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
23086 (find_inv_vars_cb): New.
23087 (find_depends): Renamed to ...
23088 (find_inv_vars): ... this.
23089 (add_candidate_1, force_var_cost): Call find_inv_vars.
23090 (split_address_cost, determine_group_iv_cost_cond): Ditto.
23092 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23094 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
23095 inv_vars. Add inv_exprs.
23096 (struct iv_cand): Rename depends_on to inv_vars.
23097 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
23098 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
23099 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
23100 (dump_cand): Dump inv_vars.
23101 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
23102 (record_invariant, find_depends, add_candidate_1): Ditto.
23103 (set_group_iv_cost, force_var_cost): Ditto.
23104 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
23105 (get_computation_cost_at, get_computation_cost): Ditto.
23106 (determine_group_iv_cost_generic): Ditto.
23107 (determine_group_iv_cost_address): Ditto.
23108 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
23109 (determine_group_iv_costs): Ditto.
23110 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
23111 (iv_ca_set_remove_invariants): Renamed to ...
23112 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
23113 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
23114 (iv_ca_set_add_invariants): Renamed to ...
23115 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
23116 (iv_ca_set_cp): Use iv_ca_set_add_invs.
23117 (iv_ca_has_deps): Support inv_vars and inv_exprs.
23118 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
23119 (create_new_ivs): Remove useless dump.
23121 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23123 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
23125 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
23126 (iv_ca_set_no_cp, create_new_iv): Ditto.
23128 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23130 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23132 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
23134 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
23135 function tree_check2.
23137 2017-05-02 Martin Liska <mliska@suse.cz>
23139 * doc/gcov.texi: Add missing preposition.
23140 * gcov.c (function_info::function_info): Properly fill up
23141 all member variables.
23143 2017-05-02 Tamar Christina <tamar.christina@arm.com>
23145 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
23147 2017-05-02 Tamar Christina <tamar.christina@arm.com>
23149 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
23151 2017-05-02 Martin Liska <mliska@suse.cz>
23154 * lto-streamer-in.c (lto_read_tree_1): Remove
23155 LTO_STREAMER_DEBUG.
23156 * lto-streamer.c (struct tree_hash_entry): Likewise.
23157 (struct tree_entry_hasher): Likewise.
23158 (tree_entry_hasher::hash): Likewise.
23159 (tree_entry_hasher::equal): Likewise.
23160 (lto_streamer_init): Likewise.
23161 (lto_orig_address_map): Likewise.
23162 (lto_orig_address_get): Likewise.
23163 (lto_orig_address_remove): Likewise.
23164 * lto-streamer.h: Likewise.
23165 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
23166 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
23168 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
23170 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
23171 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
23172 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
23173 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
23174 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
23175 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
23176 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
23177 (mm_maskz_sub_ss): New intrinsics.
23178 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23179 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23180 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
23181 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
23182 (__builtin_ia32_subss_mask_round): New builtins.
23183 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23184 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23185 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
23187 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
23188 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23190 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23193 2017-05-02 Martin Jambor <mjambor@suse.cz>
23195 PR tree-optimization/78687
23196 * tree-sra.c (access): New field parent.
23197 (process_subtree_disqualification): New function.
23198 (disqualify_candidate): Call it.
23199 (build_accesses_from_assign): Reset write flag if creating an
23201 (build_access_subtree): Fill in parent field and also prpagate
23202 down grp_write flag.
23203 (create_artificial_child_access): New parameter set_grp_write, set
23204 grp_write to its value.
23205 (propagate_subaccesses_across_link): Also propagate grp_write flag
23207 (propagate_all_subaccesses): Push the closest parent back to work
23208 queue if add_access_to_work_queue returned true.
23210 2017-05-02 Richard Biener <rguenther@suse.de>
23212 * common.opt (fstrict-overflow): Alias negative to fwrapv.
23213 * doc/invoke.texi (fstrict-overflow): Remove all traces of
23214 -fstrict-overflow documentation.
23215 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
23216 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
23217 flag_strict_overflow.
23218 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
23219 * lto-opts.c (lto_write_options): Do not stream it.
23220 * lto-wrapper.c (merge_and_complain): Do not handle it.
23221 * opts.c (default_options_table): Do not set -fstrict-overflow.
23222 (finish_options): Likewise do not clear it when sanitizing.
23223 * simplify-rtx.c (simplify_const_relational_operation): Do not
23224 test flag_strict_overflow.
23226 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
23228 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
23229 using enabled attribute.
23230 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
23231 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
23232 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
23233 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
23234 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
23235 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
23236 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
23237 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
23238 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
23239 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
23241 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
23243 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
23245 2017-05-02 Richard Biener <rguenther@suse.de>
23247 PR tree-optimization/80591
23249 2017-04-10 Richard Biener <rguenther@suse.de>
23251 * tree-ssa-structalias.c (find_func_aliases): Properly handle
23254 2017-05-02 Richard Biener <rguenther@suse.de>
23256 PR tree-optimization/80549
23257 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
23258 (cleanup_tree_cfg_noloop): Create forwarders to known loop
23259 headers if they do not have a preheader.
23261 2017-05-02 Martin Liska <mliska@suse.cz>
23264 * common.opt: Fix typo.
23265 * doc/invoke.texi: Likewise.
23267 2017-05-01 Jan Beulich <jbeulich@suse.com>
23269 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
23270 swapping, add (x,x,m,x,n) alternative.
23272 2017-05-01 Nathan Sidwell <nathan@acm.org>
23274 * calls.c (combine_pending_stack_adjustment_and_call): Remove
23275 unnecessary unadjusted_alignment check.
23277 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
23280 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
23282 * gimplify.c (gimplify_cilk_detach): New function.
23283 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
23284 * tree-core.h: Document EXPR_CILK_SPAWN.
23285 * tree.h (EXPR_CILK_SPAWN): Define.
23287 2017-05-01 David Malcolm <dmalcolm@redhat.com>
23289 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
23290 to use new fixit_hint representation, using the "replace" logic.
23291 (get_line_span_for_fixit_hint): Likewise.
23292 (layout::print_any_fixits): Likewise.
23293 (selftest::test_one_liner_many_fixits): Rename to...
23294 (selftest::test_one_liner_many_fixits_1): ...this, and update
23295 comment and expected output to reflect that the multiple fix-it
23296 hints are now consolidated into one insertion.
23297 (selftest::test_one_liner_many_fixits_2): New test.
23298 (selftest::test_diagnostic_show_locus_one_liner): Update for
23300 (selftest::test_fixit_consolidation): Update for fix-it API
23302 * diagnostic.c (print_parseable_fixits): Likewise.
23303 * edit-context.c (edited_line::m_line_events): Convert from
23304 auto_vec <line_event *> to auto_vec <line_event>.
23305 (class line_event): Convert from abstract base class to a concrete
23306 class, taking over the role of replace_event.
23307 (class insert_event): Delete.
23308 (class replace_event): Rename to class line_event. Convert to
23310 (edit_context::add_fixits): Reimplement.
23311 (edit_context::apply_insert): Delete.
23312 (edit_context::apply_replace): Rename to...
23313 (edit_context::apply_fixit): ...this. Convert to half-open range.
23314 (edited_file::apply_insert): Delete.
23315 (edited_file::apply_replace): Rename to...
23316 (edited_file::apply_fixit): ...this.
23317 (edited_line::~edited_line): Drop deletion of events.
23318 (edited_line::apply_insert): Delete.
23319 (edited_line::apply_replace): Rename to...
23320 (edited_line::apply_fixit): ...this. Convert to half-open range.
23321 Update for change to type of m_line_events.
23322 * edit-context.h (edit_context::apply_insert): Delete.
23323 (edit_context::apply_replace): Rename to...
23324 (edit_context::apply_fixit): ...this.
23326 2017-05-01 Martin Sebor <msebor@redhat.com>
23328 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
23331 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
23334 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
23335 __get_cpuid_max returns 0.
23336 (__get_cpuid_count): Ditto.
23338 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
23340 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
23341 replacement expression is another instance of one of its arguments.
23343 2017-05-01 Jakub Jelinek <jakub@redhat.com>
23346 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
23347 check for stack push/pop autoinc.
23348 * config/i386/i386.c (ix86_agi_dependent): Return false
23349 if the only reason why modified_in_p returned true is that
23350 addr is SP based and set_insn is a push or pop.
23352 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
23354 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
23357 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
23360 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
23361 (account_size_time): Use two predicates - exec_pred and
23363 (evaluate_conditions_for_known_args): Compute both clause and
23365 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
23366 (inline_summary_t::duplicate): Update.
23367 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
23369 (compute_inline_parameters): Likewise.
23370 (estimate_edge_size_and_time): Update caluclation of time.
23371 (estimate_node_size_and_time): Compute both time and nonspecialized
23373 (estimate_ipcp_clone_size_and_time): Update.
23374 (inline_merge_summary): Update.
23375 (do_estimate_edge_time): Update.
23376 (do_estimate_edge_size): Update.
23377 (do_estimate_edge_hints): Update.
23378 (inline_read_section, inline_write_summary): Stream both new predicates.
23379 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
23381 (compute_inlined_call_time): Cleanup.
23382 (big_speedup_p): Update.
23383 (edge_badness): Update.
23384 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
23385 (size_time_entry): Replace predicate by exec_predicate and
23386 nonconst_predicate.
23387 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
23388 (estimate_edge_time): Return also nonspec_time.
23389 (reset_edge_growth_cache): Update.
23391 2017-04-29 Jakub Jelinek <jakub@redhat.com>
23393 PR rtl-optimization/80491
23394 * ifcvt.c (noce_process_if_block): When looking for x setter
23395 with missing else_bb, don't check only the insn right before
23396 cond_earliest, but look for the last insn that x is modified in
23397 within the same bb.
23399 PR rtl-optimization/80491
23400 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
23402 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
23404 PR tree-optimization/80487
23405 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
23407 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23409 PR tree-optimization/79697
23410 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
23411 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
23412 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
23414 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
23415 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
23417 2017-04-28 Martin Sebor <msebor@redhat.com>
23419 PR tree-optimization/80523
23420 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
23421 (init_target_to_host_charmap, target_to_host, target_strtol10): New
23423 (maybe_warn, format_directive, parse_directive): Use new functions.
23424 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
23426 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
23428 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
23430 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
23432 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
23433 target_header_dir): Set correctly.
23434 * configure: Regenerated.
23435 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23436 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23437 instead of SYSTEM_HEADER_DIR.
23439 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
23441 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
23442 (estimate_local_effects): Likewise.
23443 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
23444 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
23445 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
23446 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
23447 do_estimate_edge_time, estimate_edge_time): Likewise.
23448 * ipa-inline-analysis.c (estimate_node_size_and_time,
23449 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
23450 (estimate_time_after_inlining): Remove.
23452 2017-04-28 Martin Liska <mliska@suse.cz>
23454 * doc/gcov.texi: Enhance documentation of gcov.
23456 2017-04-28 Martin Liska <mliska@suse.cz>
23458 * doc/gcov.texi: Sort options in alphabetic order.
23459 * doc/gcov-dump.texi: Likewise.
23460 * doc/gcov-tool.texi: Likewise.
23461 * gcov.c (print_usage): Likewise.
23462 * gcov-dump.c (print_usage): Likewise.
23463 * gcov-tool.c (print_merge_usage_message): Likewise.
23464 (print_rewrite_usage_message): Likewise.
23465 (print_overlap_usage_message): Likewise.
23467 2017-04-28 Martin Liska <mliska@suse.cz>
23469 PR gcov-profile/53915
23470 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
23472 2017-04-28 Martin Liska <mliska@suse.cz>
23474 PR gcov-profile/79891
23475 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
23476 is marked by compiler as living on a line.
23477 (get_cycles_count): Remove usage of the union.
23478 (output_intermediate_file): Likewise.
23479 (find_source): Fix GNU coding style.
23480 (accumulate_line_counts): Remove old non-all block mode.
23481 (output_lines): Remove usage of the union.
23482 * profile.c (output_location): Include all BBs, even if
23483 belonging to a same line (and file) as a previous BB.
23485 2017-04-28 Martin Liska <mliska@suse.cz>
23487 * gcov.c (process_args): Handle new argument 'w'.
23488 (read_graph_file): Assign ID to BBs.
23489 (output_branch_count): Display BB # if verbose flag is set.
23490 (output_lines): Likewise for arcs.
23491 (print_usage): Add '--verbose' option help.
23492 * doc/gcov.texi: Document --verbose (-w) option.
23494 2017-04-28 Martin Liska <mliska@suse.cz>
23496 * gcov.c (struct block_location_info): New struct.
23497 (process_file): Fill up the new structure.
23498 (read_graph_file): Replace usage of encoding by the newly added
23500 (add_line_counts): Likewise.
23501 (accumulate_line_counts): Remove usage of the union.
23502 (function_info::function_info): New function.
23503 (function_info::~function_info): Likewise.
23504 (process_file): Call delete instead of release_function.
23505 (release_function): Release the function.
23506 (release_structures): Call delete instead of release_function.
23507 (solve_flow_graph): Replace usage of num_blocks.
23508 (find_exception_blocks): Likewise.
23509 (output_lines): Fix GNU coding style.
23511 2017-04-28 Martin Liska <mliska@suse.cz>
23514 * coverage.c (coverage_remove_note_file): New function.
23515 * coverage.h: Declare the function.
23516 * toplev.c (finalize): Clean if an error has been seen.
23518 2017-04-28 Martin Liska <mliska@suse.cz>
23520 PR gcov-profile/80031
23521 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
23522 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
23523 * gcov.c (read_graph_file): Read just number of blocks.
23524 * profile.c (branch_prob): Do not stream 0 flags per a basic
23527 2017-04-28 Martin Liska <mliska@suse.cz>
23529 * gcov-dump.c (tag_*): Add new argument to declarations.
23530 (dump_gcov_file): Likewise.
23531 (tag_blocks): Add and use new argument depth.
23532 (tag_arcs): Likewise.
23533 (tag_lines): Likewise.
23534 (tag_counters): Likewise.
23535 (tag_summary): Likewise.
23536 (dump_working_sets): Use depth to do a proper indentation.
23538 2017-04-28 Jakub Jelinek <jakub@redhat.com>
23541 * cgraph.h (symtab_node::debug_symtab): No longer inline.
23542 * symtab.c (symtab_node::debug_symtab): Move definition here.
23544 2017-04-28 Richard Biener <rguenther@suse.de>
23546 * lto-streamer.h (LTO_major_version): Bump to 7.
23548 2017-04-28 Richard Biener <rguenther@suse.de>
23550 * tree-vrp.c (assert_info): New struct.
23551 (add_assert_info): New helper.
23552 (register_edge_assert_for_2): Refactor to add asserts to a vector
23554 (register_edge_assert_for_1): Likewise.
23555 (register_edge_assert_for): Likewise.
23556 (finish_register_edge_assert_for): New helper actually registering
23557 asserts where live on edge.
23558 (find_conditional_asserts): Adjust.
23559 (find_switch_asserts): Likewise.
23560 (evrp_dom_walker::try_find_new_range): Generalize.
23561 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
23563 2017-04-27 Marek Polacek <polacek@redhat.com>
23566 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
23567 arg10 and arg11 to itype.
23569 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
23571 * doc/extend.texi (Object Size Checking): Improve grammar.
23573 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
23576 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
23577 that the logic for permitting reciprocal estimates matches that
23580 2017-04-27 Jakub Jelinek <jakub@redhat.com>
23583 * tree.c (type_cache_hasher::equal): Only compare
23584 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
23585 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
23586 non-aggregate element types.
23587 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
23588 about the flag on ARRAY_TYPEs in the comment, formatting fix.
23590 2017-04-27 Richard Biener <rguenther@suse.de>
23592 PR middle-end/80533
23593 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
23594 stripping ARRAY_REFs from MEM_EXPR make sure we're not
23595 keeping a reference to a trailing array.
23597 2017-04-27 Richard Biener <rguenther@suse.de>
23599 PR middle-end/80539
23600 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
23601 being in loop-closed SSA form conservatively.
23602 (chrec_fold_multiply_poly_poly): Likewise.
23604 2017-04-27 Tamar Christina <tamar.christina@arm.com>
23606 PR middle-end/79665
23607 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
23608 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
23610 2017-04-27 Jakub Jelinek <jakub@redhat.com>
23613 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
23614 (aarch64_function_arg_alignment): Return unsigned int again, but still
23615 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
23616 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
23617 Don't emit -Wpsabi note.
23618 (aarch64_function_arg_boundary): Likewise.
23619 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
23622 2017-04-26 Nathan Sidwell <nathan@acm.org>
23624 * tree.h (crc32_unsigned_n): Declare.
23625 (crc32_unsigned, crc32_unsigned): Make inline.
23626 * tree.c (crc32_unsigned_bits): Replace with ...
23627 (crc32_unsigned_n): ... this.
23628 (crc32_unsigned, crc32_byte): Remove.
23629 (crc32_string): Remove unnecessary braces.
23631 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
23633 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
23634 * ipa-inline-analysis.c (MAX_TIME): Remove.
23635 (account_size_time): Use sreal for time.
23636 (dump_inline_summary): Update.
23637 (estimate_function_body_sizes): Update.
23638 (estimate_edge_size_and_time): Update.
23639 (estimate_calls_size_and_time): Update.
23640 (estimate_node_size_and_time): Update.
23641 (inline_merge_summary): Update.
23642 (inline_update_overall_summary): Update.
23643 (estimate_time_after_inlining): Update.
23644 (inline_read_section): Update.
23645 (inline_write_summary): Update.
23646 * ipa-inline.c (compute_uninlined_call_time): Update.
23647 (compute_inlined_call_time): Update.
23648 (recursive_inlining): Update.
23649 (inline_small_functions): Update.
23650 (dump_overall_stats): Update.
23651 * ipa-inline.h: Include sreal.h.
23652 (size_time_entry): Turn time to sreal.
23653 (inline_summary): Turn self_time nad time to sreal.
23655 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
23657 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
23659 (sreal::stream_out, sreal::stream_in): New.
23660 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
23662 2017-04-25 Jakub Jelinek <jakub@redhat.com>
23664 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
23667 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
23670 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
23672 (dimode_scalar_chain::compute_convert_gain): Ditto.
23673 (dimode_scalar_chain::make_vector_copies): Ditto.
23674 (dimode_scalar_chain::convert_reg): Ditto.
23675 (dimode_scalar_chain::convert_insn): Ditto.
23676 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
23677 (VI248_AVX512BW_1): New mode iterator.
23678 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
23679 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
23682 2017-04-25 Martin Sebor <msebor@redhat.com>
23684 PR tree-optimization/80497
23685 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
23686 constants are representable in HOST_WIDE_INT.
23687 (parse_directive): Ditto.
23689 2017-04-25 Martin Sebor <msebor@redhat.com>
23692 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
23693 (new_zero_array): Adjust signature.
23694 (dom_info::dom_init): Used unsigned rather that size_t.
23695 (dom_info::dom_info): Same.
23697 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23698 Jakub Jelinek <jakub@redhat.com>
23701 * config/arm/arm.c: Include gimple.h.
23702 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
23703 returns negative, increment ncrn only if it returned positive.
23704 (arm_needs_doubleword_align): Return int instead of bool,
23705 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
23706 members, but if there is any such non-FIELD_DECL
23707 > PARM_BOUNDARY aligned decl, return -1 instead of false.
23708 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
23709 returns negative, increment nregs only if it returned positive.
23710 (arm_setup_incoming_varargs): Likewise.
23711 (arm_function_arg_boundary): Emit -Wpsabi note if
23712 arm_needs_doubleword_align returns negative, return
23713 DOUBLEWORD_ALIGNMENT only if it returned positive.
23715 2017-04-25 Marek Polacek <polacek@redhat.com>
23718 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
23719 first argument to type.
23721 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
23724 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
23725 type checks to test for compatibility instead of equality.
23727 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23728 Jakub Jelinek <jakub@redhat.com>
23731 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
23733 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
23734 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
23735 the alignment computation, but return their maximum in warn_alignment.
23736 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
23737 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
23739 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
23740 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
23743 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23745 * config/arc/simdext.md (dmpyh): Fix typo.
23747 2017-04-25 Richard Biener <rguenther@suse.de>
23749 PR tree-optimization/80492
23750 * alias.c (compare_base_decls): Handle registers with asm
23751 specification conservatively.
23752 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
23753 compare_base_decls returning dont-know properly.
23755 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23757 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
23758 (legitimate_offset_address_p): New function.
23759 (arc_legitimate_address_p): Use above function.
23761 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23763 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
23765 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23767 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
23768 ACCH registers whenever they are available.
23770 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23772 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
23773 double regs fix when not used.
23775 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23777 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
23779 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
23780 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
23782 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23784 * config/arc/arc.c (arc_output_addsi): Check for h-register class
23785 when emitting short ADD instructions.
23787 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23789 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
23791 (cmpsi_cc_c_insn): Likewise.
23792 (cbranchsi4_scratch): Compute proper instruction length using
23793 compact_hreg_operand.
23794 * config/arc/predicates.md (compact_hreg_operand): New predicate.
23796 2017-04-25 Richard Biener <rguenther@suse.de>
23798 PR middle-end/80509
23799 * passes.c (pass_manager::pass_manager): Initialize
23800 m_name_to_pass_map.
23802 2017-04-25 Richard Biener <rguenther@suse.de>
23804 PR tree-optimization/79201
23805 * tree-ssa-sink.c (statement_sink_location): Handle calls.
23807 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23810 * config/s390/vector.md: Split MEM->GPR vector moves for
23811 non-s_operand addresses.
23813 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23816 * config/s390/predicates.md (reload_const_wide_int_operand): New
23818 * config/s390/s390.md ("movti"): Remove d/P alternative.
23819 ("movti_bigconst"): New pattern definition.
23821 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23824 * s390-protos.h (s390_expand_cs_hqi): Removed.
23825 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
23826 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
23827 modes as well as CCZ1mode and CCZmode.
23828 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
23829 signature of s390_emit_compare_and_swap.
23830 (s390_expand_cs_hqi): Likewise, make static.
23831 (s390_expand_cs_tdsi): Generate an explicit compare before trying
23832 compare-and-swap, in some cases.
23833 (s390_expand_cs): Wrapper function.
23834 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
23836 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
23837 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
23838 patterns for small and large integers. Forbid symref memory operands.
23839 Move expander to s390.c. Require cc register.
23840 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
23841 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
23842 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
23843 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
23844 symref memory operands. Remove CC mode and call s390_match_ccmode
23846 ("atomic_exchange<mode>"): Allow and implement all integer modes.
23848 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23850 * config/s390/s390.md (define_peephole2): New peephole to help
23851 combining the load-and-test pattern with volatile memory.
23853 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23855 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
23856 with CCZmode for TARGET_Z196.
23858 2017-04-25 Jakub Jelinek <jakub@redhat.com>
23860 PR rtl-optimization/80501
23861 * combine.c (make_compound_operation_int): Set subreg_code to SET
23862 even for AND with mask of the sign bit of mode.
23864 PR rtl-optimization/80500
23865 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
23866 sum's initial value.
23868 2017-04-25 Julian Brown <julian@codesourcery.com>
23869 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23871 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
23873 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
23875 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
23877 2017-04-25 Julian Brown <julian@codesourcery.com>
23878 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23880 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
23881 (thunderx2t99_sha): New Reservation.
23883 2017-04-25 Julian Brown <julian@codesourcery.com>
23884 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23886 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
23887 type for 1-element load.
23889 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
23891 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
23893 2017-04-24 Martin Jambor <mjambor@suse.cz>
23895 PR tree-optimization/80293
23896 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
23897 char arrays not totally scalarizable if it is false.
23898 (analyze_all_variable_accesses): Pass correct value in the new
23899 parameter. Add a statistics counter.
23901 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
23903 PR middle-end/79931
23904 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
23906 2017-04-24 Richard Biener <rguenther@suse.de>
23908 PR tree-optimization/80494
23909 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
23910 out for complex types.
23912 2017-04-24 Richard Biener <rguenther@suse.de>
23914 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
23915 * tree-ssa-sccvn.c (print_scc): Print SCC size.
23916 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
23917 (DFS): Adjust and never fail.
23918 (sccvn_dom_walker::fail): Remove.
23919 (sccvn_dom_walker::before_dom_children): Adjust.
23920 (run_scc_vn): Likewise and never fail.
23921 * tree-ssa-pre.c (pass_pre::execute): Adjust.
23922 (pass_fre::execute): Likewise.
23924 2017-04-24 Richard Biener <rguenther@suse.de>
23926 PR tree-optimization/79725
23927 * tree-ssa-sink.c (statement_sink_location): Return whether
23928 failure reason was zero uses. Move that check later.
23929 (sink_code_in_bb): Deal with zero uses by removing the stmt
23932 2017-04-24 Richard Biener <rguenther@suse.de>
23935 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
23936 pointer-based references.
23938 2017-04-24 Richard Biener <rguenther@suse.de>
23941 * pass_manager.h (pass_manager::operator new): Remove.
23942 (pass_manager::operator delete): Likewise.
23943 * passes.c (pass_manager::operator new): Remove.
23944 (pass_manager::operator delete): Likewise.
23945 (pass_manager::pass_manager): Zero individual pass members.
23947 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
23950 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
23951 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
23952 Check "XEXP (src, 1)" operand here.
23953 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
23954 Check "XEXP (src, 1)" operand here.
23955 (dimode_scalar_chain::make_vector_copies): Detect count register
23956 of a shift instruction. Zero extend count register from QImode
23957 to DImode to satisfy vector shift pattern count operand predicate.
23958 Substitute vector shift count operand with a DImode copy.
23959 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
23962 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
23964 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
23965 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
23966 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
23967 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
23968 (UNSPEC_NOREX_MEM): Remove definition.
23970 2017-04-21 Richard Biener <rguenther@suse.de>
23972 PR tree-optimization/79547
23973 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
23974 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
23975 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
23976 without any constraints.
23978 2017-04-21 Richard Biener <rguenther@suse.de>
23980 PR tree-optimization/78847
23981 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
23983 2017-04-21 Richard Biener <rguenther@suse.de>
23985 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
23986 (build_distinct_type_copy): Likewise.
23987 (build_variant_type_copy): Likewise.
23988 * tree.c (build_qualified_type): Pass down mem-stat info.
23989 (build_distinct_type_copy): Likewise.
23990 (build_variant_type_copy): Likewise.
23992 2017-04-21 Richard Biener <rguenther@suse.de>
23994 PR tree-optimization/80237
23995 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
23997 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
23998 for a simplified result.
24000 2016-04-21 Richard Biener <rguenther@suse.de>
24002 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
24003 sth as strict as a simple_iv but a chrec without symbols and an
24004 operand defined in the loop we are peeling (and not some subloop).
24005 (propagate_constants_for_unrolling): Propagate all constants.
24007 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
24010 * config/i386/i386.c (print_reg): Remove assert for disalowed
24011 regno values, call output_operand_lossage instead.
24013 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
24016 * config/i386/constraints.md (Yc): New register constraint.
24017 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
24018 Use Yc constraint for alternative 2 of operand 0. Remove
24019 preferred_for_speed attribute.
24021 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
24023 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
24024 lastprivate clauses in SIMT case.
24026 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
24028 * doc/invoke.texi (-Wextra-semi): Document new warning option.
24030 2017-04-20 Richard Biener <rguenther@suse.de>
24032 PR tree-optimization/57796
24033 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
24034 as N scalar stores.
24035 (vect_model_load_cost): Cost gathers as N scalar loads.
24037 2017-04-20 Richard Biener <rguenther@suse.de>
24039 * ggc-page.c (ggc_allocated_p): Rename to ...
24040 (safe_lookup_page_table_entry): ... this and return the lookup
24042 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
24044 2017-04-20 Richard Biener <rguenther@suse.de>
24046 PR tree-optimization/80453
24047 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
24048 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
24049 from the conditions.
24050 (vn_phi_eq): Pass them down.
24051 (vn_phi_lookup): Record them.
24052 (vn_phi_insert): Likewise.
24054 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
24056 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
24057 uninitialized variable warning to avoid buffer overrun.
24059 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
24062 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
24063 is suppressed for '{ 0 }' in C.
24065 2017-04-20 Jakub Jelinek <jakub@redhat.com>
24067 * BASE-VER: Set to 8.0.0.
24069 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
24071 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
24072 priority .init_array and .fini_array section with SECTION_NOTYPE
24075 2017-04-20 Jakub Jelinek <jakub@redhat.com>
24077 PR middle-end/80423
24078 * tree.h (build_array_type): Add typeless_storage default argument.
24079 * tree.c (type_cache_hasher::equal): Also compare
24080 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
24081 (build_array_type): Add typeless_storage argument, set
24082 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
24084 (build_nonshared_array_type): Adjust build_array_type_1 caller.
24085 (build_array_type): Likewise. Add typeless_storage argument.
24087 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
24088 Jakub Jelinek <jakub@redhat.com>
24090 PR tree-optimization/80426
24091 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
24092 operation on symbolic operands, also compute the overflow for the
24093 invariant part when the operation degenerates into a negation.
24095 2017-04-19 Jakub Jelinek <jakub@redhat.com>
24098 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
24099 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
24102 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
24104 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
24107 * config/avr/avr.c (tree.h): Include it.
24108 (cgraph.h): Include it.
24109 (avr_encode_section_info): Don't warn for uninitialized progmem
24110 variable if it's just an alias.
24112 2017-04-19 Richard Biener <rguenther@suse.de>
24115 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
24116 when needed by AutoPGO.
24118 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
24121 * doc/lto.texi: Remove an extra 'that'.
24123 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
24125 PR rtl-optimization/80429
24126 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
24127 are only used in debug insns.
24129 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
24130 Vladimir Makarov <vmakarov@redhat.com>
24132 * config/sparc/predicates.md (input_operand): Add comment. Return
24133 true for any memory operand when LRA is in progress.
24134 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
24136 2017-04-18 Jeff Law <law@redhat.com>
24139 * mips.md ({return,simple_return}_internal): Do not overwrite
24142 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24144 PR tree-optimization/80443
24145 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
24146 instead of adding 1, subtract -1 and similarly instead of subtracting
24149 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
24151 PR rtl-optimization/80357
24152 * haifa-sched.c (tmp_bitmap): New variable.
24153 (model_recompute): Handle duplicate use records.
24154 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
24155 (free_global_sched_pressure_data): Free it.
24157 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
24160 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
24161 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24162 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24163 instead of SYSTEM_HEADER_DIR.
24165 2017-04-18 Jeff Law <law@redhat.com>
24167 PR middle-end/80422
24168 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
24169 predecessors after walking up the insn chain.
24171 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24174 * dwarf2out.c (modified_type_die): Try harder not to emit internal
24175 sizetype type into debug info.
24177 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24180 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
24181 unneeded test for TARGET_UPPER_REGS_SF.
24182 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
24184 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24187 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
24188 instead of gsi_after_labels.
24190 2017-04-18 Jeff Law <law@redhat.com>
24192 * regcprop.c (maybe_mode_change): Avoid creating copies of the
24196 2017-04-13 Jeff Law <law@redhat.com>
24197 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
24198 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
24200 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
24203 * config/avr/avr.c (intl.h): Include it.
24204 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
24206 2017-04-18 Martin Liska <mliska@suse.cz>
24208 PR gcov-profile/78783
24209 * gcov-tool.c (gcov_output_files): Validate that destination
24210 file is either removed by the tool or by a user.
24212 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
24213 Guy Benyei <guybe@mellanox.com>
24215 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
24216 block, and do not negate it, the stored id is already negative.
24218 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
24220 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
24222 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
24225 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
24226 masks of options that should be turned off if the VSX vector
24227 options are turned off.
24228 (OTHER_P8_VECTOR_MASKS): Likewise.
24229 (OTHER_VSX_VECTOR_MASKS): Likewise.
24230 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
24231 rs6000_disable_incompatible_switches to validate no type switches
24233 (rs6000_incompatible_switch): New function to disallow turning on
24234 other vector options if -mno-vsx, -mno-power8-vector, or
24235 -mno-power9-vector are specified.
24237 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24239 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
24241 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24243 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
24244 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
24245 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
24246 (ARG_POINTER_CFA_OFFSET): Likewise.
24248 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24250 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
24251 conditions to take advantage of various optimizations.
24253 2017-04-13 Jeff Law <law@redhat.com>
24255 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
24256 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
24257 (zero_extendsidi2_dext): Likewise.
24259 2017-04-13 Jakub Jelinek <jakub@redhat.com>
24262 * fold-const.c (fold_ternary_loc): Revert
24263 use op0 instead of fold_convert_loc (loc, type, arg0) part of
24266 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
24268 PR rtl-optimization/80343
24269 * lra-remat.c (update_scratch_ops): Assign original hard reg to
24270 new scratch pseudo.
24272 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
24275 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
24276 to ubsan_encode_value.
24278 2017-04-13 Jeff Law <law@redhat.com>
24280 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
24281 appearing in DEBUG_INSNs.
24283 2017-04-13 Martin Liska <mliska@suse.cz>
24285 PR gcov-profile/80413
24286 * gcov-io.c (gcov_write_string): Copy to buffer just when
24287 allocated size is greater than zero.
24289 2017-04-13 Jakub Jelinek <jakub@redhat.com>
24292 * dwarf2out.c (decls_for_scope): Ignore declarations of
24293 current_function_decl in BLOCK_NONLOCALIZED_VARS.
24295 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
24298 * ipa-visibility.c (non_local_p): Fix typos.
24299 (localize_node): When localizing symbol in same comdat group,
24300 dissolve the group only when we know external symbols are going
24302 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
24304 2017-04-12 Jakub Jelinek <jakub@redhat.com>
24306 PR tree-optimization/79390
24307 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
24308 order does not result in usable sequence, retry with reversed operand
24314 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
24315 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
24316 op0 instead of fold_convert_loc (loc, type, arg0).
24318 2017-04-12 Jeff Law <law@redhat.com>
24320 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
24321 has a delay slot in the generated code.
24323 * config/cris/cris.md (cris_preferred_reload_class): Return
24324 GENNONACR_REGS rather than GENERAL_REGS.
24326 2017-04-12 Jakub Jelinek <jakub@redhat.com>
24329 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
24330 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
24331 signedness of the result type.
24333 2017-04-12 Richard Biener <rguenther@suse.de>
24334 Jeff Law <law@redhat.com>
24336 PR tree-optimization/80359
24337 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
24338 trim stores to TARGET_MEM_REFs.
24340 2017-04-12 Richard Biener <rguenther@suse.de>
24342 PR tree-optimization/79390
24343 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
24344 threading case even more.
24346 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
24349 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
24350 for quad_address_p for TImode, instead of just not indexed_address.
24352 2017-04-12 Richard Biener <rguenther@suse.de>
24353 Bernd Edlinger <bernd.edlinger@hotmail.de>
24355 PR middle-end/79671
24356 * alias.c (component_uses_parent_alias_set_from): Handle
24357 TYPE_TYPELESS_STORAGE.
24358 (get_alias_set): Likewise.
24359 * tree-core.h (tree_type_common): Add typeless_storage flag.
24360 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
24361 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
24362 for types containing members with TYPE_TYPELESS_STORAGE.
24363 (place_field): Likewise.
24364 (layout_type): Likewise for ARRAY_TYPE.
24365 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
24366 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
24367 TYPE_TYPELESS_STORAGE.
24368 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
24370 2017-04-12 Jakub Jelinek <jakub@redhat.com>
24373 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
24374 first argument to type.
24376 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24380 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
24381 CONST0_RTX (mode) rather than const0_rtx where appropriate.
24382 (rs6000_expand_binop_builtin): Likewise.
24383 (rs6000_expand_ternop_builtin): Likewise; also add missing
24384 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
24385 vshasigma built-ins.
24386 * doc/extend.texi: Document that vec_xxpermdi's third argument
24387 must be a constant.
24389 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
24391 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
24392 Use shift_const cost parameter when calculating gain of STV shifts.
24394 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
24396 PR rtl-optimization/70478
24397 * lra-constraints.c (process_alt_operands): Check memory for
24398 disfavoring memory insn operand.
24400 2017-04-11 Jakub Jelinek <jakub@redhat.com>
24402 PR middle-end/80100
24403 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
24404 left shift in unsigned HOST_WIDE_INT type.
24406 PR rtl-optimization/80385
24407 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
24408 (not (neg X)) into (plus X -1) for complex or non-integral modes.
24411 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
24412 if they have any depend clauses.
24414 2017-04-11 Martin Liska <mliska@suse.cz>
24417 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
24418 * ipa-split.c (split_function): Create a local comdat symbol
24419 if caller is in a comdat group.
24421 2017-04-11 Martin Liska <mliska@suse.cz>
24424 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
24427 2017-04-11 Martin Sebor <msebor@redhat.com>
24429 PR middle-end/80364
24430 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
24431 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
24433 (directive::set_width, directive::set_precision, format_character):
24435 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
24438 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
24441 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
24442 conflict, set target->arch_name instead of target->cpu_name.
24444 2017-04-11 Richard Biener <rguenther@suse.de>
24446 PR tree-optimization/80374
24447 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
24448 build_zero_cst, remove fold_convertible_p check again.
24450 2017-04-11 Martin Liska <mliska@suse.cz>
24453 * ubsan.c (instrument_object_size): Do not instrument register
24456 2017-04-11 Jakub Jelinek <jakub@redhat.com>
24459 * config/i386/i386-builtin-types.def
24460 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
24461 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
24462 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
24463 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
24464 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
24465 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
24466 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
24467 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
24468 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
24469 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
24470 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
24471 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
24472 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
24473 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
24474 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
24475 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
24476 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
24477 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
24478 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
24479 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
24480 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
24481 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
24482 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
24483 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
24484 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
24485 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
24486 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
24487 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
24488 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
24489 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
24490 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
24491 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
24492 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
24493 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
24494 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
24495 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
24496 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
24497 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
24498 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
24499 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
24500 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
24501 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
24502 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
24503 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
24504 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
24505 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
24507 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
24508 flag to second_arg_count, handle 4 argument function type _COUNT
24509 aliases, handle second_arg_count on second argument rather than last.
24511 2017-04-10 Jeff Law <law@redhat.com>
24513 PR tree-optimization/80374
24514 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
24515 record anything if we can not convert integer_zero_node to the
24518 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24521 * config/rs6000/rs6000.c (rs6000_option_override_internal):
24522 Enhance special handling given to the TARGET_P9_MINMAX option in
24523 relation to certain other options.
24525 2017-04-10 Bin Cheng <bin.cheng@arm.com>
24527 PR tree-optimization/80153
24528 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
24529 remove POINTER_PLUS_EXPR's base part directly, rather than through
24532 2017-04-10 Richard Biener <rguenther@suse.de>
24533 Bin Cheng <bin.cheng@arm.com>
24535 PR tree-optimization/80153
24536 * tree-affine.c (aff_combination_to_tree): Get base pointer from
24537 the first element of pointer type aff_tree. Build result expr in
24539 (add_elt_to_tree): Convert to type unconditionally. Remove other
24540 fold_convert calls.
24541 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
24542 (rewrite_use_nonlinear_expr): Check invariant using iv information.
24544 2017-04-10 Richard Biener <rguenther@suse.de>
24546 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24549 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
24551 PR rtl-optimization/70478
24552 * lra-constraints.c (curr_small_class_check): New.
24553 (update_and_check_small_class_inputs): New.
24554 (process_alt_operands): Update curr_small_class_check. Disfavor
24555 alternative insn memory operands. Check available regs for small
24558 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
24561 * config/mips/mips.opt (-mvirt): Update description.
24562 * doc/invoke.texi (-mvirt): Likewise.
24564 2017-04-10 Richard Biener <rguenther@suse.de>
24566 PR middle-end/80362
24567 * fold-const.c (fold_binary_loc): Look at unstripped ops when
24568 looking for NEGATE_EXPR in -A / -B to A / B folding.
24570 2017-04-10 Martin Liska <mliska@suse.cz>
24572 PR gcov-profile/80224
24573 * gcov.c (print_usage): Fix usage string.
24574 (get_gcov_intermediate_filename): Remove.
24575 (output_gcov_file): Use both for normal and intermediate format.
24576 (generate_results): Do not initialize special file for
24577 intermediate format.
24579 2017-04-10 Richard Biener <rguenther@suse.de>
24581 PR tree-optimization/80304
24582 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
24585 2017-04-10 Nathan Sidwell <nathan@acm.org>
24588 * config/rs6000/rs6000.c (rs6000_vector_type): New.
24589 (rs6000_init_builtins): Use it.
24591 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24593 * config/arm/arm.md (<mrc>): Add mode to SET source.
24594 (<mrrc>): Likewise.
24596 2017-04-10 Richard Biener <rguenther@suse.de>
24598 PR middle-end/80344
24599 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
24601 2017-04-10 Jakub Jelinek <jakub@redhat.com>
24604 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
24605 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
24606 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
24607 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
24608 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
24609 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
24610 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
24611 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
24612 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
24613 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
24614 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
24615 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
24616 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
24617 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
24618 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
24619 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
24620 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
24621 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
24622 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
24623 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
24624 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
24625 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
24626 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
24628 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
24630 PR rtl-optimization/70478
24631 * lra-constraints.c: Reverse the last patch.
24633 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
24635 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
24636 Add comment for WCHAR_T.
24638 2017-04-08 Martin Liska <mliska@suse.cz>
24641 2017-04-07 Martin Liska <mliska@suse.cz>
24644 * ipa-split.c (split_function): Add function part to a same comdat
24647 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24650 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
24652 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
24654 * rs6000/rs6000.c (vec_load_pendulum): Rename...
24655 (vec_pairing): ...to this.
24656 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
24657 (rs6000_sched_init): Adjust for name change.
24658 (struct rs6000_sched_context): Likewise.
24659 (rs6000_init_sched_context): Likewise.
24660 (rs6000_set_sched_context): Likewise.
24662 2017-04-07 Jakub Jelinek <jakub@redhat.com>
24668 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
24670 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
24671 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
24672 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
24674 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
24676 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
24678 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
24680 PR rtl-optimization/70703
24681 * ira-color.c (update_conflict_hard_regno_costs): Use
24682 int64_t instead of HOST_WIDE_INT.
24684 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
24686 PR rtl-optimization/70478
24687 * lra-constraints.c (process_alt_operands): Disfavor alternative
24688 insn memory operands.
24690 2017-04-07 Jeff Law <law@redhat.com>
24692 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
24693 CALL and NOTE_INSN_CALL_ARG_LOCATION.
24695 2017-04-07 Martin Liska <mliska@suse.cz>
24698 * config/aarch64/aarch64.c (aarch64_process_target_attr):
24699 Show error message instead of an ICE.
24701 2017-04-07 Martin Liska <mliska@suse.cz>
24704 * ipa-split.c (split_function): Add function part to a same comdat
24707 2017-04-07 Richard Biener <rguenther@suse.de>
24709 PR middle-end/80341
24710 * tree.c (get_unwidened): Also handle ! for_type case for
24712 * convert.c (do_narrow): Split out from ...
24713 (convert_to_integer_1): ... here. Do not pass final truncation
24714 type to get_unwidened for TRUNC_DIV_EXPR.
24716 2017-04-07 Richard Biener <rguenther@suse.de>
24718 * tree-affine.c (wide_int_ext_for_comb): Take type rather
24720 (aff_combination_const): Adjust.
24721 (aff_combination_scale): Likewise.
24722 (aff_combination_add_elt): Likewise.
24723 (aff_combination_add_cst): Likewise.
24724 (aff_combination_convert): Likewise.
24725 (add_elt_to_tree): Likewise. Remove unused argument.
24726 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
24728 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
24730 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
24732 * config/arm/arm.c (arm_default_short_enums): Use
24733 ARM_DEFAULT_SHORT_ENUMS.
24734 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
24736 2017-04-06 Jakub Jelinek <jakub@redhat.com>
24739 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
24740 members with redundant out-of-class redeclaration.
24742 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
24745 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
24746 * config/i386/i386.md (*zero_extendsidi2):
24747 Add (?*x,*x) and (?*v,*v) alternatives.
24749 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
24752 * config/i386/i386.c (ix86_expand_builtin)
24753 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
24754 mode from insn data. Convert operands to insn operand mode.
24755 Copy operands that don't satisfy insn predicate to a register.
24757 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
24759 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
24762 2017-04-06 Richard Biener <rguenther@suse.de>
24764 PR tree-optimization/80334
24765 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
24766 preserve alignment of accesses.
24768 2017-04-06 Richard Biener <rguenther@suse.de>
24770 PR tree-optimization/80262
24771 * tree-sra.c (build_ref_for_offset): Preserve address-space
24773 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
24774 Drop useless address-space information on MEM_REF offsets.
24776 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
24778 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
24780 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
24782 PR rtl-optimization/70703
24783 * ira-color.c (update_conflict_hard_regno_costs): Use
24784 HOST_WIDE_INT instead of long.
24786 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
24789 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
24790 not defined for x86_64 target. Add -mmmx target option when __SSE2__
24792 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
24793 for x86_64 target. Handle -m3dnowa option.
24795 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
24797 PR rtl-optimization/70703
24798 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
24799 (update_conflict_hard_regno_costs): Use long instead of unsigned
24800 arithmetic for cost calculation.
24802 2017-04-05 Jakub Jelinek <jakub@redhat.com>
24803 Bernd Edlinger <bernd.edlinger@hotmail.de>
24806 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
24809 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
24812 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
24813 ptr_mode with Pmode throughout.
24814 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
24815 into probe_stack_range and use DImode.
24817 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
24820 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
24821 call_eh_return is true.
24823 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24825 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
24826 Initialize last_match_fntype_index.
24828 2017-04-05 Jakub Jelinek <jakub@redhat.com>
24831 * tree-nvr.c: Include internal-fn.h.
24832 (pass_return_slot::execute): Ignore internal calls without
24835 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24836 Richard Biener <rguenther@suse.de>
24839 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
24840 captures used multiple times, except for the last use.
24841 * generic-match-head.c: Include gimplify.h.
24843 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24845 PR tree-optimization/79390
24846 * target.h (struct noce_if_info): Declare.
24847 * targhooks.h (default_noce_conversion_profitable_p): Declare.
24848 * target.def (noce_conversion_profitable_p): New target hook.
24849 * ifcvt.h (struct noce_if_info): New type, moved from ...
24850 * ifcvt.c (struct noce_if_info): ... here.
24851 (noce_conversion_profitable_p): Renamed to ...
24852 (default_noce_conversion_profitable_p): ... this. No longer
24854 (noce_try_store_flag_constants, noce_try_addcc,
24855 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
24856 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
24857 instead of noce_conversion_profitable_p.
24858 * config/i386/i386.c: Include ifcvt.h.
24859 (ix86_option_override_internal): Don't override
24860 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
24861 (ix86_noce_conversion_profitable_p): New function.
24862 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
24863 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
24864 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
24865 * doc/tm.texi: Regenerated.
24867 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24869 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
24872 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24875 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
24876 instructions for small multiply cores.
24878 2017-04-04 Jeff Law <law@redhat.com>
24880 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
24882 (mips_expand_vec_perm_const): Initialize elements in orig_perm
24883 that are not set by the loop over the elements.
24885 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24888 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
24889 int mode, convert_modes it to mode as unsigned, otherwise use
24890 lowpart_subreg to mode rather than SImode.
24891 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
24892 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
24893 Use DImode instead of SImode for the shift count operand.
24894 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
24897 2017-04-04 Richard Biener <rguenther@suse.de>
24899 PR middle-end/80281
24900 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
24901 arithmetic done for the negate or the plus. Simplify.
24902 (A - (-B) -> A + B): Likewise.
24903 * fold-const.c (split_tree): Make sure to not negate pointers.
24905 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
24907 PR rtl-optimization/60818
24908 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
24909 a compare of comparisons with the thing compared if this results
24910 in a different machine mode.
24912 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
24914 * alias.c (base_alias_check): Fix typo in comment.
24915 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
24916 * cgraphunit.c (symbol_table::compile): Likewise.
24917 * collect2.c (maybe_run_lto_and_relink): Likewise.
24918 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
24919 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
24920 * config/avr/avr.c (avr_map_op_t): Likewise.
24921 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
24922 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
24923 * config/epiphany/epiphany.md (movcc): Likewise.
24924 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
24925 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
24927 * config/mips/mips.c (mips_save_restore_reg): Likewise.
24928 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
24929 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
24930 * config/sh/sh.c (sh_rtx_costs): Likewise.
24931 * fold-const.c (fold_truth_andor): Likewise.
24932 * genautomata.c (collapse_flag): Likewise.
24933 * gengtype.h (struct type::u::s): Likewise.
24934 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
24935 * input.c (FORMAT_AMOUNT): Likewise.
24936 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
24937 (known_aggs_to_agg_replacement_list): Likewise.
24938 * ipa-inline-analysis.c: Likewise.
24939 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
24940 * ipa-polymorphic-call.c
24941 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
24942 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
24943 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
24945 * modulo-sched.c (apply_reg_moves): Likewise.
24946 * omp-expand.c (build_omp_regions_1): Likewise.
24947 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
24948 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
24949 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
24950 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
24951 * value-prof.c: Likewise.
24952 * var-tracking.c (val_reset): Likewise.
24954 2017-04-03 Richard Biener <rguenther@suse.de>
24956 PR tree-optimization/80275
24957 * fold-const.c (split_address_to_core_and_offset): Handle
24960 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
24962 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
24963 descriptors is at least equal to that of functions.
24965 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
24967 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
24969 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
24972 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
24973 (mov<IMOD4:mode>): New expander.
24974 (*mov<IMOD4:mode>_internal): New insn and split pattern.
24976 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
24978 PR rtl-optimization/79405
24979 * fwprop.c (propagations_left): New variable.
24980 (forward_propagate_into): Decrement it.
24981 (fwprop_init): Initialize it.
24982 (fw_prop): If the variable has reached zero, stop propagating.
24983 (fwprop_addr): Ditto.
24985 2017-03-31 Jakub Jelinek <jakub@redhat.com>
24988 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
24989 a FUNCTION_DECL, pass it as decl instead of origin to
24992 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
24994 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
24997 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
25000 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
25001 TARGET_VSX_SMALL_INTEGER.
25003 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25005 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25006 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
25008 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
25010 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
25011 extraction from odd-numbered MSA register.
25013 2017-03-31 Jakub Jelinek <jakub@redhat.com>
25015 PR middle-end/80173
25016 * expmed.c (store_bit_field_1): Don't attempt to create
25017 a word subreg out of hard registers wider than word if they
25018 have HARD_REGNO_NREGS of 1 for their mode.
25020 PR middle-end/80163
25021 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
25022 conversions to integer types wider than word and pointer.
25025 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
25026 (rtx_equal_for_cselib_p): Pass 0 to it.
25027 * cselib.c (cselib_hasher::equal): Likewise.
25028 (rtx_equal_for_cselib_1): Add depth argument. If depth
25029 is 128, don't look up VALUE locs and punt. Increment
25030 depth in recursive calls when walking VALUE locs.
25032 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
25034 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
25035 (make_gcov_file_name): Use the canonical path name for generating
25037 (read_line): Fix handling of files with ascii null bytes.
25039 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
25041 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
25042 to initialise a vector register instead
25043 of using a const_int.
25045 2017-03-30 Jakub Jelinek <jakub@redhat.com>
25047 PR translation/80189
25048 * gimplify.c (omp_default_clause): Use %qs instead of %s in
25049 diagnostic messages.
25051 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
25054 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
25055 (dfp_diex_<mode>): Update mode of operand 1.
25056 * doc/extend.texi (dxex, dxexq): Document change to return type.
25057 (diex, diexq): Document change to argument type.
25059 2017-03-30 Martin Jambor <mjambor@suse.cz>
25062 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
25063 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
25064 it reflects the signature changes performed at the callee side.
25065 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
25066 to cgraph_build_function_type_skip_args.
25067 (build_function_decl_skip_args): Adjust call to the above function.
25069 2017-03-30 Jakub Jelinek <jakub@redhat.com>
25072 * config/i386/sse.md
25073 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
25074 register as dest whenever it is a MEM not rtx_equal_p to the
25075 corresponding dup operand, and when forcing into reg move the
25076 reg into the memory afterwards.
25077 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
25078 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
25079 for the force_reg mode.
25080 (avx512vl_vextractf128<mode>): Use register as dest either
25081 always when a MEM, or when it is a MEM not rtx_equal_p to the
25082 corresponding dup operand, or even not when it is a CONST_VECTOR
25083 depending on the mode and lo vs. hi.
25084 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
25086 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
25087 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
25088 Likewise. Require that operands[2] is even.
25089 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
25090 Remove extraneous parens. Require that operands[2] is a multiple
25092 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
25093 operands[0] is a MEM if <mask_applied>, the predicates/constraints
25094 disallow memory then.
25096 2017-03-30 Richard Biener <rguenther@suse.de>
25098 PR tree-optimization/77498
25099 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
25100 to non-constants over backedges.
25102 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
25104 PR rtl-optimization/80233
25105 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
25106 as last_combined_insn. Do not test for BARRIER_P separately.
25108 2017-03-29 Andreas Schwab <schwab@suse.de>
25111 * calls.c (prepare_call_address): Convert funexp to Pmode before
25112 copying to temp reg.
25114 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25116 PR tree-optimization/80158
25117 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
25118 Handle possible future case of more than one alternate
25120 (replace_rhs_if_not_dup): Likewise.
25121 (replace_one_candidate): Likewise.
25123 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
25125 PR rtl-optimization/80193
25126 * ira.c (ira): Do not check allocation for LRA.
25128 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
25130 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
25131 (nvptx_output_simt_exit): Declare.
25132 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
25133 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
25134 (init_softstack_frame): Move initialization of crtl->is_leaf to...
25135 (nvptx_declare_function_name): ...here. Emit declaration of local
25136 memory space buffer for omp_simt_enter insn.
25137 (nvptx_output_unisimt_switch): New.
25138 (nvptx_output_softstack_switch): New.
25139 (nvptx_output_simt_enter): New.
25140 (nvptx_output_simt_exit): New.
25141 * config/nvptx/nvptx.h (struct machine_function): New fields
25142 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
25143 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
25144 (UNSPECV_SIMT_EXIT): Ditto.
25145 (omp_simt_enter_insn): New insn.
25146 (omp_simt_enter): New expansion.
25147 (omp_simt_exit): New insn.
25148 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
25150 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
25151 (expand_GOMP_SIMT_ENTER_ALLOC): New.
25152 (expand_GOMP_SIMT_EXIT): New.
25153 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
25154 (GOMP_SIMT_ENTER_ALLOC): Ditto.
25155 (GOMP_SIMT_EXIT): Ditto.
25156 * target-insns.def (omp_simt_enter): New insn.
25157 (omp_simt_exit): Ditto.
25158 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
25160 (lower_rec_simd_input_clauses): Implement SIMT privatization.
25161 (lower_rec_input_clauses): Likewise.
25162 (lower_lastprivate_clauses): Handle SIMT privatization.
25164 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
25165 (ompdevlow_adjust_simt_enter): New.
25166 (find_simtpriv_var_op): New.
25167 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
25168 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
25170 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
25171 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
25172 (copy_decl_for_dup_finish): Ditto.
25174 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
25176 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
25179 * config/i386/i386.c (ix86_option_override_internal): Always
25180 allow -mpreferred-stack-boundary=3 for 64-bit targets.
25182 2017-03-28 Bin Cheng <bin.cheng@arm.com>
25184 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
25186 2017-03-28 Bin Cheng <bin.cheng@arm.com>
25188 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
25189 mark new edge's irreducible flag accordign to it.
25190 (vect_do_peeling): Check loop preheader edge's irreducible flag
25191 and pass it to function slpeel_add_loop_guard.
25193 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
25195 PR tree-optimization/80218
25196 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
25197 Update block frequencies and counts.
25199 2017-03-28 Richard Biener <rguenther@suse.de>
25201 PR tree-optimization/78644
25202 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
25203 of a simplification result we may not use it at all.
25205 2017-03-28 Richard Biener <rguenther@suse.de>
25208 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
25209 without arguments, generate default definition of a SSA name.
25211 2017-03-28 Richard Biener <rguenther@suse.de>
25213 PR middle-end/80222
25214 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
25215 TYPE_REF_CAN_ALIAS_ALL references.
25216 * fold-const.c (fold_indirect_ref_1): Likewise.
25218 2017-03-28 Martin Liska <mliska@suse.cz>
25221 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
25222 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
25224 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
25225 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
25227 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
25228 (EXTRA_SPECS): Define.
25229 (SUBTARGET_EXTRA_SPECS): Likewise.
25230 (SUBTARGET_CPP_SPEC): Likewise.
25231 * config/arc/elf.h (EXTRA_SPECS): Renamed to
25232 SUBTARGET_EXTRA_SPECS.
25233 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
25235 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
25237 * config/arc/simdext.md (vst64_insn): Update pattern.
25238 (vld32wh_insn): Likewise.
25239 (vld32wl_insn): Likewise.
25240 (vld64_insn): Likewise.
25241 (vld32_insn): Likewise.
25243 2017-03-28 Marek Polacek <polacek@redhat.com>
25246 * fold-const.c (fold_comparison): Use protected_set_expr_location
25247 instead of SET_EXPR_LOCATION.
25249 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
25251 * tree.c (add_expr): Avoid name lookup warning.
25253 2017-03-27 Jeff Law <law@redhat.com>
25255 PR tree-optimization/80216
25256 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
25257 function name. Limit recursion depth.
25258 (record_temporary_equivalences): Corresponding changes.
25260 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
25262 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
25265 2017-03-27 Jakub Jelinek <jakub@redhat.com>
25268 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
25270 * cfgcleanup.c (reg_note_cfa_p): New array.
25271 (insns_have_identical_cfa_notes): New function.
25272 (old_insns_match_p): Don't cross-jump in between /f
25273 and non-/f instructions. If both i1 and i2 are frame related,
25274 verify all CFA notes, their order and content.
25276 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
25279 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
25280 HImode and SImode with zero extend to DImode to one insn.
25281 (bswap<mode>2_extenddi): Likewise.
25282 (bswapsi2_extenddi): Likewise.
25283 (bswaphi2_extendsi): Likewise.
25284 (bswaphi2): Combine bswap HImode and SImode into one insn.
25285 Separate memory insns from swapping register.
25286 (bswapsi2): Likewise.
25287 (bswap<mode>2): Likewise.
25288 (bswaphi2_internal): Delete, no longer used.
25289 (bswapsi2_internal): Likewise.
25290 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
25291 store, and gpr<-gpr swap insns.
25292 (bswap<mode>2_store): Likewise.
25293 (bswaphi2_reg): Register only splitter, combine with the splitter.
25294 (bswaphi2 splitter): Likewise.
25295 (bswapsi2_reg): Likewise.
25296 (bswapsi2 splitter): Likewise.
25297 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
25298 the insns into load, store, and register/register insns.
25299 (bswapdi2_ldbrx): Likewise.
25300 (bswapdi2_load): Likewise.
25301 (bswapdi2_store): Likewise.
25302 (bswapdi2_reg): Likewise.
25304 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
25306 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
25307 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
25309 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
25312 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
25314 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25315 special handling for target option conflicts between dform
25316 options (-mpower9-dform, -mpower9-dform-vector,
25317 -mpower9-dform-scalar) and -mno-direct-move.
25319 2017-03-27 Richard Biener <rguenther@suse.de>
25321 PR tree-optimization/80181
25322 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
25324 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25326 * config/arc/predicates.md (move_double_src_operand): Replace the
25327 call to move_double_src_operand with a call to address_operand.
25329 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25331 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
25332 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
25333 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
25335 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25337 * config/arc/predicates.md (long_immediate_loadstore_operand):
25338 Consider scaled addresses cases.
25340 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25342 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
25343 restored when in interrupt.
25344 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
25345 doesn't have delay slot.
25347 2017-03-27 Richard Biener <rguenther@suse.de>
25350 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
25351 inlined thunk clones.
25353 2017-03-27 Jakub Jelinek <jakub@redhat.com>
25356 * asan.c (instrument_derefs): Copy over last operand from
25357 original COMPONENT_REF to the new COMPONENT_REF with
25358 DECL_BIT_FIELD_REPRESENTATIVE.
25359 * ubsan.c (instrument_object_size): Likewise.
25361 2017-03-27 Richard Biener <rguenther@suse.de>
25363 PR tree-optimization/80170
25364 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
25365 sure DR/SCEV didnt fold in constants we do not see when looking
25366 at the reference base alignment.
25368 2017-03-27 Richard Biener <rguenther@suse.de>
25370 PR middle-end/80171
25371 * gimple-fold.c (fold_ctor_reference): Properly guard against
25372 NULL return value from canonicalize_constructor_val.
25374 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
25377 * config/i386/i386.c (ix86_expand_builtin)
25378 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
25379 flags reg setting and flags reg using instructions.
25380 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
25381 clobbering instructions to zero extend op2.
25383 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
25385 * doc/install.texi (Configuration) <--with-aix-soname>:
25386 Update link to AIX ld.
25388 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
25390 PR rtl-optimization/80160
25391 PR rtl-optimization/80159
25392 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
25393 reg_alternate_class into account.
25395 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
25398 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
25399 to consider in curr_insn_transform.
25401 2017-03-24 Jakub Jelinek <jakub@redhat.com>
25403 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
25404 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
25405 and emit_mode_inner.
25407 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25409 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
25410 argument to the overloaded builtin variants. Use the new flag to
25411 deprecate certain builtin variants.
25412 * config/s390/s390-builtin-types.def: Add new builtin types.
25413 * config/s390/s390-builtins.h: Support new flags field for
25414 overloaded builtins.
25415 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
25416 (s390_macro_to_expand): Enable vector float data type.
25417 (s390_cpu_cpp_builtins_internal): Indicate support of the new
25418 builtins by incrementing the __VEC__ version number.
25419 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
25421 (s390_resolve_overloaded_builtin): Emit error messages depending
25422 on the builtin flags.
25423 * config/s390/s390.c (s390_expand_builtin): Support additional
25424 flags argument. Change error message to match the messages
25425 emitted in s390-c.c.
25426 * config/s390/s390.md: New UNSPEC_* constants.
25427 (op_type): Add new instruction types.
25428 * config/s390/vecintrin.h: Add new builtins and test data class
25430 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
25431 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
25432 (VEC_INEXACT, VEC_NOINEXACT): New constants.
25433 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
25434 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
25435 ("vec_mergel<mode>"): V_HW -> VEC_HW.
25437 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
25438 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
25439 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
25440 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
25442 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
25443 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
25444 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
25445 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
25447 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
25448 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
25449 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
25450 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
25451 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
25452 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
25453 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
25455 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
25456 ("vec_scatter_element<V_HW_4:mode>_DI")
25457 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
25458 ("vec_fpint<mode>", "vflls")
25459 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
25460 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
25461 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
25462 ("*vec_cmphe<mode>_cc"): ... these.
25464 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
25465 mode constant instead of magic value.
25467 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25469 * config/s390/s390.c (s390_expand_vec_compare): Support other
25470 vector floating point modes than just V2DF.
25471 (s390_expand_vcond): Likewise.
25472 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
25473 (s390_cannot_change_mode_class): Prevent mode changes between TF
25474 and V1TF in vector registers.
25475 * config/s390/s390.md (DF, SF): New mode attributes.
25476 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
25477 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
25478 SFmode support for VRs.
25479 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
25481 (VFT, VF_HW): New mode iterators.
25482 (vw, sdx): New mode attributes.
25483 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
25484 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
25485 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
25486 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
25487 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
25488 also the new vector floating point modes. Renaming to ...
25490 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
25491 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
25492 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
25493 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
25494 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
25495 ("vec_unordered<mode>"): ... these.
25497 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
25498 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
25499 ("*vec_extendv2df"): New insn definitions.
25501 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25503 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
25504 ("mulditi3_2", "*muldi3_sign"): New patterns.
25505 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
25506 rename the pattern definition.
25508 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25510 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
25512 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
25514 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25516 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
25517 instruction if possible.
25518 * config/s390/vector.md (vec_halfnumelts): New mode
25520 ("*vec_vllezlf<mode>"): New pattern.
25522 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25524 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
25525 ("popcountv4si2", "popcountv2di2"): Rename to ...
25526 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
25527 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
25529 ("popcount<mode>2_vxe"): New pattern.
25531 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25533 * common/config/s390/s390-common.c (processor_flags_table): Add
25535 * config.gcc: Add arch12.
25536 * config/s390/driver-native.c (s390_host_detect_local_cpu):
25537 Default to arch12 for unknown CPU model numbers.
25538 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
25539 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
25540 PROCESSOR_max sanity check.
25541 * config/s390/s390-opts.h (enum processor_type): Add
25543 * config/s390/s390.c (processor_table): Add arch12.
25544 (s390_expand_builtin): Add check for B_VXE flag.
25545 (s390_issue_rate): Add PROCESSOR_ARCH12.
25546 (s390_get_sched_attrmask): Likewise.
25547 (s390_get_unit_mask): Likewise.
25548 (s390_sched_score): Enable z13 scheduling for arch12.
25549 (s390_sched_reorder): Likewise.
25550 (s390_sched_variable_issue): Likewise.
25551 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
25553 (s390_tune_attr): Use z13 scheduling also for arch12.
25554 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
25555 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
25556 (TARGET_VXE_P): New macros.
25557 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
25558 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
25559 * config/s390/s390.opt: Add arch12 as processor_type.
25561 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25563 * config/s390/s390.md
25564 ("fixuns_truncdddi2", "fixuns_trunctddi2")
25565 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
25566 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
25568 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
25569 Rename expanders to ...
25571 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
25572 ("fixuns_truncdddi2_emu"): ... these.
25574 ("fixuns_trunc<mode>si2_emu"): New expander.
25576 ("*fixuns_truncdfdi2_z13"): Rename to ...
25577 ("*fixuns_truncdfdi2_vx"): ... this.
25579 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25581 * config/s390/2964.md: Remove the single element vector compare
25582 instructions which are no longer used.
25583 * config/s390/s390.c (s390_select_ccmode): Remove handling of
25585 (s390_canonicalize_comparison): Remove handling of DFmode
25587 (s390_expand_vec_compare_scalar): Remove function.
25588 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
25589 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
25591 ("*cmp<mode>_ccs"): Add wfcdb instruction.
25593 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25595 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
25597 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
25598 will anyway by matched by mov<mode>_64dfp.
25600 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25602 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
25603 vlef/vstef. Add missing operand to vleif.
25605 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25607 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
25608 pair for all vector types with 64 bit elements.
25609 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
25610 * config/s390/vector.md (V_HW_64): ... here.
25611 (V_128_NOSINGLE): New mode iterator.
25612 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
25613 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
25614 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
25615 ("*vec_load_pairv2di"): Change to ...
25616 ("*vec_load_pair<mode>"): ... this one.
25618 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25620 * config/s390/constraints.md: Add comments.
25621 (jKK): Reject element sizes > 8 bytes.
25622 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
25624 * config/s390/s390.md: Add the s_operand checks formerly in
25625 s390_split_ok_p to various splitters where they are still
25627 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
25628 for 128 bit vectors. Plus two splitters.
25630 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25632 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
25635 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25638 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
25639 error if the boundary argument is not constant.
25641 2017-03-24 Jakub Jelinek <jakub@redhat.com>
25643 PR rtl-optimization/80112
25644 * loop-doloop.c (doloop_condition_get): Don't check condition
25645 if cmp isn't SET with IF_THEN_ELSE src.
25647 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25649 PR tree-optimization/80158
25650 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
25651 replacing a candidate statement, also replace it for the
25652 candidate's alternate interpretation.
25653 (replace_rhs_if_not_dup): Likewise.
25654 (replace_one_candidate): Likewise.
25656 2017-03-24 Richard Biener <rguenther@suse.de>
25658 PR tree-optimization/80167
25659 * graphite-isl-ast-to-gimple.c
25660 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
25662 (translate_isl_ast_to_gimple::get_rename): Likewise.
25664 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
25666 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
25667 handling of certain combinations of target options, including the
25668 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
25669 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
25671 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25674 * config/arm/arm.md (*load_multiple): Add reload_completed to
25675 matching condition.
25677 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25678 Richard Biener <rguenth@suse.de>
25680 PR tree-optimization/79908
25681 PR tree-optimization/80136
25682 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
25683 been cast away, gimplify_and_add suffices.
25685 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
25687 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
25689 2017-03-23 Richard Biener <rguenther@suse.de>
25691 PR tree-optimization/80032
25692 * gimplify.c (gimple_push_cleanup): Forced unconditional
25693 cleanups still have to go to the conditional_cleanups
25696 2017-03-22 Jakub Jelinek <jakub@redhat.com>
25698 PR tree-optimization/80072
25699 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
25701 (next_operand_entry_id): Change type to unsigned int.
25702 (sort_by_operand_rank): Make sure to return the right return value
25703 even if unsigned fields are bigger than INT_MAX.
25704 (struct oecount): Change cnt and id type to unsigned int.
25705 (oecount_hasher::equal): Formatting fix.
25706 (oecount_cmp): Make sure to return the right return value
25707 even if unsigned fields are bigger than INT_MAX.
25708 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
25711 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
25712 TREE_READONLY on result if writing it more than once.
25715 * doc/invoke.texi (-fsanitize=thread): Document that with
25716 -fnon-call-exceptions atomics are not able to throw
25720 * tsan.c: Include tree-eh.h.
25721 (instrument_builtin_call): Call maybe_clean_eh_stmt or
25722 maybe_clean_or_replace_eh_stmt where needed.
25723 (instrument_memory_accesses): Add cfg_changed argument.
25724 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
25725 if it returned true.
25726 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
25728 PR rtl-optimization/63191
25729 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
25730 wrapper function, moved the whole old content into ...
25731 (ix86_delegitimize_address_1): ... this. New inline function.
25732 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
25733 true as last argument instead of ix86_delegitimize_address.
25735 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
25737 * config/aarch64/aarch64.c (generic_branch_cost): Copy
25738 cortexa57_branch_cost.
25740 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
25742 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
25744 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25747 * doc/md.texi (Constraints): Document wA constraint.
25748 * config/rs6000/constraints.md (wA): New.
25749 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
25750 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
25751 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
25752 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
25754 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
25757 * gimplify.c (is_oacc_declared): New function.
25758 (oacc_default_clause): Use it to set default flags for acc declared
25759 variables inside parallel regions.
25760 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
25761 declared variables.
25762 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
25763 declare attribute to any decl as necessary.
25765 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
25768 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
25769 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
25770 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
25771 (arm_arch_lpae): This.
25772 * config/arm/arm.c (arm_arch7ve): Rename into ...
25773 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
25774 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
25777 2017-03-22 Martin Liska <mliska@suse.cz>
25780 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
25781 error message instead of an ICE.
25783 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25785 * doc/extend.texi (6.11 Additional Floating Types): Revise.
25787 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
25789 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
25791 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25794 2017-03-21 Martin Sebor <msebor@redhat.com>
25796 * doc/extend.texi: Use "cannot" instead of "can't."
25797 * doc/hostconfig.texi: Same.
25798 * doc/install.texi: Same.
25799 * doc/invoke.texi: Same.
25800 * doc/loop.texi: Same.
25801 * doc/md.texi: Same.
25802 * doc/objc.texi: Same.
25803 * doc/rtl.texi: Same.
25804 * doc/tm.texi: Same.
25805 * doc/tm.texi.in: Same.
25806 * doc/trouble.texi: Same.
25808 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
25811 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
25812 (collect_checksum_attributes): Set it.
25813 (die_checksum_ordered): Use it.
25815 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25817 PR tree-optimization/79908
25818 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
25819 change: For a VA_ARG whose LHS has been cast away, use
25820 force_gimple_operand to construct the side effects.
25822 2017-03-21 David Malcolm <dmalcolm@redhat.com>
25824 PR translation/80001
25825 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
25826 more amenable to translation.
25827 (oacc_loop_auto_partitions): Likewise.
25829 2017-03-21 Marek Polacek <polacek@redhat.com>
25830 Martin Sebor <msebor@redhat.com>
25832 PR tree-optimization/80109
25833 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
25834 on INTEGRAL_TYPE_P.
25836 2017-03-21 Jakub Jelinek <jakub@redhat.com>
25837 Segher Boessenkool <segher@kernel.crashing.org>
25840 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
25841 check reg_used_between_p between insn and one of succ or succ2
25842 depending on if succ is artificial insn not inserted into insn
25845 2017-03-21 Martin Liska <mliska@suse.cz>
25847 PR gcov-profile/80081
25848 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
25849 * doc/gcc.texi: Include gcov-dump stuff.
25850 * doc/gcov-dump.texi: New file.
25852 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
25854 PR rtl-optimization/79150
25855 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
25856 conditional jump, if the jump is the last insn of the loop.
25858 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25859 Richard Biener <rguenth@suse.de>
25861 PR tree-optimization/79908
25862 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
25863 been cast away, use force_gimple_operand to construct the side
25866 2017-03-21 Martin Liska <mliska@suse.cz>
25868 PR libfortran/79956
25869 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
25872 2017-03-21 Brad Spengler <spender@grsecurity.net>
25875 * plugin.c (htab_hash_plugin): New function.
25876 (add_new_plugin): Use it and adjust.
25877 (parse_plugin_arg_opt): Adjust.
25878 (init_one_plugin): Likewise.
25880 2017-03-21 Richard Biener <rguenther@suse.de>
25882 PR tree-optimization/80032
25883 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
25884 if set force the cleanup to happen unconditionally.
25885 (gimplify_target_expr): Push inserted clobbers with force_uncond
25886 to avoid them being removed by control-dependent DCE.
25888 2017-03-21 Richard Biener <rguenther@suse.de>
25890 PR tree-optimization/80122
25891 * tree-inline.c (copy_bb): Do not expans va-arg packs or
25892 va_arg_pack_len when the inlined call stmt requires pack
25894 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
25896 2017-03-21 Jakub Jelinek <jakub@redhat.com>
25899 * tsan.c (instrument_builtin_call): If the memory model argument
25900 is not a constant, assume it is valid.
25903 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
25906 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
25908 PR rtl-optimization/79910
25909 * combine.c (can_combine_p): Do not allow combining an I0 or I1
25910 if its dest is used by an insn before I2 (other than the combined
25911 insns themselves, which are properly handled already).
25913 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
25916 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
25918 * combine.c (record_used_regs): New static function.
25919 (try_combine): Handle situations where there is an additional
25920 instruction between I2 and I3 which needs to have a LOG_LINK
25924 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
25926 * combine.c (try_combine): Delete redundant i1 test. Call
25927 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
25929 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25932 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
25933 alternatives 13/14.
25935 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25937 PR tree-optimization/80054
25938 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
25939 the optimization if a PHI or any of its arguments is not dominated
25940 by the candidate's basis. Use gphi* rather than gimple* as
25942 (replace_profitable_candidates): Clean up a gimple* variable that
25943 should be a gphi* variable.
25945 2017-03-20 Martin Sebor <msebor@redhat.com>
25948 * doc/extend.texi (attribute constructor): Document present limitation.
25950 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
25953 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
25954 __POWER9_VECTOR__ #ifdef control, change template definition to
25955 use Power9-specific built-in function.
25956 (vec_any_eq): Likewise.
25957 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
25958 to control outcomes from this test.
25959 (vector_ae_<mode>p): For VEC_F modes, likewise.
25961 2017-03-20 Ian Lance Taylor <iant@google.com>
25963 * config/i386/i386.c (ix86_function_regparm): Save an extra
25964 register for -fsplit-stack with DECL_STATIC_CHAIN.
25966 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
25969 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
25970 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
25972 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
25974 * config/riscv/riscv.c (riscv_print_operand): Use "fence
25976 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
25979 2017-03-20 Marek Polacek <polacek@redhat.com>
25982 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
25984 2017-03-20 Richard Biener <rguenther@suse.de>
25986 PR tree-optimization/80113
25987 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
25988 allocate extra SSA name for PHI def.
25989 (add_close_phis_to_outer_loops): Likewise.
25990 (add_close_phis_to_merge_points): Likewise.
25991 (copy_loop_close_phi_args): Likewise.
25992 (copy_cond_phi_nodes): Likewise.
25994 2017-03-20 Martin Liska <mliska@suse.cz>
25996 PR middle-end/79753
25997 * tree-chkp.c (chkp_build_returned_bound): Do not build
25998 returned bounds for a LHS that's not a BOUNDED_P type.
26000 2017-03-20 Martin Liska <mliska@suse.cz>
26004 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
26005 COMPLEX_CST and VECTOR_CST.
26007 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26010 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
26011 target operand. A new splitter adds the clobber statement in case
26012 the target operand is dead anyway.
26014 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
26016 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
26017 to age-old versions of binutils and glibc.
26019 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
26021 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
26023 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
26025 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
26027 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
26029 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
26030 requirement for binutils 2.13.
26032 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
26034 * combine.c (try_combine): Delete redundant i1 test. Call
26035 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
26037 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
26039 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
26040 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
26042 <riscv64-*-elf>: Re-arrange section
26043 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
26044 <riscv32-*-linux>: Likewise.
26045 <riscv64-*-elf>: Likewise
26046 <riscv64-*-linux>: Likewise.
26048 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
26051 * aarch64.opt(verbose-cost-dump): Fix typo.
26053 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
26056 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
26057 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
26059 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
26061 * reload.c (find_reloads): When reloading a nonoffsettable address,
26062 use RELOAD_OTHER for it and its address reloads.
26064 PR rtl-optimization/79910
26065 * combine.c (record_used_regs): New static function.
26066 (try_combine): Handle situations where there is an additional
26067 instruction between I2 and I3 which needs to have a LOG_LINK
26070 2017-03-17 Jeff Law <law@redhat.com>
26072 PR tree-optimization/71437
26073 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
26074 conditional in the hash table first.
26075 (vrp_dom_walker::before_dom_children): Extract condition from
26076 ASSERT_EXPR. Record condition, its inverion and any implied
26077 conditions as well.
26079 2017-03-17 Marek Polacek <polacek@redhat.com>
26080 Markus Trippelsdorf <markus@trippelsdorf.de>
26082 PR tree-optimization/80079
26083 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
26086 2017-03-17 Richard Biener <rguenther@suse.de>
26088 PR middle-end/80075
26089 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
26090 Properly verify the LHS before the RHS possibly claims to be
26092 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
26095 2017-03-17 Martin Jambor <mjambor@suse.cz>
26097 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
26098 (List of -O2 options): Likewise.
26099 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
26102 2017-03-17 Tom de Vries <tom@codesourcery.com>
26104 * gcov-dump.c (print_usage): Print bug_report_url.
26106 2017-03-17 Richard Biener <rguenther@suse.de>
26108 PR middle-end/80050
26109 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
26110 (parser::peek): Likewise.
26112 2017-03-17 Richard Biener <rguenther@suse.de>
26114 PR tree-optimization/80048
26115 * sese.c (free_sese_info): Properly release rename_map and
26116 copied_bb_map elements.
26118 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
26120 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
26121 Add linked-list forward and backlinks. Insert on
26122 construction, remove on destruction.
26123 (class pass_store_merging): Add m_stores_head field.
26124 (pass_store_merging::terminate_and_process_all_chains):
26125 Iterate over m_stores_head list.
26126 (pass_store_merging::terminate_all_aliasing_chains):
26128 (pass_store_merging::execute): Check for debug stmts first.
26129 Push new chains onto the m_stores_head stack.
26131 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
26134 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
26135 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
26136 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
26138 2017-03-16 Jeff Law <law@redhat.com>
26140 PR tree-optimization/71437
26141 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
26142 member function. Implementation moved into after_dom_children
26143 member function and into the threader's thread_outgoing_edges
26145 (dom_opt_dom_walker::after_dom_children): Simplify by moving
26146 some code into new thread_outgoing_edges.
26147 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
26148 definition. Simplify marker handling (do it here). Assume we always
26149 have the available expression and the const/copies tables.
26150 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
26152 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
26153 * tree-vrp.c (equiv_stack): No longer file scoped.
26154 (vrp_dom_walker): New class.
26155 (vrp_dom_walker::before_dom_children): New member function.
26156 (vrp_dom_walker::after_dom_children): Likewise.
26157 (identify_jump_threads): Setup domwalker. Use it rather than
26158 walking edges in a random order by hand. Simplify setup/finalization.
26159 (finalize_jump_threads): Remove.
26160 (vrp_finalize): Do not call identify_jump_threads here.
26161 (execute_vrp): Do it here instead and call thread_through_all_blocks
26164 PR tree-optimization/71437
26165 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
26167 (simplify_stmt_for_jump_threading): Add basic_block argument. All
26169 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
26170 (dom_opt_dom_walker::thread_across_edge): Remove
26171 handle_dominating_asserts argument. All callers changed.
26172 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
26173 changes. Remove calls to lhs_of_dominating_assert. Other
26174 uses of handle_dominating_asserts turn into unconditional code
26175 (simplify_control_stmt_condition_1): Likewise.
26176 (simplify_control_stmt_condition): Likewise.
26177 (thread_through_normal_block, thread_across_edge): Likewise.
26178 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
26179 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
26180 object if it is not an SSA_NAME.
26181 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
26182 before calling into the VRP specific simplifiers.
26183 (identify_jump_threads): Remove handle_dominating_asserts
26186 2017-03-16 Jakub Jelinek <jakub@redhat.com>
26189 * tree-diagnostic.c (default_tree_printer): No longer static.
26190 * tree-diagnostic.h (default_tree_printer): New prototype.
26192 2017-03-16 Tamar Christina <tamar.christina@arm.com>
26194 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
26195 Change ins into fmov.
26197 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26199 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
26200 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
26201 Use h_con constraint for operand 1.
26202 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
26203 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
26205 2017-03-15 Jeff Law <law@redhat.com>
26207 PR tree-optimization/71437
26208 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
26209 (record_temporary_equivalences): Use it.
26211 PR tree-optimization/71437
26212 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
26213 tree-ssa-scopedtables.
26214 (lookup_avail_expr, build_and_record_new_cond): Likewise.
26215 (record_conditions, record_cond, vuse_eq): Likewise.
26216 (record_edge_info): Adjust to API tweak of record_conditions.
26217 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
26218 (record_temporary_equivalences, optimize_stmt): Likewise.
26219 (eliminate_redundant_computations): Likewise.
26220 (record_equivalences_from_stmt): Likewise.
26221 * tree-ssa-scopedtables.c: Include options.h and params.h.
26222 (vuse_eq): New function, moved from tree-ssa-dom.c
26223 (build_and_record_new_cond): Likewise.
26224 (record_conditions): Likewise. Accept vector of conditions rather
26225 than edge_equivalence structure for first argument.
26226 for the first argument.
26227 (avail_exprs_stack::lookup_avail_expr): New member function, moved
26228 from tree-ssa-dom.c.
26229 (avail_exprs_stack::record_cond): Likewise.
26230 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
26231 from tree-ssa-dom.c.
26232 (avail_exprs_stack): Add new member functions lookup_avail_expr
26234 (record_conditions): Declare.
26236 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
26239 * lra-constraints.c (process_alt_operands): Increase reject for
26240 reloading an input/output operand.
26242 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
26245 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
26246 insns to convert from signed/unsigned char/short to IEEE 128-bit
26248 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
26250 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
26253 * config/i386/i386.c (ix86_vector_duplicate_value): Create
26254 subreg of inner mode for values already in registers.
26256 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
26258 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
26259 iteration reg is used after the loop.
26261 2017-03-14 Martin Sebor <msebor@redhat.com>
26263 PR tree-optimization/79800
26264 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
26265 precision in negative-positive range.
26266 (format_floating): Call non-const overload with adjusted precision.
26268 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26271 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
26274 2017-03-14 Martin Sebor <msebor@redhat.com>
26276 PR middle-end/80020
26277 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
26278 * builtins.def (aligned_alloc): Use it.
26281 * Makefile.in (GTFILES): Add calls.c.
26282 * calls.c: Include "gt-calls.h".
26284 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
26286 PR rtl-optimization/79728
26287 * regs.h (struct target_regs): New field
26288 x_contains_allocatable_regs_of_mode.
26289 (contains_allocatable_regs_of_mode): New macro.
26290 * reginfo.c (init_reg_sets_1): Initialize it, and change
26291 contains_reg_of_mode so it includes global regs as well.
26292 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
26293 rather than contains_regs_of_mode.
26295 2017-03-14 Martin Liska <mliska@suse.cz>
26297 * doc/invoke.texi: Document options that can't be combined with
26298 -fcheck-pointer-bounds.
26300 2017-03-14 Martin Liska <mliska@suse.cz>
26302 PR middle-end/79831
26303 * doc/invoke.texi (-Wchkp): Document the option.
26305 2017-03-14 Martin Liska <mliska@suse.cz>
26307 * Makefile.in: Install gcov-dump.
26309 2017-03-14 Martin Liska <mliska@suse.cz>
26311 * multiple_target.c (expand_target_clones): Bail out for
26312 an invalid attribute.
26314 2017-03-14 Richard Biener <rguenther@suse.de>
26316 * alias.c (struct alias_set_entry): Pack properly.
26317 * cfgloop.h (struct loop): Likewise.
26318 * cse.c (struct set): Likewise.
26319 * ipa-utils.c (struct searchc_env): Likewise.
26320 * loop-invariant.c (struct invariant): Likewise.
26321 * lra-remat.c (struct cand): Likewise.
26322 * recog.c (struct change_t): Likewise.
26323 * rtl.h (struct address_info): Likewise.
26324 * symbol-summary.h (function_summary): Likewise.
26325 * tree-loop-distribution.c (struct partition): Likewise.
26326 * tree-object-size.c (struct object_size_info): Likewise.
26327 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
26328 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
26329 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
26330 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
26331 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
26332 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
26333 (struct _stmt_vec_info): Likewise.
26335 2017-03-14 Martin Liska <mliska@suse.cz>
26338 * multiple_target.c (create_dispatcher_calls): Check that
26339 a target can create a function dispatcher.
26341 2017-03-14 Martin Liska <mliska@suse.cz>
26344 * multiple_target.c (expand_target_clones): Drop local.local
26345 flag for default implementation.
26347 2017-03-14 Richard Biener <rguenther@suse.de>
26349 PR tree-optimization/80030
26350 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
26352 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
26354 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
26355 gcc_fallthrough() instead of __attribute__((fallthrough));
26357 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
26359 * doc/gcc.texi: Remove "up" link to (DIR).
26360 * doc/gccint.texi: Ditto.
26362 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
26364 * doc/install.texi (Specific) <avr>: Remove reference to
26367 2017-03-13 Jeff Law <law@redhat.com>
26369 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
26370 attribute rather than comments.
26372 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
26373 match_scratch operand is highest.
26375 2017-03-13 Martin Liska <mliska@suse.cz>
26377 PR middle-end/78339
26378 * ipa-pure-const.c (warn_function_noreturn): If the declarations
26379 is a CHKP clone, use original declaration.
26381 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26383 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
26384 (arc_conditional_register_usage): Use a different allocation order
26385 when optimizing for size.
26386 * common/config/arc/arc-common.c (arc_option_optimization_table):
26387 Section anchors default on when optimizing for size.
26389 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26391 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
26393 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26395 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
26396 * config/arc/arc.md (cpu_facility): Add cd variant.
26397 (*movqi_insn): Add code density variant.
26398 (*movhi_insn): Likewise.
26399 (*movqi_insn): Likewise.
26400 (*addsi3_mixed): Likewise.
26401 (subsi3_insn): Likewise.
26403 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26405 * config/arc/arc.md (movsi_cond_exec): Update constraint.
26407 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26409 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
26410 expressions with MINUS and UNARY ops.
26412 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26415 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
26417 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
26418 between vec_select and vector argument.
26419 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
26420 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
26421 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
26422 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
26423 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
26424 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
26426 2017-03-13 Richard Biener <rguenther@suse.de>
26429 * params.def (vect-max-peeling-for-alignment): Fix typo.
26431 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26433 * doc/install.texi (Specific) <mips-*-*>: Remove description of
26434 issue that only occurred with binutils below 2.18.
26436 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26438 * doc/install.texi (Specific) <cris-axis-elf>: No longer
26439 refer to binutils 2.11/2.12 minimum.
26441 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26443 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
26444 ftp.kernel.org and simplify binutils requirement.
26446 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
26448 * doc/invoke.texi (Warning Options): Fix spelling of link-time
26450 (Optimize Options): Ditto. Also remove redundancy.
26452 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26454 PR translation/79848
26455 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
26457 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
26458 to G_ to avoid double translation.
26460 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26462 PR translation/79923
26463 * auto-profile.c (get_combined_location): Convert leading
26464 character of diagnostics to lower case and remove trailing period.
26465 (read_profile): Likewise for various diagnostics.
26466 * config/arm/arm.c (arm_option_override): Remove trailing period
26467 from various diagnostics.
26468 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
26469 (msp430_expand_delay_cycles): Likewise.
26471 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26474 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
26475 full command-line argument, rather than just "str".
26476 (aarch64_validate_march): Likewise.
26477 (aarch64_validate_mtune): Likewise.
26479 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
26481 PR rtl-optimization/78911
26482 * lra-assigns.c (must_not_spill_p): New function.
26483 (spill_for): Use it.
26485 2017-03-10 Jakub Jelinek <jakub@redhat.com>
26487 PR tree-optimization/79981
26488 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
26489 ATOMIC_COMPARE_EXCHANGE ifn result.
26490 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
26491 IFN_ATOMIC_COMPARE_EXCHANGE.
26493 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26496 * opts.c (parse_sanitizer_options): Add missing question mark to
26497 "did you mean" message.
26499 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26501 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
26503 (VMULEUH_UNS): Likewise.
26504 (VMULOUB_UNS): Likewise.
26505 (VMULOUH_UNS): Likewise.
26506 * config/rs6000/rs6000.c (builtin_function_type): Remove
26507 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
26509 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26512 * read-rtl-function.c (function_reader::read_rtx_operand): Update
26513 x with result of extra_parsing_for_operand_code_0.
26514 (function_reader::extra_parsing_for_operand_code_0): Convert
26515 return type from void to rtx, returning x. When reading
26516 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
26517 larger size containing struct block_symbol.
26519 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
26521 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
26522 -mfloat128-hardware without -m64.
26524 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
26527 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
26528 entries to the case statement that marks unsigned arguments to
26529 overloaded functions.
26531 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26533 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
26534 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
26536 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
26539 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
26540 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
26542 2017-03-10 Martin Liska <mliska@suse.cz>
26546 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
26547 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
26550 2017-03-10 Olivier Hainque <hainque@adacore.com>
26552 * tree-switch-conversion (array_value_type): Start by resetting
26553 candidate type to it's main variant.
26555 2017-03-10 Jakub Jelinek <jakub@redhat.com>
26557 PR rtl-optimization/79909
26558 * combine.c (try_combine): Use simplify_replace_rtx on individual
26559 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
26560 of the whole CALL_INSN_FUNCTION_USAGE.
26562 PR tree-optimization/79972
26563 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
26564 get_range_info on SSA_NAMEs. Formatting fixes.
26566 2017-03-10 Richard Biener <rguenther@suse.de>
26567 Jakub Jelinek <jakub@redhat.com>
26569 PR tree-optimization/77975
26570 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
26571 edge to be constant.
26572 (get_val_for): For constant x return it. Formatting fix.
26573 (loop_niter_by_eval): Avoid pointless looping if the next iteration
26574 would use the same bases as the current one.
26576 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26578 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
26579 instead of vec_select for V1TImode.
26580 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
26582 (VSX_LE_128): Add V1TI to this mode iterator.
26583 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
26584 (*vsx_le_perm_store_<mode>): Likewise.
26585 (pre-reload splitter for VSX stores): Likewise.
26586 (post-reload splitter for VSX stores): Likewise.
26587 (*vsx_xxpermdi2_le_<mode>): Likewise.
26588 (*vsx_lxvd2x2_le_<mode>): Likewise.
26589 (*vsx_stxvd2x2_le_<mode>): Likewise.
26591 2017-03-09 Michael Eager <eager@eagercon.com>
26593 Correct failures with --enable-checking=yes,rtl.
26595 * config/microblaze/microblaze.c (microblaze_expand_shift):
26596 Replace GET_CODE test with CONST_INT_P and INTVAL test with
26597 test for const0_rtx.
26598 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
26599 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
26601 2017-03-09 Richard Biener <rguenther@suse.de>
26603 PR tree-optimization/79977
26604 * graphite-scop-detection.c (scop_detection::merge_sese):
26605 Handle the case of extra exits to blocks dominating the entry.
26607 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
26609 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26612 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
26614 PR rtl-optimization/79949
26615 * lra-constraints.c (process_alt_operands): Check memory when
26616 trying to predict a cycle. Print about the overall increase.
26618 2017-03-09 Richard Biener <rguenther@suse.de>
26620 PR middle-end/79971
26621 * gimple-expr.c (useless_type_conversion_p): Preserve
26622 TYPE_SATURATING for fixed-point types.
26624 2017-03-09 Richard Biener <rguenther@suse.de>
26627 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
26628 alignment of BLKmode params.
26630 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26633 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
26634 (VALL_NO_V2Q): Likewise.
26636 * config/aarch64/aarch64-simd.md
26637 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
26639 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
26640 VALL_NO_V2Q mode iterator.
26641 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
26643 2017-03-09 Martin Liska <mliska@suse.cz>
26645 PR tree-optimization/79631
26646 * tree-chkp-opt.c (chkp_is_constant_addr): Call
26647 tree_int_cst_sign_bit just for INTEGER constants.
26649 2017-03-09 Martin Liska <mliska@suse.cz>
26653 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
26656 2017-03-09 Marek Polacek <polacek@redhat.com>
26659 * tree.c (inchash::add_expr): Handle TREE_VEC.
26661 2017-03-09 Martin Liska <mliska@suse.cz>
26664 (chkp_narrow_size_and_offset): New function.
26665 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
26666 (void chkp_parse_bit_field_ref): New function.
26667 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
26668 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
26670 2017-03-09 Martin Liska <mliska@suse.cz>
26673 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
26674 (chkp_find_bounds_1): Remove gcc_unreachable.
26676 2017-03-09 Jakub Jelinek <jakub@redhat.com>
26679 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
26680 BUILT_IN_SYNC*, determine the access type from the size suffix and
26681 always build a MEM_REF with that type. Handle forgotten
26682 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
26685 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
26686 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
26687 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
26688 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
26689 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
26690 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
26691 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
26692 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
26693 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
26694 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
26695 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
26696 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
26697 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
26698 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
26699 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
26700 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
26701 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
26702 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
26703 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
26704 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
26705 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
26706 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
26707 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
26708 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
26709 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
26710 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
26711 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
26712 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
26713 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
26714 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
26715 definitions outside of __OPTIMIZE__ guarded section.
26718 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
26719 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
26720 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
26721 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
26724 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26726 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
26727 ("vfenez<mode>"): Add missing constraints.
26729 2017-03-08 Martin Sebor <msebor@redhat.com>
26732 * config/nds32/nds32.c (nds32_option_override):
26733 Fix misspelled diagnostic.
26735 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26738 * gimplify.c (gimplify_omp_for): Replace index var in outer
26739 taskloop statement with an artificial variable and add
26740 OMP_CLAUSE_PRIVATE clause for it.
26742 2017-03-08 Richard Biener <rguenther@suse.de>
26744 PR tree-optimization/79955
26745 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
26746 for accesses that are completely outside of the variable.
26748 2017-03-08 Andrew Haley <aph@redhat.com>
26750 PR tree-optimization/79943
26751 * tree-ssa-loop-split.c (compute_new_first_bound): When
26752 calculating the new upper bound, (END-BEG) should be added, not
26755 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26757 * config/avr/avr.md (setmemhi): Make sure match_dup
26758 operand number comes before match_scratch.
26760 2017-03-08 Richard Biener <rguenther@suse.de>
26762 PR tree-optimization/79920
26763 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
26764 with ncopies == 1 to ...
26765 (vect_transform_slp_perm_load): ... here. Properly compute
26766 all element loads by iterating VF times over the group. Do
26767 not handle ncopies (computed in a broken way) in
26768 vect_create_mask_and_perm.
26770 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26773 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
26774 is a uniform vector, use uniform_vector_p return value instead of
26775 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
26777 2017-03-07 Marek Polacek <polacek@redhat.com>
26779 PR middle-end/79809
26780 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
26781 (alloca_call_type): Likewise.
26783 2017-03-07 Martin Liska <mliska@suse.cz>
26785 * gcov.c (process_args): Put comment to correct location.
26787 2017-03-07 Martin Liska <mliska@suse.cz>
26789 PR middle-end/68270
26790 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
26791 Use array_at_struct_end_p instead of DECL_CHAIN (field).
26792 (chkp_narrow_bounds_for_field): Likewise.
26793 (chkp_parse_array_and_component_ref): Pass one more argument to
26796 2017-03-07 Richard Biener <rguenther@suse.de>
26798 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
26801 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
26803 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
26804 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
26806 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26809 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
26810 to end of description.
26811 (PARAM_MAX_STORES_TO_MERGE): Likewise.
26813 2017-03-07 Jakub Jelinek <jakub@redhat.com>
26815 PR rtl-optimization/79901
26816 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
26818 (*avx512f_<code><mode>3<mask_name>): ... this.
26819 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
26820 iterator instead of VI8_AVX2_AVX512BW.
26822 PR rtl-optimization/79901
26823 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
26824 min/max expander, expand it using expand_vec_cond_expr.
26827 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
26830 2017-03-06 Jakub Jelinek <jakub@redhat.com>
26833 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
26834 to void * for PCH reasons.
26835 * dwarf2out.c (output_loc_operands, output_die): Cast
26836 v.val_vec.array to unsigned char *.
26838 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
26841 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
26844 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
26846 PR rtl-optimization/79571
26847 * lra-constraints.c (process_alt_operands): Calculate static
26848 reject and subtract it from overall when only addresses will be
26851 2017-03-06 Julia Koval <julia.koval@intel.com>
26854 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
26855 incoming stack boundary to 128 for 64-bit targets.
26857 2017-03-06 Richard Biener <rguenther@suse.de>
26859 PR tree-optimization/79894
26860 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
26861 to NULL after folding it.
26863 2017-03-06 Richard Biener <rguenther@suse.de>
26865 PR tree-optimization/79824
26866 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
26867 check disabling peeling for gaps.
26869 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
26871 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
26872 attributes): Document gettimeofday.
26874 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
26876 * config/s390/s390.c (s390_option_override_internal): Set
26877 PARAM_MIN_VECT_LOOP_BOUND
26879 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
26881 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
26882 * config/s390/s390.md: Likewise.
26884 2017-03-06 Jakub Jelinek <jakub@redhat.com>
26887 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
26888 (<avx2_avx512>_perm<mode>): Rename to ...
26889 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
26891 (<avx512>_perm<mode>_mask): Rename to ...
26892 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
26894 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
26895 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
26896 instead of VI8F_256_512.
26897 (avx512f_perm<mode>): New define_expand.
26898 (avx512f_perm<mode>_mask): Likewise.
26899 (avx512f_perm<mode>_1<mask_name>): New define_insn.
26900 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
26902 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26904 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
26905 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
26907 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
26909 2017-03-06 Martin Liska <mliska@suse.cz>
26912 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
26913 when having a SSA NAME w/o VAR_DECL assigned to it.
26915 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26917 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
26918 msa_dpsub_<su>_d): Fix MODE for vec_select.
26920 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26922 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
26924 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
26926 2017-03-06 Richard Biener <rguenther@suse.de>
26928 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
26929 * plugin.c (register_plugin_info): Likewise.
26930 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
26932 2017-03-05 Jakub Jelinek <jakub@redhat.com>
26934 * config/i386/sse.md (sse_storehps, sse_storelps,
26935 avx_<castmode><avxsizesuffix>_<castmode>,
26936 avx512f_<castmode><avxsizesuffix>_<castmode>,
26937 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
26938 in condition that at least one operand is not a MEM.
26940 2017-03-03 Jakub Jelinek <jakub@redhat.com>
26942 PR middle-end/79805
26943 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
26944 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
26946 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
26947 gimple_call_nothrow_p flag based on whether original builtin can throw.
26948 If it can, emit following stmts on the fallthrough edge.
26949 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
26950 don't create new bb if inserting just debug stmts on the edge, try to
26951 insert them on the fallthru bb or just reset debug stmts.
26953 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
26956 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
26957 restore recog_data (including the operand rtxes inside it) around
26958 the call to get_insn_template.
26960 2017-03-03 Martin Sebor <msebor@redhat.com>
26962 PR tree-optimization/79699
26963 * context.c (context::~context): Free MPFR caches to avoid
26964 a memory leak on program exit.
26966 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26968 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
26969 Use wide_int::ulow () instead of .elt (0).
26971 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
26973 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
26974 (*pushxf): Limit oF constraint to 32bit targets and add oC
26975 constraint for 64bit targets.
26976 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
26977 (*pushdf): Change rmF constraint to rmC.
26979 2017-03-03 Martin Liska <mliska@suse.cz>
26981 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
26982 Remove unused variable.
26984 2017-03-03 Jakub Jelinek <jakub@redhat.com>
26987 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
26988 is a memory operand, increase num_memory.
26989 (ix86_expand_args_builtin): Likewise.
26991 2017-03-03 Jan Hubicka <jh@suse.cz>
26994 * ipa-devirt.c (maybe_record_node): Properly handle
26995 __cxa_pure_virtual visibility.
26997 2017-03-03 Martin Liska <mliska@suse.cz>
26999 PR tree-optimization/79803
27000 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
27002 (pass_loop_prefetch::execute): Disabled optimization if an
27003 assumption about L1 cache size is not met.
27005 2017-03-03 Martin Liska <mliska@suse.cz>
27007 PR rtl-optimization/79574
27008 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
27009 (hash_scan_set): Likewise.
27010 (dump_hash_table): Likewise.
27011 (hoist_code): Likewise.
27013 2017-03-03 Richard Biener <rguenther@suse.de>
27015 * fixed-value.c (fixed_from_string): Restore use of elt (1)
27016 in place of uhigh ().
27017 (fixed_convert_from_real): Likewise.
27019 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
27022 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
27024 2017-03-03 Richard Biener <rguenther@suse.de>
27026 PR middle-end/79818
27027 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
27028 TYPE_OVERFLOW_UNDEFINED check.
27030 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27032 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
27034 (vector_ae_<mode>_p): Likewise.
27035 (vector_nez_<mode>_p): Likewise.
27036 (vector_ne_v2di_p): Likewise.
27037 (vector_ae_v2di_p): Likewise.
27038 (vector_ne_<mode>_p): Likewise.
27039 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
27041 (vsx_tsqrt<mode>2_fe): Likewise.
27043 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
27046 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
27048 2017-03-02 Jakub Jelinek <jakub@redhat.com>
27050 PR rtl-optimization/79780
27051 * cprop.c (one_cprop_pass): When second and further conditional trap
27052 in a single basic block is turned into an unconditional trap, turn it
27053 into a deleted note to avoid RTL verification failures.
27055 2017-03-02 Richard Biener <rguenther@suse.de>
27057 * fold-const.c (const_binop): Use ulow () instead of elt (0).
27059 2017-03-02 Richard Biener <rguenther@suse.de>
27061 PR tree-optimization/79345
27063 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
27064 param and abort the walk, returning -1 if it is hit.
27065 (walk_aliased_vdefs): Take a limit param and pass it on.
27066 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
27067 defaulting to 0 and return a signed int.
27068 * tree-ssa-uninit.c (struct check_defs_data): New struct.
27069 (check_defs): New helper.
27070 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
27071 about uninitialized memory.
27072 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
27073 bogus uninitialized warning.
27074 (fixed_convert_from_real): Likewise.
27076 2017-03-02 Bin Cheng <bin.cheng@arm.com>
27078 PR tree-optimization/66768
27079 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
27080 iv_use if base object can't be determined.
27082 2017-03-02 Jakub Jelinek <jakub@redhat.com>
27084 PR tree-optimization/79345
27085 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
27086 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
27087 (get_pattern_stats): Initialize it.
27088 * genemit.c (gen_expand): Verify match_scratch numbers come after
27089 match_operand/match_dup numbers.
27090 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
27091 match_scratch numbers.
27092 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
27094 * config/s390/s390.md (trunctdsd2): Likewise.
27096 2017-03-02 Richard Biener <rguenther@suse.de>
27098 * wide-int.h (wide_int_storage::operator=): Implement in terms
27101 2017-03-02 Richard Biener <rguenther@suse.de>
27103 PR tree-optimization/79777
27104 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
27105 the to insert expression to sth existing.
27107 2017-03-01 Martin Sebor <msebor@redhat.com>
27109 PR middle-end/79692
27110 * gimple-ssa-sprintf.c
27111 (directive::known_width_and_precision): New function.
27112 (format_integer): Use it.
27113 (get_mpfr_format_length): Consider the full range of precision
27114 when computing %g output with the # flag. Set the likely byte
27115 count to 3 rather than 1 when precision is indeterminate.
27116 (format_floating): Correct the lower bound of precision.
27118 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27120 * doc/invoke.texi: Document default code model for 64-bit Linux.
27122 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27125 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
27126 udiv rather than div since input pattern is unsigned.
27128 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
27130 * config/i386/i386.c (print_reg): Warn for values of
27131 unsupported size in integer register.
27133 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
27136 * config/rs6000/predicates.md (current_file_function_operand): Do
27137 not allow self calls to be local if the function is replaceable.
27139 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
27142 * config/rs6000/altivec.h (vec_ctz and others): Change the
27143 preprocessor macro that controls conditional compilation from
27144 _ARCH_PWR9 to __POWER9_VECTOR__.
27145 (vec_all_ne): Change parameterization of __altivec_scalar_pred
27146 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
27147 control (instead of _ARCH_PWR9 control) so that template
27148 definition uses power9-specific function.
27149 (vec_any_eq): Likewise.
27150 (vec_all_ne): Change macro definition to use a power9-specific
27151 expansion under #ifdef __POWER9_VECTOR__ control (instead of
27152 _ARCH_PWR9 control).
27153 (vec_any_eq) Likewise.
27154 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
27155 expansion for CMPNEF to remove support for xvcmpnesp instruction.
27156 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
27157 support for xvcmpnedp instruction.
27158 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
27159 macro expansion so that Power9 implementation of vec_all_ne does
27160 not use the AltiVec predicate framework.
27161 (VCMPNEH_P): Likewise.
27162 (VCMPNEW_P): Likewise.
27163 (VCMPNED_P): Likewise.
27164 (VCMPNEFP_P): Likewise.
27165 (VCMPNEDP_P): Likewise.
27166 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
27167 implementation of vec_any_eq to not use AltiVec predicate
27169 (VCMPAEH_P): Likewise.
27170 (VCMPAEW_P): Likewise.
27171 (VCMPAED_P): Likewise.
27172 (VCMPAEFP_P): Likewise.
27173 (VCMPAEDP_P): Likewise.
27174 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
27175 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
27176 not use the AltiVec predicate framework.
27177 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
27178 of vec_any_eq to not use AltiVec predicate framework.
27179 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27180 support for predefined __POWER9_VECTOR__ macro to indicate that
27181 Power9 instruction selection is enabled.
27182 (altivec_overloaded_builtins): Remove extraneous
27183 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
27184 function argument types RS6000_BTI_bool_V16QI and
27185 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
27186 entry for overloaded function argument types RS6000_BTI_bool_V4SI
27187 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
27188 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
27189 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
27190 Power9 for implementations of vec_cmpne. Change the signature for
27191 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
27192 (representing vec_all_ne) to remove the previously described first
27193 argument of type RS6000_BTI_INTSI, as this was an artifact of
27194 reliance on the AltiVec predicate framework, which is no longer
27195 used in the implementation of these functions. Add
27196 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
27197 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
27198 since, unlike the AltiVec predicate framework implementation, we
27199 do not share function descriptors between vec_alle and vec_anyeq.
27200 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
27201 set of modes that receive special treatment even when
27202 TARGET_P9_VECTOR is true. The special treatment emits code that
27203 does not depend on Power9 instructions.
27204 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
27205 define_expand to not rely on AltiVec predicate framework.
27206 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
27208 (vector_ne_v2di_p): Change this define_expand to not rely on
27209 AltiVec predicate framework.
27210 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
27212 (vector_ne_<mode>_p): Change this define_expand to not rely on
27213 AltiVec predicate framework.
27214 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
27216 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
27217 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
27218 define_insn pattern.
27219 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
27220 define_insn pattern because the xvcmpne<VSs>. instruction is not
27222 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
27223 instruction is not supported.
27225 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27227 * config/nvptx/nvptx.c: Include intl.h.
27229 2017-03-01 Martin Jambor <mjambor@suse.cz>
27232 * ipa-prop.h (ipa_bits): Removed field known.
27233 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
27234 to pointers. Adjusted their comments to warn about their sharing.
27235 (ipcp_transformation_summary): Change bits to a vector of pointers.
27236 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
27237 (ipa_get_ipa_bits_for_value): Declare.
27238 * tree-vrp.h (value_range): Mark as GTY((for_user)).
27239 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
27240 (ipa_bits_hash_table): Likewise.
27241 (ipa_vr_ggc_hash_traits): Likewise.
27242 (ipa_vr_hash_table): Likewise.
27243 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
27244 being pointers and vr_known being removed.
27245 (ipa_set_jf_unknown): Likewise.
27246 (ipa_get_ipa_bits_for_value): New function.
27247 (ipa_set_jfunc_bits): Likewise.
27248 (ipa_get_value_range): New overloaded functions.
27249 (ipa_set_jfunc_vr): Likewise.
27250 (ipa_compute_jump_functions_for_edge): Use the above functions to
27251 construct bits and vr parts of jump functions.
27252 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
27253 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
27255 (ipcp_grow_transformations_if_necessary): Also allocate
27256 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
27258 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
27259 them. Fix too long lines.
27260 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
27261 vr_known being removed.
27262 (ipa_read_jump_function): Use new setter functions to construct bits
27263 and vr parts of jump functions or set them to NULL.
27264 (write_ipcp_transformation_info): Adjust for bits being pointers.
27265 (read_ipcp_transformation_info): Likewise.
27266 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
27268 Include gt-ipa-prop.h.
27269 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
27271 (ipcp_store_bits_results): Likewise.
27272 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
27273 Do not write to existing jump functions but use a temporary instead.
27275 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27278 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
27279 attempt to use its first operand as BIT_FIELD_REF base.
27281 2017-03-01 Richard Biener <rguenther@suse.de>
27283 PR middle-end/79721
27284 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
27285 interpolating formula in wrapping arithmetic.
27286 (chrec_apply): Convert chrec_evaluate return value to wanted type.
27288 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27290 PR tree-optimization/79734
27291 * tree-vect-generic.c (expand_vector_condition): Optimize
27292 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
27293 Handle VEC_COND_EXPR where comparison has different inner width from
27294 type's inner width.
27296 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
27298 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
27299 markup, and similar issues. Remove @opindex entries for things
27300 that aren't options. Add missing -mmpy-option entries.
27302 2017-02-28 Jakub Jelinek <jakub@redhat.com>
27304 PR tree-optimization/79737
27305 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
27306 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
27307 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
27308 instead of byte_size. Formatting fix.
27309 (shift_bytes_in_array_right): Formatting fix.
27311 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
27314 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
27315 condition on optimize for the leaf function test.
27317 2017-02-28 Martin Liska <mliska@suse.cz>
27320 * read-rtl-function.c (function_reader::handle_unknown_directive):
27321 Bail out when one uses -flto.
27323 2017-02-28 Martin Liska <mliska@suse.cz>
27325 * common.opt: Replace space with tabular for options of <number>
27327 * config/i386/i386.opt: Show <number> value for
27328 -mlarge-data-threshold.
27329 * opts.c (print_filtered_help): Do not display number in hexadecimal
27332 2017-02-28 Martin Liska <mliska@suse.cz>
27334 * common.opt: Fix --help=option -Q for options which are of
27337 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
27339 * config/i386/i386.c (print_reg): Error out for values
27340 of 8-bit size in invalid integer register.
27342 2017-02-28 Martin Sebor <msebor@redhat.com>
27344 PR tree-optimization/79691
27345 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
27347 2017-02-28 Jakub Jelinek <jakub@redhat.com>
27350 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
27351 gcc_unreachable with output_operand_lossage.
27353 2017-02-28 Richard Biener <rguenther@suse.de>
27355 PR tree-optimization/79740
27356 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
27358 (visit_nary_op): Insert the nary into the hashtable if we
27359 pattern-matched sth.
27360 * tree-ssa-pre.c (eliminate_insert): Robustify.
27362 2017-02-28 Richard Biener <rguenther@suse.de>
27364 PR middle-end/79731
27365 * fold-const.c (decode_field_reference): Reject out-of-bound
27368 2017-02-28 Jakub Jelinek <jakub@redhat.com>
27370 * config/i386/i386.c: Include intl.h.
27371 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
27372 instead of just cond ? "..." : "...".
27373 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
27374 * coverage.c (read_counts_file): Likewise.
27375 * omp-offload.c: Include intl.h.
27376 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
27377 of just cond ? "..." : "...".
27378 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
27379 of just cond ? "..." : "...".
27381 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
27384 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
27386 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
27387 'tune for' CPU name.
27388 * config/arm/arm-cpu-data.h: Regenerated.
27390 2017-02-28 Richard Biener <rguenther@suse.de>
27392 PR tree-optimization/79732
27393 * tree-inline.c (expand_call_inline): Do not shadow var.
27395 2017-02-28 Richard Biener <rguenther@suse.de>
27397 PR tree-optimization/79723
27398 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
27399 address-space properly.
27401 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
27403 * doc/optinfo.texi (Optimization groups): Fix option used for
27405 * doc/invoke.texi (-fopt-info): Document "omp".
27406 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
27407 (OPTGROUP_ALL): Add OPTGROUP_OMP.
27408 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
27409 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
27410 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
27412 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
27414 * dumpfile.c (optgroup_options): Instead of "openmp", associate
27415 OPTGROUP_OMP with "omp".
27417 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
27420 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
27421 for arithmetic shift of unsigned V2DI.
27423 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
27425 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
27426 arc/linux.h headers.
27427 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
27428 (LINK_SPEC): Likewise.
27429 (ARC_TLS_EXTRA_START_SPEC): Likewise.
27430 (EXTRA_SPECS): Likewise.
27431 (STARTFILE_SPEC): Likewise.
27432 (ENDFILE_SPEC): Likewise.
27433 (LIB_SPEC): Likewise.
27434 (TARGET_SDATA_DEFAULT): Likewise.
27435 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
27436 (MULTILIB_DEFAULTS): Likewise.
27437 (DWARF2_UNWIND_INFO): Likewise.
27438 * config/arc/big.h: New file.
27439 * config/arc/elf.h: Likewise.
27440 * config/arc/linux.h: Likewise.
27441 * config/arc/t-uClibc: Remove.
27443 2017-02-27 Bin Cheng <bin.cheng@arm.com>
27445 PR tree-optimization/77536
27446 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
27447 (tree_transform_and_unroll_loop): Use above function to compute the
27448 estimated niter of unrolled loop and use it when scaling profile.
27449 Also use count info rather than frequency if it's non-zero.
27450 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
27451 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
27452 (vect_transform_loop): Call above function.
27454 2017-02-27 Richard Biener <rguenther@suse.de>
27456 PR tree-optimization/45397
27457 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
27458 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
27459 (visit_nary_op): Add pattern matching for CSEing sign-changed
27460 or truncated operations with wider ones.
27462 2017-02-27 Richard Biener <rguenther@suse.de>
27464 PR tree-optimization/79690
27465 * tree-vect-stmts.c (vectorizable_store): Use vector type
27466 built from the DR with address-space.
27468 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
27470 * doc/invoke.texi (Optimize Options): Refine the description
27471 of asan-use-after-return.
27473 2017-02-25 Alan Modra <amodra@gmail.com>
27475 PR rtl-optimization/79584
27476 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
27477 base, not ad->base_term, the reg within base. Remove assertion
27478 that ad->base == ad->base_term. Replace gen_int_mode using
27479 bogus mode with const0_rtx.
27481 2017-02-25 Jakub Jelinek <jakub@redhat.com>
27483 PR middle-end/79396
27484 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
27485 FMA_EXPR like tcc_binary or tcc_unary.
27487 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
27490 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
27492 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
27493 (output_loc_operands): Handle DW_OP_call_ref and
27494 DW_OP_GNU_variable_value.
27495 (struct variable_value_struct): New type.
27496 (struct variable_value_hasher): Likewise.
27497 (variable_value_hash): New variable.
27498 (string_types): Remove.
27499 (copy_loc_descr): New function.
27500 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
27501 (prepend_loc_descr_to_each): New function.
27502 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
27503 instead of add_loc_descr_to_each if the first argument is single
27504 location list and the second has multiple.
27505 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
27506 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
27507 when looking for variable value which doesn't have other location info.
27508 (loc_list_from_tree): Formatting fix.
27509 (gen_array_type_die): Simplify DW_AT_string_length handling.
27510 (adjust_string_types): Remove.
27511 (gen_subprogram_die): Don't call adjust_string_types nor test/set
27512 string_types. Call resolve_variable_values.
27513 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
27514 (resolve_addr_in_expr): Likewise. Add A argument.
27515 (copy_deref_exprloc): Remove deref argument. Adjust for the
27516 original expression being DW_OP_GNU_variable_value with optionally
27517 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
27518 optionally after it.
27519 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
27520 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
27521 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
27522 (variable_value_hasher::hash, variable_value_hasher::equal): New
27524 (resolve_variable_value_in_expr, resolve_variable_value,
27525 resolve_variable_values, note_variable_value_in_expr,
27526 note_variable_value): New functions.
27527 (dwarf2out_early_finish): Call note_variable_value on all toplevel
27530 2017-02-24 Jakub Jelinek <jakub@redhat.com>
27533 * opts.h (handle_generated_option): Add GENERATED_P argument.
27534 * opts-common.c (handle_option): Adjust function comment.
27535 (handle_generated_option): Add GENERATED_P argument, pass it to
27537 (control_warning_option): Pass false to handle_generated_option
27539 * opts.c (maybe_default_option): Pass true to handle_generated_option
27541 * optc-gen.awk: Likewise.
27543 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27545 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
27546 a REG, look at the REG it is a SUBREG of.
27547 (splitter for cmpeqsi_t): Ditto.
27549 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27551 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
27552 the special USEs with the pattern of the insn, not the insn itself.
27554 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
27557 * doc/invoke.texi: Document -mload-store-pairs.
27559 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27560 Sandra Loosemore <sandra@codesourcery.com>
27562 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
27563 argument isn't a CONST_INT.
27564 (nios2_alternate_compare_const): Assert op is a CONST_INT.
27565 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
27566 (nios2_validate_compare): Bypass alternate compare logic if *op2
27567 is not a CONST_INT.
27568 (ldstwm_operation_p): Return false if first_base is not a REG or
27569 if first_offset is not a CONST_INT.
27571 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27573 * config/cris/cris.md: Use correct operand in a define_peephole2.
27575 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27577 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
27579 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27581 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
27582 this_insn if it is an INSN or JUMP_INSN.
27583 (force_offsettable): Look at base, not at addr.
27584 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
27585 on things that aren't necessarily CONST_INTs.
27587 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
27589 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
27590 -mfpmath=sse is the default also for x86-32 targets with SSE2
27591 instruction set when @option{-ffast-math} is enabled
27593 2017-02-24 Jeff Law <law@redhat.com>
27595 PR rtl-optimizatoin/79286
27596 * ira.c (update_equiv_regs): Drop may_trap_p exception to
27599 2017-02-24 Richard Biener <rguenther@suse.de>
27601 PR tree-optimization/79389
27602 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
27605 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
27607 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
27608 function comment to reflect reality.
27609 (loop_exits_before_overflow): Fix typo in function description.
27611 2017-02-24 Richard Biener <rguenther@suse.de>
27613 PR tree-optimization/79389
27614 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
27615 properly that a threading opportunity exists. Detect conditional
27616 copy/constant propagation opportunities.
27618 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
27620 * config/visium/visium.md (type): Add trap.
27621 (b): New mode attribute.
27622 (*btst): Rename into...
27623 (*btst<mode>): ...this and adjust.
27624 (*cbranchsi4_btst_insn): Rename into...
27625 (*cbranch<mode>4_btst_insn): ...this and adjust.
27626 (trap): New define_insn.
27628 2017-02-23 Jakub Jelinek <jakub@redhat.com>
27630 PR tree-optimization/79389
27631 * ifcvt.c (struct noce_if_info): Add rev_cond field.
27632 (noce_reversed_cond_code): New function.
27633 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
27634 reversed_comparison_code. Formatting fix.
27635 (noce_try_store_flag): Test rev_cond != NULL in addition to
27636 reversed_comparison_code.
27637 (noce_try_store_flag_constants): Likewise.
27638 (noce_try_store_flag_mask): Likewise.
27639 (noce_try_addcc): Use rev_cond if non-NULL instead of
27640 reversed_comparison_code.
27641 (noce_try_cmove_arith): Likewise. Formatting fixes.
27642 (noce_try_minmax, noce_try_abs): Clear rev_cond.
27643 (noce_find_if_block): Initialize rev_cond.
27644 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
27645 instead of false as last argument never attempt to reverse it
27648 2017-02-23 Bin Cheng <bin.cheng@arm.com>
27650 PR tree-optimization/79663
27651 * tree-predcom.c (combine_chains): Process refs in reverse order
27652 only for ZERO length chains, and add explaining comment.
27654 2017-02-23 Jeff Law <law@redhat.com>
27656 PR tree-optimization/79578
27657 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
27658 in call to operand_equal_p.
27660 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
27663 * config/i386/cpuid.h: Fix another undefined behavior.
27665 2017-02-23 Richard Biener <rguenther@suse.de>
27667 PR tree-optimization/79683
27668 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
27669 vector types for data-refs.
27671 2017-02-23 Martin Liska <mliska@suse.cz>
27673 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
27675 2017-02-23 Jakub Jelinek <jakub@redhat.com>
27677 PR middle-end/79665
27678 * internal-fn.c (get_range_pos_neg): Moved to ...
27679 * tree.c (get_range_pos_neg): ... here. No longer static.
27680 * tree.h (get_range_pos_neg): New prototype.
27681 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
27682 are known to be in between 0 and signed maximum inclusive, try to
27683 expand both unsigned and signed divmod and use the cheaper one from
27686 2017-02-22 Jeff Law <law@redhat.com>
27688 PR tree-optimization/79578
27689 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
27690 to compare base operands.
27692 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
27695 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
27696 gpc_reg_operand instead of fpr_reg_operand.
27698 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
27700 * config/mips/mips.c (mips_return_in_memory): Force FP
27701 vector types to be returned in memory for o32 ABI.
27703 2017-02-22 Jakub Jelinek <jakub@redhat.com>
27705 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
27706 instead of DW_TAG_member for static data member declarations and don't
27707 set no_linkage_name for static inline data members.
27708 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
27711 2017-02-22 Martin Liska <mliska@suse.cz>
27713 * doc/invoke.texi: Replace inequality signs with square brackets
27716 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27718 PR tree-optimization/68644
27719 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
27721 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
27724 * lra-constraints.c (simplify_operand_subreg): Handle
27725 WORD_REGISTER_OPERATIONS targets.
27727 2017-02-22 Jakub Jelinek <jakub@redhat.com>
27730 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
27731 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
27732 elimination by swapping fld*.
27734 2017-02-22 Richard Biener <rguenther@suse.de>
27736 PR tree-optimization/79673
27737 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
27738 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
27739 irrelevant address-space qualifiers and avoiding a
27740 ADDR_SPACE_CONVERT_EXPR from fold_convert.
27742 2017-02-22 Richard Biener <rguenther@suse.de>
27744 PR tree-optimization/79666
27745 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
27746 to not symbolically negate if that may introduce undefined
27749 2017-02-22 Martin Liska <mliska@suse.cz>
27752 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
27753 * data-streamer-out.c (streamer_write_gcov_count_stream):
27755 * value-prof.c (stream_out_histogram_value): Make assert more
27756 precise based on type of counter.
27758 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
27761 * config/i386/i386.md (standard_x87sse_constant_load splitter):
27762 Use nonimmediate_operand instead of memory_operand for operand 1.
27763 (float-extend standard_x87sse_constant_load splitter): Ditto.
27765 2017-02-21 Jeff Law <law@redhat.com>
27767 PR tree-optimization/79621
27768 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
27769 blocks with edges to themselves.
27771 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27774 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
27775 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
27776 Use gimple_call_builtin_p.
27779 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
27780 on temporarily removed DEBUG_INSNs.
27782 PR tree-optimization/79649
27783 * tree-loop-distribution.c (classify_partition): Give up on
27784 non-generic address space loads/stores.
27786 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
27788 * doc/loop.texi (Loop manipulation): Remove nonexistent
27789 tree_ssa_loop_version from the documentation.
27790 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
27792 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27795 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
27796 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
27797 * config/rs6000/rs6000.c: Include except.h.
27798 (rs6000_expand_split_stack_prologue): Call
27799 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
27801 2017-02-21 Martin Jambor <mjambor@suse.cz>
27804 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
27805 have been analyzed.
27807 2017-02-21 Martin Jambor <mjambor@suse.cz>
27809 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
27810 for backward compatibility only.
27811 * doc/invoke.texi (Option Summary): Remove all references to
27812 -fipa-cp-alignment.
27814 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
27818 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27820 * lra-constraints.c (curr_insn_transform): Handle
27821 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
27823 2017-02-21 Martin Liska <mliska@suse.cz>
27825 * config/i386/i386.opt: Replace -masm-dialect with -masm.
27827 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
27829 PR translation/79638
27830 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
27832 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
27835 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
27836 (arm_function_ok_for_sibcall): Return false for an indirect call by
27837 descriptor if all the argument registers are used.
27838 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
27839 alignment of the function.
27841 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27843 PR tree-optimization/61441
27844 * simplify-rtx.c (simplify_const_unary_operation): For
27845 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
27846 the sNaN unmodified.
27848 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27850 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27851 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27852 instead of SYSTEM_HEADER_DIR.
27854 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
27855 Martin Liška <mliska@suse.cz>
27857 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
27858 Fix typos and grammar, use active voice, and clarify.
27860 2017-02-20 Marek Polacek <polacek@redhat.com>
27862 PR middle-end/79537
27863 * gimplify.c (gimplify_expr): Handle unused *&&L;.
27866 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
27868 2017-02-20 Jakub Jelinek <jakub@redhat.com>
27871 * config/i386/i386.c (ix86_expand_builtin): Handle
27872 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
27873 ix86_builtins_isa[fcode].isa as a requirement of those
27874 flags and any other flag in the bitmask.
27875 (ix86_init_mmx_sse_builtins): Use 0 instead of
27876 ~OPTION_MASK_ISA_64BIT as mask.
27877 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
27878 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
27879 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
27880 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
27882 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27885 * lra-constraints.c (split_reg): Check requested split mode
27886 is supported by the register.
27888 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27890 * lra-constraints.c (simplify_operand_subreg): Remove early
27893 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27896 * lra-constraints.c (curr_insn_transform): Tighten condition
27897 for converting SUBREG reloads from OP_OUT to OP_INOUT.
27899 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27902 * lra-constraints.c (curr_insn_transform): Handle
27903 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
27905 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
27908 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
27910 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
27912 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
27915 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
27918 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27920 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
27923 2017-02-18 Jakub Jelinek <jakub@redhat.com>
27925 * final.c (last_columnnum, override_columnnum): New variables.
27926 (final_start_function): Set last_columnnum, pass it to begin_prologue
27927 hook and pass 0 to dwarf2out_begin_prologue.
27928 (final_scan_insn): Update override_columnnum. Pass last_columnnum
27929 to source_line debug hook.
27930 (notice_source_line): Compute last_columnnum and for debug_column_info
27931 return true on column changes.
27932 * debug.h (struct gcc_debug_hooks): Add column argument to
27933 source_line and begin_prologue hooks.
27934 (debug_nothing_int_charstar_int_bool): Remove prototype.
27935 (debug_nothing_int_int_charstar,
27936 debug_nothing_int_int_charstar_int_bool): New prototypes.
27937 (dwarf2out_begin_prologue): Add column argument.
27938 * debug.c (do_nothing_debug_hooks): Adjust source_line and
27939 begin_prologue hooks.
27940 (debug_nothing_int_charstar_int_bool): Remove.
27941 (debug_nothing_int_int_charstar,
27942 debug_nothing_int_int_charstar_int_bool): New functions.
27943 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
27944 through to dwarf2out_source_line.
27945 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
27946 (dwarf2out_source_line): Add column argument, emit it if requested.
27947 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
27949 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
27950 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
27951 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
27952 through to dwarf2out_begin_prologue.
27953 (vmsdbgout_source_line): Add column argument, pass it through to
27954 dwarf2out_source_line.
27955 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
27956 dbxout_source_line caller.
27957 (dbxout_source_line): Add column argument.
27959 * common.opt (gno-column-info, gcolumn-info): New options.
27960 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
27961 (check_die): Also test for multiple DW_AT_decl_column attributes.
27962 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
27963 DW_AT_decl_column if requested.
27964 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
27966 (gen_variable_die): Likewise.
27967 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
27968 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
27971 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
27972 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
27973 (ix86_handle_option): Handle OPT_m3dnowa.
27974 * doc/invoke.texi (-m3dnowa): Document.
27975 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
27976 -m3dnowa instead of -m3dnow -march=athlon.
27979 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
27980 instead of gcc_assert for K, r and R code checks. Formatting fixes.
27982 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27985 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
27986 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
27987 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
27988 generator for vsx_xxpermdi_<mode>_be.
27989 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
27990 force big-endian semantics.
27991 (vsx_xxpermdi_<mode>_be): New define_expand with same
27992 implementation as previous version of vsx_xxpermdi_<mode>.
27994 2017-02-17 Jakub Jelinek <jakub@redhat.com>
27996 PR tree-optimization/79327
27997 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
27998 variable, its initialization and use.
28000 2017-02-17 Julia Koval <julia.koval@intel.com>
28002 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
28003 (OPTION_MASK_ISA_PKU_UNSET): New.
28004 (ix86_handle_option): Handle -mrdpid.
28005 * config/i386/cpuid.h (bit_RDPID): New.
28006 * config/i386/driver-i386.c (host_detect_local_cpu):
28007 Detect RDPID feature.
28008 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
28009 * config/i386/i386-c.c (ix86_target_macros_internal):
28011 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
28012 (ix86_valid_target_attribute_inner_p): Add "rdpid".
28013 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
28014 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
28015 * config/i386/i386.md (define_insn "rdpid"): New.
28016 * config/i386/i386.opt Add -mrdpid.
28017 * config/i386/immintrin.h (_rdpid_u32): New.
28019 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
28021 PR rtl-optimization/79541
28022 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
28023 instead of transforming it into USE.
28025 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
28027 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
28028 If HONOR_SNANS (SFmode) force the input to a register.
28029 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
28030 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
28031 an frsp or similar insn.
28033 2017-02-17 Martin Liska <mliska@suse.cz>
28035 PR rtl-optimization/79577
28036 * params.def (selsched-max-sched-times): Increase minimum to 1.
28038 2017-02-17 Martin Liska <mliska@suse.cz>
28040 PR rtl-optimization/79574
28041 * gcse.c (want_to_gcse_p): Prevent integer overflow.
28043 2017-02-17 Martin Liska <mliska@suse.cz>
28045 PR tree-optimization/79529
28046 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
28047 ssa_defined_default_def_p to handle cases which are implicitly
28049 * tree-ssa.c (ssa_defined_default_def_p): New function.
28050 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
28051 which are implicitly defined.
28052 * tree-ssa.h (ssa_defined_default_def_p): Declare.
28054 2017-02-17 Richard Biener <rguenther@suse.de>
28056 PR middle-end/79576
28057 * params.def (max-ssa-name-query-depth): Limit to 10.
28059 2017-02-17 Richard Biener <rguenther@suse.de>
28061 PR tree-optimization/79552
28062 * tree-ssa-structalias.c (visit_loadstore): Properly verify
28065 2017-02-17 Richard Biener <rguenther@suse.de>
28068 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
28070 2017-02-17 Marek Polacek <polacek@redhat.com>
28072 PR middle-end/79536
28073 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
28074 (fold_negate_expr): New wrapper.
28076 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
28078 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
28079 Correct terminology and de-emphasize pre-standard behavior.
28081 2017-02-16 Alan Modra <amodra@gmail.com>
28083 PR rtl-optimization/79286
28084 * ira.c (def_dominates_uses): New function.
28085 (update_equiv_regs): Don't create an equivalence for insns that
28086 may trap where the register def does not dominate the use.
28088 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
28090 PR rtl-optimization/78127
28091 * lra.c (lra): Call lra_eliminate before finish the loop after
28094 2017-02-16 Richard Biener <rguenther@suse.de>
28096 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
28098 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
28099 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
28100 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
28101 (isl_val_int_from_wi): New function.
28102 (extract_affine_gmp): Rename to ...
28103 (extract_affine_wi): ... this, take a widest_int.
28104 (extract_affine_int): Just wrap extract_affine_wi.
28105 (add_param_constraints): Use isl_val_int_from_wi.
28106 (add_loop_constraints): Likewise, and extract_affine_wi.
28108 2017-02-15 Jeff Law <law@redhat.com>
28110 PR middle-end/79521
28111 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
28112 ira_init_register_move_cost_if_necessary.
28114 2017-02-15 Martin Sebor <msebor@redhat.com>
28116 PR middle-end/32003
28117 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
28118 removed in a prior commit.
28120 2017-02-15 Bin Cheng <bin.cheng@arm.com>
28122 PR tree-optimization/79347
28123 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
28124 counters during peeling.
28126 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
28128 * Makefile.in (site.exp): Remove "set ISLVER".
28130 2017-02-15 Jakub Jelinek <jakub@redhat.com>
28133 * real.c (real_from_integer): Call real_convert even for decimal.
28135 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
28138 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
28140 2017-02-14 Andrew Pinski <apinski@cavium.com>
28142 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
28143 cores and change the partno/implementer to be correct.
28144 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
28145 the 'B" as the implementer.
28146 * config/aarch64/aarch64-tune.md: Regenerate.
28148 2017-02-14 Carl Love <cel@us.ibm.com>
28150 * config/rs6000/rs6000.c: Add case statement entry to make the
28151 xvcvuxdsp built-in argument unsigned.
28152 * config/rs6000/vsx.md: Fix the source and return operand types so they
28153 match the instruction definitions from the ISA document. Fix typo
28154 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
28157 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
28160 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
28161 member early_clobber_alts.
28162 * lra-lives.c (reg_early_clobber_p): New.
28163 (process_bb_lives): Use it.
28164 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
28165 (debug_operand_data): Initialize early_clobber_alts.
28166 (setup_operand_alternative): Set up early_clobber_alts.
28167 (collect_non_operand_hard_regs): Ditto. Pass early clobber
28168 alternatives to new_insn_reg.
28169 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
28171 (lra_update_insn_regno_info): Pass the new arg.
28173 2017-02-14 Jakub Jelinek <jakub@redhat.com>
28175 PR middle-end/79505
28176 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
28177 (new_oacc_loop_raw): Don't clear already cleared fields.
28180 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
28181 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
28182 _mm512_prefetch_i64gather_ps): New inline functions and macros.
28184 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
28187 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
28189 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
28192 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
28193 the extra instruction to the right place to store 128-bit constant
28196 2017-02-14 Martin Sebor <msebor@redhat.com>
28198 PR middle-end/79448
28199 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
28200 warning for strings of unknown length.
28202 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
28204 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
28206 2017-02-14 Jeff Law <law@redhat.com>
28209 * ira-costs.c (scan_one_insn): Initialize register move costs
28210 for pseudos seen in USE/CLOBBER insns.
28212 PR tree-optimization/79095
28213 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
28214 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
28215 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
28216 if the operands are known to be not equal, then the resulting range
28218 (intersect_ranges): If the new range is ~[0,0] and the old range is
28219 wide, then prefer ~[0,0].
28220 * tree-vrp.c (overflow_comparison_p_1): New function.
28221 (overflow_comparison_p): New function.
28222 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
28223 if NAME is used in an overflow test.
28224 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
28225 overflow check that can be expressed as an equality test, then adjust
28226 ops to be that equality test.
28228 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28230 * config/s390/s390-builtin-types.def: Remove flags argument.
28231 * config/s390/s390.c (s390_init_builtins): Likewise.
28233 2017-02-14 Martin Liska <mliska@suse.cz>
28235 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
28236 vector. Fix trailing white spaces.
28238 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
28240 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
28243 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28245 PR rtl-optimization/68664
28246 * config/arm/arm.c (arm_sched_can_speculate_insn):
28247 New function. Declare prototype.
28248 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
28250 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28252 PR rtl-optimization/68664
28253 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
28255 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
28257 2017-02-14 Amit Pawar <amit.pawar@amd.com>
28259 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
28260 max skip bytes for function, loop and jump.
28262 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28264 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
28265 ABS_EXPR for gimple dump.
28267 2017-02-14 Jakub Jelinek <jakub@redhat.com>
28270 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
28272 PR tree-optimization/79408
28273 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
28274 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
28275 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
28276 also if rhs1 is INTEGER_CST.
28278 2017-02-14 Richard Biener <rguenther@suse.de>
28280 PR middle-end/79432
28281 * tree-into-ssa.c (insert_phi_nodes): When the function can
28282 have abnormal edges rewrite SSA names with broken use-def
28283 dominance out of SSA and register them for PHI insertion.
28285 2017-02-13 Martin Sebor <msebor@redhat.com>
28287 PR middle-end/79496
28288 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
28289 clearing info.nowrite flag when snprintf size argument is a range.
28291 2017-02-13 Jakub Jelinek <jakub@redhat.com>
28293 * cprop.c (cprop_jump): Add missing space in string literal.
28294 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
28295 (get_constraint_for_component_ref): Likewise.
28296 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
28297 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
28298 * lra-constraints.c (process_alt_operands): Likewise.
28299 * ipa-inline.c (inline_small_functions): Likewise.
28300 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
28301 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
28302 * trans-mem.c (diagnose_tm_1_op): Likewise.
28303 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
28304 (grid_parallel_clauses_gridifiable): Likewise.
28306 * config/nvptx/mkoffload.c (process): Add space in between
28309 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
28310 "MOD4_SSE_REGS" and "ALL_REGS".
28312 * spellcheck.c (test_data): Add , in between "foo" and "food".
28314 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28317 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
28318 boundary crossing check and subsequent code generation agree.
28320 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28322 * config/aarch64/aarch64.c (has_memory_op): Delete.
28323 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
28326 2017-02-13 Jakub Jelinek <jakub@redhat.com>
28328 PR rtl-optimization/79388
28329 PR rtl-optimization/79450
28330 * combine.c (distribute_notes): When removing TEM_INSN for which
28331 corresponding dest has last value recorded, invalidate that last
28334 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28336 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
28337 of explicit '@'. Add missing assembly comment marker on branch costs
28340 2017-02-13 Nathan Sidwell <nathan@acm.org>
28342 * gengtype-lex.l (<in_struct>): Add '/'.
28344 2017-02-13 Martin Liska <mliska@suse.cz>
28347 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
28349 2017-02-13 Richard Biener <rguenther@suse.de>
28351 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
28353 * configure: Re-generate.
28354 * config.in: Likewise.
28355 * graphite-dependences.c: Simplify as if
28356 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
28357 * graphite-isl-ast-to-gimple.c: Likewise.
28358 * graphite-optimize-isl.c: Likewise.
28359 * graphite-poly.c: Likewise.
28360 * graphite-sese-to-poly.c: Likewise.
28361 * graphite.h: Likewise.
28362 * toplev.c: Include isl/version.h and use isl_version () for
28363 printing the ISL version.
28364 * doc/install.texi: Update ISL requirement.
28366 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
28368 * doc/standards.texi (Standards): Update reference to
28371 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
28373 * doc/extend.texi (Named Address Spaces): sourceware.org now
28375 * doc/install.texi (Binaries): Ditto.
28378 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
28380 * doc/cpp.texi: Replace "stringify"/"stringification" with C
28381 standard terminology "stringize"/"stringizing" throughout.
28382 * doc/cppinternals.texi: Likewise.
28384 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
28386 * doc/extend.texi: Fix some spelling mistakes and typos.
28387 * doc/invoke.texi: Likewise.
28389 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
28392 * params.def (inline-min-speedup) Change from 10 to 8.
28394 2017-02-11 Jakub Jelinek <jakub@redhat.com>
28396 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
28399 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
28402 * ipa-inline-analysis.c (get_minimal_bb): New function.
28403 (record_modified): Use it.
28404 (remap_edge_change_prob): Handle also ancestor functions.
28406 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
28408 * doc/contrib.texi (Contributors): Remove broken link into
28409 the Mauve CVS repository.
28411 2017-02-11 Jakub Jelinek <jakub@redhat.com>
28413 PR middle-end/79454
28414 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
28415 result computation whenever lhs doesn't have vector mode, not
28416 just when it has BLKmode.
28418 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
28420 * doc/makefile.texi (profiledbootstrap): Refer to the
28421 installation instructions only in textual form.
28423 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28426 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
28428 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
28430 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
28431 (Specific): Update mingw-w64 reference.
28433 (Specific): Remove broken link to Renesas RX processor.
28435 2017-02-10 Richard Biener <rguenther@suse.de>
28437 * toplev.c (process_options): Do not mention obsolete graphite
28438 options when printing sorry message about missing graphite support.
28439 Mention -floop-nest-optimize.
28441 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
28443 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
28444 (vtst_p16): Likewise.
28445 (vtstq_p8): Likewise.
28446 (vtstq_p16): Likewise.
28448 (vtstq_p64): Likewise.
28449 * config/arm/arm_neon.h (vgetq_lane_p64): New.
28450 (vset_lane_p64): New.
28451 (vsetq_lane_p64): New.
28453 2017-02-10 Jakub Jelinek <jakub@redhat.com>
28455 PR tree-optimization/79411
28456 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
28457 stmt operands are SSA_NAMEs used in abnormal phis.
28458 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
28461 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
28464 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
28467 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
28469 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
28471 2017-02-09 Jakub Jelinek <jakub@redhat.com>
28473 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
28477 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
28478 not arbitrary TREE_CONSTANT.
28481 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
28482 "omp declare target link" attribute unless is_global_var.
28483 * omp-offload.c (find_link_var_op): Likewise.
28485 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
28486 Chung-Lin Tang <cltang@codesourcery.com>
28488 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
28490 (gimplify_adjust_omp_clauses): Don't delete TILE.
28491 (gimplify_omp_for): Deal with TILE.
28492 * internal-fn.c (expand_GOACC_TILE): New function.
28493 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
28495 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
28496 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
28498 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
28499 avoid DIV for outermost collapse var.
28500 (expand_oacc_for): Insert tile element loop as needed. Adjust.
28501 Remove out of date comments, fix whitespace.
28502 * omp-general.c (omp_extract_for_data): Deal with tiling.
28503 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
28504 adjust OLF_DIM_BASE value.
28505 (struct omp_for_data): Add tiling field.
28506 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
28507 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
28508 for auto loops. Remove default auto determining, moved to
28509 oacc_loop_fixed_partitions.
28510 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
28511 stmts, add e_mask field.
28512 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
28513 (oacc_thread_numbers): Use oacc_dim_call.
28514 (oacc_xform_tile): New.
28515 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
28516 (finish_oacc_loop): Adjust for ifns vector.
28517 (oacc_loop_discover_walk): Append loop abstraction sites to list,
28518 add case for GOACC_TILE fns.
28519 (oacc_loop_xform_loop): Delete.
28520 (oacc_loop_process): Iterate over call list directly, and add
28521 handling for GOACC_TILE fns.
28522 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
28524 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
28525 vector partitioning to outer loops. Assign 2 partitions to loops
28526 when available. Add TILE handling.
28527 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
28528 (execite_oacc_device_lower): Process GOACC_TILE fns,
28529 ignore unknown specs.
28530 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
28531 * tree.c (omp_clause_num_ops): Adjust TILE ops.
28532 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
28534 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
28536 * configure.ac (ACX_BUGURL): Update.
28537 * configure: Regenerate.
28539 2017-02-09 Richard Biener <rguenther@suse.de>
28541 PR tree-optimization/69823
28542 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
28543 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
28545 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
28547 * config/arc/arc-c.def: Add __NPS400__ definition.
28548 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
28549 (TARGET_NPS400): Define.
28551 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
28553 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
28555 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
28556 pointer, arch_info.
28557 (arc_cpu_types): Fill the arch_info field with a pointer into the
28558 arc_arch_types table.
28559 (arc_selected_cpu): Declare.
28560 * config/arc/arc.c (arc_selected_cpu): Make global.
28561 (arc_selected_arch): Delete.
28562 (arc_base_cpu): Delete.
28563 (arc_override_options): Remove references to deleted variables,
28564 update access to arch information.
28565 (ARC_OPT): Update access to arch information.
28566 (ARC_OPTX): Likewise.
28567 * config/arc/arc.h (arc_base_cpu): Remove declaration.
28568 (TARGET_ARC600): Update access to arch information.
28569 (TARGET_ARC601): Likewise.
28570 (TARGET_ARC700): Likewise.
28571 (TARGET_EM): Likewise.
28572 (TARGET_HS): Likewise.
28573 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
28576 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
28579 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
28580 condition/operands for integer GE/LE/GEU/LEU operations.
28582 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
28584 PR translation/79397
28585 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
28588 2017-02-08 Martin Jambor <mjambor@suse.cz>
28591 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
28592 whether allocation happened.
28593 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
28594 nothing was allocated.
28596 2017-02-08 Jakub Jelinek <jakub@redhat.com>
28598 PR tree-optimization/79408
28599 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
28600 constant, but SSA_NAME with a known integer range, use the minimum
28601 of that range instead of op1 to determine if modulo can be replaced
28602 with its first operand.
28604 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28606 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
28608 2017-02-08 Richard Biener <rguenther@suse.de>
28610 PR tree-optimization/71824
28611 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
28612 Check all loops contained in the merged region.
28614 2017-02-07 Andrew Pinski <apinski@cavium.com>
28616 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
28618 2017-02-07 Andrew Pinski <apinski@cavium.com>
28620 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
28621 (thunderxt88): Likewise.
28622 (thunderxt81): Disable LSE and change v8.1 to v8.
28623 (thunderxt83): Likewise.
28625 2017-02-07 Jakub Jelinek <jakub@redhat.com>
28626 Richard Biener <rguenther@suse.de>
28628 PR middle-end/79399
28629 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
28630 type from int to size_t.
28631 * ira-costs.c (struct_costs_size): Change type from int to size_t.
28633 2017-02-07 Jakub Jelinek <jakub@redhat.com>
28635 PR rtl-optimization/79386
28636 * cprop.c (bypass_conditional_jumps): Initialize
28637 bypass_last_basic_block already before splitting bbs after
28638 unconditional traps...
28639 (bypass_conditional_jumps): ... rather than here.
28642 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
28643 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
28644 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
28645 fix -masm=intel patterns.
28647 2017-02-07 Richard Biener <rguenther@suse.de>
28649 PR tree-optimization/79256
28650 PR middle-end/79278
28651 * builtins.c (get_object_alignment_2): Use min_align_of_type
28652 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
28653 and ADJUST_FIELD_ALIGN.
28655 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
28657 * doc/tm.texi: Regenerate.
28658 * stor-layout.c (layout_decl): Adjust.
28659 (update_alignment_for_field): Likewise.
28660 (place_field): Likewise.
28661 (min_align_of_type): Likewise.
28662 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
28663 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
28664 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
28665 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
28666 * config/frv/frv.c (frv_adjust_field_align): Likewise.
28667 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
28668 * config/i386/i386.c (x86_field_alignment): Likewise.
28669 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
28670 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
28671 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
28672 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
28673 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
28674 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
28678 2017-01-30 Richard Biener <rguenther@suse.de>
28680 PR tree-optimization/79256
28681 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
28682 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
28685 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
28687 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
28688 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
28689 builtins to SImode and emit a zero-extend, if necessary.
28691 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28693 * docs/invoke.texi (RISC-V Options): Alphabetize.
28695 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28697 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
28700 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28702 * config/riscv/riscv.c: New file.
28703 * common/config/riscv/riscv-common.c: Likewise.
28704 * config.gcc: Likewise.
28705 * config/riscv/constraints.md: Likewise.
28706 * config/riscv/elf.h: Likewise.
28707 * config/riscv/generic.md: Likewise.
28708 * config/riscv/linux.h: Likewise.
28709 * config/riscv/multilib-generator: Likewise.
28710 * config/riscv/peephole.md: Likewise.
28711 * config/riscv/pic.md: Likewise.
28712 * config/riscv/predicates.md: Likewise.
28713 * config/riscv/riscv-builtins.c: Likewise.
28714 * config/riscv/riscv-c.c: Likewise.
28715 * config/riscv/riscv-ftypes.def: Likewise.
28716 * config/riscv/riscv-modes.def: Likewise.
28717 * config/riscv/riscv-opts.h: Likewise.
28718 * config/riscv/riscv-protos.h: Likewise.
28719 * config/riscv/riscv.h: Likewise.
28720 * config/riscv/riscv.md: Likewise.
28721 * config/riscv/riscv.opt: Likewise.
28722 * config/riscv/sync.md: Likewise.
28723 * config/riscv/t-elf-multilib: Likewise.
28724 * config/riscv/t-linux: Likewise.
28725 * config/riscv/t-linux-multilib: Likewise.
28726 * config/riscv/t-riscv: Likewise.
28727 * configure.ac: Likewise.
28728 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
28729 Waterman as RISC-V maintainers.
28730 * doc/install.texi: Add RISC-V entries.
28731 * doc/invoke.texi: Add RISC-V options section.
28732 * doc/md.texi: Add RISC-V constraints section.
28733 * configure: Regenerated.
28735 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
28738 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
28739 false values to be constant vectors with all 0 or all 1 bits set.
28740 (vcondu<mode><mode>): Likewise.
28741 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
28743 (fpmask_comparison_operator): Update comment.
28744 (vecint_comparison_operator): New predicate.
28745 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
28746 vector conditionals when the true and false values are constant
28747 vectors with all 0 bits or all 1 bits set.
28749 2017-02-06 Martin Sebor <msebor@redhat.com>
28751 PR tree-optimization/79376
28752 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
28754 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
28756 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
28757 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
28758 to simplify split condition.
28760 2017-02-06 Jakub Jelinek <jakub@redhat.com>
28762 * omp-expand.c (oxpand_omp_atomic_fetch_op,
28763 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
28766 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
28768 PR rtl-optimization/68664
28769 * target.def (can_speculate_insn): New hook.
28770 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
28771 * doc/tm.texi: Regenerate.
28772 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
28773 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
28774 (rs6000_sched_can_speculate_insn): New function.
28776 2017-02-06 Jakub Jelinek <jakub@redhat.com>
28778 PR tree-optimization/79284
28779 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
28780 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
28781 vectorizable_mask_load_store, vectorizable_operation,
28782 vect_is_simple_cond, get_same_sized_vectype): Use it instead
28783 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
28784 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
28785 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
28786 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
28787 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28788 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
28789 is_gimple_assign (stmt). Replace another such test with
28790 is_gimple_assign (stmt).
28792 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
28795 * config/avr/avr.c (rtl-iter.h): Include it.
28796 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
28797 (avr_legitimate_combined_insn): ...and implementation.
28799 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
28801 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
28802 * config/s390/s390.c (s390_const_operand_ok)
28803 (s390_canonicalize_comparison, s390_extract_part)
28804 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
28805 (s390_contiguous_bitmask_p, s390_rtx_costs)
28806 (legitimize_pic_address): Likewise.
28807 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
28808 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
28809 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
28810 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
28811 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
28813 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
28815 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
28816 REGNO($0) == REGNO($1).
28818 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28820 * config/s390/linux.h(SIZE_TYPE): Add comment.
28822 2017-02-06 Julian Brown <julian@codesourcery.com>
28823 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28824 Virendra Pathak <virendra.pathak@broadcom.com>
28826 * config/aarch64/aarch64-cores.def: Change the scheduler
28828 * config/aarch64/aarch64.md: Include thunderx2t99.md.
28829 * config/aarch64/thunderx2t99.md: New file.
28831 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28833 * doc/standards.texi (Go Language): Update link to language
28836 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
28838 * tree-eh.c (lower_resx): Sanitize profile.
28839 (cleanup_empty_eh_move_lp): Likewise.
28841 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
28844 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
28846 * cfgloopmanip.h (loop_version): Update prototype.
28847 * modulo-sched.c (sms_schedule): Update call of loop_version.
28848 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
28849 * tree-parloops.c (gen_parallel_loop): Likewise.
28850 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
28851 * tree-ssa-loop-split.c (split_loop): Likewise.
28852 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
28853 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
28855 2017-02-05 Martin Liska <mliska@suse.cz>
28858 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
28860 (print_operand_address): Initialize a struct to zero.
28862 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28864 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
28865 garbage collector only in textual form.
28867 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28869 * doc/extend.texi (x86 specific memory model extensions for
28870 transactional memory): Simplify a phrase.
28872 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
28875 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
28876 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
28877 (atomic_storedi_1): Likewise.
28879 2017-02-04 Jakub Jelinek <jakub@redhat.com>
28881 PR tree-optimization/79338
28882 * tree-parloops.c (gather_scalar_reductions): Don't call
28883 vect_analyze_loop_form for loop->inner before destroying loop's
28886 2017-02-03 Martin Sebor <msebor@redhat.com>
28888 PR tree-optimization/79327
28889 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
28890 when precision has resulted in leading zeros.
28891 (format_integer): Adjust the likely counter to assume an unknown
28892 argument that may be zero is non-zero.
28894 2017-02-03 Jason Merrill <jason@redhat.com>
28897 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
28898 avoid copying non-taken branch.
28900 2017-02-03 Jakub Jelinek <jakub@redhat.com>
28902 PR tree-optimization/79340
28903 * tree-vect-loop.c (vectorizable_reduction): Release
28904 vec_defs elements after safe_splicing them into other vectors.
28907 PR tree-optimization/79327
28908 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
28909 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
28911 (format_integer): Use wide_int_to_tree instead of build_int_cst
28912 + to_?hwi. If argmin is NULL, just set argmin and argmax to
28913 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
28914 of shortest and longest sequence.
28916 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
28918 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
28919 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
28921 2017-02-03 Walter Lee <walt@tilera.com>
28924 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
28925 after initial stackframe link reg save.
28926 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
28928 2017-02-03 Jakub Jelinek <jakub@redhat.com>
28931 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
28932 wu for stxssp alternative.
28934 2017-02-03 Martin Sebor <msebor@redhat.com>
28936 PR tree-optimization/79352
28937 * gimple-fold.c (get_range_strlen): Add argument.
28938 (get_range_strlen): Change return type to bool.
28939 (get_maxval_strlen): Pass in a dummy argument.
28940 * gimple-fold.h (get_range_strlen): Change return type to bool.
28941 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
28942 * tree.h (array_at_struct_end_p): Add argument.
28943 * tree.c (array_at_struct_end_p): Handle it.
28945 2017-02-03 Martin Liska <mliska@suse.cz>
28948 * multiple_target.c (create_dispatcher_calls): Redirect edge
28949 from a caller of a dispatcher.
28950 (expand_target_clones): Make the clones local.
28951 (ipa_target_clone): Do both target clones and resolvers.
28952 (ipa_dispatcher_calls): Remove the pass.
28953 (pass_dispatcher_calls::gate): Likewise.
28954 (make_pass_dispatcher_calls): Likewise.
28955 * passes.def (pass_target_clone): Put as very first IPA early
28958 2017-02-03 Martin Liska <mliska@suse.cz>
28960 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
28961 in case of a function with ifunc attribute.
28963 2017-02-03 Martin Liska <mliska@suse.cz>
28965 * cgraph.c (cgraph_node::dump): Dump function version info.
28966 * symtab.c (symtab_node::dump_base): Add missing new line.
28968 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
28970 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
28971 (ifcombine_ifandif): Use it.
28973 2017-02-03 Martin Liska <mliska@suse.cz>
28975 * doc/invoke.texi: Document default value for
28976 use-after-scope-direct-emission-threshold.
28978 2017-02-03 Martin Liska <mliska@suse.cz>
28980 PR tree-optimization/79339
28981 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
28982 (format_floating): Likewise.
28984 2017-02-03 Martin Liska <mliska@suse.cz>
28987 * ipa-prop.c (ipa_node_params_t::insert): Remove current
28989 (ipa_node_params_t::remove): Likewise.
28990 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
28991 initialization from removed ipa_node_params_t::insert.
28992 (ipa_node_params::~ipa_node_params): Move from removed
28993 ipa_node_params_t::release.
28994 * symbol-summary.h (symbol_summary::m_released): New member.
28995 Do not release a summary twice. Do not allow to call finalizer
28996 for types of a summary that live in GGC memory.
28998 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29000 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
29003 2017-02-02 Martin Sebor <msebor@redhat.com>
29005 PR middle-end/79275
29006 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
29007 (format_string): Tighten up the range of output for non-constant
29008 strings and correct the expected range for wide non-constant strings.
29010 2017-02-02 Martin Sebor <msebor@redhat.com>
29012 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
29014 PR middle-end/32003
29015 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
29017 (-fdump-tree-@var): Add to index and document how to come up
29018 with pass-specific option and dump file names.
29019 (-fdump-passes): Clarify where to look for output.
29021 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
29023 PR middle-end/77445
29024 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
29025 statistics of the analyzed path; allow threading for speed when
29026 any of BBs along the path are optimized for speed.
29028 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
29030 PR middle-end/78468
29031 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
29032 settings of the virtual registers.
29035 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
29037 * explow.c (get_dynamic_stack_size): Take known alignment of stack
29038 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
29041 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29043 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
29044 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
29046 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29048 * config/s390/s390.md: Add missing comments with the expanded
29050 * config/s390/vector.md: Likewise.
29051 * config/s390/vx-builtins.md: Likewise.
29053 2017-02-02 Jakub Jelinek <jakub@redhat.com>
29056 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
29057 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
29058 conditions on a single line.
29060 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29062 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
29063 __S390_VX__ to __VX__.
29065 2017-02-01 Andrew Pinski <apinski@cavium.com>
29067 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
29068 stmt_info to record_stmt_cost.
29069 (vect_get_known_peeling_cost): Pass stmt_info if known to
29071 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
29072 cpu_vector_cost field into
29073 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
29074 field into vec_int_stmt_cost and vec_fp_stmt_cost.
29075 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
29076 splitting of scalar_stmt_cost and vec_stmt_cost.
29077 (thunderx_vector_cost): Likewise.
29078 (cortexa57_vector_cost): LIkewise.
29079 (exynosm1_vector_cost): Likewise.
29080 (xgene1_vector_cost): Likewise.
29081 (thunderx2t99_vector_cost): Improve after the splitting of the two
29083 (aarch64_builtin_vectorization_cost): Update for the splitting of
29084 scalar_stmt_cost and vec_stmt_cost.
29086 2017-02-01 Torvald Riegel <triegel@redhat.com>
29087 Richard Henderson <rth@redhat.com>
29089 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
29090 conditional on existance of a fast atomic load.
29091 * optabs-query.c (can_atomic_load_p): New function.
29092 * optabs-query.h (can_atomic_load_p): Declare it.
29093 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
29094 no fast atomic load is available for the particular size of access.
29095 (expand_atomic_compare_and_swap): Likewise.
29096 (expand_atomic_load): Likewise.
29097 (expand_atomic_store): Likewise.
29098 (expand_atomic_fetch_op): Likewise.
29099 * testsuite/lib/target-supports.exp
29100 (check_effective_target_sync_int_128): Remove x86 because it provides
29101 no fast atomic load.
29102 (check_effective_target_sync_int_128_runtime): Likewise.
29104 2017-02-01 Richard Biener <rguenther@suse.de>
29106 * graphite.c: Include tree-vectorizer.h for find_loop_location.
29107 (graphite_transform_loops): Provide opt-info for optimized nests.
29108 * tree-parloop.c (parallelize_loops): Provide opt-info for
29109 parallelized loops.
29111 2017-02-01 Richard Biener <rguenther@suse.de>
29113 PR middle-end/79315
29114 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
29115 was not set before.
29117 2017-02-01 Richard Biener <rguenther@suse.de>
29119 PR tree-optimization/71824
29120 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
29121 Verify the loops are valid in the merged SESE region.
29122 (scop_detection::can_represent_loop_1): Check analyzing the
29123 evolution of the number of iterations in the region succeeds.
29125 2017-01-31 Ian Lance Taylor <iant@golang.org>
29127 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
29128 REG_ARGS_SIZE note to 32-bit push insns and call insn.
29130 2017-01-31 David Malcolm <dmalcolm@redhat.com>
29132 PR preprocessor/79210
29133 * input.c (get_substring_ranges_for_loc): Replace line_width
29134 assertion with error-handling.
29136 2017-01-31 Richard Biener <rguenther@suse.de>
29138 PR tree-optimization/77318
29139 * graphite-sese-to-poly.c (extract_affine): Fix assert.
29140 (create_pw_aff_from_tree): Take loop parameter.
29141 (add_condition_to_pbb): Pass loop of the condition to
29142 create_pw_aff_from_tree.
29144 2017-01-31 Jakub Jelinek <jakub@redhat.com>
29146 * config/s390/s390.c (s390_asan_shadow_offset): New function.
29147 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
29149 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
29153 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
29155 (convert_int_to_float128): Likewise.
29156 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
29157 (convert_int_to_float128): Likewise.
29158 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
29159 (UNSPEC_IEEE128_CONVERT): Likewise.
29160 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
29161 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
29162 Use local variables for IBM extended format.
29163 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
29164 (fix_trunc<mode>si2_fprs): Likewise.
29165 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
29166 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
29167 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
29168 to know that we can now have integers of all sizes in vector
29170 (fix<uns>_<mode>di2_hw): Likewise.
29171 (float<uns>_<mode>si2_hw): Likewise.
29172 (fix_<mode>si2_hw): Likewise.
29173 (fixuns_<mode>si2_hw): Likewise.
29174 (float<uns>_<mode>di2_hw): Likewise.
29175 (float_<mode>di2_hw): Likewise.
29176 (float_<mode>si2_hw): Likewise.
29177 (floatuns_<mode>di2_hw): Likewise.
29178 (floatuns_<mode>si2_hw): Likewise.
29179 (xscvqp<su>wz_<mode>): Delete, no longer used.
29180 (xscvqp<su>dz_<mode>): Likewise.
29181 (xscv<su>dqp_<mode>): Likewise.
29182 (ieee128_mfvsrd_64bit): Likewise.
29183 (ieee128_mfvsrd_32bit): Likewise.
29184 (ieee128_mfvsrwz): Likewise.
29185 (ieee128_mtvsrw): Likewise.
29186 (ieee128_mtvsrd_64bit): Likewise.
29187 (ieee128_mtvsrd_32bit): Likewise.
29189 2017-01-31 Martin Liska <mliska@suse.cz>
29192 * ipa-prop.c (ipa_free_all_node_params): Call release method
29193 instead of ~sumbol_summary to not to trigger double times
29196 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
29198 PR tree-optimization/71691
29199 * bitmap.h (class auto_bitmap): New.
29200 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
29201 is_maybe_undefined instead of ssa_undefined_value_p.
29203 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29205 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
29206 __S390_ARCH_LEVEL__ to __ARCH__.
29208 2017-01-31 Jakub Jelinek <jakub@redhat.com>
29210 PR tree-optimization/79267
29211 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
29212 if should_remove_lhs_p is true.
29214 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
29217 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
29218 (add_alignment_attribute): New.
29219 (base_type_die): Add alignment attribute.
29220 (subrange_type_die): Likewise.
29221 (modified_type_die): Likewise.
29222 (gen_array_type_die): Likewise.
29223 (gen_descr_array_type_die: Likewise.
29224 (gen_enumeration_type_die): Likewise.
29225 (gen_subprogram_die): Likewise.
29226 (gen_variable_die): Likewise.
29227 (gen_field_die): Likewise.
29228 (gen_ptr_to_mbr_type_die): Likewise.
29229 (gen_struct_or_union_type_die): Likewise.
29230 (gen_subroutine_type_die): Likewise.
29231 (gen_typedef_die): Likewise.
29232 (base_type_cmp): Compare alignment attribute.
29234 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29237 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
29238 (setb_unsigned) New pattern for setb with CCUNS.
29239 * config/rs6000/rs6000.c (expand_block_compare): Use a different
29240 subfc./subfe sequence to avoid overflow problems. Generate a
29241 shorter sequence with cmpld/setb for power9.
29242 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
29243 for generating subfc. instruction.
29244 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
29245 now uses this instruction.
29247 2017-01-30 Ian Lance Taylor <iant@google.com>
29250 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
29251 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
29253 2017-01-30 Martin Sebor <msebor@redhat.com>
29255 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
29256 Move constant to the right of a relational operator.
29257 (get_mpfr_format_length, format_character, format_string): Ditto.
29258 (should_warn_p, maybe_warn): Same.
29260 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
29262 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
29265 * asan.c (get_translation_unit_decl): Remove function.
29266 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
29268 2017-01-30 Martin Liska <mliska@suse.cz>
29270 PR gcov-profile/79259
29271 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
29272 -fprofile-generate.
29274 2017-01-30 Martin Liska <mliska@suse.cz>
29277 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
29278 Initialize variables with NULL value.
29280 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
29283 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
29285 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
29287 2017-01-30 Richard Biener <rguenther@suse.de>
29289 PR tree-optimization/79276
29290 * tree-vrp.c (process_assert_insertions): Properly adjust common
29291 when removing a duplicate.
29293 * gcc.dg/torture/pr79276.c: New testcase.
29295 2017-01-30 Richard Biener <rguenther@suse.de>
29297 PR tree-optimization/79256
29298 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
29299 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
29301 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
29303 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
29306 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
29307 ("*r<noxa>sbg_<mode>_sll_bitmask")
29308 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
29309 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
29310 Use contiguous_bitmask_nowrap_operand.
29312 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29315 * config/rs6000/altivec.h (vec_xl): Revise #define.
29316 (vec_xst): Likewise.
29318 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
29320 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
29322 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
29324 PR rtl-optimization/79194
29325 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
29326 traps before call to bypass_conditional_jumps.
29328 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
29330 PR tree-optimization/71374
29331 * lra-constraints.c (check_conflict_input_operands): New.
29332 (match_reload): Use it.
29334 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
29337 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
29338 account to calculate conflict_set.
29340 2017-01-27 Bin Cheng <bin.cheng@arm.com>
29342 PR rtl-optimization/78559
29343 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
29344 other_insn in combine.
29346 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29348 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
29349 uint16_type_node for BT_UINT16.
29351 2017-01-27 David Malcolm <dmalcolm@redhat.com>
29353 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
29354 "RTL Tests" to menu.
29355 (GIMPLE Tests): New node.
29356 (RTL Tests): New node.
29358 2017-01-27 Richard Biener <rguenther@suse.de>
29360 PR tree-optimization/79245
29361 * tree-loop-distribution.c (distribute_loop): Apply cost
29362 modeling also to detected patterns.
29364 2017-01-27 Richard Biener <rguenther@suse.de>
29366 PR tree-optimization/71433
29367 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
29368 (compare_assert_loc): New function.
29369 (process_assert_insertions): Sort and optimize assert locations
29370 to remove duplicates and push down identical assertions on
29371 edges to their destination block.
29373 2017-01-27 Richard Biener <rguenther@suse.de>
29375 PR tree-optimization/79244
29376 * tree-vrp.c (remove_range_assertions): Forcefully propagate
29377 out SSA names even if abnormal.
29379 2017-01-27 Jakub Jelinek <jakub@redhat.com>
29381 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
29382 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
29383 instead of MPFR_RNDN.
29385 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
29388 * arm.c (arm_option_override): Don't call build_target_option_node
29389 until after doing all option overrides.
29390 (arm_valid_target_attribute_tree): Likewise.
29392 2017-01-27 Martin Liska <mliska@suse.cz>
29394 * doc/invoke.texi (-fprofile-arcs): Document profiling support
29395 for {cd}tors and C++ {cd}tors.
29397 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
29399 * config/s390/s390.md ("*setmem_long_and")
29400 ("*setmem_long_and_31z"): Use zero_extend instead of and.
29402 2017-01-26 Martin Sebor <msebor@redhat.com>
29404 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
29407 2017-01-26 Martin Sebor <msebor@redhat.com>
29409 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
29410 HAVE_DFmode before using XFmode or DFmode.
29411 (parse_directive): Avoid using the z length modifier to avoid
29412 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
29414 PR middle-end/78703
29415 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
29416 to accept adjustment as an array.
29417 (get_int_range): New function.
29418 (struct directive): Make width and prec arrays.
29419 (directive::set_width, directive::set_precision): Call get_int_range.
29420 (format_integer, format_floating): Handle width and precision ranges.
29421 (format_string, parse_directive): Same.
29423 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29426 * dwarf2out.c (generate_skeleton_bottom_up): For children with
29427 comdat_type_p set, just clone them, but keep the children in the
29431 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
29432 which have direct callers with -fvar-tracking-assignments enabled
29434 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
29435 inside of type units.
29437 2017-01-26 Martin Sebor <msebor@redhat.com>
29439 PR middle-end/78703
29440 * gimple-ssa-sprintf.c (struct result_range): Add likely and
29442 (struct format_result): Replace number_chars, number_chars_min,
29443 and number_chars_max with a single member of struct result_range.
29445 (format_result::operator+=): Adjust.
29446 (struct fmtresult): Remove bounded. Handle likely and unlikely
29448 (fmtresult::adjust_for_width_or_precision): New function.
29449 (fmtresult:type_max_digits): New function.
29450 (bytes_remaining): Handle likely and unlikely counters.
29451 (min_bytes_remaining): Remove.
29452 (format_percent): Simplify.
29453 (format_integer, format_floating): Set likely and unlikely counters.
29454 (get_string_length, format_character, format_string): Same.
29455 (format_plain, should_warn_p): New function.
29456 (maybe_warn): Call should_warn_p. Update diagnostic messages
29457 and handle those for all directives, including plain strings.
29458 (format_directive): Handle likely and unlikely counters.
29459 Remove unnecessary quoting from diagnostics. Add an informational
29461 (add_bytes): Remove.
29462 (pass_sprintf_length::compute_format_length): Simplify.
29463 (try_substitute_return_value): Handle likely and unlikely counters.
29465 2017-01-26 Carl Love <cel@us.ibm.com>
29467 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
29468 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
29470 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
29473 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
29474 endianess for subregs into account.
29475 * lra-constraints.c (lra_constraints): Do risky transformations
29476 always on the first iteration.
29477 * lra-lives.c (check_pseudos_live_through_calls): Add arg
29478 last_call_used_reg_set.
29479 (process_bb_lives): Define and use last_call_used_reg_set.
29480 * lra.c (lra): Always continue after lra_constraints on the first
29483 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
29485 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
29487 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
29489 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29491 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
29492 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
29493 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
29494 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
29495 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
29496 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
29497 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
29498 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
29499 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
29501 2017-01-26 Marek Polacek <polacek@redhat.com>
29504 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
29505 for the third operand.
29507 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29509 PR middle-end/79236
29510 * omp-low.c (struct omp_context): Add simt_stmt field.
29511 (scan_omp_for): Return omp_context *.
29512 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
29513 context to the _simt_ SIMD stmt.
29514 (lower_omp_for): For combined SIMD with sibling _simt_
29515 SIMD, make sure to use the same decls in _looptemp_
29516 clauses as in the sibling.
29518 2017-01-26 David Sherwood <david.sherwood@arm.com>
29520 PR middle-end/79212
29521 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
29524 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29527 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
29528 emit fld b; fld a; if possible.
29530 * brig-builtins.def: Update copyright years.
29531 * config/arm/arm_acle_builtins.def: Update copyright years.
29533 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
29536 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
29537 constraint instead of o for the stxsd instruction.
29539 2017-01-25 Carl Love <cel@us.ibm.com>
29541 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
29542 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
29544 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
29546 * doc/invoke.texi (C++ Dialect Options): Fix typo.
29548 2017-01-25 Richard Biener <rguenther@suse.de>
29550 PR tree-optimization/69264
29551 * target.def (vector_alignment_reachable): Improve documentation.
29552 * doc/tm.texi: Regenerate.
29553 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
29555 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
29556 earlier changes with respect to TYPE_USER_ALIGN.
29557 (vector_alignment_reachable_p): Likewise. Improve dumping.
29559 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29562 * config/arm/arm.md (xordi3): Force constant operand into a register
29565 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29567 * doc/invoke.texi (-fstore-merging): Correct default optimization
29568 levels at which it is enabled.
29569 (-O): Move -fstore-merging from list to...
29572 2017-01-25 Richard Biener <rguenther@suse.de>
29575 * omp-expand.c: Include debug.h.
29576 (expand_omp_taskreg): Make sure to generate early debug before
29577 outlining anything from a function.
29578 (expand_omp_target): Likewise.
29579 (grid_expand_target_grid_body): Likewise.
29581 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
29584 * asan.c (get_translation_unit_decl): New function.
29585 (asan_add_global): Extract modules file name from globals
29586 TRANSLATION_UNIT_DECL name.
29588 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
29591 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
29592 for long calls with APCS frame and VFP.
29594 2017-01-24 David Malcolm <dmalcolm@redhat.com>
29596 * cfg.c (original_copy_tables_initialized_p): New function.
29597 * cfg.h (original_copy_tables_initialized_p): New decl.
29598 * cfgrtl.c (relink_block_chain): Guard the call to
29599 free_original_copy_tables with a call to
29600 original_copy_tables_initialized_p.
29601 * cgraph.h (symtab_node::native_rtl_p): New decl.
29602 * cgraphunit.c (symtab_node::native_rtl_p): New function.
29603 (symtab_node::needed_p): Don't assert for early assembly output
29604 for __RTL functions.
29605 (cgraph_node::finalize_function): Set "force_output" for __RTL
29607 (cgraph_node::analyze): Bail out early for __RTL functions.
29608 (analyze_functions): Update assertion to support __RTL functions.
29609 (cgraph_node::expand): Bail out early for __RTL functions.
29610 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
29612 * function.h (struct function): Update comment for field
29614 * gimple-expr.c: Include "tree-pass.h".
29615 (gimple_has_body_p): Return false for __RTL functions.
29616 * Makefile.in (OBJS): Add run-rtl-passes.o.
29617 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
29619 (gcc::pass_manager::get_clean_slate): New accessor.
29620 * passes.c: Include "insn-addr.h".
29621 (should_skip_pass_p): Add logging. Update logic for running
29622 "expand" to be compatible with both __GIMPLE and __RTL. Guard
29623 property-provider override so it is only done for gimple passes.
29625 (skip_pass): New function.
29626 (execute_one_pass): Call skip_pass when skipping passes.
29627 * read-md.c (md_reader::read_char): Support filtering
29628 the input to a subset of line numbers.
29629 (md_reader::md_reader): Initialize fields
29630 m_first_line and m_last_line.
29631 (md_reader::read_file_fragment): New function.
29632 * read-md.h (md_reader::read_file_fragment): New decl.
29633 (md_reader::m_first_line): New field.
29634 (md_reader::m_last_line): New field.
29635 * read-rtl-function.c (function_reader::create_function): Only
29636 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
29637 curr_properties. Set DECL_INITIAL to a dummy block.
29638 (read_rtl_function_body_from_file_range): New function.
29639 * read-rtl-function.h (read_rtl_function_body_from_file_range):
29641 * run-rtl-passes.c: New file.
29642 * run-rtl-passes.h: New file.
29644 2017-01-24 Jeff Law <law@redhat.com>
29646 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
29649 2017-01-24 Bin Cheng <bin.cheng@arm.com>
29651 PR tree-optimization/79159
29652 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
29653 (record_nonwrapping_iv): Improve boundary using above function if no
29654 value range information.
29656 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
29657 Martin Jambor <mjambor@suse.cz>
29659 * brig-builtins.def: New file.
29660 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
29661 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
29662 (DEF_HSAIL_SAT_BUILTIN): Likewise.
29663 (DEF_HSAIL_INTR_BUILTIN): Likewise.
29664 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
29665 * builtin-types.def (BT_INT8): New.
29666 (BT_INT16): Likewise.
29667 (BT_UINT8): Likewise.
29668 (BT_UINT16): Likewise.
29669 (BT_FN_ULONG): Likewise.
29670 (BT_FN_UINT_INT): Likewise.
29671 (BT_FN_UINT_ULONG): Likewise.
29672 (BT_FN_UINT_LONG): Likewise.
29673 (BT_FN_UINT_PTR): Likewise.
29674 (BT_FN_ULONG_PTR): Likewise.
29675 (BT_FN_INT8_FLOAT): Likewise.
29676 (BT_FN_INT16_FLOAT): Likewise.
29677 (BT_FN_UINT32_FLOAT): Likewise.
29678 (BT_FN_UINT16_FLOAT): Likewise.
29679 (BT_FN_UINT8_FLOAT): Likewise.
29680 (BT_FN_UINT64_FLOAT): Likewise.
29681 (BT_FN_UINT16_UINT32): Likewise.
29682 (BT_FN_UINT32_UINT16): Likewise.
29683 (BT_FN_UINT16_UINT16_UINT16): Likewise.
29684 (BT_FN_INT_PTR_INT): Likewise.
29685 (BT_FN_UINT_PTR_UINT): Likewise.
29686 (BT_FN_LONG_PTR_LONG): Likewise.
29687 (BT_FN_ULONG_PTR_ULONG): Likewise.
29688 (BT_FN_VOID_UINT64_UINT64): Likewise.
29689 (BT_FN_UINT8_UINT8_UINT8): Likewise.
29690 (BT_FN_INT8_INT8_INT8): Likewise.
29691 (BT_FN_INT16_INT16_INT16): Likewise.
29692 (BT_FN_INT_INT_INT): Likewise.
29693 (BT_FN_UINT_FLOAT_UINT): Likewise.
29694 (BT_FN_FLOAT_UINT_UINT): Likewise.
29695 (BT_FN_ULONG_UINT_UINT): Likewise.
29696 (BT_FN_ULONG_UINT_PTR): Likewise.
29697 (BT_FN_ULONG_ULONG_ULONG): Likewise.
29698 (BT_FN_UINT_UINT_UINT): Likewise.
29699 (BT_FN_VOID_UINT_PTR): Likewise.
29700 (BT_FN_UINT_UINT_PTR: Likewise.
29701 (BT_FN_UINT32_UINT64_PTR): Likewise.
29702 (BT_FN_INT_INT_UINT_UINT): Likewise.
29703 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
29704 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
29705 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
29706 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
29707 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
29708 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
29709 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
29710 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
29711 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
29712 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
29713 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
29714 * doc/frontends.texi: List BRIG FE.
29715 * doc/install.texi (Testing): Add BRIG tesring requirements.
29716 * doc/invoke.texi (Overall Options): Mention BRIG.
29717 * doc/standards.texi (Standards): Doucment BRIG HSA version.
29719 2017-01-24 Richard Biener <rguenther@suse.de>
29721 PR translation/79208
29722 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
29724 2017-01-24 Martin Jambor <mjambor@suse.cz>
29727 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
29728 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
29729 and known_contexts.
29731 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
29733 PR middle-end/79123
29734 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
29735 casts from signed to unsigned really don't have a range.
29737 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
29739 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
29740 GMP_RNDx for compatiblity.
29742 2017-01-24 Martin Liska <mliska@suse.cz>
29745 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
29746 that would prevent us to call alloca with -1 as argument.
29748 2017-01-24 Jakub Jelinek <jakub@redhat.com>
29750 * dwarf2out.c (output_compilation_unit_header, output_file_names):
29751 Avoid -Wformat-security warning.
29753 2017-01-23 Andrew Pinski <apinski@cavium.com>
29755 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
29758 2017-01-23 Martin Sebor <msebor@redhat.com>
29760 PR middle-end/78703
29761 * gimple-ssa-sprintf.c (warn_level): New global.
29762 (format_integer): Use it here and throughout the rest of the file.
29763 Use the same switch to compute sign as base.
29764 (maybe_warn): New function.
29765 (format_directive): Factor out warnings into maybe_warn.
29766 Add debugging output. Use warn_level.
29767 (add_bytes): Use warn_level.
29768 (pass_sprintf_length::compute_format_length): Add debugging output.
29769 (try_substitute_return_value): Same.
29770 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
29772 PR middle-end/78703
29773 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
29774 (struct fmtresult, format_integer, format_floating): Adjust.
29775 (fmtresult::fmtresult): Set max correctly in two argument ctor.
29776 (get_string_length, format_string,format_directive): Same.
29777 (pass_sprintf_length::compute_format_length): Same.
29778 (try_substitute_return_value): Simplify slightly.
29780 PR middle-end/78703
29781 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
29782 (fmtresult::operator+=): Outlined.
29783 (struct fmtresult): Add ctors.
29784 (struct conversion_spec): Rename...
29785 (struct directive): ...to this. Add and remove data members.
29786 (directive::set_width, directive::set_precision): New functions.
29787 (format_percent): Use fmtresult ctor.
29788 (get_width_and_precision): Remove.
29789 (format_integer): Make naming changes. Avoid computing width and
29791 (format_floating): Same. Adjust indentation.
29792 (format_character, format_none): New functions.
29793 (format_string): Moved character handling to format_character.
29794 (format_directive): Remove arguments, change return type.
29795 (parse_directive): New function.
29796 (pass_sprintf_length::compute_format_length): Move directive
29797 parsing to parse_directive.
29799 2017-01-23 Jakub Jelinek <jakub@redhat.com>
29801 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
29802 (assign_assembler_name_if_needed): ... this.
29803 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
29804 (assign_assembler_name_if_needed): ... this.
29805 (free_lang_data_in_cgraph): Adjust callers.
29806 * cgraphunit.c (cgraph_node::analyze): Likewise.
29807 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
29810 2017-01-23 Richard Biener <rguenther@suse.de>
29812 PR tree-optimization/79088
29813 PR tree-optimization/79188
29814 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
29815 resetting loop bounds after last path deletion. Reset loop
29816 bounds of the target loop, make code match the comments.
29817 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
29818 Make sure loops need no fixups.
29820 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29822 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
29823 exponent support with double type for first argument.
29824 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
29825 type returned by __builtin_vec_extract_sig,
29826 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
29827 functions from "vector int" to "vector unsigned int" or from
29828 "vector long long int" to "vector unsigned long long int".
29829 Changed type returned by __builtin_vec_extract_exp,
29830 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
29831 functions from "vector int" to "vector unsigned int" or from
29832 "vector long long int" to "vector unsigned long long int".
29833 Changed return type of __builtin_vec_test_data_class,
29834 __builtin_vec_test_data_class_sp, and
29835 __builtin_vec_test_data_class_dp from "vector int" to
29836 "vector bool int" or from "vector long long int" to "vector bool
29837 long long int" and changed second argument type from "unsigned
29838 int" to "int". Added new overloaded function forms "vector float
29839 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
29840 "vector float __builtin_vec_insert_exp_sp (vector float, vector
29841 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
29842 double, vector unsigned long long int)" and "vector double
29843 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
29844 long int)". Changed return type of
29845 __builtin_scalar_test_data_class and
29846 __builtin_scalar_test_data_class_sp and
29847 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
29848 int" and changed second argument from "unsigned int" to "int".
29849 Changed type returned by __builtin_scalar_test_neg,
29850 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
29851 from "int" to "bool int". Added new overloaded function form
29852 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
29853 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
29854 exponent double-precision with floating point first argument.
29855 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
29856 documentation of scalar_test_data_class, scalar_test_neg,
29857 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
29858 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
29859 vec_test_data_class built-in functions to reflect refinements in
29860 their type signatures.
29862 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
29864 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
29866 (aarch64_elf_asm_destructor): Likewise.
29868 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
29870 PR rtl-optimization/78634
29871 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
29872 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
29873 * ifcvt.c (noce_try_cmove): Add missing cost check.
29875 PR rtl-optimization/71724
29876 * combine.c (if_then_else_cond): Look for situations where it is
29877 beneficial to undo the work of one of the recursive calls.
29879 2017-01-23 Bin Cheng <bin.cheng@arm.com>
29881 PR tree-optimization/70754
29882 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
29883 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
29884 combined stmt before it if not NULL.
29885 (combine_chains): Process refs reversely and compute dominance point
29888 2017-01-23 Martin Liska <mliska@suse.cz>
29890 PR tree-optimization/79196
29891 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
29892 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
29894 (strlen_optimize_stmt): Call the renamed function.
29896 2017-01-23 Michael Matz <matz@suse.de>
29898 PR tree-optimization/78384
29899 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
29901 2017-01-23 Richard Biener <rguenther@suse.de>
29903 PR tree-optimization/79186
29904 * tree-vrp.c (register_new_assert_for): Make sure we've seen
29905 both incoming edges before moving an assert.
29907 2017-01-23 Martin Jambor <mjambor@suse.cz>
29909 * ipa-prop.c (load_from_param_1): Removed.
29910 (load_from_unmodified_param): Bits from load_from_param_1 put back
29912 (load_from_param): Removed.
29913 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
29914 with stmt. Reverted back to use of load_from_unmodified_param.
29916 2017-01-23 Martin Jambor <mjambor@suse.cz>
29919 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
29920 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
29921 field a pointer to garbage collected vector, mark lattices and
29922 ipcp_orig_node with GTY((skip)).
29923 (ipa_get_param_count): Adjust to descriptors being a pointer.
29924 (ipa_get_param): Likewise.
29925 (ipa_get_type): Likewise.
29926 (ipa_get_param_move_cost): Likewise.
29927 (ipa_set_param_used): Likewise.
29928 (ipa_get_controlled_uses): Likewise.
29929 (ipa_set_controlled_uses): Likewise.
29930 (ipa_is_param_used): Likewise.
29931 (ipa_node_params_t): Move into garbage collector. New methods insert
29933 (ipa_node_params_sum): Annotate wth GTY(()).
29934 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
29936 (ipa_load_from_parm_agg): Adjust declaration.
29937 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
29938 * ipa-profile.c (ipa_profile): Likewise.
29939 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
29940 (ipa_populate_param_decls): Make descriptors parameter garbage
29942 (ipa_dump_param): Adjust to descriptors being a pointer.
29943 (ipa_alloc_node_params): Likewise.
29944 (ipa_initialize_node_params): Likewise.
29945 (load_from_param_1): Make descriptors parameter garbage collected.
29946 (load_from_unmodified_param): Likewise.
29947 (load_from_param): Likewise.
29948 (ipa_load_from_parm_agg): Likewise.
29949 (ipa_node_params::~ipa_node_params): Removed.
29950 (ipa_free_all_node_params): Remove call to delete operator.
29951 (ipa_node_params_t::insert): New.
29952 (ipa_node_params_t::remove): Likewise.
29953 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
29954 copy known_csts and known_contexts vectors.
29955 (ipa_read_node_info): Adjust to descriptors being a pointer.
29956 (ipcp_modif_dom_walker): Make m_descriptors field garbage
29958 (ipcp_transform_function): Make descriptors variable garbage
29961 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
29963 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
29964 * config/i386/avx512dqintrin.h: Ditto.
29965 * config/i386/avx512fintrin.h: Ditto.
29966 * config/i386/i386.c: Handle new builtins.
29967 * config/i386/i386-builtin.def: Add new builtins.
29968 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
29969 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
29971 2017-01-23 Jakub Jelinek <jakub@redhat.com>
29972 Martin Liska <mliska@suse.cz>
29974 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
29975 * asan.c (asan_expand_poison_ifn): Support stores and use
29976 appropriate ASAN report function.
29977 * internal-fn.c (expand_ASAN_POISON_USE): New function.
29978 * internal-fn.def (ASAN_POISON_USE): Declare.
29979 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
29980 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
29981 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
29982 ASAN_POISON calls w/o LHS.
29983 * tree-ssa.c (execute_update_addresses_taken): Create clobber
29984 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
29985 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
29986 * gimplify.c (asan_poison_variables): Add attribute
29987 use_after_scope_memory to variables that really needs to live
29989 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
29990 having the attribute.
29992 2017-01-23 Martin Liska <mliska@suse.cz>
29994 * asan.c (create_asan_shadow_var): New function.
29995 (asan_expand_poison_ifn): Likewise.
29996 * asan.h (asan_expand_poison_ifn): New declaration.
29997 * internal-fn.c (expand_ASAN_POISON): Likewise.
29998 * internal-fn.def (ASAN_POISON): New builtin.
29999 * sanopt.c (pass_sanopt::execute): Expand
30000 asan_expand_poison_ifn.
30001 * tree-inline.c (copy_decl_for_dup_finish): Make function
30003 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
30004 * tree-ssa.c (is_asan_mark_p): New function.
30005 (execute_update_addresses_taken): Rewrite local variables
30006 (identified just by use-after-scope as addressable) into SSA.
30008 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
30010 * doc/install.texi (Specific): opensource.apple.com uses https
30011 now. Remove trailing slash.
30013 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
30015 * README.Portability: Remove note on an Irix compatibility issue.
30017 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
30019 * gcov.c (INCLUDE_ALGORITHM): Define.
30020 (INCLUDE_VECTOR): Define.
30021 No longer include <vector> and <algorithm> directly.
30023 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
30025 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
30027 * doc/invoke.texi (Code Gen Options): Ditto.
30029 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
30032 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
30034 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
30036 rtl-optimization/79125
30037 * cprop.c (local_cprop_pass): Handle cases where we make an
30038 unconditional trap.
30040 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
30044 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
30045 read from, for big endian.
30047 2017-01-20 Jiong Wang <jiong.wang@arm.com>
30049 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
30050 register pauth builtins for LP64 only.
30052 2017-01-20 Marek Polacek <polacek@redhat.com>
30055 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
30058 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
30060 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
30062 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
30063 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
30064 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
30066 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30069 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
30070 in big-endian mode when they are not a single duplicated value.
30072 2017-01-20 Richard Biener <rguenther@suse.de>
30074 * BASE-VER: Bump to 7.0.1.
30076 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
30078 * omp-low.c (omplow_simd_context): New struct. Use it...
30079 (lower_rec_simd_input_clauses): ...here and...
30080 (lower_rec_input_clauses): ...here to hold common data. Adjust all
30081 references to idx, lane, max_vf, is_simt.
30083 2017-01-20 Graham Markall <graham.markall@embecosm.com>
30085 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
30088 2017-01-20 Martin Jambor <mjambor@suse.cz>
30090 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
30091 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
30093 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
30094 (GTFILES): Rename hsa.c to hsa-common.c.
30095 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
30096 * hsa-dump.c: Likewise.
30097 * hsa-gen.c: Likewise.
30098 * hsa-regalloc.c: Likewise.
30099 * ipa-hsa.c: Likewise.
30100 * omp-expand.c: Likewise.
30101 * omp-low.c: Likewise.
30102 * toplev.c: Likewise.
30104 2017-01-20 Marek Polacek <polacek@redhat.com>
30107 * doc/invoke.texi: Document -Wduplicated-branches.
30108 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
30109 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
30110 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
30111 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
30112 return 0 only when not OEP_LEXICOGRAPHIC.
30113 (fold_build_cleanup_point_expr): Use the expression
30114 location when building CLEANUP_POINT_EXPR.
30115 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
30116 * tree.c (add_expr): Handle error_mark_node.
30118 2017-01-20 Martin Liska <mliska@suse.cz>
30121 * tree-profile.c (init_ic_make_global_vars): Do not call
30123 (gimple_init_gcov_profiler): Likewise.
30125 2017-01-20 Martin Liska <mliska@suse.cz>
30128 * cgraph.h (maybe_create_reference): Remove argument and
30130 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
30132 * ipa-cp.c (create_specialized_node): Likewise.
30133 * symtab.c (symtab_node::maybe_create_reference): Handle
30134 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
30136 2017-01-20 Martin Liska <mliska@suse.cz>
30138 * read-rtl-function.c (function_reader::create_function): Use
30139 build_decl instread of build_decl_stat.
30141 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
30143 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
30144 * config/i386/avx512dqintrin.h: Ditto.
30145 * config/i386/avx512fintrin.h: Ditto.
30146 * config/i386/i386-builtin-types.def: Add new types.
30147 * config/i386/i386.c: Handle new types.
30148 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
30149 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
30150 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
30151 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
30152 (__builtin_ia32_kshiftridi): New.
30153 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
30155 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
30159 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
30160 define to rs6000_init_stack_protect_guard.
30161 (rs6000_init_stack_protect_guard): New function.
30163 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
30164 Yunqiang Su <yunqiang.su@imgtec.com>
30166 * config.gcc (supported_defaults): Add madd4.
30167 (with_madd4): Add validation.
30168 (all_defaults): Add madd4.
30169 * config/mips/mips.opt (mmadd4): New option.
30170 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
30172 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
30174 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
30175 (ISA_HAS_FUSED_MADD4): Likewise.
30176 * doc/invoke.texi (-mmadd4): Document the new option.
30177 * doc/install.texi (--with-madd4): Document the new option.
30179 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30181 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
30182 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
30183 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
30184 (aarch64_init_pauth_hint_builtins): New.
30185 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
30186 (aarch64_expand_builtin): Expand new builtins.
30188 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30190 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
30191 * combine-stack-adj.c (no_unhandled_cfa): Handle
30192 REG_CFA_TOGGLE_RA_MANGLE.
30193 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
30194 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
30195 info for return address signing.
30196 (aarch64_expand_epilogue): Likewise.
30198 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30200 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
30201 * config/aarch64/aarch64-protos.h
30202 (aarch64_return_address_signing_enabled): New declaration.
30203 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
30205 (aarch64_expand_prologue): Sign return address before it's pushed onto
30207 (aarch64_expand_epilogue): Authenticate return address fetched from
30209 (aarch64_override_options): Sanity check for ILP32 and ISA level.
30210 (aarch64_attributes): New function attributes for "sign-return-address".
30211 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
30212 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
30213 ("*do_return"): Generate combined instructions according to key index.
30214 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
30215 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
30217 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
30218 * config/aarch64/aarch64.opt (msign-return-address=): New.
30219 * doc/extend.texi (AArch64 Function Attributes): Documents
30220 "sign-return-address=".
30221 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
30223 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
30225 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
30226 overall option summary.
30228 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30230 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
30231 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
30232 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
30233 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
30235 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
30237 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
30238 -mpower9-minmax by default for -mcpu=power9.
30239 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
30240 128-bit floating point.
30242 2017-01-20 Alan Modra <amodra@gmail.com>
30244 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
30245 optimizing for size.
30247 2017-01-20 Alan Modra <amodra@gmail.com>
30250 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
30251 for strcmp and strncmp from corresponding builtin decl.
30253 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
30255 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
30256 instead of i386/rtems-64.h.
30257 * config/i386/rtems-64.h: Remove.
30259 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
30263 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
30265 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
30267 2017-01-19 Tamar Christina <tamar.christina@arm.com>
30269 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
30270 Change int to HOST_WIDE_INT.
30271 * config/aarch64/aarch64-protos.h
30272 (aarch64_simd_gen_const_vector_dup): Likewise.
30273 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
30275 2017-01-19 David Malcolm <dmalcolm@redhat.com>
30277 * langhooks-def.h (lhd_type_for_size): New decl.
30278 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
30279 * langhooks.c (lhd_type_for_size): New function, taken from
30282 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
30284 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
30285 define_bypass for CR latency.
30286 (power9-cracked-alu): Update bypass latency and remove power9-branch.
30287 (power9-alu2): Add define_bypass for CR latency.
30289 (power9-mul): Update insn latency.
30290 (power9-mul-compare): Update insn latency, bypass latency and remove
30293 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30295 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
30297 * config/aarch64/aarch64.md
30298 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
30299 aarch64_nopcrelative_literal_loads.
30300 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
30302 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
30304 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
30305 TARGET_LOONGSON_3A.
30306 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
30308 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
30311 * config.gcc (supported_defaults): Add lxc1-sxc1.
30312 (with_lxc1_sxc1): Add validation.
30313 (all_defaults): Add lxc1-sxc1.
30314 * config/mips/mips.opt (mlxc1-sxc1): New option.
30315 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
30317 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
30318 __mips_no_lxc1_sxc1.
30319 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
30320 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
30321 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
30323 2017-01-19 Richard Biener <rguenther@suse.de>
30325 PR tree-optimization/72488
30326 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
30327 sure to restore SSA info.
30328 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
30330 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
30332 PR rtl-optimization/79121
30333 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
30334 of the inner type when shifting an extended value.
30336 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
30339 * symtab.c (symtab_node::equal_address_to): Fix comparing of
30340 interposable aliases.
30342 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
30345 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
30346 Use the evmergelohi instruction.
30347 (mov_si<mode>_e500_subreg4_2_le): Likewise.
30348 (mov_sitf_e500_subreg8_2_be): Likewise.
30349 (mov_sitf_e500_subreg12_2_le): Likewise.
30350 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
30351 (mov_si<mode>_e500_subreg4_2_be): Likewise.
30352 (mov_sitf_e500_subreg8_2_le): Likewise.
30353 (mov_sitf_e500_subreg12_2_be): Likewise.
30355 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30357 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
30358 attribute from vecsimple to vecperm.
30359 (altivec_vbpermq2): Likewise.
30361 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30364 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
30366 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30367 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
30368 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
30369 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
30370 case where N arg is SIZE_MAX.
30371 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
30372 (cmpstrsi): Add pattern.
30374 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30376 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30377 __builtin_vec_revb builtins.
30378 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
30379 built-in functions to support generation of the ISA 3.0 XXBR<x>
30380 vector byte reverse instructions.
30381 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
30382 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
30383 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
30384 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
30385 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
30386 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
30387 (P9V_BUILTIN_VEC_REVB): Likewise.
30388 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
30389 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
30390 (p9_xxbrq_v16qi): Likewise.
30391 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
30392 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
30393 (p9_xxbrh_v8hi): Likewise.
30394 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
30395 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
30396 vec_revb built-in functions.
30398 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
30400 PR rtl-optimization/78952
30401 * config/i386/i386.md (any_extract): New code iterator.
30402 (*insvqi_2): Use any_extract for source operand.
30403 (*insvqi_3): Use any_shiftrt for source operand.
30405 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
30407 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
30409 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
30411 2017-01-18 Matthias Klose <doko@ubuntu.com>
30413 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
30415 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30417 * config/rs6000/altivec.h (vec_bperm): Change #define.
30418 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
30419 (altivec_vbpermq2): New define_insn.
30420 (altivec_vbpermd): Likewise.
30421 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
30422 function interface.
30423 (VBPERMD): Likewise.
30424 (VBPERM): New polymorphic function interface.
30425 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
30426 Add entries for P9V_BUILTIN_VEC_VBPERM.
30427 * doc/extend.texi: Add interfaces for vec_bperm.
30429 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30431 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
30432 first letter of error messages.
30433 (s390_resolve_overloaded_builtin): Likewise.
30434 * config/s390/s390.c (s390_expand_builtin): Likewise.
30435 (s390_invalid_arg_for_unprototyped_fn): Likewise.
30436 (s390_valid_target_attribute_inner_p): Likewise.
30437 * config/s390/s390.md ("tabort"): Likewise.
30439 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
30441 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
30442 (ISA_AVOID_DIV_HILO): New macro.
30443 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
30444 (ISA_HAS_DDIV): Likewise.
30446 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30448 * doc/invoke.texi (fabi-version): Correct number of occurrences.
30450 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30452 * doc/invoke.texi (fabi-version): Spelling fix.
30454 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30457 * doc/invoke.texi (fabi-version): Mention mangling fix for
30460 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30463 * doc/invoke.texi (fabi-version): Document discriminator mangling.
30465 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
30468 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
30469 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
30471 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
30473 (stack_protect_test): Ditto.
30474 * config/rs6000/rs6000.opt (mstack-protector-guard=,
30475 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
30477 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
30478 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
30479 -mstack-protector-guard-offset=.
30480 (RS/6000 and PowerPC Options): Ditto.
30482 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
30484 * config/i386/i386.h (MASK_CLASS_P): New define.
30485 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
30486 there are no registers from different register sets also when
30487 mask registers are used. Update function comment.
30488 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
30489 to (*k/*r) and (*k/*km) alternatives.
30491 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
30493 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
30494 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
30495 (EH_RETURN_HANDLER_RTX): New define.
30496 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
30497 Force frame pointer in EH return functions.
30498 (aarch64_expand_epilogue): Add barrier for eh_return.
30499 (aarch64_final_eh_return_addr): Remove.
30500 (aarch64_eh_return_handler_rtx): New function.
30501 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
30503 (aarch64_eh_return_handler_rtx): New prototype.
30505 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30507 * config/rs6000/altivec.h (vec_rlmi): New #define.
30508 (vec_vrlnm): Likewise.
30509 (vec_rlnm): Likewise.
30510 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
30511 (UNSPEC_VRLNM): Likewise.
30512 (VIlong): New mode iterator.
30513 (altivec_vrl<VI_char>mi): New define_insn.
30514 (altivec_vrl<VI_char>nm): Likewise.
30515 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
30517 (VRLDNM): Likewise.
30518 (RLNM): New polymorphic function entry.
30519 (VRLWMI): New monomorphic function entry.
30520 (VRLDMI): Likewise.
30521 (RLMI): New polymorphic function entry.
30522 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
30523 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
30524 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
30527 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30530 * dwarf2out.c (field_byte_offset): Restore the
30531 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
30532 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
30533 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
30536 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
30539 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
30541 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30544 * dwarf2out.c (add_data_member_location_attribute): For constant
30545 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
30546 instead of DW_AT_data_member_location, DW_AT_bit_offset and
30547 DW_AT_byte_size attributes.
30549 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
30551 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
30552 after forcing to constant memory when the code model is medium.
30554 2017-01-17 Julia Koval <julia.koval@intel.com>
30557 * config/i386/avx512fintrin.h
30558 (_mm512_i32gather_ps): Change __addr type to void const*.
30559 (_mm512_mask_i32gather_ps): Ditto.
30560 (_mm512_i32gather_pd): Ditto.
30561 (_mm512_mask_i32gather_pd): Ditto.
30562 (_mm512_i64gather_ps): Ditto.
30563 (_mm512_mask_i64gather_ps): Ditto.
30564 (_mm512_i64gather_pd): Ditto.
30565 (_mm512_mask_i64gather_pd): Ditto.
30566 (_mm512_i32gather_epi32): Ditto.
30567 (_mm512_mask_i32gather_epi32): Ditto.
30568 (_mm512_i32gather_epi64): Ditto.
30569 (_mm512_mask_i32gather_epi64): Ditto.
30570 (_mm512_i64gather_epi32): Ditto.
30571 (_mm512_mask_i64gather_epi32): Ditto.
30572 (_mm512_i64gather_epi64): Ditto.
30573 (_mm512_mask_i64gather_epi64): Ditto.
30574 (_mm512_i32scatter_ps): Change __addr type to void*.
30575 (_mm512_mask_i32scatter_ps): Ditto.
30576 (_mm512_i32scatter_pd): Ditto.
30577 (_mm512_mask_i32scatter_pd): Ditto.
30578 (_mm512_i64scatter_ps): Ditto.
30579 (_mm512_mask_i64scatter_ps): Ditto.
30580 (_mm512_i64scatter_pd): Ditto.
30581 (_mm512_mask_i64scatter_pd): Ditto.
30582 (_mm512_i32scatter_epi32): Ditto.
30583 (_mm512_mask_i32scatter_epi32): Ditto.
30584 (_mm512_i32scatter_epi64): Ditto.
30585 (_mm512_mask_i32scatter_epi64): Ditto.
30586 (_mm512_i64scatter_epi32): Ditto.
30587 (_mm512_mask_i64scatter_epi32): Ditto.
30588 (_mm512_i64scatter_epi64): Ditto.
30589 (_mm512_mask_i64scatter_epi64): Ditto.
30590 * config/i386/avx512pfintrin.h
30591 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
30592 (_mm512_mask_prefetch_i32gather_ps): Ditto.
30593 (_mm512_mask_prefetch_i64gather_pd): Ditto.
30594 (_mm512_mask_prefetch_i64gather_ps): Ditto.
30595 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
30596 (_mm512_prefetch_i32scatter_ps): Ditto.
30597 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
30598 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
30599 (_mm512_prefetch_i64scatter_pd): Ditto.
30600 (_mm512_prefetch_i64scatter_ps): Ditto.
30601 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
30602 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
30603 * config/i386/avx512vlintrin.h
30604 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
30605 (_mm_mmask_i32gather_ps): Ditto.
30606 (_mm256_mmask_i32gather_pd): Ditto.
30607 (_mm_mmask_i32gather_pd): Ditto.
30608 (_mm256_mmask_i64gather_ps): Ditto.
30609 (_mm_mmask_i64gather_ps): Ditto.
30610 (_mm256_mmask_i64gather_pd): Ditto.
30611 (_mm_mmask_i64gather_pd): Ditto.
30612 (_mm256_mmask_i32gather_epi32): Ditto.
30613 (_mm_mmask_i32gather_epi32): Ditto.
30614 (_mm256_mmask_i32gather_epi64): Ditto.
30615 (_mm_mmask_i32gather_epi64): Ditto.
30616 (_mm256_mmask_i64gather_epi32): Ditto.
30617 (_mm_mmask_i64gather_epi32): Ditto.
30618 (_mm256_mmask_i64gather_epi64): Ditto.
30619 (_mm_mmask_i64gather_epi64): Ditto.
30620 (_mm256_i32scatter_ps): Change __addr type to void*.
30621 (_mm256_mask_i32scatter_ps): Ditto.
30622 (_mm_i32scatter_ps): Ditto.
30623 (_mm_mask_i32scatter_ps): Ditto.
30624 (_mm256_i32scatter_pd): Ditto.
30625 (_mm256_mask_i32scatter_pd): Ditto.
30626 (_mm_i32scatter_pd): Ditto.
30627 (_mm_mask_i32scatter_pd): Ditto.
30628 (_mm256_i64scatter_ps): Ditto.
30629 (_mm256_mask_i64scatter_ps): Ditto.
30630 (_mm_i64scatter_ps): Ditto.
30631 (_mm_mask_i64scatter_ps): Ditto.
30632 (_mm256_i64scatter_pd): Ditto.
30633 (_mm256_mask_i64scatter_pd): Ditto.
30634 (_mm_i64scatter_pd): Ditto.
30635 (_mm_mask_i64scatter_pd): Ditto.
30636 (_mm256_i32scatter_epi32): Ditto.
30637 (_mm256_mask_i32scatter_epi32): Ditto.
30638 (_mm_i32scatter_epi32): Ditto.
30639 (_mm_mask_i32scatter_epi32): Ditto.
30640 (_mm256_i32scatter_epi64): Ditto.
30641 (_mm256_mask_i32scatter_epi64): Ditto.
30642 (_mm_i32scatter_epi64): Ditto.
30643 (_mm_mask_i32scatter_epi64): Ditto.
30644 (_mm256_i64scatter_epi32): Ditto.
30645 (_mm256_mask_i64scatter_epi32): Ditto.
30646 (_mm_i64scatter_epi32): Ditto.
30647 (_mm_mask_i64scatter_epi32): Ditto.
30648 (_mm256_i64scatter_epi64): Ditto.
30649 (_mm256_mask_i64scatter_epi64): Ditto.
30650 (_mm_i64scatter_epi64): Ditto.
30651 (_mm_mask_i64scatter_epi64): Ditto.
30652 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
30653 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
30654 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
30655 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
30656 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
30657 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
30658 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
30659 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
30660 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
30661 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
30662 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
30663 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
30664 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
30665 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
30666 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
30667 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
30668 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
30669 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
30670 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
30671 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
30672 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
30673 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
30674 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
30675 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
30676 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
30677 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
30678 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
30679 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
30680 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
30681 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
30682 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
30683 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
30684 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
30685 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
30686 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
30687 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
30688 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
30689 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
30690 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
30691 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
30692 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
30693 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
30694 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
30695 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
30696 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
30697 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
30698 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
30699 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
30700 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
30701 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
30702 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
30703 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
30704 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
30705 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
30706 definitions accordingly.
30708 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
30709 Kuan-Lin Chen <kuanlinchentw@gmail.com>
30712 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
30715 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
30718 * ira-conflicts.c (ira_build_conflicts): Update total conflict
30719 hard regs for inner regno.
30721 2017-01-17 Martin Liska <mliska@suse.cz>
30724 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
30725 assumption and add comment.
30727 2017-01-17 Nathan Sidwell <nathan@acm.org>
30729 * ipa-visibility.c (localize_node): New function, broken out of ...
30730 (function_and_variable_visibility): ... here. Call it.
30732 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
30734 PR middle-end/77445
30735 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
30736 correctly set frequency of oudgoing edge.
30737 (duplicate_thread_path): Fix profile updating.
30739 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30742 * configure.ac: Add GCC_BASE_VER.
30743 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
30744 version from BASE-VER file.
30745 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
30746 (gcc.o): Depend on $(BASEVER).
30747 * common.opt (dumpfullversion): New option.
30748 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
30749 * doc/invoke.texi: Document -dumpfullversion.
30750 * doc/install.texi: Document --with-gcc-major-version-only.
30751 * configure: Regenerated.
30753 2017-01-17 Richard Biener <rguenther@suse.de>
30755 PR tree-optimization/71433
30756 * tree-vrp.c (register_new_assert_for): Merge same asserts
30757 on all incoming edges.
30758 (process_assert_insertions_for): Handle insertions at the
30761 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
30763 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
30764 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
30766 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
30769 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
30772 2017-01-17 Alan Modra <amodra@gmail.com>
30775 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
30776 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
30777 symbolic stack limit when pic.
30779 2017-01-16 Martin Sebor <msebor@redhat.com>
30781 PR tree-optimization/78608
30782 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
30784 2017-01-16 Jeff Law <law@redhat.com>
30787 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
30788 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
30789 for several include directories that may be relative to sysroot.
30790 * config/i386/x-mingw32 (gplus_includedir): Define.
30791 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
30792 (native_system_includedir): Likewise.
30793 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
30794 override if TARGET_SYSTEM_ROOT is defined.
30795 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
30797 PR tree-optimization/79090
30798 PR tree-optimization/33562
30799 PR tree-optimization/61912
30800 PR tree-optimization/77485
30801 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
30802 and computed trims into the dump file.
30804 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
30806 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
30808 2017-01-16 Jakub Jelinek <jakub@redhat.com>
30811 * gimplify.c (gimplify_init_constructor): If want_value and
30812 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
30816 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
30817 sequence. Formatting fixes.
30818 (doloop_optimize): Formatting fixes.
30821 * gcc.c (debug_level_greater_than_spec_func): New function.
30822 (static_spec_functions): Add debug-level-gt spec function.
30823 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
30825 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
30826 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
30827 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
30828 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
30829 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
30830 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
30832 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
30834 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
30835 QImode fixups to general and mask registers only.
30837 2017-01-16 Carl Love <cel@us.ibm.com>
30839 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
30840 for built-in functions
30841 vector signed char vec_nabs (vector signed char)
30842 vector signed short vec_nabs (vector signed short)
30843 vector signed int vec_nabs (vector signed int)
30844 vector signed long long vec_nabs (vector signed long long)
30845 vector float vec_nabs (vector float)
30846 vector double vec_nabs (vector double)
30847 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
30849 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
30850 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
30851 * doc/extend.texi: Update the documentation file for the new built-in
30854 2017-01-16 Martin Sebor <msebor@redhat.com>
30856 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
30859 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30861 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
30862 UNSPEC_VSX__XXSPLTD to require special splat handling.
30864 2017-01-16 David Malcolm <dmalcolm@redhat.com>
30867 * system.h: Poison strndup.
30869 2017-01-16 Alan Modra <amodra@gmail.com>
30872 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
30875 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
30877 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
30879 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
30881 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
30882 call recog here. Assert that INSN_CODE (insn) is non-negative.
30884 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30887 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
30889 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
30890 in the currently scheduled RTL fragment.
30892 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30894 PR rtl-optimization/78751
30895 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
30898 2017-01-14 Jeff Law <law@redhat.com>
30900 PR tree-optimization/79090
30901 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
30902 variable length stores.
30903 (compute_trims): Delete dead assignment to *trim_tail.
30904 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
30907 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
30909 PR rtl-optimization/78626
30910 PR rtl-optimization/78727
30911 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
30912 of a block, and split such blocks after everything else is finished.
30914 2017-01-14 Alan Modra <amodra@gmail.com>
30917 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
30918 target legitimate_combined_insn.
30919 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
30920 (rs6000_legitimate_combined_insn): New function.
30921 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
30923 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
30924 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
30925 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
30927 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
30929 * doc/frontends.texi (G++ and GCC): Remove references to Java.
30931 2017-01-13 Jeff Law <law@redhat.com>
30933 PR tree-optimization/33562
30934 PR tree-optimization/61912
30935 PR tree-optimization/77485
30936 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
30938 (delete_dead_assignment): Likewise.
30939 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
30940 statement to delete_dead_call and delete_dead_assignment.
30942 2017-01-13 David Malcolm <dmalcolm@redhat.com>
30945 * substring-locations.c (format_warning_va): Strengthen case 1 so
30946 that both endpoints of the substring must be within the format
30947 range for just the substring to be printed.
30949 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
30951 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
30952 * config/i386/i386.c (ix86_target_string): Add missing options
30953 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
30954 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
30955 flags_other and ix86_target_other to flags2_other. Display unknown
30957 (ix86_valid_target_attribute_inner_p): Add missing options and
30958 reorder options by implied ISAs, as in ix86_target_string.
30960 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
30962 * hash-table.h (hash_table::too_empty_p): New function.
30963 (hash_table::expand): Use it.
30964 (hash_table::traverse): Likewise.
30965 (hash_table::empty_slot): Use sizeof (value_type) instead of
30966 sizeof (PTR) to convert bytes to elements. Shrink the table
30967 if the current size is excessive for the current number of
30970 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
30972 * ira-costs.c (record_reg_classes): Break from the inner loop
30973 early once alt_fail is known to be true. Update outer loop
30974 handling accordingly.
30976 2017-01-13 Jeff Law <law@redhat.com>
30978 * tree-ssa-dse.c (decrement_count): New function.
30979 (increment_start_addr, maybe_trim_memstar_call): Likewise.
30980 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
30981 when we know the partially dead statement is a mem* function.
30983 PR tree-optimization/61912
30984 PR tree-optimization/77485
30985 * tree-ssa-dse.c: Include expr.h.
30986 (maybe_trim_constructor_store): New function.
30987 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
30989 PR tree-optimization/33562
30990 PR tree-optimization/61912
30991 PR tree-optimization/77485
30992 * doc/invoke.texi: Document new dse-max-object-size param.
30993 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
30994 * tree-ssa-dse.c: Include params.h.
30995 (dse_store_status): New enum.
30996 (initialize_ao_ref_for_dse): New, partially extracted from
30998 (valid_ao_ref_for_dse, normalize_ref): New.
30999 (setup_live_bytes_from_ref, compute_trims): Likewise.
31000 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
31001 (maybe_trim_partially_dead_store): Likewise.
31002 (maybe_trim_complex_store): Likewise.
31003 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
31004 Track what bytes live from the original store. Return tri-state
31005 for dead, partially dead or live.
31006 (dse_dom_walker): Add constructor, destructor and new private members.
31007 (delete_dead_call, delete_dead_assignment): New extracted from
31009 (dse_optimize_stmt): Make a member of dse_dom_walker.
31010 Use initialize_ao_ref_for_dse.
31012 PR tree-optimization/33562
31013 PR tree-optimization/61912
31014 PR tree-optimization/77485
31015 * sbitmap.h (bitmap_count_bits): Prototype.
31016 (bitmap_clear_range, bitmap_set_range): Likewise.
31017 * sbitmap.c (bitmap_clear_range): New function.
31018 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
31020 2017-01-13 Martin Liska <mliska@suse.cz>
31023 * function.c (set_cfun): Add new argument force.
31024 * function.h (set_cfun): Likewise.
31025 * ipa-inline-transform.c (inline_call): Use the function when
31026 strict alising from is dropped for function we inline to.
31028 2017-01-13 Richard Biener <rguenther@suse.de>
31030 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
31031 for dumping GIMPLE INTEGER_CSTs.
31033 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31035 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
31036 to 201112L since C++17.
31038 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
31041 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
31042 if -fsanitize=kernel-address is present.
31044 2017-01-13 Richard Biener <rguenther@suse.de>
31046 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
31047 as _Literal ( type ) number in case usual suffixes do not
31048 preserve all information.
31050 2017-01-13 Richard Biener <rguenther@suse.de>
31052 PR tree-optimization/77283
31053 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
31054 and ssa-iterators.h.
31055 (is_feasible_trace): Implement a cost model based on joiner
31058 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
31061 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
31062 char or short to __float128/_Float128 directly.
31064 2017-01-12 Martin Sebor <msebor@redhat.com>
31066 to -Wformat-overflow.
31067 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
31068 (min_bytes_remaining): Same.
31069 (get_string_length): Same.
31070 (format_string): Same.
31071 (format_directive): Same.
31073 (pass_sprintf_length::handle_gimple_call): Same.
31075 2017-01-12 Jakub Jelinek <jakub@redhat.com>
31077 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
31078 info.nowrite calls with no lhs that can't throw. Return bool
31079 whether gsi_remove has been called or not.
31080 (pass_sprintf_length::handle_gimple_call): Return bool whether
31081 try_substitute_return_value called gsi_remove. Formatting fix.
31082 (pass_sprintf_length::execute): Don't use gsi_remove if
31083 handle_gimple_call returned true.
31086 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
31087 be removed due to side-effects, don't remove following barrier nor
31088 turn the successor edge into fallthru edge.
31090 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31093 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
31094 element-reversing loads and stores as not swappable.
31096 2017-01-12 Nathan Sidwell <nathan@acm.org>
31097 Nicolai Stange <nicstange@gmail.com>
31099 * combine.c (try_combine): Don't ignore result of overlap checking
31100 loop. Combine overlap & asm check into single loop.
31102 2017-01-12 Richard Biener <rguenther@suse.de>
31104 * tree-pretty-print.c (dump_generic_node): Provide -gimple
31105 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
31107 2017-01-12 Richard Biener <rguenther@suse.de>
31109 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
31110 and TS_TARGET_OPTION directly derive from TS_BASE.
31111 * tree-core.h (tree_optimization_option): Derive from tree_base.
31112 (tree_target_option): Likewise.
31114 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
31116 * config/i386/i386.c (memory_address_length): Increase len
31117 only when rip_relative_addr_p returns false.
31119 2017-01-11 Julia Koval <julia.koval@intel.com>
31121 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
31122 (OPTION_MASK_ISA_SGX_SET): New.
31123 (ix86_handle_option): Handle OPT_msgx.
31124 * config.gcc: Added sgxintrin.h.
31125 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
31126 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
31127 * config/i386/i386.c (ix86_target_string): Add -msgx.
31129 (ix86_option_override_internal): Handle new options.
31130 (ix86_valid_target_attribute_inner_p): Add sgx.
31131 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
31132 * config/i386/i386.opt: Add msgx.
31133 * config/i386/sgxintrin.h: New file.
31134 * config/i386/x86intrin.h: Add sgxintrin.h.
31136 2017-01-11 Jakub Jelinek <jakub@redhat.com>
31139 * fold-const.c (maybe_nonzero_address): Return 1 for function
31141 (tree_single_nonzero_warnv_p): Don't handle function local objects
31145 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
31148 2017-01-11 David Malcolm <dmalcolm@redhat.com>
31151 * opts.c: Include "spellcheck.h"
31152 (struct string_fragment): New struct.
31153 (struct edit_distance_traits<const string_fragment &>): New
31155 (get_closest_sanitizer_option): New function.
31156 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
31158 2017-01-11 Jakub Jelinek <jakub@redhat.com>
31160 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
31162 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
31163 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
31164 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
31165 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
31166 for initial die_offset if dwarf_split_debug_info.
31167 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
31168 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
31170 (output_skeleton_debug_sections): Formatting fix. Use
31171 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
31172 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
31174 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
31176 * config/arm/cortex-a53.md: Add bypasses for
31177 cortex_a53_r2f_cvt.
31178 (cortex_a53_r2f): Only use for transfers.
31179 (cortex_a53_f2r): Likewise.
31180 (cortex_a53_r2f_cvt): Add reservation for conversions.
31181 (cortex_a53_f2r_cvt): Likewise.
31183 2017-01-11 Tamar Christina <tamar.christina@arm.com>
31185 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
31186 to all inlined functions, change static to extern.
31188 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
31191 * config/arm/arm.c (legitimize_pic_address): Handle reference to
31193 (arm_assemble_integer): Likewise.
31195 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
31197 * config.gcc: Use new awk script to check CPU, FPU and architecture
31198 parameters for --with-... options.
31199 * config/arm/parsecpu.awk: New file
31200 * config/arm/arm-cpus.in: New file.
31201 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
31203 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
31205 * config/arm/t-arm: Update dependency rules.
31206 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
31207 of processing .def files.
31208 * config/arm/genopt.sh: Deleted.
31209 * config/arm/gentune.sh: Deleted.
31210 * config/arm/arm-cores.def: Deleted.
31211 * config/arm/arm-arches.def: Deleted.
31212 * config/arm/arm-fpus.def: Deleted.
31213 * config/arm/arm-tune.md: Regenerated.
31214 * config/arm/arm-tables.opt: Regenerated.
31215 * config/arm/arm-cpu.h: New generated file.
31216 * config/arm/arm-cpu-data.h: New generated file.
31217 * config/arm/arm-cpu-cdata.h: New generated file.
31219 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
31222 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
31224 (input_varpool_node): Unpack dynamically_initialized bit.
31226 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
31228 PR rtl-optimization/79032
31229 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
31230 the alignment of the adjusted memory reference against that of MODE,
31231 instead of the alignment of the original memory reference.
31233 2017-01-11 Martin Jambor <mjambor@suse.cz>
31235 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
31237 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
31238 decorated functions.
31240 2017-01-11 Richard Biener <rguenther@suse.de>
31242 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
31243 set range/nonnull info for PHI results. Do not set it on
31244 stmts marked for removal.
31246 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
31248 * expr.c (store_field): In the bitfield case, fetch the return value
31249 from the registers before applying a single big-endian adjustment.
31250 Always do a final load for a BLKmode value not larger than a word.
31252 2017-01-10 David Malcolm <dmalcolm@redhat.com>
31255 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
31256 that we correctly handle column numbers greater than
31257 LINE_MAP_MAX_COLUMN_NUMBER.
31259 2017-01-10 Martin Sebor <msebor@redhat.com>
31261 PR middle-end/78245
31262 * gimple-ssa-sprintf.c (get_destination_size): Call
31263 {init,fini}object_sizes.
31264 * tree-object-size.c (addr_object_size): Adjust.
31265 (pass_through_call): Adjust.
31266 (pass_object_sizes::execute): Adjust.
31267 * tree-object-size.h (fini_object_sizes): Declare.
31269 2017-01-10 Martin Sebor <msebor@redhat.com>
31271 PR tree-optimization/78775
31272 * builtins.c (get_size_range): Move...
31273 * calls.c: ...to here.
31274 (alloc_max_size): Accept zero argument.
31275 (operand_signed_p): Remove.
31276 (maybe_warn_alloc_args_overflow): Call get_size_range.
31277 * calls.h (get_size_range): Declare.
31279 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
31281 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
31282 from TI's devices.csv file as of September 2016.
31283 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
31285 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
31287 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
31288 * doc/invoke.texi: Likewise.
31289 * doc/md.texi: Likewise.
31290 * doc/objc.texi: Likewise.
31292 2017-01-10 Joshua Conner <joshconner@google.com>
31294 * config/arm/fuchsia-elf.h: New file.
31295 * config/fuchsia.h: New file.
31296 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
31297 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
31299 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
31301 2016-01-10 Richard Biener <rguenther@suse.de>
31303 PR tree-optimization/79034
31304 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
31305 Propagate out degenerate PHIs in the joiner.
31307 2017-01-10 Martin Liska <mliska@suse.cz>
31309 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
31310 (sort_congruence_classes_by_decl_uid): Likewise.
31311 (sort_congruence_class_groups_by_decl_uid): Likewise.
31312 (sem_item_optimizer::merge_classes): Sort class, groups in these
31313 classes and members in the groups by DECL_UID of declarations.
31314 This would make merge operations stable.
31316 2017-01-10 Martin Liska <mliska@suse.cz>
31318 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
31319 usage of m_classes_vec.
31320 (sem_item_optimizer::~sem_item_optimizer): Likewise.
31321 (sem_item_optimizer::get_group_by_hash): Likewise.
31322 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
31323 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
31324 (sem_item_optimizer::verify_classes): Likewise.
31325 (sem_item_optimizer::process_cong_reduction): Likewise.
31326 (sem_item_optimizer::dump_cong_classes): Likewise.
31327 (sem_item_optimizer::merge_classes): Likewise.
31328 * ipa-icf.h (congruence_class_hash): Rename from
31329 congruence_class_group_hash. Remove declaration of m_classes_vec.
31331 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
31333 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
31334 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
31335 * config.gcc: Add avx512vpopcntdqintrin.h.
31336 * config/i386/avx512vpopcntdqintrin.h: New.
31337 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
31338 * config/i386/i386-builtin-types.def: Add new types.
31339 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
31340 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
31341 __builtin_ia32_vpopcountq_v8di_mask): New.
31342 * config/i386/i386-c.c (ix86_target_macros_internal): Define
31343 __AVX512VPOPCNTDQ__.
31344 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
31345 (PTA_AVX512VPOPCNTDQ): Define.
31346 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
31347 TARGET_AVX512VPOPCNTDQ_P): Define.
31348 * config/i386/i386.opt: Add mavx512vpopcntdq.
31349 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
31350 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
31352 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31354 PR middle-end/77484
31355 * predict.def (PRED_CALL): Set to 67.
31357 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
31359 * expr.c (store_field): In the bitfield case, if the value comes from
31360 a function call and is of an aggregate type returned in registers, do
31361 not modify the field mode; extract the value in all cases if the mode
31362 is BLKmode and the size is not larger than a word.
31364 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
31367 * config/i386/cpuid.h: Fix undefined behavior.
31369 2017-01-04 Jeff Law <law@redhat.com>
31371 PR tree-optimization/79007
31372 PR tree-optimization/67955
31373 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
31374 conservative for pt.null when flag_non_call_exceptions is on.
31376 2017-01-09 Jakub Jelinek <jakub@redhat.com>
31378 PR translation/79019
31379 PR translation/79020
31380 * params.def (PARAM_INLINE_MIN_SPEEDUP,
31381 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
31382 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
31384 * config/avr/avr.opt (maccumulate-args): Likewise.
31385 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
31386 * common.opt (freport-bug): Likewise.
31387 * cif-code.def (CIF_FINAL_ERROR): Likewise.
31388 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
31389 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
31390 translatable string.
31391 * config/i386/i386.c (function_value_32): Likewise.
31392 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
31393 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
31395 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
31396 * common/config/msp430/msp430-common.c (msp430_handle_option):
31398 * symtab.c (symtab_node::verify_base): Likewise.
31399 * opts.c (set_debug_level): Likewise.
31400 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
31401 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
31402 missing whitespace to translatable strings.
31403 * config/avr/avr.md (bswapsi2): Fix typo in comment.
31404 * config/sh/superh.h: Likewise.
31405 * config/i386/xopintrin.h: Likewise.
31406 * config/i386/znver1.md: Likewise.
31407 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
31408 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
31409 * double-int.h (struct double_int): Likewise.
31410 * double-int.c (div_and_round_double): Likewise.
31411 * wide-int.cc: Likewise.
31412 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
31413 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
31414 * cfgcleanup.c (crossjumps_occured): Renamed to ...
31415 (crossjumps_occurred): ... this.
31416 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
31419 PR tree-optimization/78899
31420 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
31421 returning bool return struct loop *, NULL for failure and the new
31423 (versionable_outer_loop_p): Don't version outer loop if it has
31424 dont_vectorized bit set.
31425 (tree_if_conversion): When versioning outer loop, ensure
31426 tree_if_conversion is performed also on the inner loop of the
31427 non-vectorizable outer loop copy.
31428 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
31429 LOOP_VECTORIZED in inner loop of the scalar outer loop and
31430 prevent vectorization of it.
31431 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
31432 the outer loop vectorization of the non-scalar version is attempted
31433 before vectorization of the inner loop in scalar version. If
31434 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
31435 vectorization of its inner loop.
31436 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
31437 has 2 inner loops, rename also on edges from bb whose single pred
31438 is outer_loop->header. Fix typo in function comment.
31440 2017-01-09 Martin Sebor <msebor@redhat.com>
31443 * asan.c (asan_emit_stack_protection): Increase local buffer size
31444 to avoid snprintf truncation warning.
31446 2017-01-09 Andrew Pinski <apinski@cavium.com>
31448 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
31449 to reference thunderx2t99 for the tuning structure
31450 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
31452 (thunderx2t99_extra_costs): This.
31453 * config/aarch64/aarch64-tune.md: Regenerate.
31454 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
31455 (vulcan_addrcost_table): This.
31456 (vulcan_regmove_cost): Rename to ...
31457 (thunderx2t99_regmove_cost): This.
31458 (vulcan_vector_cost): Rename to ...
31459 (thunderx2t99_vector_cost): this.
31460 (vulcan_branch_cost): Rename to ...
31461 (thunderx2t99_branch_cost): This.
31462 (vulcan_tunings): Rename to ...
31463 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
31464 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
31466 2017-01-09 Martin Jambor <mjambor@suse.cz>
31470 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
31471 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
31472 (propagate_vr_accross_jump_function): Use the above function for all
31473 value range computations for pass-through jump functions and type
31474 converasion from explicit value range values.
31475 (ipcp_propagate_stage): Do not attempt to deduce types of formal
31476 parameters from TYPE_ARG_TYPES.
31477 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
31478 (ipa_write_node_info): Stream type of the actual argument.
31479 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
31481 2017-01-09 Martin Liska <mliska@suse.cz>
31484 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
31485 (lookup_compiler): Do not show error message with have_E.
31487 2017-01-09 Jakub Jelinek <jakub@redhat.com>
31489 PR tree-optimization/78938
31490 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
31491 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
31492 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
31493 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
31496 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31498 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
31501 2017-01-09 Richard Biener <rguenther@suse.de>
31503 PR tree-optimization/78997
31504 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
31505 name condition properly.
31507 2017-01-09 Richard Biener <rguenther@suse.de>
31510 * dwarf2out.c (is_cxx): New overload with context.
31511 (is_naming_typedef_decl): Use it.
31513 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31515 * invoke.texi (Option Summary): Correct spacing in option lists
31516 and add line breaks to fix over-long lines.
31518 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31520 PR middle-end/17660
31522 * extend.texi (Common Variable Attributes): Add xref to GCC
31523 Internals manual to explain mode attribute keywords.
31525 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31528 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
31529 and Preprocessor Options.
31530 (Options for Linking): Document -pthread here....
31531 (RS/6000 and PowerPC Options): ...not here.
31532 (Solaris 2 Options): ...or here.
31533 * doc/cppopts.texi: Document -pthread.
31535 2017-01-08 Martin Sebor <msebor@redhat.com>
31537 PR middle-end/77708
31538 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
31539 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
31540 New member functions.
31541 (format_directive): Used them.
31543 (pass_sprintf_length::handle_gimple_call): Same.
31544 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
31545 to avoid truncation for any argument.
31546 (extract_affine_mul): Same.
31547 * tree.c (get_file_function_name): Same.
31549 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31551 PR middle-end/77484
31552 * predict.def (PRED_INDIR_CALL): Set to 86.
31554 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31556 PR preprocessor/54124
31557 * doc/cppopts.texi: Reformat -d subtable to list the full name
31558 of the options. Add cross-reference to the docs for the general
31559 compiler -d options.
31560 * doc/invoke.texi (Developer Options): Add cross-reference to the
31561 preprocessor-specific -d option documentation.
31563 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31565 PR preprocessor/13498
31566 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
31567 redudant material, and reflect new command-line options.
31568 (System Headers): Likewise.
31570 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31572 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
31573 -isystem, and -idirafter. Copy-edit.
31574 * doc/cppopts.texi: Copy-edit. Remove contradiction about
31575 default for -ftrack-macro-expansion. Delete obsolete and
31576 badly-formatted implementation details about -fdebug-cpp output.
31577 * doc/cppwarnopts.texi: Copy-edit.
31579 2017-01-07 David Malcolm <dmalcolm@redhat.com>
31582 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
31583 that the transition from a max line width >= 1<<10 to narrower
31584 lines works correctly.
31586 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
31588 * doc/options.texi (PerFunction): New.
31589 * opt-functions.awk (switch_flags): Map both Optimization and
31590 PerFunction to CL_OPTIMIZATION.
31591 * opth-gen.awk: Test for PerFunction flag along with
31593 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
31594 it only when the latter is present. Skip those that don't in
31595 the hash function generator.
31596 * common.opt (fvar-tracking): Mark as PerFunction instead of
31598 (fvar-tracking-assignments): Likewise.
31599 (fvar-tracking-assignments-toggle): Likewise.
31600 (fvar-tracking-uninit): Likewise.
31602 2017-01-07 Jakub Jelinek <jakub@redhat.com>
31604 PR translation/79018
31605 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
31608 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
31611 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
31612 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
31613 TARGET_LONG_JUMP_TABLE_OFFSETS.
31614 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
31615 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
31616 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
31617 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
31618 * config/m68k/m68k.md (tablejump expander): Likewise.
31619 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
31620 TARGET_LONG_JUMP_TABLE_OFFSETS.
31621 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
31622 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
31624 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
31625 David Holsgrove <david.holsgrove@xilinx.com>
31627 * common/config/microblaze/microblaze-common.c
31628 (TARGET_EXCEPT_UNWIND_INFO): Remove.
31629 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
31631 * config/microblaze/microblaze.c (microblaze_must_save_register)
31632 (microblaze_expand_epilogue, microblaze_return_addr): Handle
31634 (microblaze_eh_return): New function.
31635 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
31636 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
31637 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
31638 * config/microblaze/microblaze.md (eh_return): New pattern.
31640 2017-01-06 Jakub Jelinek <jakub@redhat.com>
31642 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
31643 GCC_DIAGNOSTIC_STRINGIFY): Define.
31645 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
31647 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31649 * config/arm/arm.md (<mcrr>): New.
31651 * config/arm/arm.c (arm_arch5te): New.
31652 (arm_option_override): Set arm_arch5te.
31653 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
31655 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
31656 (arm_mcrr_qualifiers): ... this. New.
31657 (MRRC_QUALIFIERS): Define to...
31658 (arm_mrrc_qualifiers): ... this. New.
31659 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
31661 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
31662 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
31663 (MRRCI, mrrc, MRRC): New.
31664 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
31665 VUNSPEC_MRRC2): New.
31667 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31669 * config/arm/arm.md (<mcr>): New.
31671 * config/arm/arm.c (arm_coproc_builtin_available): Add
31672 support for mcr, mrc, mcr2 and mrc2.
31673 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
31674 (arm_mcr_qualifiers): ... this. New.
31675 (MRC_QUALIFIERS): Define to ...
31676 (arm_mrc_qualifiers): ... this. New.
31677 (MCR_QUALIFIERS): Define to ...
31678 (arm_mcr_qualifiers): ... this. New.
31679 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
31681 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
31682 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
31683 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
31684 VUNSPEC_MRC2): New.
31686 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31688 * config/arm/arm.md (*ldc): New.
31692 * config/arm/arm.c (arm_coproc_builtin_available): Add
31693 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
31694 (arm_coproc_ldc_stc_legitimate_address): New.
31695 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
31696 'qualifier_const_pointer'.
31697 (LDC_QUALIFIERS): Define to...
31698 (arm_ldc_qualifiers): ... this. New.
31699 (STC_QUALIFIERS): Define to...
31700 (arm_stc_qualifiers): ... this. New.
31701 * config/arm/arm-protos.h
31702 (arm_coproc_ldc_stc_legitimate_address): New.
31703 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
31704 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
31705 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
31706 stc2, stcl, stc2l): New.
31707 * config/arm/constraints.md (Uz): New.
31708 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
31709 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
31710 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
31711 VUNSPEC_STC2L): New.
31713 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31715 * config/arm/arm.md (<cdp>): New.
31716 * config/arm/arm.c (neon_const_bounds): Rename this ...
31717 (arm_const_bounds): ... this.
31718 (arm_coproc_builtin_available): New.
31719 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
31720 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
31721 (CDP_QUALIFIERS): Define to...
31722 (arm_cdp_qualifiers): ... this. New.
31724 (arm_expand_builtin_args): Add case for 6 arguments.
31725 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
31726 (arm_const_bounds): ... this.
31727 (arm_coproc_builtin_available): New.
31728 * config/arm/arm_acle.h (__arm_cdp): New.
31730 * config/arm/arm_acle_builtins.def (cdp): New.
31732 * config/arm/iterators.md (CDPI,CDP,cdp): New.
31733 * config/arm/neon.md: Rename all 'neon_const_bounds' to
31734 'arm_const_bounds'.
31735 * config/arm/types.md (coproc): New.
31736 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
31737 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
31738 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
31739 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
31741 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31743 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
31744 (UBINOP_QUALIFIERS): New.
31746 (acle_builtin_data): New. Change comment.
31747 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
31748 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
31749 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
31750 arm_acle_builtins.def.
31751 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
31752 (arm_init_acle_builtins): New.
31753 (CRC32_BUILTIN): Remove.
31754 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
31755 crc32cb, crc32ch and crc32cw.
31756 (arm_init_crc32_builtins): Remove.
31757 (arm_init_builtins): Use arm_init_acle_builtins rather
31758 than arm_init_crc32_builtins.
31759 (arm_expand_acle_builtin): New.
31760 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
31761 * config/arm/arm_acle_builtins.def: New.
31763 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31765 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
31766 (arm_builtin_datum): ... this.
31767 (arm_init_neon_builtin): Rename to ...
31768 (arm_init_builtin): ... this. Add a new parameters PREFIX
31769 and USE_SIG_IN_NAME.
31770 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
31771 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
31772 'arm_builtin_datum'.
31773 (arm_init_vfp_builtins): Likewise.
31774 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
31775 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
31776 (arm_expand_neon_args): Rename to ...
31777 (arm_expand_builtin_args): ... this. Rename builtin_arg
31778 enum values and differentiate between ARG_BUILTIN_MEMORY
31779 and ARG_BUILTIN_NEON_MEMORY.
31780 (arm_expand_neon_builtin_1): Rename to ...
31781 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
31782 values, arm_expand_builtin_args and add bool parameter NEON.
31783 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
31784 (arm_expand_vfp_builtin): Likewise.
31785 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
31787 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31789 PR middle-end/77484
31790 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
31791 * predict.c (tree_estimate_probability_bb): Reverse direction of
31792 polymorphic call predictor.
31794 2017-01-06 David Malcolm <dmalcolm@redhat.com>
31796 * passes.c (execute_one_pass): Split out pass-skipping logic into...
31797 (determine_pass_name_match): ...this new function and...
31798 (should_skip_pass_p): ...this new function.
31800 2017-01-06 Nathan Sidwell <nathan@acm.org>
31802 * ipa-visibility.c (function_and_variable_visibility): Reformat
31803 comments and long lines. Remove extrneous if.
31804 * symtab.c (symtab_node::make_decl_local): Fix code format.
31805 (symtab_node::set_section_for_node): Fix comment typo.
31807 2017-01-06 Martin Liska <mliska@suse.cz>
31810 * lra-constraints.c: Rename invariant to lra_invariant.
31811 * predict.c (set_even_probabilities): Initialize e to NULL.
31813 2017-01-05 Martin Sebor <msebor@redhat.com>
31815 PR tree-optimization/78910
31816 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
31817 (format_integer): Correct off-by-one error in the handling
31818 of precision with negative numbers in signed conversions..
31820 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
31822 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
31824 2017-01-05 Jakub Jelinek <jakub@redhat.com>
31826 PR tree-optimization/71016
31827 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
31828 factor_out_conditional_conversion. Formatting fix.
31829 (factor_out_conditional_conversion): Add cond_stmt argument.
31830 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
31831 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
31834 2017-01-05 David Malcolm <dmalcolm@redhat.com>
31836 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
31837 read-rtl-function.o, and selftest-rtl.o.
31838 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
31839 (selftest::aarch64_test_loading_full_dump): New function.
31840 (selftest::aarch64_run_selftests): New function.
31841 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
31842 selftest::aarch64_run_selftests.
31843 * config/i386/i386.c
31844 (selftest::ix86_test_loading_dump_fragment_1): New function.
31845 (selftest::ix86_test_loading_call_insn): New function.
31846 (selftest::ix86_test_loading_full_dump): New function.
31847 (selftest::ix86_test_loading_unspec): New function.
31848 (selftest::ix86_run_selftests): Call the new functions.
31849 * emit-rtl.c (maybe_set_max_label_num): New function.
31850 * emit-rtl.h (maybe_set_max_label_num): New decl.
31851 * function.c (instantiate_decls): Guard call to
31852 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
31853 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
31855 * gensupport.c (gen_reader::gen_reader): Pass "false"
31856 for new "compact" param of rtx_reader.
31857 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
31858 rather than an empty string for NULL strings.
31859 * read-md.c: Potentially include config.h rather than bconfig.h.
31860 Wrap include of errors.h with #ifdef GENERATOR_FILE.
31861 (have_error): New global, copied from errors.c.
31862 (md_reader::read_name): Rename to...
31863 (md_reader::read_name_1): ...this, adding "out_loc" param,
31864 and converting "missing name or number" to returning false, rather
31866 (md_reader::read_name): Reimplement in terms of read_name_1.
31867 (md_reader::read_name_or_nil): New function.
31868 (md_reader::read_string): Handle "(nil)" by returning NULL.
31869 (md_reader::md_reader): Add new param "compact".
31870 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
31871 (md_reader::read_file): New method.
31872 * read-md.h (md_reader::md_reader): Add new param "compact".
31873 (md_reader::read_file): New method.
31874 (md_reader::is_compact): New accessor.
31875 (md_reader::read_name): Convert return type from void to file_location.
31876 (md_reader::read_name_or_nil): New decl.
31877 (md_reader::read_name_1): New decl.
31878 (md_reader::m_compact): New field.
31879 (noop_reader::noop_reader): Pass "false" for new "compact" param
31881 (rtx_reader::rtx_reader): Add new "compact" param.
31882 (rtx_reader::read_rtx_operand): Make virtual and convert return
31883 type from void to rtx.
31884 (rtx_reader::read_until): New decl.
31885 (rtx_reader::handle_any_trailing_information): New virtual function.
31886 (rtx_reader::postprocess): New virtual function.
31887 (rtx_reader::finalize_string): New virtual function.
31888 (rtx_reader::m_in_call_function_usage): New field.
31889 (rtx_reader::m_reuse_rtx_by_id): New field.
31890 * read-rtl-function.c: New file.
31891 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
31892 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
31893 (selftest::verify_three_block_rtl_cfg): New decl.
31894 * read-rtl-function.h: New file.
31895 * read-rtl.c: Potentially include config.h rather than bconfig.h.
31896 For host, include function.h, memmodel.h, and emit-rtl.h.
31897 (one_time_initialization): New function.
31898 (struct compact_insn_name): New struct.
31899 (compact_insn_names): New array.
31900 (find_code): Handle insn codes in compact dumps.
31901 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
31902 (bind_subst_iter_and_attr): Likewise.
31903 (add_condition_to_string): Likewise.
31904 (add_condition_to_rtx): Likewise.
31905 (apply_attribute_uses): Likewise.
31906 (add_current_iterators): Likewise.
31907 (apply_iterators): Likewise.
31908 (initialize_iterators): Guard usage of apply_subst_iterator with
31909 #ifdef GENERATOR_FILE.
31910 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
31911 (md_reader::read_mapping): Likewise.
31912 (add_define_attr_for_define_subst): Likewise.
31913 (add_define_subst_attr): Likewise.
31914 (read_subst_mapping): Likewise.
31915 (check_code_iterator): Likewise.
31916 (rtx_reader::read_rtx): Likewise. Move one-time initialization
31918 (one_time_initialization): New function.
31919 (rtx_reader::read_until): New method.
31920 (read_flags): New function.
31921 (parse_reg_note_name): New function.
31922 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
31923 Handle reuse_rtx ids.
31924 Wrap iterator lookup within #ifdef GENERATOR_FILE.
31925 Add parsing support for RTL dumps, mirroring the special-cases in
31926 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
31927 values, and calling handle_any_trailing_information.
31928 (rtx_reader::read_rtx_operand): Convert return type from void
31929 to rtx, returning return_rtx. Handle case 'e'. Call
31930 finalize_string on XSTR and XTMPL fields.
31931 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
31932 "(nil)" values were omitted. Call the postprocess vfunc on the
31934 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
31935 class ctor. Initialize m_in_call_function_usage. Call
31936 one_time_initialization.
31937 * rtl-tests.c (selftest::test_uncond_jump): Call
31938 set_new_first_and_last_insn.
31939 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
31940 * selftest-rtl.c: New file.
31941 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
31942 (selftest::get_insn_by_uid): New decl.
31943 * selftest-run-tests.c (selftest::run_tests): Call
31944 read_rtl_function_c_tests.
31945 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
31946 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
31949 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
31951 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
31952 operands in a special way. Assert that pos+len <= mode precision.
31954 2017-01-05 Jakub Jelinek <jakub@redhat.com>
31956 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
31957 3 argument Alias with unlimited for the negative form.
31958 (fno-vect-cost-model): Removed.
31960 2017-01-05 Martin Liska <mliska@suse.cz>
31962 * hsa-gen.c (gen_hsa_divmod): New function.
31963 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
31965 2017-01-05 Martin Liska <mliska@suse.cz>
31968 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
31971 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31973 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
31974 small constant length operands.
31976 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31978 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
31979 between loop iterations.
31981 2017-01-05 Martin Liska <mliska@suse.cz>
31984 * gimplify.c (gimplify_decl_expr): Compare to
31985 asan_poisoned_variables instread of checking flags.
31986 (gimplify_target_expr): Likewise.
31987 (gimplify_expr): Likewise.
31988 (gimplify_function_tree): Conditionally initialize
31989 asan_poisoned_variables.
31991 2017-01-04 Jeff Law <law@redhat.com>
31993 PR tree-optimizatin/78812
31994 * rtl.h (contains_mem_rtx_p): Prototype.
31995 * ifcvt.c (containts_mem_rtx_p): Move from here to...
31996 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
31997 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
31998 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
31999 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
32001 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32003 * input.c (assert_char_at_range): Default-initialize actual_range.
32005 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32007 * df-scan.c (df_ref_create_structure): Make regno unsigned,
32008 to match the caller.
32010 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32012 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
32013 insns after final jump in test to emit dummy move.
32015 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32017 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
32018 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
32020 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32022 * multiple_target.c (create_dispatcher_calls): Init e_next.
32023 * tree-ssa-loop-split.c (split_loop): Init border.
32024 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
32027 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
32032 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
32033 (altivec_register_operand): Do not return true if the operand
32034 contains a SUBREG mixing SImode and SFmode.
32035 (vsx_register_operand): Likewise.
32036 (vsx_reg_sfsubreg_ok): New predicate.
32037 (vfloat_operand): Do not return true if the operand contains a
32038 SUBREG mixing SImode and SFmode.
32039 (vint_operand): Likewise.
32040 (vlogical_operand): Likewise.
32041 (gpc_reg_operand): Likewise.
32042 (int_reg_operand): Likewise.
32043 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
32044 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
32045 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
32047 (rs6000_emit_move_si_sf_subreg): New helper function.
32048 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
32049 fixup SUBREGs involving SImode and SFmode.
32050 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
32051 numbers for the new peephole2 optimization.
32052 (peephole2 for SFmode unions): New peephole2 to optimize cases in
32053 the GLIBC math library that do AND/IOR/XOR operations on single
32054 precision floating point.
32055 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
32056 target macros to say whether we need to avoid SUBREGs mixing
32058 (TARGET_ALLOW_SF_SUBREG): Likewise.
32059 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
32060 (UNSPEC_SI_FROM_SF): Likewise.
32061 (iorxor): Change spacing.
32062 (and_ior_xor): New iterator for AND, IOR, and XOR.
32063 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
32064 (movdi_from_sf_zero_ext): Likewise.
32065 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
32066 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
32067 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
32068 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
32069 (fms<mode>4): Likewise.
32070 (fnma<mode>4): Likewise.
32071 (fnms<mode>4): Likewise.
32072 (nfma<mode>4): Likewise.
32073 (nfms<mode>4): Likewise.
32075 2017-01-04 Marek Polacek <polacek@redhat.com>
32078 * doc/invoke.texi: Document -Wpointer-compare.
32080 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32082 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
32085 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
32086 descriptions for -gdwarf-5 and emit them as uleb128 instead of
32089 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
32092 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
32093 documentation of the powerpc_popcntb_ok attribute.
32094 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32095 code to issue warning messages if a requested CPU configuration is
32096 not supported by the binary (assembler and loader) toolchain.
32097 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
32098 made to define a built-in function that has been disabled.
32099 (paired_init_builtins): Add assertion to prevent ICE if attempt is
32100 made to define a built-in function that has been disabled.
32101 (altivec_init_builtins): Add comment explaining why definition
32102 of the DST built-in functions is not preceded by an assertion
32103 check. Add assertions to prevent ICE if attempts are made to
32104 define an altivec predicate or an abs* built-in function that has
32106 (htm_init_builtins): Add comment explaining why definition of the
32107 htm built-in functions is not preceded by an assertion check.
32109 2017-01-04 Jeff Law <law@redhat.com>
32111 PR tree-optimizatin/67955
32112 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
32113 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
32114 the points-to solution does not include pt_null. Use DECL_PT_UID
32117 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
32119 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
32120 Use gen_int_mode instead of gen_lopwart for const_int operands.
32122 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32124 PR tree-optimization/71563
32125 * match.pd: Simplify X << Y into X if Y is known to be 0 or
32126 out of range value - has low bits known to be zero.
32128 2017-01-04 Alan Modra <amodra@gmail.com>
32130 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
32131 * configure: Regenerate.
32132 * config.in: Regenerate.
32134 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32137 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
32138 a substring of the message, but strcmp with the whole message. Ifdef
32139 ENABLE_NLS, translate the message first using dgettext.
32141 2017-01-03 Jeff Law <law@redhat.com>
32143 PR tree-optimizatin/78856
32144 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
32145 (mark_threaded_blocks): Remove code to truncate thread paths that
32146 cross multiple loop headers. Instead invalidate the cached loop
32147 iteration information and handle case of a thread path walking
32148 into an irreducible region.
32150 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
32153 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
32154 assertions. Add support for doing the signbit if the IEEE 128-bit
32155 floating point value is in a GPR.
32156 * config/rs6000/rs6000.md (Fsignbit): Delete.
32157 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
32158 Update the length attribute if the value is in a GPR.
32159 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
32160 the sign or zero extension instruction, since the value is always 0/1.
32161 (signbit<mode>2_dm2): Delete using <Fsignbit>.
32164 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
32165 extracting SImode to a GPR register so that we can generate a
32166 store, limit the vector to be in a traditional Altivec register
32167 for the vextuwrx instruction.
32169 2017-01-03 Ian Lance Taylor <iant@google.com>
32171 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
32173 2017-01-03 Martin Sebor <msebor@redhat.com>
32175 PR tree-optimization/78696
32176 * gimple-ssa-sprintf.c (format_floating): Correct handling of
32177 precision. Use MPFR for %f for greater fidelity. Correct handling
32179 (pass_sprintf_length::compute_format_length): Set width and precision
32180 specified by asrerisk to void_node for vararg functions.
32181 (try_substitute_return_value): Adjust dump output.
32183 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
32185 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
32187 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
32189 * doc/invoke.texi (SPARC options): Document -mlra as the default.
32190 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
32191 -mlra/-mno-lra was passed to the compiler.
32193 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
32195 PR rtl-optimization/65618
32196 * emit-rtl.c (try_split): Move initialization of "before" and
32197 "after" to just before the call to emit_insn_after_setloc.
32199 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
32201 * doc/md.texi (Standard Names): Remove reference to Java frontend.
32203 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
32205 * dwarf2out.c (gen_enumeration_type_die): When
32206 -gno-strict-dwarf, add a DW_AT_encoding attribute.
32208 2017-01-03 Jakub Jelinek <jakub@redhat.com>
32210 PR tree-optimization/78965
32211 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
32212 Change first argument from const call_info & to call_info &. For %n
32213 set info.nowrite to false.
32215 PR middle-end/78901
32216 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
32217 possibly throwing calls.
32219 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
32220 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
32221 and fns handling, rather than in a separate case SSA_NAME.
32223 2017-01-02 Jeff Law <law@redhat.com>
32225 * config/darwin-driver.c (darwin_driver_init): Const-correctness
32226 fixes for first_period and second_period variables.
32228 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
32231 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
32232 (*insvqi_1): New insn pattern.
32233 (*insvqi_1_mem_rex64): Ditto.
32234 (*insvqi_2): Ditto.
32235 (*insvqi_3): Rename from *insvqi.
32237 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
32239 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
32241 * doc/cfg.texi (Edges): Remove reference to Java.
32242 (Maintaining the CFG): Ditto.
32244 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32246 PR middle-end/77674
32247 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
32248 transparent aliases.
32250 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32252 PR middle-end/77484
32253 * predict.def (PRED_CALL): Update hitrate.
32254 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
32255 * predict.c (tree_estimate_probability_bb): Split CALL predictor
32256 into direct/indirect/polymorphic variants.
32258 2017-01-01 Jakub Jelinek <jakub@redhat.com>
32260 Update copyright years.
32262 * gcc.c (process_command): Update copyright notice dates.
32263 * gcov-dump.c (print_version): Ditto.
32264 * gcov.c (print_version): Ditto.
32265 * gcov-tool.c (print_version): Ditto.
32266 * gengtype.c (create_file): Ditto.
32267 * doc/cpp.texi: Bump @copying's copyright year.
32268 * doc/cppinternals.texi: Ditto.
32269 * doc/gcc.texi: Ditto.
32270 * doc/gccint.texi: Ditto.
32271 * doc/gcov.texi: Ditto.
32272 * doc/install.texi: Ditto.
32273 * doc/invoke.texi: Ditto.
32275 Copyright (C) 2017 Free Software Foundation, Inc.
32277 Copying and distribution of this file, with or without modification,
32278 are permitted in any medium without royalty provided the copyright
32279 notice and this notice are preserved.