1 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4 gimple folding of vec_madd() intrinsics.
5 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
6 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
7 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
9 2017-10-30 Richard Biener <rguenther@suse.de>
11 PR tree-optimization/82762
13 2017-10-23 Richard Biener <rguenther@suse.de>
15 PR tree-optimization/82129
17 2017-08-01 Richard Biener <rguenther@suse.de>
19 PR tree-optimization/81181
20 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21 (compute_antic): ... end of iteration here.
23 2017-10-30 Joseph Myers <joseph@codesourcery.com>
25 * doc/invoke.texi (C Dialect Options): Document -std=c17,
26 -std=iso9899:2017 and -std=gnu17.
27 * doc/standards.texi (C Language): Document C17 support.
28 * doc/cpp.texi (Overview): Mention -std=c17.
29 (Standard Predefined Macros): Document C11 and C17 values of
30 __STDC_VERSION__. Do not refer to C99 support as incomplete.
31 * doc/extend.texi (Inline): Do not list individual options for
32 standards newer than C99.
33 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
35 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
38 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
40 * asan.c (asan_finish_file): Align asan globals array by shadow
43 2017-10-30 Jakub Jelinek <jakub@redhat.com>
46 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
47 (struct store_immediate_info): Add bitregion_start and bitregion_end
49 (store_immediate_info::store_immediate_info): Add brs and bre
50 arguments and initialize bitregion_{start,end} from those.
51 (struct merged_store_group): Add bitregion_start, bitregion_end,
52 align_base and mask fields. Drop unnecessary struct keyword from
53 struct store_immediate_info. Add do_merge method.
54 (clear_bit_region_be): Use memset instead of loop storing zeros.
55 (merged_store_group::do_merge): New method.
56 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
57 stores as long as the surrounding bitregions have no gaps.
58 (merged_store_group::merge_overlapping): Use do_merge.
59 (merged_store_group::apply_stores): Test that bitregion_{start,end}
60 is byte aligned, rather than requiring that start and width are
61 byte aligned. Drop unnecessary struct keyword from
62 struct store_immediate_info. Allocate and populate also mask array.
63 Make start of the arrays relative to bitregion_start rather than
64 start and size them according to bitregion_{end,start} difference.
65 (struct imm_store_chain_info): Drop unnecessary struct keyword from
66 struct store_immediate_info.
67 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
68 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
69 struct keyword from struct store_immediate_info.
70 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
71 between stores as long as the surrounding bitregions have no gaps.
73 (struct split_store): Add orig non-static data member.
74 (split_store::split_store): Initialize orig to false.
75 (find_constituent_stmts): Return store_immediate_info *, non-NULL
76 if there is exactly a single original stmt. Change stmts argument
77 to pointer from reference, if NULL, don't push anything to it. Add
78 first argument, use it to optimize skipping over orig stmts that
79 are known to be before bitpos already. Simplify.
80 (split_group): Return unsigned int count how many stores are or
81 would be needed rather than a bool. Add allow_unaligned argument.
82 Change split_stores argument from reference to pointer, if NULL,
83 only do a dry run computing how many stores would be produced.
84 Rewritten algorithm to use both alignment and misalign if
85 !allow_unaligned and handle bitfield stores with gaps.
86 (imm_store_chain_info::output_merged_store): Set start_byte_pos
87 from bitregion_start instead of start. Compute allow_unaligned
88 here, if true, do 2 split_group dry runs to compute which one
89 produces fewer stores and prefer aligned if equal. Punt if
90 new count is bigger or equal than original before emitting any
91 statements, rather than during that. Remove no longer needed
92 new_ssa_names tracking. Replace num_stmts with
93 split_stores.length (). Use 32-bit stack allocated entries
94 in split_stores auto_vec. Try to reuse original store lhs/rhs1
95 if possible. Handle bitfields with gaps.
96 (pass_store_merging::execute): Ignore bitsize == 0 stores.
97 Compute bitregion_{start,end} for the stores and construct
98 store_immediate_info with that. Formatting fixes.
100 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
103 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
104 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
106 2017-10-29 Jim Wilson <wilson@tuliptree.org>
108 * gimplify.c: Include tm_p.h.
110 * common.opt (gcoff): Re-add as ignored option.
111 (gcoff1, gcoff2, gcoff3): Likewise.
113 * Makefile.in (OBJS): Delete sdbout.o.
114 (GTFILES): Delete $(srcdir)/sdbout.c.
115 * debug.h: Delete sdb_debug_hooks.
116 * final.c: Delete sdbout.h include.
117 (final_scan_insn): Delete SDB_DEBUG check.
118 (rest_of_clean_state): Likewise.
119 * output.h: Delete sdb_begin_function_line.
122 * toplev.c: Delete sdbout.h include.
123 (process_options): Delete SDB_DEBUG check.
124 * tree-core.h (tree_type_common): Delete pointer field of
126 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
128 * tree.h (TYPE_SYMTAB_POINTER): Delete.
129 (TYPE_SYMTAB_IS_POINTER): Delete.
130 (TYPE_SYMTAB_IS_DIE): Renumber.
131 * xcoffout.c: Refer to former sdbout.c file.
132 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
134 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
135 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
136 (Debugging Options): Delete -gcoff.
137 (-gstabs): Delete SDB reference.
139 (-gcoff@var{level}): Delete.
140 * doc/passes.texi (Debugging information output): Delete SDB and
142 * doc/tm.texi: Regenerate.
143 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
144 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
146 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
147 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
148 (SDB_DEBUGGING_INFO): Delete.
149 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
150 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
151 * target.def (output_source_filename): Delete COFF reference.
153 * common.opt (gcoff): Delete.
154 (gxcoff+): Update Negative chain.
155 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
157 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
158 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
159 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
161 (expand_function_start): Change sdb reference to past tense.
162 (expand_function_end): Change sdb reference to past tense.
163 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
164 * opts.c (debug_type_names): Delete coff entry.
165 (common_handle_option): Delete OPT_gcoff case.
166 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
168 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
169 * config/cris/cris.h: Delete SDB reference in comment.
170 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
171 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
172 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
173 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
175 (ix86_expand_prologue): Likewise.
176 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
177 * config/ia64/ia64.h: Likewise.
178 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
179 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
181 * config/mmix/mmix.h: Likewise.
182 * config/nds32/nds32.c: Likewise.
183 * config/stormy/storym16.h: Likewise.
184 * config/visium/visium.h: Likewise.
185 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
187 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
189 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
190 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
191 FRAME_POINTER_REGNUM point at high end of local var area.
193 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
195 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
196 Move comment around. Do not reset best_edge for a copiable
197 destination if the copy would cause a partition change.
198 (better_edge_p): Remove redundant check.
200 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
202 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
204 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
206 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
207 for math functions that have _Float<N> and _Float<N>X variants.
208 (mathfn_built_in_2): Add support for math functions that have
209 _Float<N> and _Float<N>X variants.
210 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
211 (expand_builtin_mathfn_ternary): Add support for fma with
212 _Float<N> and _Float<N>X variants.
213 (expand_builtin): Likewise.
214 (fold_builtin_3): Likewise.
215 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
216 create math function _Float<N> and _Float<N>X variants as external
218 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
219 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
220 the __builtin_ prefix and if not strict ansi, without the prefix.
221 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
222 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
223 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
224 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
225 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
226 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
227 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
228 function signatures for fma _Float<N> and _Float<N>X variants.
229 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
230 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
231 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
232 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
233 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
234 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
235 * gencfn-macros.c (print_case_cfn): Add support for math functions
236 that have _Float<N> and _Float<N>X variants.
237 (print_define_operator_list): Likewise.
238 (fltfn_suffixes): Likewise.
240 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
241 for math functions that have _Float<N> and _Float<N>X variants.
242 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
243 and _Float<N>X variants.
244 (COPYSIGN): Likewise.
247 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
248 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
250 (integer_valued_read_call_p): Likewise.
251 * fold-const-call.c (fold_const_call_ss): Likewise.
252 (fold_const_call_sss): Add support for copysign, fmin, and fmax
253 _Float<N> and _Float<N>X variants.
254 (fold_const_call_ssss): Add support for fma _Float<N> and
256 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
257 support for copysign and fma _Float<N> and _Float<N>X variants.
258 (backprop::process_builtin_call_use): Likewise.
259 * tree-call-cdce.c (can_test_argument_range); Add support for
260 sqrt _Float<N> and _Float<N>X variants.
261 (edom_only_function): Likewise.
262 (get_no_error_domain): Likewise.
263 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
264 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
265 copysign _Float<N> and _Float<N>X variants.
266 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
267 handled by machine independent code.
269 * doc/cpp.texi (Common Predefined Macros): Document defining
270 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
271 fma _Float<N> and _Float<N>X variants.
273 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
276 * config/i386/i386-modes.def (CCFPU): Remove definition.
277 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
278 (ix86_cc_modes_compatible): Ditto.
279 (ix86_expand_carry_flag_compare): Ditto.
280 (ix86_expand_int_movcc): Ditto.
281 (ix86_expand_int_addcc): Ditto.
282 (ix86_reverse_condition): Ditto.
283 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
284 Return true/false for unordered/ordered fp comparisons.
285 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
286 (ix86_prepare_fp_compare_args): Update for rename.
287 (ix86_expand_fp_compare): Update for rename. Generate unordered
288 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
289 (ix86_expand_sse_compare_and_jump): Ditto.
290 * config/i386/predicates.md (fcmov_comparison_operator):
291 Remove CCFPU mode handling.
292 (ix86_comparison_operator): Ditto.
293 (ix86_carry_flag_operator): Ditto.
294 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
295 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
296 (*cmpu<mode>_cc_i387): Ditto.
297 (FPCMP): Remove mode iterator.
298 (unord): Remove mode attribute.
299 (unord_subst): New define_subst transformation
300 (unord): New define_subst attribute.
302 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
303 (*cmpi<unord>xf_i387): Ditto.
304 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
305 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
306 using unord_subst transformation.
307 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
308 (round_saeonly): Also handle CCFP mode.
309 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
310 Remove UNSPEC_SAHF unspec handling.
312 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
314 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
316 2017-10-27 Jeff Law <law@redhat.com>
318 * gimple-ssa-sprintf.c: Include domwalk.h.
319 (class sprintf_dom_walker): New class, derived from dom_walker.
320 (sprintf_dom_walker::before_dom_children): New function.
321 (struct call_info): Moved into sprintf_dom_walker class
322 (compute_formath_length, handle_gimple_call): Likewise.
323 (sprintf_length::execute): Call the dominator walker rather
324 than walking the statements.
326 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
327 gimple statement locations.
328 (check_array_bounds): Corresponding changes. Get the statement's
329 location directly from wi->stmt.
331 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
334 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
336 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
338 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
339 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
341 2017-10-27 Jakub Jelinek <jakub@redhat.com>
344 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
345 * config/i386/i386.c (maybe_get_pool_constant): Removed.
346 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
347 maybe_get_pool_constant.
348 * config/i386/predicates.md (zero_extended_scalar_load_operand):
351 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
353 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
354 2.26 caveat. Update gas and gld versions.
355 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
358 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
360 * cgraph.h (set_malloc_flag): Declare.
361 * cgraph.c (set_malloc_flag_1): New function.
362 (set_malloc_flag): Likewise.
363 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
364 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
366 (read_ipa_call_summary): Add support for reading is_return_callee.
367 (write_ipa_call_summary): Stream is_return_callee.
368 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
369 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
370 ipa-prop.h, ipa-fnsummary.h.
371 (pure_const_names): Change to static.
372 (malloc_state_e): Define.
373 (malloc_state_names): Define.
374 (funct_state_d): Add field malloc_state.
375 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
376 (check_retval_uses): New function.
377 (malloc_candidate_p): Likewise.
378 (analyze_function): Add support for malloc attribute.
379 (pure_const_write_summary): Stream malloc_state.
380 (pure_const_read_summary): Add support for reading malloc_state.
381 (dump_malloc_lattice): New function.
382 (propagate_malloc): New function.
383 (warn_function_malloc): New function.
384 (ipa_pure_const::execute): Call propagate_malloc and
386 (pass_local_pure_const::execute): Add support for malloc attribute.
387 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
388 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
390 2017-10-27 Martin Liska <mliska@suse.cz>
392 PR gcov-profile/82457
393 * doc/invoke.texi: Document that one needs a non-strict ISO mode
394 for fork-like functions to be properly instrumented.
396 2017-10-27 Richard Biener <rguenther@suse.de>
399 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
400 info when we redirected EH.
402 2017-10-26 Michael Collison <michael.collison@arm.com>
404 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
406 (<optab>_trunchf<GPI:mode>2: New pattern.
407 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
408 * config/aarch64/iterators.md (wv): New mode attribute.
409 (vf, VF): New mode attributes.
410 (vgp, VGP): New mode attributes.
411 (s): Update attribute with SImode and DImode prefixes.
413 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
415 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
416 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
417 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
418 (nios2_option_overide): Initialize it. Don't allow R0-relative
420 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
421 (nios2_symbolic_constant_p): Likewise.
422 (nios2_legitimate_address_p): Likewise.
423 (nios2_r0rel_section_name_p): New.
424 (nios2_symbol_ref_in_r0rel_data_p): New.
425 (nios2_emit_move_sequence): Handle r0rel_constant_p.
426 (r0rel_constant_p): New.
427 (nios2_print_operand_address): Handle r0rel_constant_p.
428 (nios2_cdx_narrow_form_p): Likewise.
429 * config/nios2/nios2.opt (mr0rel-sec=): New option.
430 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
431 (Nios II Options): Document -mr0rel-sec.
433 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
435 * config/nios2/nios2.c: Include xregex.h.
436 (nios2_gprel_sec_regex): New.
437 (nios2_option_overide): Initialize it. Don't allow GP-relative
439 (nios2_small_section_name_p): Check for regex match.
440 * config/nios2/nios2.opt (mgprel-sec=): New option.
441 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
442 (Nios II Options): Document -mgprel-sec.
444 2017-10-26 Jim Wilson <wilson@tuliptree.org>
446 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
448 2017-10-26 Tom de Vries <tom@codesourcery.com>
450 PR tree-optimization/82707
451 * gimple.c (gimple_copy): Fix unsharing of
452 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
454 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
456 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
457 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
458 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
459 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
460 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
461 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
462 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
463 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
464 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
465 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
466 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
467 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
468 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
469 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
470 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
471 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
472 _mm512_mask_cmpunord_ps_mask): New intrinsics.
474 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
476 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
478 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
479 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
480 warning to rs6000.c. Remove the Undocumented flag, since it has
482 (-mabi=ibmlongdouble): Likewise.
483 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
484 already set, set the default format for long double.
485 (rs6000_debug_reg_global): Print whether long double is IBM or
487 (rs6000_option_override_internal): Rework setting long double
488 format. Only warn if the user is changing the long double default
489 and they did not use -Wno-psabi.
490 * doc/invoke.texi (PowerPC options): Update the documentation for
491 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
493 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
494 Alan Hayward <alan.hayward@arm.com>
495 David Sherwood <david.sherwood@arm.com>
497 * rtl.h (wider_subreg_mode): New function.
498 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
499 rather than an unsigned int *.
500 * ira-color.c (regno_max_ref_width): Replace with...
501 (regno_max_ref_mode): ...this new variable.
502 (coalesced_pseudo_reg_slot_compare): Update accordingly.
503 Use wider_subreg_mode.
504 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
505 rather than an unsigned int *.
506 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
507 (process_alt_operands): Likewise.
508 (invariant_p): Likewise.
509 * lra-spills.c (assign_mem_slot): Likewise.
510 (add_pseudo_to_slot): Likewise.
511 * lra.c (collect_non_operand_hard_regs): Likewise.
512 (add_regs_to_insn_regno_info): Likewise.
513 * reload1.c (regno_max_ref_width): Replace with...
514 (regno_max_ref_mode): ...this new variable.
515 (reload): Update accordingly. Update call to
516 ira_sort_regnos_for_alter_reg.
517 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
518 (init_eliminable_invariants): Update to use regno_max_ref_mode.
519 (scan_paradoxical_subregs): Likewise.
521 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
523 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
524 (aarch64_frame): Add emit_frame_chain boolean.
525 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
526 Move eh_return case to aarch64_layout_frame.
527 (aarch64_layout_frame): Initialize emit_frame_chain.
528 (aarch64_expand_prologue): Use emit_frame_chain.
530 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
532 * config/aarch64/aarch64.c (aarch64_layout_frame):
533 Ensure LR is always stored at the bottom of the callee-saves.
534 Remove rarely used frame layout which saves callee-saves at top of
535 frame, so the store of LR can be used as a valid probe in all cases.
537 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
539 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
540 Improve unaligned TImode/TFmode base/offset split.
542 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
543 Alan Hayward <alan.hayward@arm.com>
544 David Sherwood <david.sherwood@arm.com>
546 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
547 * combine.c (find_single_use_1): Likewise.
548 (expand_field_assignment): Likewise.
549 (move_deaths): Likewise.
550 * lra-constraints.c (simplify_operand_subreg): Likewise.
551 (curr_insn_transform): Likewise.
552 * lra.c (collect_non_operand_hard_regs): Likewise.
553 (add_regs_to_insn_regno_info): Likewise.
554 * rtlanal.c (reg_referenced_p): Likewise.
555 (covers_regno_no_parallel_p): Likewise.
557 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
559 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
560 Don't print any bits outside the precision of the value.
561 * wide-int.cc (test_printing): Add some new tests.
563 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
565 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
566 supports -xbrace_comment option.
567 * configure: Regenerate.
568 * config.in: Regenerate.
569 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
570 (ASM_CPU_SPEC): Use it.
572 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
574 * target.def (static_rtx_alignment): New hook.
575 * targhooks.h (default_static_rtx_alignment): Declare.
576 * targhooks.c (default_static_rtx_alignment): New function.
577 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
578 * doc/tm.texi: Regenerate.
579 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
580 instead of targetm.constant_alignment. Remove call to
582 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
583 (cris_preferred_mininum_alignment): New function, split out from...
584 (cris_constant_alignment): ...here.
585 (cris_static_rtx_alignment): New function.
586 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
588 (ix86_constant_alignment): ...here.
589 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
590 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
591 (mmix_static_rtx_alignment): New function.
592 * config/spu/spu.c (spu_static_rtx_alignment): New function.
593 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
595 2017-10-26 Tamar Christina <tamar.christina@arm.com>
598 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
599 Add flag_trapping_math and flag_fp_int_builtin_inexact.
601 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
603 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
604 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
605 mark as a sign-extending load.
606 (local_pic_load_u): Define.
608 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
611 * fold-const.c (operand_equal_for_comparison_p): Also return true
612 if ARG0 is a simple variant of ARG1 with narrower precision.
613 (fold_ternary_loc): Always pass unstripped operands to the predicate.
615 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
617 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
619 * i386.h (processor_costs): Add gather_static, gather_per_elt,
620 scatter_static, scatter_per_elt.
621 * x86-tune-costs.h: Add new cost entries.
623 2017-10-25 Richard Biener <rguenther@suse.de>
625 * tree-ssa-sccvn.h (vn_eliminate): Declare.
626 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
627 class pass_fre): Move to ...
628 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
629 class pass_fre): ... here and adjust for statistics.
631 2017-10-25 Jakub Jelinek <jakub@redhat.com>
634 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
635 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
636 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
637 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
640 2017-10-25 Richard Biener <rguenther@suse.de>
642 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
643 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
644 eliminate_push_avail, eliminate_insert): Move inside...
645 (class eliminate_dom_walker): ... this class in preparation
647 (fini_eliminate): Remove by merging with ...
648 (eliminate): ... this function. Adjust for class changes.
649 (pass_pre::execute): Remove fini_eliminate call.
650 (pass_fre::execute): Likewise.
652 2017-10-24 Jakub Jelinek <jakub@redhat.com>
655 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
656 (VPERMI2, VPERMI2I): New mode iterators.
657 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
658 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
660 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
661 mode iterator. Remove 3 old define_insn patterns.
662 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
663 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
664 VPERMI2 mode iterator, remove the other two expanders.
665 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
666 to use VPERMI2 mode iterator, add another alternative for vpermi2*
667 instructions, remove the other two patterns.
668 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
669 mode iterator, remove the other two patterns.
670 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
671 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
672 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
673 and adjust argument order accordingly.
674 (ix86_expand_vec_perm): Adjust caller.
675 (expand_vec_perm_1): Likewise.
676 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
677 (expand_vec_perm_vpermt2_vpshub2): ... this.
678 (ix86_expand_vec_perm_const_1): Adjust caller.
679 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
682 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
683 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
684 (vec_shl_<mode>): Remove unused expander.
685 (avx512bw_<shift_insn><mode>3): New define_insn.
686 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
687 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
689 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
692 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
695 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
697 PR rtl-optimization/82396
698 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
699 (autopref_multipass_init): Simplify initialization.
700 (autopref_rank_data): Simplify sort order.
701 * gcc/sched-int.h (autopref_multipass_data_): Remove
702 multi_mem_insn_p, min_offset and max_offset.
704 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
707 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
708 Check special value of flag_omit_frame_pointer.
709 (aarch64_can_eliminate): Likewise.
710 (aarch64_override_options_after_change_1): Simplify handling of
711 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
713 2017-10-24 Richard Biener <rguenther@suse.de>
715 PR tree-optimization/82697
716 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
717 zero for conditional load and unconditional store.
719 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
721 * doc/install.texi: Document bootstrap-cet.
723 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
726 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
727 ENDBR instruction at function entrance if function is only
730 2017-10-24 Jakub Jelinek <jakub@redhat.com>
733 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
734 patterns to better describe from which operation the CF is computed.
735 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
736 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
737 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
738 is 0, use _0 suffixed expanders instead of emitting a comparison
741 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
743 * config/i386/i386.md(*movsf_internal, *movdf_internal):
744 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
746 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
749 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
750 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
751 * loop-iv.c (iv_get_reaching_def): Likewise.
752 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
753 variable is promoted and the partition contains undefined values.
755 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
757 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
759 (nios2_address_cost): Define.
760 (nios2_legitimize_address): Recognize (exp + constant) directly.
761 (TARGET_ADDRESS_COST): Define.
763 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
765 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
766 (nios2_symbolic_memory_operand_p): Declare.
767 (nios2_split_large_constant): Declare.
768 (nios2_split_symbolic_memory_operand): Declare.
769 * config/nios2/nios2.c: Adjust includes.
770 (nios2_symbolic_constant_allowed): New.
771 (nios2_symbolic_constant_p): New.
772 (nios2_plus_symbolic_constant_p): New.
773 (nios2_valid_addr_expr_p): Recognize addresses involving
775 (nios2_legitimate_address_p): Likewise, also LO_SUM.
776 (nios2_symbolic_memory_operand_p): New.
777 (nios2_large_constant_p): New.
778 (nios2_split_large_constant): New.
779 (nios2_split_plus_large_constant): New.
780 (nios2_split_symbolic_memory_operand): New.
781 (nios2_legitimize_address): Code refactoring. Handle addresses
782 involving symbolic constants.
783 (nios2_emit_move_sequence): Likewise.
784 (nios2_print_operand): Improve error output.
785 (nios2_print_operand_address): Handle LO_SUM.
786 (nios2_cdx_narrow_form_p): Likewise.
787 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
788 operands involving symbolic constants.
789 (movhi_internal, movsi_internal): Likewise.
790 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
791 (extendhisi2, extendqi<mode>2): Likewise.
793 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
795 * tree-pass.h (PROP_rtl_split_insns): Define.
796 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
798 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
800 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
802 2017-10-23 Jakub Jelinek <jakub@redhat.com>
805 * target.def (const_not_ok_for_debug_p): Default to
806 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
807 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
808 * targhooks.c (default_const_not_ok_for_debug_p): New function.
809 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
810 which targetm.const_not_ok_for_debug_p returned true.
811 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
813 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
815 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
816 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
817 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
819 (ix86_const_not_ok_for_debug_p): New function.
820 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
821 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
823 2017-10-23 David Malcolm <dmalcolm@redhat.com>
826 * system.h: Conditionally include "unique-ptr.h" if
827 INCLUDE_UNIQUE_PTR is defined.
828 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
829 of defining INCLUDE_UNIQUE_PTR before including "system.h".
831 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
833 * config/rl78/rl78.md: New define_expand "subdi3".
835 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
838 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
839 DF_REF_INSN if DF_REF_INSN_INFO is false.
841 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
843 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
844 xmm_move instead of sse_move.
845 (sse_store_index): New function.
846 (ix86_register_move_cost): Be more sensible about mismatch stall;
847 model AVX moves correctly; make difference between sse->integer and
849 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
850 moves; make difference between SSE and AVX.
851 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
852 and zmm_move. Increase size of sse load and store tables;
853 add unaligned load and store tables; add ssemmx_to_integer.
854 * x86-tune-costs.h: Update all entries according to real
855 move latencies from Agner Fog's manual and chip documentation.
857 2017-10-23 Jakub Jelinek <jakub@redhat.com>
860 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
861 * config/i386/constraints.md (Wf): New constraint.
862 * config/i386/i386.md (UNSPEC_SBB): New unspec.
863 (cmp<dwi>_doubleword): Removed.
864 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
865 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
866 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
867 expand with cmp<dwi>_doubleword. For LTU and GEU use
868 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
870 * common.opt (gcolumn-info): Enable by default.
871 * doc/invoke.texi (gcolumn-info): Document new default.
873 2017-10-23 Richard Biener <rguenther@suse.de>
875 PR tree-optimization/82672
876 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
877 Fold the stmt if we propagated into it.
879 2017-10-23 Richard Biener <rguenther@suse.de>
881 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
882 (bitmap_remove_expr_from_set): ... this. All callers call this
883 for non-constant values.
884 (bitmap_set_subtract): Rename to...
885 (bitmap_set_subtract_expressions): ... this. Adjust and
887 (bitmap_set_contains_value): Remove superfluous check.
888 (bitmap_set_replace_value): Inline into single caller ...
889 (bitmap_value_replace_in_set): ... here and simplify.
890 (dependent_clean): Merge into ...
891 (clean): ... this using an overload. Adjust.
892 (prune_clobbered_mems): Adjust.
893 (compute_antic_aux): Likewise.
894 (compute_partial_antic_aux): Likewise.
896 2017-10-23 Richard Biener <rguenther@suse.de>
898 PR tree-optimization/82129
900 2017-08-01 Richard Biener <rguenther@suse.de>
902 PR tree-optimization/81181
903 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
904 (compute_antic): ... end of iteration here.
906 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
908 * target.def (starting_frame_offset): New hook.
909 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
910 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
911 * doc/tm.texi.in: Regenerate.
912 * hooks.h (hook_hwi_void_0): Declare.
913 * hooks.c (hook_hwi_void_0): New function.
914 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
915 STARTING_FRAME_OFFSET.
916 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
917 * reload1.c (reload): Likewise.
918 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
919 instead of STARTING_FRAME_OFFSET.
920 * function.c (try_fit_stack_local): Likewise.
921 (assign_stack_local_1): Likewise
922 (instantiate_virtual_regs): Likewise.
923 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
924 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
925 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
926 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
927 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
928 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
929 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
930 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
931 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
932 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
933 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
934 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
935 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
936 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
937 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
938 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
939 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
940 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
941 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
942 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
943 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
944 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
945 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
946 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
947 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
948 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
949 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
950 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
951 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
952 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
953 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
954 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
955 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
956 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
957 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
958 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
959 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
960 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
961 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
962 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
963 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
964 * config/avr/avr.c (avr_starting_frame_offset): Make static and
965 return a HOST_WIDE_INT.
966 (avr_builtin_setjmp_frame_value): Use it instead of
967 STARTING_FRAME_OFFSET.
968 (TARGET_STARTING_FRAME_OFFSET): Redefine.
969 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
970 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
972 (TARGET_STARTING_FRAME_OFFSET): Redefine.
973 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
974 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
975 (TARGET_CONSTANT_ALIGNMENT): Redefine.
976 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
977 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
978 (TARGET_STARTING_FRAME_OFFSET): Redefine.
979 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
980 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
981 (TARGET_STARTING_FRAME_OFFSET): Redefine.
982 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
983 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
985 (TARGET_STARTING_FRAME_OFFSET): Redefine.
986 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
987 * config/mips/mips.c (mips_compute_frame_info): Refer to
988 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
989 (mips_starting_frame_offset): New function.
990 (TARGET_STARTING_FRAME_OFFSET): Redefine.
991 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
992 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
993 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
994 and return a HOST_WIDE_INT.
995 (TARGET_STARTING_FRAME_OFFSET): Redefine.
996 (mmix_initial_elimination_offset): Refer to
997 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
998 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
999 * config/pa/pa.c (pa_starting_frame_offset): New function.
1000 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
1001 (pa_expand_prologue): Likewise.
1002 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1003 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
1004 !FRAME_GROWS_DOWNWARD handling to...
1005 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1006 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
1007 !FRAME_GROWS_DOWNWARD handling to...
1008 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1009 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
1010 !FRAME_GROWS_DOWNWARD handling to...
1011 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1012 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
1014 (rs6000_starting_frame_offset): New function.
1015 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
1016 !FRAME_GROWS_DOWNWARD handling to...
1017 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1018 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
1019 !FRAME_GROWS_DOWNWARD handling to...
1020 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1021 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
1022 !FRAME_GROWS_DOWNWARD handling to...
1023 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1024 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
1025 (rs6000_starting_frame_offset): New function.
1026 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
1027 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
1028 * config/vax/vax.c (vax_starting_frame_offset): New function.
1029 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
1030 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1031 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
1032 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
1033 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1034 * system.h (STARTING_FRAME_OFFSET): Poison.
1036 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1038 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
1039 SCALAR_TYPE_MODE instead of TYPE_MODE.
1041 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1042 Alan Hayward <alan.hayward@arm.com>
1043 David Sherwood <david.sherwood@arm.com>
1045 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
1047 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1048 Alan Hayward <alan.hayward@arm.com>
1049 David Sherwood <david.sherwood@arm.com>
1051 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
1053 2017-10-23 Richard Biener <rguenther@suse.de>
1055 PR tree-optimization/82129
1056 * tree-ssa-pre.c (bitmap_set_and): Remove.
1057 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
1058 canonicalizing expressions in the set to those with lowest
1059 ID rather than taking that from the first edge.
1061 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1063 * combine.c (rtx_equal_for_field_assignment_p): Use
1064 byte_lowpart_offset.
1066 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1067 Alan Hayward <alan.hayward@arm.com>
1068 David Sherwood <david.sherwood@arm.com>
1070 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
1071 to a SUBREG_PROMOTED_VAR.
1073 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1074 Alan Hayward <alan.hayward@arm.com>
1075 David Sherwood <david.sherwood@arm.com>
1077 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
1078 (expand_debug_source_expr): Likewise.
1079 * combine.c (combine_simplify_rtx): Likewise.
1080 * cse.c (fold_rtx): Likewise.
1081 * optabs.c (expand_float): Likewise.
1082 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1083 (simplify_binary_operation_1): Likewise.
1085 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1086 Alan Hayward <alan.hayward@arm.com>
1087 David Sherwood <david.sherwood@arm.com>
1089 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
1090 (record_promoted_value): Likewise.
1091 * expr.c (expand_expr_real_2): Likewise.
1092 * ree.c (update_reg_equal_equiv_notes): Likewise.
1093 (combine_set_extension): Likewise.
1094 * rtlanal.c (low_bitmask_len): Likewise.
1095 * simplify-rtx.c (neg_const_int): Likewise.
1096 (simplify_binary_operation_1): Likewise.
1098 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1099 Alan Hayward <alan.hayward@arm.com>
1100 David Sherwood <david.sherwood@arm.com>
1102 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
1103 * regcprop.c (maybe_mode_change): Likewise.
1104 * reload1.c (alter_reg): Likewise.
1106 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1108 * inchash.h (inchash::hash::add_wide_int): New function.
1109 * lto-streamer-out.c (hash_tree): Use it.
1111 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1113 * inchash.h (inchash::hash::add_wide_int): Rename to...
1114 (inchash::hash::add_hwi): ...this.
1115 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
1116 (polymorphic_call_target_hasher::hash): Likewise.
1117 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
1118 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
1119 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
1120 * lto-streamer-out.c (hash_tree): Likewise.
1121 * optc-save-gen.awk: Likewise.
1122 * tree.c (add_expr): Likewise.
1124 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
1127 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
1128 for ordered inequality comparisons even with TARGET_IEEE_FP.
1130 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
1133 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
1134 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
1135 Expand with cmp<dwi>_doubleword.
1137 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1139 * extend.texi: Add x86 specific to 'nocf_check' attribute.
1140 List CET intrinsics.
1141 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
1142 specific to -fcf-protection option.
1144 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1146 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
1147 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
1148 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
1149 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
1150 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
1151 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
1152 (extra_objs): Add cet.o for Linux/x86 targets.
1153 (tmake_file): Add i386/t-cet for Linux/x86 targets.
1154 * config/i386/cet.c: New file.
1155 * config/i386/cetintrin.h: Likewise.
1156 * config/i386/t-cet: Likewise.
1157 * config/i386/cpuid.h (bit_SHSTK): New.
1158 (bit_IBT): Likewise.
1159 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
1160 pass IBT and SHSTK bits.
1161 * config/i386/i386-builtin-types.def
1162 (VOID_FTYPE_UNSIGNED_PVOID): New.
1163 (VOID_FTYPE_UINT64_PVOID): Likewise.
1164 * config/i386/i386-builtin.def: Add CET intrinsics.
1165 * config/i386/i386-c.c (ix86_target_macros_internal): Add
1166 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
1167 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
1168 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
1170 * config/i386/i386.c (rest_of_insert_endbranch): New.
1171 (pass_data_insert_endbranch): Likewise.
1172 (pass_insert_endbranch): Likewise.
1173 (make_pass_insert_endbranch): Likewise.
1174 (ix86_notrack_prefixed_insn_p): Likewise.
1175 (ix86_target_string): Add -mibt, -mshstk flags.
1176 (ix86_option_override_internal): Add flag_cf_protection
1178 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
1179 (ix86_print_operand): Add 'notrack' prefix output.
1180 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
1181 (ix86_expand_builtin): Expand CET intrinsics.
1182 (x86_output_mi_thunk): Add 'endbranch' instruction.
1183 * config/i386/i386.h (TARGET_IBT): New.
1184 (TARGET_IBT_P): Likewise.
1185 (TARGET_SHSTK): Likewise.
1186 (TARGET_SHSTK_P): Likewise.
1187 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
1188 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
1189 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
1190 (builtin_setjmp_setup): New pattern.
1191 (builtin_longjmp): Likewise.
1192 (rdssp<mode>): Likewise.
1193 (incssp<mode>): Likewise.
1194 (saveprevssp): Likewise.
1195 (rstorssp): Likewise.
1196 (wrss<mode>): Likewise.
1197 (wruss<mode>): Likewise.
1198 (setssbsy): Likewise.
1199 (clrssbsy): Likewise.
1200 (nop_endbr): Likewise.
1201 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
1203 * config/i386/immintrin.h: Include <cetintrin.h>.
1204 * config/i386/linux-common.h
1205 (file_end_indicate_exec_stack_and_cet): New prototype.
1206 (TARGET_ASM_FILE_END): New.
1208 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
1210 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
1211 latencies instead of having separate table; make difference between
1212 integer and float costs.
1213 * i386.h (processor_costs): Remove scalar_stmt_cost,
1214 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
1215 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
1217 * x86-tune-costs.h: Remove entries which has been removed in
1218 procesor_costs from all tables; make cond_taken_branch_cost
1219 and cond_not_taken_branch_cost COST_N_INSNS based.
1221 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
1223 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
1225 2017-10-20 Jakub Jelinek <jakub@redhat.com>
1227 * config/i386/i386.md (isa): Remove fma_avx512f.
1228 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
1229 <avx512>_fmadd_<mode>_mask3<round_name>,
1230 <avx512>_fmsub_<mode>_mask<round_name>,
1231 <avx512>_fmsub_<mode>_mask3<round_name>,
1232 <avx512>_fnmadd_<mode>_mask<round_name>,
1233 <avx512>_fnmadd_<mode>_mask3<round_name>,
1234 <avx512>_fnmsub_<mode>_mask<round_name>,
1235 <avx512>_fnmsub_<mode>_mask3<round_name>,
1236 <avx512>_fmaddsub_<mode>_mask<round_name>,
1237 <avx512>_fmaddsub_<mode>_mask3<round_name>,
1238 <avx512>_fmsubadd_<mode>_mask<round_name>,
1239 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
1240 (*vec_widen_umult_even_v16si<mask_name>,
1241 *vec_widen_smult_even_v16si<mask_name>): Likewise.
1242 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
1244 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1246 * extend.texi: Add 'nocf_check' documentation.
1247 * gimple.texi: Add second parameter to
1248 gimple_build_call_from_tree.
1249 * invoke.texi: Add -fcf-protection documentation.
1250 * rtl.texi: Add REG_CALL_NOTRACK documenation.
1252 2017-10-20 Richard Biener <rguenther@suse.de>
1254 PR tree-optimization/82473
1255 * tree-vect-loop.c (vectorizable_reduction): Properly get at
1256 the largest input type.
1258 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1260 * c-attribs.c (handle_nocf_check_attribute): New function.
1261 (c_common_attribute_table): Add 'nocf_check' handling.
1262 * gimple-parser.c: Add second argument NULL to
1263 gimple_build_call_from_tree.
1264 * attrib.c (comp_type_attributes): Check nocf_check attribute.
1265 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
1267 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
1268 * common.opt: Add fcf-protection flag.
1269 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
1270 * flag-types.h: Add enum cf_protection_level.
1271 * gimple.c (gimple_build_call_from_tree): Add second parameter.
1272 Add 'nocf_check' attribute propagation to gimple call.
1273 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
1274 (gimple_build_call_from_tree): Update prototype.
1275 (gimple_call_nocf_check_p): New function.
1276 (gimple_call_set_nocf_check): Likewise.
1277 * gimplify.c: Add second argument to gimple_build_call_from_tree.
1278 * ipa-icf.c: Add nocf_check attribute in statement hash.
1279 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
1280 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
1281 * toplev.c (process_options): Add flag_cf_protection handling.
1283 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
1285 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
1287 2017-10-20 Richard Biener <rguenther@suse.de>
1289 PR tree-optimization/82603
1290 * tree-if-conv.c (predicate_mem_writes): Make sure to only
1291 remove false predicated stores.
1293 2017-10-20 Richard Biener <rguenther@suse.de>
1295 * graphite-isl-ast-to-gimple.c
1296 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
1297 Remove return value and simplify, dump copied stmt after lhs
1299 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
1300 Reduce dump verbosity.
1301 (gsi_insert_earliest): Likewise.
1302 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
1303 * graphite.c (print_global_statistics): Adjust dumping.
1304 (print_graphite_scop_statistics): Likewise.
1305 (print_graphite_statistics): Do not dump loops here.
1306 (graphite_transform_loops): But here.
1308 2017-10-20 Nicolas Roche <roche@adacore.com>
1310 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
1311 * configure: Regenerate.
1313 2017-10-20 Jakub Jelinek <jakub@redhat.com>
1316 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
1317 functions when optimizing replace GIMPLE_RETURN stmts with
1318 calls to __builtin_unreachable ().
1321 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
1322 for -fsanitize=thread link of executables.
1323 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
1324 link of executables.
1327 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
1329 (<shift_insn><mode>3<mask_name>): Change the last of the 3
1330 define_insns for logical vector shifts to use VI248_AVX512BW
1331 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
1332 condition, useless isa and prefix attributes. Change the first
1333 2 of these define_insns to ...
1334 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
1335 define_insn for avx512vl.
1336 (<shift_insn><mode>3): ... and this, new define_insn without
1337 masking for non-avx512vl.
1340 * config/i386/sse.md (*andnot<mode>3,
1341 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
1342 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
1343 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
1344 not applied use empty suffix even for TARGET_AVX512VL.
1345 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
1346 is applied, supply evex,evex or evex,evex,evex instead of just
1349 2017-10-20 Julia Koval <julia.koval@intel.com>
1351 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
1352 (OPTION_MASK_ISA_GFNI_UNSET): New.
1353 (ix86_handle_option): Handle OPT_mgfni.
1354 * config/i386/cpuid.h (bit_GFNI): New.
1355 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
1356 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
1357 * config/i386/i386.c (ix86_target_string): Add -mgfni.
1358 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
1359 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
1360 * config/i386/i386.opt: Add mgfni.
1362 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
1364 * config/msp430/msp430.c (msp430_option_override): Disable
1365 -fdelete-null-pointer-checks.
1366 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
1368 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
1370 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
1371 of x87 and SSE instructions.
1373 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
1375 * asan.c (create_cond_insert_point): Do not update edge count.
1376 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
1377 (afdo_propagate_circuit): Likewise.
1378 (afdo_calculate_branch_prob): Likewise.
1379 (afdo_annotate_cfg): Likewise.
1380 * basic-block.h (struct edge_def): Remove count.
1381 (edge_def::count): New accessor.
1382 * bb-reorder.c (rotate_loop): Update.
1383 (find_traces_1_round): Update.
1384 (connect_traces): Update.
1385 (sanitize_hot_paths): Update.
1386 * cfg.c (unchecked_make_edge): Update.
1387 (make_single_succ_edge): Update.
1388 (check_bb_profile): Update.
1389 (dump_edge_info): Update.
1390 (update_bb_profile_for_threading): Update.
1391 (scale_bbs_frequencies_int): Update.
1392 (scale_bbs_frequencies_gcov_type): Update.
1393 (scale_bbs_frequencies_profile_count): Update.
1394 (scale_bbs_frequencies): Update.
1395 * cfganal.c (connect_infinite_loops_to_exit): Update.
1396 * cfgbuild.c (compute_outgoing_frequencies): Update.
1397 (find_many_sub_basic_blocks): Update.
1398 * cfgcleanup.c (try_forward_edges): Update.
1399 (try_crossjump_to_edge): Update
1400 * cfgexpand.c (expand_gimple_cond): Update
1401 (expand_gimple_tailcall): Update
1402 (construct_exit_block): Update
1403 * cfghooks.c (verify_flow_info): Update
1404 (redirect_edge_succ_nodup): Update
1405 (split_edge): Update
1406 (make_forwarder_block): Update
1407 (duplicate_block): Update
1408 (account_profile_record): Update
1409 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
1410 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
1411 * cfgloopmanip.c (scale_loop_profile): Update.
1413 (lv_adjust_loop_entry_edge): Update.
1414 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
1415 (force_nonfallthru_and_redirect): Update.
1416 (purge_dead_edges): Update.
1417 (rtl_flow_call_edges_add): Update.
1418 * cgraphunit.c (init_lowered_empty_function): Update.
1419 (cgraph_node::expand_thunk): Update.
1420 * gimple-pretty-print.c (dump_probability): Update.
1421 (dump_edge_probability): Update.
1422 * gimple-ssa-isolate-paths.c (isolate_path): Update.
1423 * haifa-sched.c (sched_create_recovery_edges): Update.
1424 * hsa-gen.c (convert_switch_statements): Update.
1425 * ifcvt.c (dead_or_predicable): Update.
1426 * ipa-inline-transform.c (inline_transform): Update.
1427 * ipa-split.c (split_function): Update.
1428 * ipa-utils.c (ipa_merge_profiles): Update.
1429 * loop-doloop.c (add_test): Update.
1430 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
1431 * lto-streamer-in.c (input_cfg): Update.
1432 (input_function): Update.
1433 * lto-streamer-out.c (output_cfg): Update.
1434 * modulo-sched.c (sms_schedule): Update.
1435 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
1436 * predict.c (maybe_hot_edge_p): Update.
1437 (unlikely_executed_edge_p): Update.
1438 (probably_never_executed_edge_p): Update.
1439 (dump_prediction): Update.
1440 (drop_profile): Update.
1441 (propagate_unlikely_bbs_forward): Update.
1442 (determine_unlikely_bbs): Update.
1443 (force_edge_cold): Update.
1444 * profile.c (compute_branch_probabilities): Update.
1445 * reg-stack.c (better_edge): Update.
1446 * shrink-wrap.c (handle_simple_exit): Update.
1447 * tracer.c (better_p): Update.
1448 * trans-mem.c (expand_transaction): Update.
1449 (split_bb_make_tm_edge): Update.
1450 * tree-call-cdce.c: Update.
1451 * tree-cfg.c (gimple_find_sub_bbs): Update.
1452 (gimple_split_edge): Update.
1453 (gimple_duplicate_sese_region): Update.
1454 (gimple_duplicate_sese_tail): Update.
1455 (gimple_flow_call_edges_add): Update.
1456 (insert_cond_bb): Update.
1457 (execute_fixup_cfg): Update.
1458 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
1459 * tree-complex.c (expand_complex_div_wide): Update.
1460 * tree-eh.c (lower_resx): Update.
1461 (unsplit_eh): Update.
1462 (cleanup_empty_eh_move_lp): Update.
1463 * tree-inline.c (copy_edges_for_bb): Update.
1464 (freqs_to_counts): Update.
1465 (copy_cfg_body): Update.
1466 * tree-ssa-dce.c (remove_dead_stmt): Update.
1467 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
1468 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
1469 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
1470 (unloop_loops): Update.
1471 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
1472 * tree-ssa-loop-split.c (connect_loops): Update.
1473 (split_loop): Update.
1474 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
1475 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
1476 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
1477 * tree-ssa-reassoc.c (branch_fixup): Update.
1478 * tree-ssa-tail-merge.c (replace_block_by): Update.
1479 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
1480 (compute_path_counts): Update.
1481 (update_profile): Update.
1482 (recompute_probabilities): Update.
1483 (update_joiner_offpath_counts): Update.
1484 (estimated_freqs_path): Update.
1485 (freqs_to_counts_path): Update.
1486 (clear_counts_path): Update.
1487 (ssa_fix_duplicate_block_edges): Update.
1488 (duplicate_thread_path): Update.
1489 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
1490 (case_bit_test_cmp): Update.
1491 (collect_switch_conv_info): Update.
1492 (gen_inbound_check): Update.
1493 (do_jump_if_equal): Update.
1494 (emit_cmp_and_jump_insns): Update.
1495 * tree-tailcall.c (decrease_profile): Update.
1496 (eliminate_tail_call): Update.
1497 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
1498 (vect_do_peeling): Update.
1499 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
1500 * ubsan.c (ubsan_expand_null_ifn): Update.
1501 (ubsan_expand_ptr_ifn): Update.
1502 * value-prof.c (gimple_divmod_fixed_value): Update.
1503 (gimple_mod_pow2): Update.
1504 (gimple_mod_subtract): Update.
1505 (gimple_ic): Update.
1506 (gimple_stringop_fixed_value): Update.
1508 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
1511 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
1513 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
1515 PR rtl-optimization/82395
1516 * ira-color.c (allocno_priority_compare_func): Fix comparison step
1517 based on non_spilled_static_chain_regno_p.
1519 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
1521 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
1522 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
1523 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
1525 2017-10-19 Martin Sebor <msebor@redhat.com>
1527 PR tree-optimization/82596
1528 * tree.c (array_at_struct_end_p): Handle STRING_CST.
1530 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
1532 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
1533 (get_mem_refs_of_builtin_call): Likewise.
1534 * builtins.c (expand_builtin_apply): Adjust call to
1535 allocate_dynamic_stack_space.
1536 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
1537 the third argument to allocate_dynamic_stack_space, otherwise -1.
1538 (expand_builtin): Deal with all alloca variants.
1539 (is_inexpensive_builtin): Likewise.
1540 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
1541 * calls.c (special_function_p): Deal with all alloca variants.
1542 (initialize_argument_information): Adjust call to
1543 allocate_dynamic_stack_space.
1544 (expand_call): Likewise.
1545 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
1546 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
1547 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
1548 use it for the stack usage computation.
1549 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
1550 * function.c (gimplify_parameters): Call build_alloca_call_expr.
1551 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
1552 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
1553 (in_loop_p): Remove first argument and useless check.
1554 (pass_walloca::execute): Remove useless test and adjust call to above.
1555 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
1556 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
1557 (gimplify_call_expr): Deal with all alloca variants.
1558 * hsa-gen.c (gen_hsa_alloca): Likewise.
1559 (gen_hsa_insns_for_call): Likewise.
1560 * ipa-pure-const.c (special_builtin_state): Likewise.
1561 * tree-chkp.c (chkp_build_returned_bound): Likewise.
1562 * tree-object-size.c (alloc_object_size): Likewise.
1563 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
1564 (call_may_clobber_ref_p_1): Likewise.
1565 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
1566 (ccp_fold_stmt): Likewise.
1567 (optimize_stack_restore): Likewise.
1568 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
1569 (mark_all_reaching_defs_necessary_1): Likewise.
1570 (propagate_necessity): Likewise.
1571 (eliminate_unnecessary_stmts): Likewise.
1572 * tree.c (build_common_builtin_nodes): Build
1573 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
1574 (build_alloca_call_expr): New function.
1575 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
1576 (CASE_BUILT_IN_ALLOCA): Likewise.
1577 (build_alloca_call_expr): Declare.
1578 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
1580 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
1583 * dwarf2out.c (new_die_raw): New static inline function.
1584 (new_die): Use it to create the DIE.
1585 (add_AT_external_die_ref): Likewise.
1586 (clone_die): Likewise.
1587 (clone_as_declaration): Likewise.
1588 (dwarf2out_vms_debug_main_pointer): Likewise.
1589 (base_type_die): Likewise. Remove early return for corner cases.
1590 Do not call add_pubtype on the DIE here.
1591 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
1592 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
1593 typedefs for base types with DW_AT_endianity. Make sure a DIE with
1594 native order exists for base types, attach the DIE manually and call
1595 add_pubtype on it. Do not equate a reverse order DIE to the type.
1597 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
1599 * config/arm/arm.c (align_ok_ldrd_strd): New function.
1600 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
1602 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
1604 2017-10-19 Jakub Jelinek <jakub@redhat.com>
1606 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
1607 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
1608 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
1609 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
1610 * opts.c (sanitizer_opts): Add builtin.
1611 * ubsan.c (instrument_builtin): New function.
1612 (pass_ubsan::execute): Call it.
1613 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
1614 * doc/invoke.texi: Document -fsanitize=builtin.
1616 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
1617 builtins, store max (log2 (align), 0) into uchar field instead of
1618 align into uptr field.
1619 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
1620 store uchar 0 field instead of uptr 0 field.
1621 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
1622 instead of passing one address of struct with 2 locations pass
1623 two addresses of structs with 1 location each.
1624 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
1625 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
1626 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
1627 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
1628 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
1629 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
1630 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
1631 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
1633 2017-10-19 Martin Liska <mliska@suse.cz>
1636 * file-find.c (remove_prefix): Remove.
1637 * file-find.h (remove_prefix): Likewise.
1638 * gcc-ar.c: Remove smartness of lookup.
1640 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
1642 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
1643 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
1644 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
1646 2017-10-19 Jakub Jelinek <jakub@redhat.com>
1649 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
1650 (setcc + and to xor + setcc): New peephole2.
1652 2017-10-19 Tom de Vries <tom@codesourcery.com>
1654 * doc/sourcebuild.texi (Test Directives, Variants of
1655 dg-require-support): Add dg-require-stack-size.
1657 2017-10-19 Martin Liska <mliska@suse.cz>
1660 * gimplify.c (gimplify_decl_expr): Do not instrument variables
1661 that have a large alignment.
1662 (gimplify_target_expr): Likewise.
1664 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
1666 PR rtl-optimization/82602
1667 * ira.c (rtx_moveable_p): Return false for volatile asm.
1669 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
1672 * config/i386/i386-modes.def (CCGZ): New CC mode.
1673 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
1674 * config/i386/predicates.md (ix86_comparison_operator):
1676 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
1677 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
1678 with double-word subtraction.
1679 (put_condition_code): Handle CCGZmode.
1681 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
1683 * wide-int.cc (debug (const wide_int &)): New.
1684 (debug (const wide_int *)): New.
1685 (debug (const widest_int &)): New.
1686 (debug (const widest_int *)): New.
1688 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
1691 * lra-constraints.c (curr_insn_transform): Use non-input operand
1692 instead of output one for matched reload.
1694 2017-10-18 Bin Cheng <bin.cheng@arm.com>
1696 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
1697 (tree-ssa-loop-ivopts.h): New header file.
1698 (struct builtin_info): New fields.
1699 (classify_builtin_1): Compute and record base and offset parts for
1700 memset builtin partition by calling strip_offset.
1701 (offset_cmp, fuse_memset_builtins): New functions.
1702 (finalize_partitions): Fuse adjacent memset partitions by calling
1704 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
1705 Expose the interface.
1706 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
1708 2017-10-18 Bin Cheng <bin.cheng@arm.com>
1710 PR tree-optimization/82574
1711 * tree-loop-distribution.c (find_single_drs): New parameter. Check
1712 that data reference must be executed exactly once per iteration
1713 against the outermost loop in nest.
1714 (classify_partition): Update call to above function.
1716 2017-10-18 Richard Biener <rguenther@suse.de>
1718 PR tree-optimization/82591
1719 * graphite.c (graphite_transform_loops): Move code gen message
1721 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1722 Here. Handle scop_to_isl_ast failing.
1723 (scop_to_isl_ast): Limit the number of ISL operations.
1725 2017-10-18 Richard Biener <rguenther@suse.de>
1727 * graphite-isl-ast-to-gimple.c
1728 (translate_isl_ast_to_gimple::set_rename): Simplify.
1729 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
1730 (graphite_copy_stmts_from_block): ... here.
1731 (copy_bb_and_scalar_dependences): Simplify.
1732 (add_parameters_to_ivs_params): Canonicalize.
1733 (generate_entry_out_of_ssa_copies): Simplify.
1734 * graphite-sese-to-poly.c (extract_affine_name): Simplify
1735 by passing in ISL dimension.
1736 (parameter_index_in_region_1): Rename to ...
1737 (parameter_index_in_region): ... this.
1738 (extract_affine): Adjust assert, pass down parameter index.
1739 (add_param_constraints): Use range-info when available.
1740 (build_scop_context): Adjust.
1741 * sese.c (new_sese_info): Adjust.
1742 (free_sese_info): Likewise.
1743 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
1744 Remove unused typedefs.
1745 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
1747 2017-10-18 Martin Liska <mliska@suse.cz>
1749 * combine.c (simplify_compare_const): Add gcc_fallthrough.
1751 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1753 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
1754 (s390_sched_init): Do not reset s390_sched_state if we entered the
1755 current basic block via a fallthru edge and all others are unlikely.
1757 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1759 * config/s390/s390.c (NUM_SIDES): New variable.
1760 (LONGRUNNING_THRESHOLD): New variable.
1761 (LATENCY_FACTOR): New variable.
1762 (s390_sched_score): Decrease score for long-running instructions on
1764 (s390_sched_variable_issue): Perform bookkeeping for long-running
1767 2017-10-18 Richard Biener <rguenther@suse.de>
1769 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
1770 Simplify with removal of the parameter rename map.
1771 (set_rename): Likewise.
1772 (should_copy_to_new_region): Likewise.
1773 (graphite_copy_stmts_from_block): Likewise.
1774 (copy_bb_and_scalar_dependences): Remove initialization of
1775 unused copied_bb_map.
1777 (copy_internal_parameters): Likewise.
1778 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
1779 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
1780 Use INTEGRAL_TYPE_P.
1781 (parameter_index_in_region_1): Rename to ...
1782 (assign_parameter_index_in_region): ... this. Assert we have
1783 a parameter we handle.
1784 (scan_tree_for_params): Adjust.
1785 * sese.h (parameter_rename_map_t): Remove.
1786 (struct sese_info_t): Remove unused parameter_rename_map and
1787 copied_bb_map members.
1788 * sese.c (new_sese_info): Adjust.
1789 (free_sese_info): Likewise.
1791 2017-10-18 Martin Liska <mliska@suse.cz>
1794 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
1795 on an abnormal edge.
1797 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
1799 * doc/invoke.texi (ffunction-sections and fdata-sections):
1802 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
1804 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
1805 the use statement can throw internally.
1807 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
1809 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
1810 any RTX present on the RHS of a SET.
1811 * compare-elim.c (try_eliminate_compare): Restore comment.
1813 2017-10-17 Jakub Jelinek <jakub@redhat.com>
1815 * langhooks.h (struct lang_hooks): Document that tree_size langhook
1816 may be also called on tcc_type nodes.
1817 * langhooks.c (lhd_tree_size): Likewise.
1819 2017-10-17 David Malcolm <dmalcolm@redhat.com>
1821 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
1822 format_warning_at_substring.
1823 (maybe_warn): Convert source_range * param to a location_t. Pass
1824 UNKNOWN_LOCATION rather than NULL to fmtwarn.
1825 (format_directive): Remove code to extract source_ranges and
1826 source_range * in favor of just a location_t.
1827 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
1829 * substring-locations.c (format_warning_va): Convert
1830 source_range * param to a location_t.
1831 (format_warning_at_substring): Likewise.
1832 * substring-locations.h (format_warning_va): Likewise.
1833 (format_warning_at_substring): Likewise.
1835 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
1837 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
1839 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
1840 and scatter/gather ops.
1842 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
1843 vec_gather_load and vec_scatter_store.
1844 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
1845 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
1846 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
1847 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
1848 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
1849 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
1851 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
1853 * reg-stack.c (compare_for_stack_reg): Add bool argument.
1854 Detect FTST instruction and handle its register pops. Only pop
1855 second operand if can_pop_second_op is true.
1856 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
1857 set can_pop_second_op to false in the compare_for_stack_reg call.
1859 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
1860 output_fp_compare for stack register operands.
1861 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
1862 instructions here. Do not emit stack register pops here. Assert
1863 that FCOMPP pops next to top stack register. Rewrite function.
1865 2017-10-17 Nathan Sidwell <nathan@acm.org>
1868 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
1869 use DECL_ASSEMBLER_NAME_RAW.
1872 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
1875 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
1876 Wilco Dijkstra <wilco.dijkstra@arm.com>
1878 * builtins.c (expand_builtin_update_setjmp_buf): Add a
1879 converstion to Pmode from the buf_addr.
1881 2017-10-17 Richard Biener <rguenther@suse.de>
1883 * graphite-dependences.c (scop_get_reads_and_writes): Change
1884 output parameters to references.
1886 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
1888 PR 71026/tree-optimization
1889 * fold-const.c (distribute_real_division): Removed.
1890 (fold_binary_loc): Remove calls to distribute_real_divison.
1892 2017-10-17 Richard Biener <rguenther@suse.de>
1894 * graphite-scop-detection.c
1895 (scop_detection::stmt_has_simple_data_refs_p): Always use
1896 the full nest as region.
1897 (try_generate_gimple_bb): Likewise.
1898 * sese.c (scalar_evolution_in_region): Simplify now that
1899 SCEV can handle instantiation in regions.
1900 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
1901 in the non-loop part of a function if requested.
1903 2017-10-17 Richard Biener <rguenther@suse.de>
1905 PR tree-optimization/82563
1906 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
1908 (graphite_regenerate_ast_isl): Call it.
1909 * graphite-scop-detection.c (build_scops): Remove entry edge split.
1911 2017-10-17 Jakub Jelinek <jakub@redhat.com>
1913 PR tree-optimization/82549
1914 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
1915 Formatting fixes. Instead of calling make_bit_field_ref with negative
1918 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
1920 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
1921 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
1922 _mm_maskz_reduce_ss): New.
1923 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
1924 __builtin_ia32_reducess_mask): Ditto..
1925 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
1926 * config/i386/sse.md (reduces<mode>): Renamed to ...
1927 (reduces<mode><mask_scalar_name>): ... this.
1928 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
1930 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
1931 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
1933 2017-10-16 David Malcolm <dmalcolm@redhat.com>
1935 * Makefile.in (OBJS): Add unique-ptr-tests.o.
1936 * selftest-run-tests.c (selftest::run_tests): Call
1937 selftest::unique_ptr_tests_cc_tests.
1938 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
1939 * unique-ptr-tests.cc: New file.
1941 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
1944 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1946 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1947 (make_hard_regno_born, make_hard_regno_dead): Update
1948 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
1950 2017-10-16 Jeff Law <law@redhat.com>
1952 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
1954 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
1956 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
1958 2017-10-16 Olivier Hainque <hainque@adacore.com>
1960 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
1961 with_cpu if we were configured for an e500v2 target cpu name.
1963 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
1965 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
1966 * doc/invoke.texi: Document +nodsp as a valid extension for
1969 2017-10-16 Martin Liska <mliska@suse.cz>
1971 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
1972 (test_set_range): Likewise.
1973 (test_range_functions): Rename to ...
1974 (test_bit_in_range): ... this.
1975 (sbitmap_c_tests): Add new test.
1977 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1979 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
1981 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
1982 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
1984 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1986 * config/aarch64/aarch64-builtins.c
1987 (aarch64_types_quadopu_lane_qualifiers): New.
1988 (TYPES_QUADOPU_LANE): New.
1989 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
1990 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
1991 (aarch64_<sur>dot_laneq<vsi2qi>): New.
1992 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
1993 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
1994 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
1995 (Vdottype, DOTPROD): New.
1996 (sur): Add SDOT and UDOT.
1998 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2000 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
2001 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
2002 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
2004 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
2005 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
2006 Enable TARGET_DOTPROD.
2007 (cortex-a75.cortex-a55): Likewise.
2008 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
2010 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2012 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
2013 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
2015 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
2017 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
2018 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
2019 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
2020 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
2021 * config/arm/types.md (neon_dot, neon_dot_q): New.
2022 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
2024 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2026 * config/arm/arm.h (TARGET_DOTPROD): New.
2027 * config/arm/arm.c (arm_arch_dotprod): New.
2028 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
2029 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
2030 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
2031 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
2032 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
2033 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
2034 * doc/invoke.texi (armv8.2-a): Document dotprod
2036 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
2038 * i386.c (ix86_vec_cost): New function.
2039 (ix86_rtx_costs): Handle vector operations better.
2040 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
2041 * x86-tune-costs.h: Add new costs to all tables.
2043 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
2045 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
2047 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
2048 divsd, sqrtss and sqrtsd
2049 * x86-tune-costs.h: Add new entries to all costs.
2050 (znver1_cost): Fix to match real instruction latencies.
2052 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2053 Michael Collison <michael.collison@arm.com>
2055 * compare-elim.c: Include emit-rtl.h.
2056 (can_merge_compare_into_arith): New function.
2057 (try_validate_parallel): Likewise.
2058 (try_merge_compare): Likewise.
2059 (try_eliminate_compare): Call the above when no previous clobber
2061 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
2064 2017-10-14 Jakub Jelinek <jakub@redhat.com>
2068 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
2069 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
2073 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
2074 to be any operand_equal_p operands. For & (B - 1) require
2075 B to be power of 2. Recognize
2076 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
2078 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2081 * optabs.c (expand_memory_blockage): Fix call of
2082 targetm.have_memory_blockage.
2084 2017-10-14 Jakub Jelinek <jakub@redhat.com>
2087 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
2088 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
2089 objects to extra_objs instead of overwriting it.
2091 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2093 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
2094 Use any_fp_register_operand as operand[3] predicate. Simplify
2095 equality test for operands[2] and operands[4] memory location.
2096 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
2097 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
2098 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
2099 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
2100 any_fp_register_operand as operand[1] predicate. Simplify
2101 equality test for operands[0] and operands[3] memory location.
2102 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
2103 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
2104 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
2106 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2108 * target-insns.def: Add memory_blockage.
2109 * optabs.c (expand_memory_blockage): New function.
2110 (expand_asm_memory_barrier): Rename ...
2111 (expand_asm_memory_blockage): ... to this.
2112 (expand_mem_thread_fence): Call expand_memory_blockage
2113 instead of expand_asm_memory_barrier.
2114 (expand_mem_singnal_fence): Ditto.
2115 (expand_atomic_load): Ditto.
2116 (expand_atomic_store): Ditto.
2117 * doc/md.texi (Standard Pattern Names For Generation):
2118 Document memory_blockage instruction pattern.
2120 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
2122 * config/rl78/rl78.c (rl78_emit_libcall): New function.
2123 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
2124 * config/rl78/rl78.md: New define_expand "adddi3".
2126 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
2128 * cfghooks.c (verify_flow_info): Disable check that all probabilities
2131 2017-10-13 Jeff Law <law@redhat.com>
2133 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
2135 2017-10-13 Jakub Jelinek <jakub@redhat.com>
2138 * internal-fn.c (expand_mul_overflow): If both operands have
2139 the same highpart of -1 or 0 and the topmost bit of lowpart
2140 is different, overflow is if res <= 0 rather than res < 0.
2142 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
2144 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
2145 TARGET_P9_VECTOR code for unaligned_load case.
2147 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
2149 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
2151 2017-10-13 Nathan Sidwell <nathan@acm.org>
2153 * tree-core.h (tree_contains_struct): Make bool.
2154 * tree.c (tree_contains_struct): Likewise.
2155 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
2156 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
2157 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
2158 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
2159 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
2161 2017-10-13 Richard Biener <rguenther@suse.de>
2163 * graphite-isl-ast-to-gimple.c
2164 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
2165 parameters and dominance check.
2166 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
2167 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
2168 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
2169 Do not update SSA form here or do intermediate IL verification.
2170 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
2171 (graphite_initialize): Remove check on the number of loops in
2172 the function and inline into graphite_transform_loops.
2173 (graphite_finalize): Inline into graphite_transform_loops.
2174 (graphite_transform_loops): Perform SSA update and IL verification
2176 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
2178 2017-10-13 Richard Biener <rguenther@suse.de>
2180 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
2181 graphite_expression_type_precision): Avoid global constructor
2183 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
2184 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
2185 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
2186 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
2187 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
2188 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
2190 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
2193 * config/i386/i386.h (ix86_red_zone_size): New.
2194 * config/i386/i386.md (push peephole2s): Replace
2195 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
2197 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2198 Alan Hayward <alan.hayward@arm.com>
2199 David Sherwood <david.sherwood@arm.com>
2201 * combine.c (can_change_dest_mode): Reject changes in
2202 REGMODE_NATURAL_SIZE.
2204 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2205 Alan Hayward <alan.hayward@arm.com>
2206 David Sherwood <david.sherwood@arm.com>
2208 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
2209 (expand_debug_source_expr): Likewise.
2210 * combine.c (combine_simplify_rtx): Likewise.
2211 * cse.c (fold_rtx): Likewise.
2212 * fwprop.c (canonicalize_address): Likewise.
2213 * targhooks.c (default_shift_truncation_mask): Likewise.
2215 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2216 Alan Hayward <alan.hayward@arm.com>
2217 David Sherwood <david.sherwood@arm.com>
2219 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
2220 (widened_mode): Likewise.
2221 (expand_unop): Likewise.
2222 * ree.c (transform_ifelse): Likewise.
2223 (merge_def_and_ext): Likewise.
2224 (combine_reaching_defs): Likewise.
2225 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2227 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2228 Alan Hayward <alan.hayward@arm.com>
2229 David Sherwood <david.sherwood@arm.com>
2231 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
2232 * combine.c (gen_lowpart_for_combine): Likewise.
2233 * dwarf2out.c (rtl_for_decl_location): Likewise.
2234 * final.c (alter_subreg): Likewise.
2235 * rtlhooks.c (gen_lowpart_general): Likewise.
2236 (gen_lowpart_if_possible): Likewise.
2238 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2239 Alan Hayward <alan.hayward@arm.com>
2240 David Sherwood <david.sherwood@arm.com>
2242 * calls.c (expand_call): Use subreg_lowpart_offset.
2243 * cse.c (cse_insn): Likewise.
2244 * regcprop.c (copy_value): Likewise.
2245 (copyprop_hardreg_forward_1): Likewise.
2247 2017-10-13 Jakub Jelinek <jakub@redhat.com>
2250 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
2251 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
2252 =Q constraints to +Q and into insn condition add check
2253 that operands[0] and operands[1] are equal.
2254 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
2255 =Q constraints to +Q and into insn condition add check
2256 that operands[0] is equal to either operands[1] or operands[2].
2259 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
2260 instead of handling MINUS_EXPR twice (once for each argument),
2261 canonicalize operand order and handle just once, use rtype where
2262 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
2265 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
2266 any values of __C while still being pattern recognizable as a simple
2269 2017-10-13 Richard Biener <rguenther@suse.de>
2271 PR tree-optimization/82451
2273 2017-10-02 Richard Biener <rguenther@suse.de>
2275 PR tree-optimization/82355
2276 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
2277 a mapping for the enclosing loop but avoid generating one for
2279 (copy_bb_and_scalar_dependences): Remove premature codegen
2280 error on PHIs in blocks duplicated into multiple places.
2281 * graphite-scop-detection.c
2282 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
2283 in the region use it as loop and nest to analyze the DR in.
2284 (try_generate_gimple_bb): Likewise.
2285 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
2286 (add_loop_constraints): For blocks in a loop not in the region
2287 create a dimension with a single iteration.
2288 * sese.h (gbb_loop_at_index): Remove assert.
2290 * cfgloop.c (loop_preheader_edge): For the loop tree root
2291 return the single successor of the entry block.
2292 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2293 Reset the SCEV hashtable and niters.
2294 * graphite-scop-detection.c
2295 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
2296 assert that we only have POLYNOMIAL_CHREC that vary in loops
2297 contained in the region.
2298 (scop_detection::graphite_can_represent_expr): Adjust.
2299 (scop_detection::stmt_has_simple_data_refs_p): For loops
2300 not in the region set loop to NULL. The nest is now the
2301 entry edge to the region.
2302 (try_generate_gimple_bb): Likewise.
2303 * sese.c (scalar_evolution_in_region): Adjust for
2304 instantiate_scev change.
2305 * tree-data-ref.h (graphite_find_data_references_in_stmt):
2306 Make nest parameter the edge into the region.
2307 (create_data_ref): Likewise.
2308 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
2309 entry edge into a region and adjust instantiate_scev calls.
2310 (create_data_ref): Likewise.
2311 (graphite_find_data_references_in_stmt): Likewise.
2312 (find_data_references_in_stmt): Pass the loop preheader edge
2313 from the nest argument.
2314 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
2315 parameter the edge into the region.
2316 (instantiate_parameters): Use the loop preheader edge as entry.
2317 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
2319 (get_instantiated_value_entry): Make instantiate_below parameter
2320 the edge into the region.
2321 (instantiate_scev_name): Likewise. Adjust dominance checks,
2322 when we cannot use loop-based instantiation instantiate by
2323 walking use-def chains.
2324 (instantiate_scev_poly): Adjust.
2325 (instantiate_scev_binary): Likewise.
2326 (instantiate_scev_convert): Likewise.
2327 (instantiate_scev_not): Likewise.
2328 (instantiate_array_ref): Remove.
2329 (instantiate_scev_3): Likewise.
2330 (instantiate_scev_2): Likewise.
2331 (instantiate_scev_1): Likewise.
2332 (instantiate_scev_r): Do not blindly handle N-operand trees.
2333 Do not instantiate array-refs. Handle all constants and invariants.
2334 (instantiate_scev): Make instantiate_below parameter
2335 the edge into the region.
2336 (resolve_mixers): Use the loop preheader edge for the region
2337 parameter to instantiate_scev_r.
2338 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
2340 2017-10-13 Richard Biener <rguenther@suse.de>
2342 PR tree-optimization/82525
2343 * graphite-isl-ast-to-gimple.c
2344 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
2346 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
2347 Fail code generation when we cannot represent the isl integer.
2348 (binary_op_to_tree): Elide modulo operations that are no-ops
2349 in the type we code generate. Remove now superfluous code
2352 2017-10-13 Richard Biener <rguenther@suse.de>
2354 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
2355 (scop_detection::harmful_loop_in_region): Remove premature
2356 IV type restriction.
2357 (scop_detection::graphite_can_represent_scev): We can handle
2358 pointer IVs just fine.
2360 2017-10-13 Alan Modra <amodra@gmail.com>
2362 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
2363 "Clobbers and Scratch Registers". Add paragraph on
2364 alternative to clobbers for scratch registers and OpenBLAS
2367 2017-10-13 Alan Modra <amodra@gmail.com>
2369 * doc/extend.texi (Clobbers): Correct vax example. Delete old
2370 example of a memory input for a string of known length. Move
2371 commentary out of table. Add a number of new examples
2372 covering array memory inputs.
2374 2017-10-12 Martin Liska <mliska@suse.cz>
2376 PR tree-optimization/82493
2377 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
2378 (test_range_functions): New function.
2379 (sbitmap_c_tests): Likewise.
2380 * selftest-run-tests.c (selftest::run_tests): Run new tests.
2381 * selftest.h (sbitmap_c_tests): New function.
2383 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
2385 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
2387 * config/rs6000/amo.h: Fix spacing issue.
2389 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2392 * config/i386/i386.md (*ashl<mode>3_mask_1,
2393 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
2394 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
2397 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
2399 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
2400 (profile_probability): Set max_probability
2401 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
2403 * profile-count.c (profile_probability::differs_from_p): Do not
2404 rely on max_probaiblity == 10000
2406 2017-10-12 Jeff Law <law@redhat.com>
2408 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
2411 2017-10-12 Martin Sebor <msebor@redhat.com>
2415 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
2416 (handle_alias_pairs): Call it.
2417 * common.opt (-Wattribute-alias): New option.
2418 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
2419 * doc/invoke.texi (-Wattribute-alias): Document.
2421 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
2424 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
2426 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2428 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2429 (make_hard_regno_born, make_hard_regno_dead): Update
2430 bb_killed_pseudos and bb_gen_pseudos.
2432 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
2434 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
2436 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
2438 * config/alpha/alpha.c (alpha_split_conditional_move):
2439 Use std::swap instead of manually swapping.
2440 (alpha_stdarg_optimize_hook): Ditto.
2441 (alpha_canonicalize_comparison): Ditto.
2443 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2445 * tree-loop-distribution.c (struct builtin_info): New struct.
2446 (struct partition): Refactor fields into struct builtin_info.
2447 (partition_free): Free struct builtin_info.
2448 (build_size_arg_loc, build_addr_arg_loc): Delete.
2449 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
2450 information from struct builtin_info.
2451 (find_single_drs): New function refactored from classify_partition.
2452 Also moved builtin validity checks to this function.
2453 (compute_access_range, alloc_builtin): New functions.
2454 (classify_builtin_st, classify_builtin_ldst): New functions.
2455 (classify_partition): Refactor code into functions find_single_drs,
2456 classify_builtin_st and classify_builtin_ldst.
2457 (distribute_loop): Don't do runtime alias check when distributing
2459 (find_seed_stmts_for_distribution): New function.
2460 (pass_loop_distribution::execute): Refactor code finding seed
2461 stmts into above function. Support distribution for the innermost
2462 two-level loop nest. Adjust dump information.
2464 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2466 * tree-loop-distribution.c: Adjust the general comment.
2467 (NUM_PARTITION_THRESHOLD): New macro.
2468 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
2469 (classify_partition): Skip builtin pattern of loop nest's inner loop.
2470 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
2471 in call to build_partition_graph.
2472 (finalize_partitions): New parameter. Make loop distribution more
2473 conservative by fusing more partitions.
2474 (distribute_loop): Don't do runtime alias check in case of loop nest
2476 (find_seed_stmts_for_distribution): New function.
2477 (prepare_perfect_loop_nest): New function.
2478 (pass_loop_distribution::execute): Refactor code finding seed stmts
2479 and loop nest into above functions. Support loop nest distribution.
2480 Adjust dump information accordingly.
2482 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2484 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
2485 and set PTYPE_SEQUENTIAL for merged partition.
2487 2017-10-12 Richard Biener <rguenther@suse.de>
2489 PR tree-optimization/69728
2491 2017-09-19 Richard Biener <rguenther@suse.de>
2493 PR tree-optimization/69728
2494 * graphite-sese-to-poly.c (schedule_error): New global.
2495 (add_loop_schedule): Handle empty domain by failing the
2497 (build_original_schedule): Handle schedule_error.
2499 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
2500 domain by returning an unchanged schedule.
2502 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2504 * genrecog.c (validate_pattern): For VEC_SELECT verify that
2505 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
2507 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
2509 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
2512 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2515 * expr.c (store_field): Don't optimize away bitsize == 0 store
2516 from CALL_EXPR with addressable return type.
2518 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
2520 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
2521 * config/rs6000/rs6000.md (sel): Delete mode attribute.
2522 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
2523 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
2524 TARGET_ISEL instead of TARGET_ISEL<sel>.
2526 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
2528 * config/rs6000/rs6000.c
2529 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
2531 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
2533 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
2535 (reg_or_cint_operand): Fix comment.
2536 (reg_or_zero_operand): New predicate.
2537 * config/rs6000/rs6000-protos.h (output_isel): Delete.
2538 * config/rs6000/rs6000.c (output_isel): Delete.
2539 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
2540 instead of reg_or_cint_operand. Output instruction directly (not via
2542 (isel_unsigned_<mode>): Ditto.
2543 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
2544 gpc_reg_operand. Add an instruction alternative for this. Output
2545 instruction directly.
2546 (*isel_reversed_unsigned_<mode>): Ditto.
2548 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2550 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
2551 (TARGET_CANONICALIZE_COMPARISON): Define.
2553 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
2556 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
2557 Check whether the dest is REG before adding REG_EQUIV note.
2559 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
2562 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2564 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2565 (make_hard_regno_born, make_hard_regno_dead): Update
2566 bb_killed_pseudos and bb_gen_pseudos.
2568 2017-10-11 Nathan Sidwell <nathan@acm.org>
2570 * incpath.h (enum incpath_kind): Name enum, prefix values.
2571 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
2572 * incpath.c (heads, tails): Use INC_MAX.
2573 (add_env_var_paths, add_standard_paths): Use incpath_kind.
2574 (merge_include_chains, split_quote_chain,
2575 register_include_chains): Update incpath_kind names.
2576 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
2577 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
2579 (add_framework_path, darwin_register_objc_includes): Likewise.
2580 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
2582 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2584 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
2585 Do not use float_operator operator predicate.
2586 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
2587 * config/i386/predicates.md (float_operator): Remove predicate.
2589 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2591 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
2592 (*jccxf_i387): Ditto.
2593 (*jcc<mode>_i387): Ditto.
2594 (*jccu<mode>_i387): Ditto.
2595 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
2596 (*jcc_*_i387 splitters): Remove.
2597 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
2598 * config/i386/i386.c (ix86_split_fp_branch): Remove.
2599 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
2602 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
2604 * profile-count.h (slow_safe_scale_64bit): New function.
2605 (safe_scale_64bit): New inline.
2606 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
2607 * profile-count.c: Include wide-int.h
2608 (slow_safe_scale_64bit): New.
2610 2017-10-11 Nathan Sidwell <nathan@acm.org>
2612 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
2613 HAS_DECL_ASSEMBLER_NAME_P.
2614 * gimple-expr.c (gimple_decl_printable_name: Check
2615 HAS_DECL_ASSEMBLER_NAME_P too.
2616 * ipa-utils.h (type_in_anonymous_namespace_p): Check
2617 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
2618 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
2619 * passes.c (rest_of_decl_compilation): Check
2620 HAS_DECL_ASSEMBLER_NAME_P too.
2621 * recog.c (verify_changes): Likewise.
2622 * tree-pretty-print.c (dump_decl_name): Likewise.
2623 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
2625 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
2626 (SET_DECL_ASSEMBLER_NAME): Use it.
2627 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
2628 (COPY_DECL_ASSEMBLER_NAME): Likewise.
2629 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
2631 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
2633 * config.gcc (i386, x86_64): Add extra objects.
2634 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
2635 (ix86_min_insn_size): Declare.
2636 (ix86_issue_rate): Declare.
2637 (ix86_adjust_cost): Declare.
2638 (ia32_multipass_dfa_lookahead): Declare.
2639 (ix86_macro_fusion_p): Declare.
2640 (ix86_macro_fusion_pair_p): Declare.
2641 (ix86_bd_has_dispatch): Declare.
2642 (ix86_bd_do_dispatch): Declare.
2643 (ix86_core2i7_init_hooks): Declare.
2644 (ix86_atom_sched_reorder): Declare.
2645 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
2646 (COSTS_N_BYTES): Move to x86-tune-costs.h.
2647 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
2648 (rip_relative_addr_p): Rename to ...
2649 (ix86_rip_relative_addr_p): ... this one; export.
2650 (memory_address_length): Update.
2651 (ix86_issue_rate): Move to x86-tune-sched.c.
2652 (ix86_flags_dependent): Move to x86-tune-sched.c.
2653 (ix86_agi_dependent): Move to x86-tune-sched.c.
2654 (exact_dependency_1): Move to x86-tune-sched.c.
2655 (exact_store_load_dependency): Move to x86-tune-sched.c.
2656 (ix86_adjust_cost): Move to x86-tune-sched.c.
2657 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
2658 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
2659 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
2660 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
2661 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
2662 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
2663 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
2664 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
2665 (min_insn_size): Rename to ...
2666 (ix86_min_insn_size): ... this one; export.
2667 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
2668 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
2669 (core2i7_first_cycle_multipass_backtrack): Move to
2670 x86-tune-sched-core.c.
2671 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
2672 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
2673 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
2674 (ix86_avoid_jump_mispredicts): Update.
2675 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
2676 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
2677 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
2678 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
2679 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
2680 (MAX_INSN): Move to ix86-tune-sched-bd.c.
2681 (MAX_IMM): Move to ix86-tune-sched-bd.c.
2682 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
2683 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
2684 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
2685 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
2686 (MAX_STORE): Move to ix86-tune-sched-bd.c.
2687 (BIG): Move to ix86-tune-sched-bd.c.
2688 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
2689 (enum insn_path): Move to ix86-tune-sched-bd.c.
2690 (get_mem_group): Move to ix86-tune-sched-bd.c.
2691 (is_cmp): Move to ix86-tune-sched-bd.c.
2692 (dispatch_violation): Move to ix86-tune-sched-bd.c.
2693 (is_branch): Move to ix86-tune-sched-bd.c.
2694 (is_prefetch): Move to ix86-tune-sched-bd.c.
2695 (init_window): Move to ix86-tune-sched-bd.c.
2696 (allocate_window): Move to ix86-tune-sched-bd.c.
2697 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
2698 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
2699 (process_end_window): Move to ix86-tune-sched-bd.c.
2700 (allocate_next_window): Move to ix86-tune-sched-bd.c.
2701 (find_constant): Move to ix86-tune-sched-bd.c.
2702 (get_num_immediates): Move to ix86-tune-sched-bd.c.
2703 (has_immediate): Move to ix86-tune-sched-bd.c.
2704 (get_insn_path): Move to ix86-tune-sched-bd.c.
2705 (get_insn_group): Move to ix86-tune-sched-bd.c.
2706 (count_num_restricted): Move to ix86-tune-sched-bd.c.
2707 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
2708 (add_insn_window): Move to ix86-tune-sched-bd.c.
2709 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
2710 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
2711 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
2712 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
2713 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
2714 (do_dispatch): Move to ix86-tune-sched-bd.c.
2715 (has_dispatch): Move to ix86-tune-sched-bd.c.
2716 * config/i386/t-i386: Add new object files.
2717 * config/i386/x86-tune-costs.h: New file.
2718 * config/i386/x86-tune-sched-atom.c: New file.
2719 * config/i386/x86-tune-sched-bd.c: New file.
2720 * config/i386/x86-tune-sched-core.c: New file.
2721 * config/i386/x86-tune-sched.c: New file.
2723 2017-10-11 Liu Hao <lh_mouse@126.com>
2725 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
2726 the generic version below instead.
2727 (should_colorize): Recognize Windows consoles as terminals
2729 * pretty-print.c [__MINGW32__] (write_all): New function.
2730 [__MINGW32__] (find_esc_head): Likewise.
2731 [__MINGW32__] (find_esc_terminator): Likewise.
2732 [__MINGW32__] (eat_esc_sequence): Likewise.
2733 [__MINGW32__] (mingw_ansi_fputs): New function that handles
2735 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
2738 2017-10-11 Richard Biener <rguenther@suse.de>
2740 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
2741 Properly call analyze_scalar_evolution with the loop of the stmt.
2743 2017-10-11 Richard Biener <rguenther@suse.de>
2745 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
2746 * tree-core.h (tree_base): Add chrec_var union member.
2747 * tree.h (CHREC_VAR): Remove.
2748 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
2749 * tree-chrec.h (build_polynomial_chrec): Adjust.
2750 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
2751 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
2753 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
2755 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
2756 * match.pd: ... here.
2757 ((T) X == (T) Y): Relax condition.
2759 2017-10-11 Bin Cheng <bin.cheng@arm.com>
2761 PR tree-optimization/82472
2762 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
2764 (break_alias_scc_partitions): Update postorder number.
2766 2017-10-11 Martin Liska <mliska@suse.cz>
2769 * opts.c (parse_no_sanitize_attribute): Do not use error_value
2771 * opts.h (parse_no_sanitize_attribute): Remove last argument.
2773 2017-10-11 Martin Liska <mliska@suse.cz>
2775 * print-rtl.c (print_insn): Move declaration of idbuf
2776 to same scope as name.
2778 2017-10-11 Martin Liska <mliska@suse.cz>
2783 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
2786 2017-10-11 Martin Liska <mliska@suse.cz>
2789 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
2792 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
2794 * config.gcc (default_gnu_indirect_function): Default to yes for
2795 arm*-*-linux* with glibc.
2797 2017-10-11 Richard Biener <rguenther@suse.de>
2799 * tree-scalar-evolution.c (get_scalar_evolution): Handle
2800 default-defs and types we do not want to analyze.
2801 (interpret_loop_phi): Replace unreachable code with an assert.
2802 (compute_scalar_evolution_in_loop): Remove and inline ...
2803 (analyze_scalar_evolution_1): ... here, replacing condition with
2804 what makes the intent clearer. Remove handling of cases
2805 get_scalar_evolution now handles.
2807 2017-10-10 Jim Wilson <wilson@tuliptree.org>
2809 PR rtl-optimization/81434
2810 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
2811 comment for main loop. In sched_group_found if, also add checks for
2812 pass and min_cost_group.
2814 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
2816 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
2817 (rs6000_insn_cost): New function.
2818 * config/rs6000/rs6000.md (cost): New attribute.
2820 2017-10-10 Jakub Jelinek <jakub@redhat.com>
2821 H.J. Lu <hongjiu.lu@intel.com>
2825 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
2826 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
2827 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
2828 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
2829 to OPTION_MASK_ISA_AVX512VL - builtins that have both
2830 OPTION_MASK_ISA_MMX and some other bit set require both
2831 mmx and the ISAs without the mmx bit.
2832 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
2833 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
2834 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
2835 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
2836 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
2837 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
2838 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
2839 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
2840 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
2841 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
2842 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
2843 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
2844 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
2845 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
2846 Add OPTION_MASK_ISA_MMX.
2848 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
2850 * config.gcc (armv7*-*-freebsd*): New target.
2851 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
2853 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
2855 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
2856 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
2859 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
2861 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
2862 a template parameter.
2863 (WIDE_INT_REF_FOR): Update accordingly.
2864 * tree.h (wi::int_traits <const_tree>): Delete.
2865 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
2866 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
2867 (wi::tree_to_wide_ref): New typedef.
2868 (wi::to_wide): New function.
2869 * calls.c (get_size_range): Use wi::to_wide when operating on
2871 * cgraph.c (cgraph_node::create_thunk): Likewise.
2872 * config/i386/i386.c (ix86_data_alignment): Likewise.
2873 (ix86_local_alignment): Likewise.
2874 * dbxout.c (stabstr_O): Likewise.
2875 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
2876 * expr.c (const_vector_from_tree): Likewise.
2877 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
2878 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
2879 (fold_negate_expr_1, int_const_binop_1, const_binop)
2880 (fold_convert_const_int_from_real, optimize_bit_field_compare)
2881 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
2882 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
2883 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
2884 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
2885 (fold_not_const, round_up_loc): Likewise.
2886 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
2887 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
2888 (alloca_call_type): Likewise.
2889 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
2890 * godump.c (go_output_typedef): Likewise.
2891 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
2892 * internal-fn.c (get_min_precision): Likewise.
2893 * ipa-cp.c (ipcp_store_vr_results): Likewise.
2894 * ipa-polymorphic-call.c
2895 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
2896 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
2897 (ipa_modify_call_arguments): Likewise.
2898 * match.pd: Likewise.
2899 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
2900 * print-tree.c (print_node_brief, print_node): Likewise.
2901 * stmt.c (expand_case): Likewise.
2902 * stor-layout.c (layout_type): Likewise.
2903 * tree-affine.c (tree_to_aff_combination): Likewise.
2904 * tree-cfg.c (group_case_labels_stmt): Likewise.
2905 * tree-data-ref.c (dr_analyze_indices): Likewise.
2906 (prune_runtime_alias_test_list): Likewise.
2907 * tree-dump.c (dequeue_and_dump): Likewise.
2908 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
2909 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
2910 * tree-pretty-print.c (dump_generic_node): Likewise.
2911 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
2912 (simple_iv_with_niters): Likewise.
2913 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2914 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
2915 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
2916 * tree-ssa-loop-niter.c (split_to_var_and_offset)
2917 (refine_value_range_using_guard, number_of_iterations_ne_max)
2918 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
2919 (get_cst_init_from_scev, record_nonwrapping_iv)
2920 (scev_var_range_cant_overflow): Likewise.
2921 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
2922 * tree-ssa-pre.c (compute_avail): Likewise.
2923 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
2924 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
2925 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
2926 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
2927 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
2928 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
2929 (dump_case_nodes, try_switch_expansion): Likewise.
2930 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
2931 (vect_do_peeling): Likewise.
2932 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
2933 * tree-vect-stmts.c (vectorizable_load): Likewise.
2934 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
2935 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
2936 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
2937 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
2938 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
2939 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
2940 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
2941 (evrp_dom_walker::before_dom_children): Likewise.
2942 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
2943 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
2944 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
2945 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
2946 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
2947 (get_range_pos_neg): Likewise.
2948 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
2949 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
2950 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
2951 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
2952 * config/avr/avr.c (avr_fold_builtin): Likewise.
2953 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
2954 * config/msp430/msp430.c (msp430_attr): Likewise.
2955 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
2956 * config/powerpcspe/powerpcspe-c.c
2957 (altivec_resolve_overloaded_builtin): Likewise.
2958 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
2959 (rs6000_expand_ternop_builtin): Likewise.
2960 * config/rs6000/rs6000-c.c
2961 (altivec_resolve_overloaded_builtin): Likewise.
2962 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
2963 (rs6000_expand_ternop_builtin): Likewise.
2964 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
2966 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2968 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
2969 when copying loop nest with only one inner loop.
2971 2017-10-10 Richard Biener <rguenther@suse.de>
2973 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
2974 blocks if SCEV is active.
2975 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
2977 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
2978 (scev_initialize): Assert we are not yet initialized.
2980 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2982 * tree-loop-distribution.c (generate_loops_for_partition): Remove
2983 inner loop's exit stmt by making it always exit the loop, otherwise
2984 we would generate an infinite empty loop.
2986 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2988 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
2989 renaming variables in new preheader if it's deleted.
2991 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2993 * tree-loop-distribution.c (struct partition): Remove unused field
2994 loops of the structure.
2995 (partition_alloc, partition_free): Ditto.
2996 (build_rdg_partition_for_vertex): Ditto.
2998 2017-10-09 Jeff Law <law@redhat.com>
3000 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
3001 return type to match prototype and documentation.
3003 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3005 * config/rs6000/rs6000.c (processor_costs): Move to ...
3006 * config/rs6000/rs6000.h: ... here.
3007 (rs6000_cost): Declare.
3009 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
3011 * except.c (setjmp_fn): New global variable.
3012 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
3013 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
3014 if DONT_USE_BUILTIN_SETJMP is defined.
3016 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3018 * target.def (insn_cost): New hook.
3019 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
3020 * doc/tm.texi: Regenerate.
3021 * rtlanal.c (insn_cost): Use the new hook.
3023 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3025 * combine.c (combine_validate_cost): Compute the new insn_cost,
3026 not just pattern_cost.
3027 (try_combine): Adjust comment.
3029 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3031 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
3033 * combine.c (uid_insn_cost): Adjust comment.
3034 (combine_validate_cost): Adjust comment. Use pattern_cost instead
3036 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
3037 * dse.c (find_shift_sequence): Ditto.
3038 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
3039 (bb_valid_for_noce_process_p): Use pattern_cost.
3040 * rtl.h (insn_rtx_cost): Delete.
3041 (pattern_cost): New prototype.
3042 (insn_cost): New prototype.
3043 * rtlanal.c (insn_rtx_cost): Rename to...
3044 (pattern_cost): ... this.
3047 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
3049 * config/i386/i386.md (*jcc_2): Remove insn pattern.
3050 (*jcc<mode>_0_r_i387): Ditto.
3051 (*jccxf_r_i387): Ditto.
3052 (*jcc<mode>_r_i387): Ditto.
3053 (*jccu<mode>_r_i387): Ditto.
3054 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
3055 (*jcc): Rename from *jcc_1.
3057 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3059 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
3060 deferred rescans after the lvx/stvx recombination pre-pass.
3062 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
3064 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
3065 memory operation instruction support.
3066 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
3067 (rs6000-ibm-aix[789]*): Likewise.
3068 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
3069 Document new functions.
3071 2017-10-09 Richard Biener <rguenther@suse.de>
3073 PR tree-optimization/82397
3074 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
3075 equality only for semantically equal trees.
3077 2017-10-09 Richard Biener <rguenther@suse.de>
3079 PR tree-optimization/82449
3080 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
3081 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
3082 allow constant addresses.
3083 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
3086 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3088 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
3091 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3094 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
3097 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3100 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
3102 2017-10-09 Jakub Jelinek <jakub@redhat.com>
3105 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
3106 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
3108 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
3110 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
3111 (WI_BINARY_PREDICATE_RESULT): Likewise.
3112 (wi::binary_traits::operator_result): New type.
3113 (wi::binary_traits::predicate_result): Likewise.
3114 (generic_wide_int::operator~, unary generic_wide_int::operator-)
3115 (generic_wide_int::operator==, generic_wide_int::operator!=)
3116 (generic_wide_int::operator&, generic_wide_int::and_not)
3117 (generic_wide_int::operator|, generic_wide_int::or_not)
3118 (generic_wide_int::operator^, generic_wide_int::operator+
3119 (binary generic_wide_int::operator-, generic_wide_int::operator*):
3121 (operator~, unary operator-, operator==, operator!=, operator&)
3122 (operator|, operator^, operator+, binary operator-, operator*): New
3124 * expr.c (get_inner_reference): Use wi::bit_and_not.
3125 * fold-const.c (fold_binary_loc): Likewise.
3126 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
3127 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
3128 (bit_value_binop): Likewise.
3129 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
3130 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
3131 (extract_range_from_binary_expr_1): Likewise.
3132 (masked_increment): Likewise.
3133 (simplify_bit_ops_using_ranges): Likewise.
3135 2017-10-09 Martin Jambor <mjambor@suse.cz>
3138 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
3139 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
3140 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
3142 (hsa_fixup_mov_insn_type): New function.
3143 (hsa_op_with_type::get_in_type): Use it.
3144 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
3145 immediates in an assert.
3146 (hsa_op_with_type::extend_int_to_32bit): New method.
3147 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
3148 types. Convert to dest type if necessary.
3149 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
3150 (reg_for_gimple_ssa): Pass false as min32int to
3151 hsa_type_for_scalar_tree_type.
3152 (gen_hsa_addr): Fixup type when creating addresable temporary.
3153 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
3154 (gen_hsa_unary_operation): Extend operands and convert to dest type if
3155 necessary. Call hsa_fixup_mov_insn_type.
3156 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
3157 extend operands and convert to dest type if necessary.
3158 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
3159 to dest type if necessary.
3160 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
3161 if conversion nt necessary and size matches.
3162 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
3163 to dest type if necessary.
3164 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
3165 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
3167 (gen_hsa_clrsb): Likewise.
3168 (gen_hsa_ffs): Likewise.
3169 (gen_hsa_divmod): Extend operands and convert to dest type if
3171 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
3173 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
3175 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
3176 Remove empty default arguments. Use a brace block as output
3178 (conditional return): Ditto.
3180 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
3181 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
3183 (group_ending_nop): Ditto.
3184 (doloop_end): Ditto.
3185 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
3186 (splitters for those): Ditto.
3188 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
3190 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
3191 a conditional jump (and the compare for it) so that pc_rtx is the
3193 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
3194 for the deleted and renamed ctr<mode>_internal[234] patterns.
3195 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
3196 Delete second conditional return pattern.
3197 (ctr<mode>_internal2): Delete this second bdnz pattern.
3198 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
3199 (ctr<mode>_internal4): Delete this second bdz pattern.
3201 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
3203 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
3204 (always_initialized_rtx_for_ssa_name_p): New predicate.
3205 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
3206 (finish_out_of_ssa): Free new field of SA.
3207 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
3208 * tree-ssa-coalesce.c: Include tree-ssa.h.
3209 (get_parm_default_def_partitions): Remove extern keyword.
3210 (get_undefined_value_partitions): New function.
3211 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
3212 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
3214 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
3216 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
3218 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
3220 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
3222 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
3223 for targets that preffer 128bit.
3225 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
3227 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
3229 2017-10-08 Olivier Hainque <hainque@adacore.com>
3231 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
3232 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
3234 (thumb_set_return_address): Likewise.
3236 2017-10-08 Olivier Hainque <hainque@adacore.com>
3238 * common/config/arm/arm-common.c (arm_except_unwind_info):
3239 Handle DWARF2_UNWIND_INFO.
3241 2017-10-07 Michael Collison <michael.collison@arm.com>
3243 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
3246 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
3248 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
3249 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
3250 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
3251 defined, force the creation of a new block for a dispatch label.
3253 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
3255 * invoke.texi (Wsuggest-attribute=cold): Document.
3256 * common.opt (Wsuggest-attribute=cold): New
3257 * ipa-pure-const.c (warn_function_cold): New function.
3258 * predict.c (compute_function_frequency): Use it.
3259 * predict.h (warn_function_cold): Declare.
3261 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
3263 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
3266 2017-10-06 Martin Liska <mliska@suse.cz>
3268 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
3269 keyword for member functions.
3270 (struct sanopt_tree_couple): New struct.
3271 (struct sanopt_tree_couple_hash): New function.
3272 (struct sanopt_ctx): Add new hash_map.
3273 (has_dominating_ubsan_ptr_check): New function.
3274 (record_ubsan_ptr_check_stmt): Likewise.
3275 (maybe_optimize_ubsan_ptr_ifn): Likewise.
3276 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
3277 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
3279 2017-10-06 Sudakshina Das <sudi.das@arm.com>
3282 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
3283 aarch64_simd_valid_immediate on CONST_VECTORs.
3284 (aarch64_reg_or_bic_imm): Likewise.
3286 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
3288 PR rtl-optimization/82396
3289 * haifa-sched.c (ready_sort_real): Disable qsort checking.
3291 2017-10-06 Sebastian Pop <sebpop@gmail.com>
3293 * graphite-dependences.c (scop_get_reads): Move code to...
3294 (scop_get_must_writes): Move code to...
3295 (scop_get_may_writes): Move code to...
3296 (scop_get_reads_and_writes): ... here.
3297 (scop_get_dependences): Call scop_get_reads_and_writes.
3299 2017-10-06 Jakub Jelinek <jakub@redhat.com>
3301 PR tree-optimization/82434
3302 * fold-const.h (can_native_encode_type_p,
3303 can_native_encode_string_p): Remove.
3304 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
3305 don't encode anything, just return what would be otherwise returned.
3306 (native_encode_fixed, native_encode_complex, native_encode_vector):
3308 (native_encode_string): Likewise. Inline by hand
3309 can_native_encode_string_p.
3310 (can_native_encode_type_p): Remove.
3311 (can_native_encode_string_p): Remove.
3312 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
3313 STRING_CSTs using can_native_encode_string_p, test all
3314 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
3315 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
3316 argument from native_encode_expr.
3317 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
3318 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
3321 2017-10-06 Richard Biener <rguenther@suse.de>
3323 PR tree-optimization/82397
3324 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
3325 operand_equal_p but rely on data_ref_compare_tree for detecting
3327 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
3328 to match up with dr_group_sort_cmp.
3330 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3333 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
3335 * config/s390/s390-builtin-types.def: Regenerate.
3337 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3340 * config/s390/s390-builtin-types.def: Regenerate.
3341 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
3342 Change flag from B_VXE to B_VX.
3343 (s390_vec_min_dbl): Remove B_VXE flag.
3345 2017-10-06 Richard Biener <rguenther@suse.de>
3347 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
3348 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
3349 translate_isl_ast_to_gimple::is_valid_rename,
3350 translate_isl_ast_to_gimple::get_rename,
3351 translate_isl_ast_to_gimple::get_def_bb_for_const,
3352 translate_isl_ast_to_gimple::get_new_name,
3353 translate_isl_ast_to_gimple::collect_all_ssa_names,
3354 translate_isl_ast_to_gimple::copy_loop_phi_args,
3355 translate_isl_ast_to_gimple::collect_all_ssa_names,
3356 translate_isl_ast_to_gimple::copy_loop_phi_args,
3357 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
3358 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
3359 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
3360 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
3361 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
3362 translate_isl_ast_to_gimple::copy_cond_phi_args,
3363 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
3364 translate_isl_ast_to_gimple::edge_for_new_close_phis,
3365 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
3366 translate_isl_ast_to_gimple::rename_uses,
3367 translate_isl_ast_to_gimple::rename_all_uses): Remove.
3368 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
3369 (set_rename_for_each_def): Likewise.
3370 (graphite_copy_stmts_from_block): Handle debug stmt resetting
3371 here. Handle rewriting SCEV analyzable uses here.
3372 (copy_bb_and_scalar_dependences): Generate code for PHI
3374 (graphite_regenerate_ast_isl): Adjust.
3375 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
3376 (add_write, add_read): New functions.
3377 (build_cross_bb_scalars_def): Use it and simplify.
3378 (build_cross_bb_scalars_use): Likewise.
3379 (graphite_find_cross_bb_scalar_vars): Inline into...
3380 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
3381 simulating out-of-SSA. Compute liveout and add dependencies.
3382 (build_scops): Force an empty entry block.
3383 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
3385 (sese_build_liveouts): Declare.
3386 (sese_trivially_empty_bb_p): Likewise.
3387 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
3388 compute liveout and debug_liveout.
3389 (sese_bad_liveouts_use): Remove.
3390 (sese_reset_debug_liveouts_bb): Likewise.
3391 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
3392 (sese_build_liveouts): Build liveout and debug_liveout and store
3394 (new_sese_info): Adjust.
3395 (free_sese_info): Likewise.
3396 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
3397 do not build liveout here.
3398 (move_sese_in_condition): Adjust region entry.
3399 (scev_analyzable_p): Match up with chrec_apply requirements.
3400 (sese_trivially_empty_bb_p): New.
3401 * tree-into-ssa.c (get_reaching_def): Properly support generating
3402 default-defs for incremental rewrite of anonymous names.
3404 2017-10-06 Richard Biener <rguenther@suse.de>
3406 * graphite-sese-to-poly.c (extract_affine): For casts increasing
3407 precision do not perform modulo reduction.
3409 2017-10-06 Richard Biener <rguenther@suse.de>
3411 PR tree-optimization/82436
3412 * tree-vect-slp.c (vect_supported_load_permutation_p): More
3413 conservatively choose the vectorization factor when checking
3414 whether we can perform the required load permutation.
3415 (vect_transform_slp_perm_load): Assert when we may not fail.
3417 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
3419 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
3420 message for incompatible -msdata=* and -mcall-* options.
3422 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3424 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
3425 rate for post-reload scheduling.
3427 2017-10-05 Tamar Christina <tamar.christina@arm.com>
3429 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
3431 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3433 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
3434 to improve monte carlo in scimark.
3436 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3438 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
3439 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
3440 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
3441 pentium4_cost, nocona_cost): Set reassociation width to 1.
3442 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
3443 width to 2 for fp operations and 1 otherwise.
3444 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
3446 (atom_cost): Set reassociation width to 2.
3447 (slm_cost, generic_cost): Set fp reassociation width
3448 to 2 and 1 otherwise.
3449 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
3450 (core_cost): Set fp reassociation width to 4 and vector to 2.
3451 (ix86_reassociation_width): Rewrite using cost table; special case
3452 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
3453 and TARGET_AVX128_OPTIMAL.
3454 * config/i386/i386.h (processor_costs): Add
3455 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
3456 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
3457 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
3458 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
3459 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
3460 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
3462 2017-10-05 Nathan Sidwell <nathan@acm.org>
3464 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
3466 2017-10-05 Tamar Christina <tamar.christina@arm.com>
3468 * config/arm/arm.c (arm_test_fpu_data): New.
3469 (arm_run_selftests): Call arm_test_fpu_data.
3471 2017-10-04 Nathan Sidwell <nathan@acm.org>
3473 * toplev.c (toplev::main): Remove excess parens on pretty_printer
3475 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
3477 2017-10-04 Sudakshina Das <sudi.das@arm.com>
3479 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
3480 check type for aarch64_simd_valid_immediate.
3481 (aarch64_output_simd_mov_immediate): Update prototype.
3482 (aarch64_simd_valid_immediate): Update prototype.
3483 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
3484 support for ORR-immediate.
3485 (and<mode>3): modified pattern to add support for BIC-immediate.
3486 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
3487 now checks for valid immediate for BIC and ORR based on new enum
3489 (aarch64_output_simd_mov_immediate): Function now used to output
3490 BIC/ORR imm as well based on new enum argument.
3491 * config/aarch64/constraints.md (Do): New vector immediate constraint.
3493 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
3494 (aarch64_reg_or_bic_imm): Likewise.
3496 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3498 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
3499 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
3501 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
3505 PR rtl-optimization/82396
3506 * haifa-sched.c (autopref_multipass_init): Simplify
3508 (autopref_rank_data): Simplify sort order.
3509 * sched-int.h (autopref_multipass_data_): Remove
3510 multi_mem_insn_p, min_offset and max_offset.
3512 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3514 * doc/sourcebuild.texi: Document vect_peeling_profitable.
3516 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3518 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
3521 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3523 * doc/sourcebuild.texi: Document vect_long_mult.
3525 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
3527 PR tree-optimization/82413
3528 * fold-const.c (build_range_check): Use widest_int when comparing
3529 the maximum ETYPE value with HIGH.
3531 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
3533 PR rtl-optimization/82396
3534 * haifa-sched.c (autopref_multipass_init): Simplify
3536 (autopref_rank_data): Simplify sort order.
3537 * sched-int.h (autopref_multipass_data_): Remove
3538 multi_mem_insn_p, min_offset and max_offset.
3540 2017-10-04 Jakub Jelinek <jakub@redhat.com>
3542 PR tree-optimization/82381
3543 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
3544 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
3547 PR tree-optimization/82374
3548 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
3549 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
3550 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
3551 current_function_decl to the new decl.
3553 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
3555 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
3556 helper macro for IEEE float128 hardware built-in functions.
3557 (SQRTF128_ODD): Add built-in functions with the round-to-odd
3559 (TRUNCF128_ODD): Likewise.
3560 (ADDF128_ODD): Likewise.
3561 (SUBF128_ODD): Likewise.
3562 (MULF128_ODD): Likewise.
3563 (DIVF128_ODD): Likewise.
3564 (FMAF128_ODD): Likewise.
3565 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
3566 UNSPEC_TRUNC_ROUND_TO_ODD.
3567 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
3568 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
3569 floating point round to odd instructions.
3570 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
3571 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
3572 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
3573 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
3574 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
3575 (trunc<mode>sf2_hw): Change the truncate with round to odd
3576 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
3577 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
3578 to odd hardware instructions.
3579 (sub<mode>3_odd): Likewise.
3580 (mul<mode>3_odd): Likewise.
3581 (div<mode>3_odd): Likewise.
3582 (sqrt<mode>2_odd): Likewise.
3583 (fma<mode>4_odd): Likewise.
3584 (fms<mode>4_odd): Likewise.
3585 (nfma<mode>4_odd): Likewise.
3586 (nfms<mode>4_odd): Likewise.
3587 (trunc<mode>df2_odd): Change the truncate with round to odd
3588 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
3590 * doc/extend.texi (PowerPC built-in functions): Update documentation
3591 for existing IEEE float128-bit built-in functions. Add built-in
3592 functions that generate the IEEE 128-bit floating point round to
3595 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
3597 PR rtl-optimization/77729
3598 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
3599 to (X&(C1&~C2))|C2 transformations.
3601 2017-10-03 Martin Jambor <mjambor@suse.cz>
3603 PR tree-optimization/82363
3604 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
3605 mismatch, mark lacc written regardless of racc.
3607 2017-10-03 Jakub Jelinek <jakub@redhat.com>
3609 PR tree-optimization/82381
3610 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
3611 stmt_to_insert nor wheather SSA_NAMEs are default defs.
3612 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
3613 fallthrough into reassoc_stmt_dominates_stmt_p.
3616 * combine.c (combine_instructions): Don't combine in unreachable
3619 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
3622 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
3623 function to not use the have_cpu variable. Do not set cpu_index,
3624 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
3626 (rs6000_valid_attribute_p): Remove duplicate initializations of
3627 old_optimize and func_optimize.
3628 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
3629 (rs6000_activate_target_options): Make global.
3630 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
3633 2017-10-02 Jakub Jelinek <jakub@redhat.com>
3635 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
3636 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
3637 Set *poffset to 0 and *psize and *pmax_size to -1 if
3638 *poffset + *psize overflows in HOST_WIDE_INT.
3640 PR tree-optimization/82387
3641 PR tree-optimization/82388
3642 PR tree-optimization/82389
3643 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
3644 instead of live_bytes non-NULL.
3646 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
3649 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
3652 2017-10-02 Richard Biener <rguenther@suse.de>
3654 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
3655 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
3656 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
3658 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
3660 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
3661 requested precision matches the type's.
3662 * calls.c (alloc_max_size): Calculate the new candidate size as
3663 a widest_int and use wi::to_widest when comparing it with the
3664 current candidate size.
3665 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
3666 zero rather than integer_zero_node.
3667 * match.pd: Check for a no-op conversion before using wi::add
3668 rather than after. Use tree_to_uhwi when summing small shift
3669 counts into an unsigned int.
3671 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
3672 Alan Hayward <alan.hayward@arm.com>
3673 David Sherwood <david.sherwood@arm.com>
3676 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
3677 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
3678 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
3679 POINTER_AND_FP_REGS.
3681 2017-10-02 Richard Biener <rguenther@suse.de>
3683 PR tree-optimization/82355
3684 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
3685 a mapping for the enclosing loop but avoid generating one for
3687 (copy_bb_and_scalar_dependences): Remove premature codegen
3688 error on PHIs in blocks duplicated into multiple places.
3689 * graphite-scop-detection.c
3690 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
3691 in the region use it as loop and nest to analyze the DR in.
3692 (try_generate_gimple_bb): Likewise.
3693 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
3694 (add_loop_constraints): For blocks in a loop not in the region
3695 create a dimension with a single iteration.
3696 * sese.h (gbb_loop_at_index): Remove assert.
3698 2017-10-01 Kevin Buettner <kevinb@redhat.com>
3700 * omp-expand.c (adjust_context_scope): New function.
3701 (expand_parallel_call): Call adjust_context_scope.
3703 2017-10-01 Jeff Law <law@redhat.com>
3705 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
3706 dom_opt_dom_walker class with direct access to private members.
3707 Add comments. Call test_for_singularity.
3708 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
3709 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
3710 m_dummy_cond anymore.
3711 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
3713 (pass_dominator:execute): Build the dummy_cond here and pass it
3714 to the dom_opt_dom_walker ctor.
3715 (test_for_singularity): New function.
3717 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
3718 Maya Rashish <coypu@sdf.org>
3720 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
3721 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
3722 (alpha*-*-netbsd*) Use nbsd_tm_file.
3723 (arm*-*-netbsdelf*) Likewise.
3724 (i[34567]86-*-netbsdelf*) Likewise.
3725 (x86_64-*-netbsd*) Likewise.
3726 (mips*-*-netbsd*) Likewise.
3727 (powerpc-*-netbsd*) Likewise.
3728 (sh*-*-netbsd*) Likewise.
3729 (sparc-*-netbsdelf*) Likewise.
3730 (sparc64-*-netbsd*) Likewise.
3731 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
3733 (vax-*-netbsdelf*) Likewise.
3734 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
3735 (UINT_FAST8_TYPE) Likewise.
3736 (INT_FAST16_TYPE) Check CHAR_FAST16.
3737 (UINT_FAST16_TYPE) Likewise.
3739 2017-09-30 Jakub Jelinek <jakub@redhat.com>
3742 * config/i386/i386.md
3743 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
3744 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
3745 *divmodsi4_zext_2): New define_insn_and_split.
3746 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
3747 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
3748 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
3749 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
3750 New define_insn_and_split.
3751 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
3752 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
3753 operands[1] having DImode when mode is SImode.
3755 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
3756 always SImode for DIV and MOD in REG_EQUAL notes.
3758 2017-09-29 Yury Gribov <tetra2005@gmail.com>
3761 * match.pd: Fix handling of NaNs in pattern.
3763 2017-09-29 Jeff Law <law@redhat.com>
3765 * sbitmap.c (bitmap_bit_in_range_p): New function.
3766 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
3767 * tree-ssa-dse.c (live_bytes_read): New function.
3768 (dse_classify_store): Ignore reads of dead bytes.
3770 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
3771 typos and whitespace errors.
3772 * config/i386/predicates.md (address_no_seg_operand): Likewise.
3773 * config/s390/s390.c (s390_emit_prologue): Likewise.
3775 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
3778 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
3779 with a symbol for LRA.
3781 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
3783 PR rtl-optimization/82338
3784 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
3786 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
3788 * genmodes.c (calc_wider_mode): Suppress qsort macro.
3789 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
3790 (qsort_chk): Declare.
3791 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
3792 (qsort_chk): New function.
3794 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3796 PR tree-optimization/82337
3797 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
3798 phi definition if the PHI result appears in an abnormal PHI.
3799 (find_basis_for_base_expr): Don't record a basis if the LHS of the
3800 basis appears in an abnormal PHI.
3802 2017-09-29 Richard Biener <rguenther@suse.de>
3804 * graphite-isl-ast-to-gimple.c
3805 (translate_isl_ast_to_gimple::set_codegen_error): New function.
3806 (binary_op_to_tree): Use it.
3807 (get_rename_from_scev): Likewise.
3808 (copy_loop_phi_nodes): Likewise.
3809 (copy_bb_and_scalar_dependences): Likewise.
3810 (translate_pending_phi_nodes): Likewise.
3812 2017-09-29 Jakub Jelinek <jakub@redhat.com>
3815 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
3816 for movabsq $(i32 << shift), r64.
3818 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
3820 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
3821 index when encoding %esp as %rsp to avoid 0x67 prefix.
3823 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
3825 * config/i386/i386.md (*movsf_internal, *movdf_internal):
3826 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
3828 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
3830 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
3831 Extensions with more than 16 double VFP registers.
3832 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
3833 to_clear_mask and all code related to it. Replace the remaining
3834 entry by a sbitmap and adapt code accordingly.
3836 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
3838 * brig-builtins.def: Change pure attributes to const.
3840 2017-09-28 Joseph Myers <joseph@codesourcery.com>
3842 * config.gcc (default_gnu_indirect_function): Default to yes for
3843 sparc*-*-linux* with glibc.
3845 2017-09-28 Joseph Myers <joseph@codesourcery.com>
3847 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
3848 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
3849 when creating .init_array and .fini_array sections with priority
3852 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
3855 * config/aarch64/aarch64.c
3856 (aarch64_builtin_support_vector_misalignment): Always return false
3857 when misalignment is unknown.
3859 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
3861 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
3862 this function to return false if the definition used by the swap
3863 instruction is artificial, or if the memory address from which the
3864 constant value is loaded is not represented by a base address held
3865 in a register or if the base address register is a frame or stack
3866 pointer. Additionally, return false if the base address of the
3867 loaded constant is a SYMBOL_REF but is not considered to be a
3869 (replace_swapped_load_constant): New function.
3870 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
3871 loaded constant vector with a load of a swapped constant vector.
3873 2017-09-27 Carl Love <cel@us.ibm.com>
3875 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
3876 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
3877 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
3880 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
3882 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
3883 first, always call autopref_rank_data otherwise.
3885 2017-09-27 Richard Biener <rguenther@suse.de>
3887 * graphite-scop-detection.c (find_scop_parameters): Move
3888 loop bound handling ...
3889 (gather_bbs::before_dom_children): ... here, avoiding the need
3890 to build scop_info->loop_nest.
3891 (record_loop_in_sese): Remove.
3892 * sese.h (sese_info_t::loop_nest): Remove.
3893 * sese.c (new_sese_info): Do not allocate loop_nest.
3894 (free_sese_info): Do not free loop_nest.
3896 2017-09-27 Jakub Jelinek <jakub@redhat.com>
3899 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
3900 lhs from calls if the lhs has addressable type.
3902 2017-09-27 Richard Biener <rguenther@suse.de>
3904 * graphite.h (scop::max_alias_set): New member.
3905 * graphite-scop-detection.c: Remove references to non-existing
3906 --param in comments.
3907 (build_alias_sets): Record the maximum alias set used for drs.
3908 (build_scops): Support zero as unlimited for
3909 --param graphite-max-arrays-per-scop.
3910 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
3912 (build_poly_sr_1): ... here. Compute alias set based on the
3913 maximum alias set used for drs rather than
3914 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
3916 2017-09-27 Richard Biener <rguenther@suse.de>
3918 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
3919 --param loop-block-tile-size=0 to disable tiling.
3921 2017-09-27 Richard Biener <rguenther@suse.de>
3923 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
3924 (graphite-max-nb-scop-params): Document special value zero.
3925 * domwalk.h (dom_walker::STOP): New symbolical constant.
3926 (dom_walker::dom_walker): Add optional parameter for bb to
3928 (dom_walker::~dom_walker): Declare.
3929 (dom_walker::before_dom_children): Document STOP return value.
3930 (dom_walker::m_user_bb_to_rpo): New member.
3931 (dom_walker::m_bb_to_rpo): Likewise.
3932 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
3933 mapping here if not provided by the user.
3934 (dom_walker::~dom_walker): Free bb to RPO mapping if not
3935 provided by the user.
3936 (dom_walker::STOP): Define.
3937 (dom_walker::walk): Do not compute bb to RPO mapping here.
3938 Support STOP return value from before_dom_children to stop
3940 * graphite-optimize-isl.c (optimize_isl): If the schedule
3941 is the same still generate code if -fgraphite-identity
3942 or -floop-parallelize-all are given.
3943 * graphite-scop-detection.c: Include cfganal.h.
3944 (gather_bbs::gather_bbs): Get and pass through bb to RPO
3946 (gather_bbs::before_dom_children): Return STOP for BBs
3948 (build_scops): Compute bb to RPO mapping and pass it to
3949 the domwalk. Treat --param graphite-max-nb-scop-params=0
3950 as not limiting the number of params.
3951 * graphite.c (graphite_initialize): Remove limit on the
3952 number of basic-blocks in a function.
3953 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
3954 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
3955 default value of 10.
3957 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3959 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
3960 Adjust code to eliminate needing to do the shift right 32-bits
3961 operation after XSCVDPSPN.
3963 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3965 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
3966 ((X / Y) != 0 -> X >= Y): Likewise.
3968 2017-09-26 Carl Love <cel@us.ibm.com>
3970 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
3971 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
3972 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
3973 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
3974 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
3975 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
3976 definitions and overloading.
3977 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
3978 statement for P9V_BUILTIN_XST_LEN_R.
3979 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
3980 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
3981 define_expand and define_insn for the instructions and builtins.
3982 * doc/extend.texi: Update the built-in documentation file for the new
3984 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
3985 define_insn for the instructions
3987 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
3990 * gcc/config/netbsd-protos.h: New file.
3991 * gcc/config/netbsd.c: New file.
3992 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
3993 * gcc/config/t-netbsd: New file.
3994 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
3995 (tmake_file) Add t-netbsd.
3996 (extra_objs) Add netbsd.o.
3998 2017-09-26 Janus Weil <janus@gcc.gnu.org>
4002 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
4004 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4006 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
4007 sign extension from a vector register to a GPR by doing a 32-bit
4008 direct move and then an EXTSW.
4009 (extendsi<mode>2 splitter): Likewise.
4010 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
4011 right or vector extract after doing XSCVDPSPN. Use
4012 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
4014 (movdi_from_sf_zero_ext): Likewise.
4015 (reload_gpr_from_vsxsf): Likewise.
4016 (p8_mfvsrd_4_disf): Delete, no longer used.
4017 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
4018 then needing to move the SFmode to a GPR to use the XSCVDPSP
4019 instruction instead of FRSP and XSCVDPSPN.
4020 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
4021 it is adjacent to the other XSCVSPDP insns.
4022 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
4023 SFmode to be in traditional Altivec registers.
4024 (vsx_xscvdpspn): Eliminate useless alternative constraint.
4025 (vsx_xscvspdpn): Likewise.
4026 (vsx_xscvspdpn_scalar): Likewise.
4028 2017-09-26 Martin Jambor <mjambor@suse.cz>
4030 * tree-sra.c (compare_access_positions): Put integral types first,
4031 stabilize sorting of integral types, remove conditions putting
4032 non-full-precision integers last.
4033 (sort_and_splice_var_accesses): Disable scalarization if a
4034 non-integert would be represented by a non-full-precision integer.
4036 2017-09-26 Joseph Myers <joseph@codesourcery.com>
4038 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
4039 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
4040 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
4041 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
4042 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
4043 conditionals inside the function instead of around it. Call
4044 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
4045 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
4047 2017-09-26 Richard Biener <rguenther@suse.de>
4049 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
4051 (scop_detection::build_scop_breadth): ... this. Removed.
4052 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
4053 (scop_detection::harmful_stmt_in_bb): Likewise.
4054 (scop_detection::graphite_can_represent_stmt): Likewise.
4055 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
4056 (scop_detection::can_represent_loop): Remove recursion, fold in ...
4057 (scop_detection::can_represent_loop_1): ... this. Removed.
4058 (scop_detection::harmful_loop_in_region): Simplify after inlining
4059 the above and remove more quadraticness.
4060 (build_scops): Adjust.
4061 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
4064 2017-09-26 Jakub Jelinek <jakub@redhat.com>
4067 * config/i386/i386.c (ix86_print_operand_address_as): Only test
4068 REGNO (base) == SP_REG if base is a REG.
4071 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
4072 if it is different SSA_NAME.
4073 (optimize_range_tests_cmp_bitwise): New function.
4074 (optimize_range_tests): Call it.
4076 2017-09-26 Richard Biener <rguenther@suse.de>
4078 PR tree-optimization/82321
4079 * graphite.c (canonicalize_loop_closed_ssa): Properly check
4080 for the def being inside the loop.
4082 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4084 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
4086 * config/s390/s390-builtins.def: Fix constraint on op4.
4088 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4090 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
4091 independent expanders.
4092 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
4093 ("vec_ordered", "vec_unordered"): New expanders.
4095 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4097 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
4100 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4102 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
4103 vec_unpacks_lo_v16qi.
4104 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
4106 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4108 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
4109 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
4110 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
4112 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4114 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
4116 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
4118 ("*vec_slb<mode>"): New insn pattern.
4119 ("vec_shr_<mode>"): New expander.
4120 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
4121 and force the shift count operand to V16QImode.
4122 ("vec_srb<mode>"): Set shift count mode to V16QI.
4124 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4126 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
4127 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
4128 ("vec_widen_smult_hi_<mode>"): New expander definitions.
4130 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
4133 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
4135 2017-09-26 Richard Biener <rguenther@suse.de>
4137 PR tree-optimization/82320
4138 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
4141 2017-09-25 Jeff Law <law@redhat.com>
4143 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
4144 prototype for new argument.
4145 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
4146 mostly extracted from rs6000_emit_allocate_stack.
4147 (rs6000_emit_probe_stack_range_stack_clash): New function.
4148 (rs6000_emit_allocate_stack): Call
4149 rs6000_emit_probe_stack_range_stack_clash as needed.
4150 (rs6000_emit_probe_stack_range): Add additional argument
4151 to call to gen_probe_stack_range{si,di}.
4152 (output_probe_stack_range): New.
4153 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
4154 (output_probe_stack_range_stack_clash): New.
4155 (rs6000_emit_prologue): Emit notes into dump file as requested.
4156 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
4157 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
4158 Add additional operand and pass it to output_probe_stack_range.
4160 2017-09-25 Bin Cheng <bin.cheng@arm.com>
4162 PR tree-optimization/82163
4163 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
4164 (checking_verify_loop_closed_ssa): New parameter.
4165 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
4166 (check_loop_closed_ssa_stmt): Delete.
4167 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
4168 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
4169 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
4172 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
4174 * brig-builtins.def: Treat HSAIL barrier builtins as
4175 setjmp/longjump style functions.
4177 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
4179 * target.def (constant_alignment): New hook.
4180 * defaults.h (CONSTANT_ALIGNMENT): Delete.
4181 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
4182 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
4183 * doc/tm.texi: Regenerate.
4184 * targhooks.h (default_constant_alignment): Declare.
4185 (constant_alignment_word_strings): Likewise.
4186 * targhooks.c (default_constant_alignment): New function.
4187 (constant_alignment_word_strings): Likewise.
4188 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
4189 instead of CONSTANT_ALIGNMENT.
4190 * varasm.c (align_variable, get_variable_align, build_constant_desc)
4191 (force_const_mem): Likewise.
4192 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
4193 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
4194 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
4195 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4196 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
4198 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
4199 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4200 constant_alignment_word_strings.
4201 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
4202 (CONSTANT_ALIGNMENT): Likewise.
4203 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4204 (arm_constant_alignment): New function.
4205 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
4206 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4207 constant_alignment_word_strings.
4208 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
4209 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4210 constant_alignment_word_strings.
4211 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
4212 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4213 (cris_constant_alignment): New function.
4214 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
4215 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4216 (epiphany_constant_alignment): New function.
4217 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
4218 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4219 constant_alignment_word_strings.
4220 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
4221 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4222 constant_alignment_word_strings.
4223 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
4224 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4225 constant_alignment_word_strings.
4226 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
4227 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
4228 * config/i386/i386.c (ix86_constant_alignment): Make static.
4229 Use the same interface as the target hook.
4230 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4231 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
4232 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4233 constant_alignment_word_strings.
4234 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
4235 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
4236 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4237 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
4238 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4239 constant_alignment_word_strings.
4240 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
4241 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4242 constant_alignment_word_strings.
4243 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
4244 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4245 constant_alignment_word_strings.
4246 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
4247 * config/microblaze/microblaze.c (microblaze_constant_alignment):
4249 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4250 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
4251 * config/mips/mips.c (mips_constant_alignment): New function.
4252 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4253 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
4254 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
4255 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4256 (mmix_constant_alignment): Make static. Use the same interface
4258 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
4259 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4260 constant_alignment_word_strings.
4261 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
4262 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4263 constant_alignment_word_strings.
4264 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
4265 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4266 constant_alignment_word_strings.
4267 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
4268 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4269 (rs6000_constant_alignment): New function.
4270 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
4271 * config/riscv/riscv.c (riscv_constant_alignment): New function.
4272 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4273 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
4274 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4275 (rs6000_constant_alignment): New function.
4276 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
4277 * config/s390/s390.c (s390_constant_alignment): New function.
4278 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4279 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
4280 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4281 constant_alignment_word_strings.
4282 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
4283 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4284 (sparc_constant_alignment): New function.
4285 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
4286 * config/spu/spu.c (spu_constant_alignment): New function.
4287 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4288 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
4289 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4290 constant_alignment_word_strings.
4291 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
4292 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4293 constant_alignment_word_strings.
4294 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
4295 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4296 constant_alignment_word_strings.
4297 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
4298 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4299 (visium_constant_alignment): New function.
4300 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
4301 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4302 (xtensa_constant_alignment): New function.
4303 * system.h (CONSTANT_ALIGNMENT): Poison.
4305 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
4307 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
4308 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
4309 (rs6000_builtin_valid_without_lhs): New helper function.
4310 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4311 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
4313 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
4315 * target.h (vec_perm_indices): Use unsigned short rather than
4317 (auto_vec_perm_indices): Likewise.
4318 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
4319 Use unsigned int rather than unsigned char.
4320 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
4322 2017-09-25 Richard Biener <rguenther@suse.de>
4324 * cfgloop.h (sort_sibling_loops): Declare.
4325 * cfgloop.c (sort_sibling_loops_cmp): New helper.
4326 (sort_sibling_loops): New function sorting the sibling loop list
4328 * graphite.c (graphite_transform_loops): Sort sibling loops.
4330 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
4332 * target.def (vec_perm_const_ok): Change sel parameter to
4334 * optabs-query.c (can_vec_perm_p): Update accordingly.
4335 * doc/tm.texi: Regenerate.
4336 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
4337 auto_vec_perm_indices and remove separate nelt field.
4338 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
4339 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
4340 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
4341 (aarch64_expand_vec_perm_const): Update accordingly.
4342 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
4343 to vec_perm_indices.
4344 * config/arm/arm.c (expand_vec_perm_d): Change perm to
4345 auto_vec_perm_indices and remove separate nelt field.
4346 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
4347 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
4348 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
4350 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
4351 to vec_perm_indices.
4352 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
4353 sel to vec_perm_indices.
4354 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
4355 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
4356 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
4358 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
4361 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
4364 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
4365 on the FUNCTION_DECL function context if it has a DIE that is a
4368 2017-09-25 Richard Biener <rguenther@suse.de>
4370 PR tree-optimization/82285
4371 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
4374 2017-09-25 Tom de Vries <tom@codesourcery.com>
4378 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
4381 2017-09-25 Richard Biener <rguenther@suse.de>
4383 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
4384 ISL errors other than isl_error_quota happen. Dump if the
4385 schedule is the same.
4386 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
4387 errors instead of aborting inside ISL.
4389 2017-09-25 Iain Sandoe <iain@codesourcery.com>
4392 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
4393 of libgcc_eh for m64.
4394 * config/i386/darwin64.h: Likewise.
4396 2017-09-25 Richard Biener <rguenther@suse.de>
4399 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
4400 attribute for incomplete types nor twice for complete ones.
4402 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
4405 * config/i386/i386.c (ix86_print_operand_address_as): Encode
4406 %esp as %rsp to avoid 0x67 prefix if there is no index or base
4409 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
4412 * config/i386/i386.opt (mprefer-avx256): Use
4413 ix86_target_flags variable.
4414 * config/i386/i386.c (ix86_target_string): Move
4415 -mprefer-avx256 to flag2_opts.
4417 2017-09-22 Jakub Jelinek <jakub@redhat.com>
4420 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
4421 and x != -1 | y != -1 into (x & y) != -1.
4423 2017-09-22 Steve Ellcey <sellcey@cavium.com>
4425 * config.gcc: Add new case statement to set
4426 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
4427 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
4428 s390x-*-linux* case statements. Added aarch64 to the list of
4429 supported architectures.
4431 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4433 PR tree-optimization/82289
4434 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
4435 STMT_VINFO_RELEVANT_P.
4437 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4438 Alan Hayward <alan.hayward@arm.com>
4439 David Sherwood <david.sherwood@arm.com>
4441 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
4442 for VR_RANGE only; don't allow VR_ANTI_RANGE.
4443 (extract_range_from_binary_expr_1): Don't call
4444 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
4446 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4447 Alan Hayward <alan.hayward@arm.com>
4448 David Sherwood <david.sherwood@arm.com>
4450 * target.def (preferred_vector_alignment): New hook.
4451 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
4453 * doc/tm.texi: Regenerate.
4454 * targhooks.h (default_preferred_vector_alignment): Declare.
4455 * targhooks.c (default_preferred_vector_alignment): New function.
4456 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
4458 (DR_TARGET_ALIGNMENT): New macro.
4459 (aligned_access_p): Update commentary.
4460 (vect_known_alignment_in_bytes): New function.
4461 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
4463 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
4464 Calculate the misalignment based on the target alignment rather than
4466 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
4467 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
4468 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
4469 the target alignment, rather than masking the element misalignment
4470 with the number of elements in a vector. Also use the target
4471 alignment when calculating the maximum number of peels.
4472 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
4473 instead of TYPE_ALIGN_UNIT.
4474 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
4475 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
4476 (vect_create_addr_base_for_vector_ref): Update call accordingly.
4477 (vect_create_data_ref_ptr): Likewise.
4478 (vect_setup_realignment): Realign by ANDing with
4479 -DR_TARGET_MISALIGNMENT.
4480 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
4481 the number of peels based on DR_TARGET_ALIGNMENT.
4482 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
4483 with the guaranteed alignment boundary when deciding whether
4485 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
4486 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
4487 (ensure_base_align): Remove stmt_info parameter. Get the
4488 target base alignment from DR_TARGET_ALIGNMENT.
4489 (vectorizable_store): Update call accordingly. Interpret
4490 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
4492 (vectorizable_load): Likewise.
4494 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4495 Alan Hayward <alan.hayward@arm.com>
4496 David Sherwood <david.sherwood@arm.com>
4498 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
4499 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
4500 (vect_enhance_data_refs_alignment): Likewise.
4502 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
4504 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
4505 error. Only quit immediately if parsing is complete.
4506 (BEGIN): Initialize fatal_err and parse_done.
4507 (begin fpu, end fpu): Check number of arguments.
4508 (begin arch, end arch): Likewise.
4509 (begin cpu, end cpu): Likewise.
4510 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
4511 (optalias): Likewise.
4513 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
4515 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
4516 * config/arm/arm-isa.h: Delete. Move definitions to ...
4517 * arm-cpus.in: ... here. Use new feature and fgroup values.
4518 * config/arm/arm.c (arm_option_override): Use lower case for feature
4520 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
4521 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
4522 * config/arm/parsecpu.awk (END): Add new command 'isa'.
4524 (print_isa_bits_for): New function.
4525 (gen_isa): New function.
4526 (gen_comm_data): Use print_isa_bits_for.
4527 (define feature): New keyword.
4528 (define fgroup): New keyword.
4529 * config/arm/t-arm (TM_H): Remove.
4530 (GTM_H): Add arm-isa.h.
4531 (arm-isa.h): Add rule to generate file.
4532 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
4533 case for feature bit names.
4535 2017-09-22 Richard Biener <rguenther@suse.de>
4537 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
4539 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
4540 print of no dependency loops ...
4541 * graphite.c (graphite_transform_loops): ... here.
4542 (canonicalize_loop_closed_ssa_form): Work from inner to outer
4544 (same_close_phi_node, remove_duplicate_close_phi,
4545 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
4546 (canonicalize_loop_closed_ssa): ... here and simplify.
4547 * graphite-optimize-isl.c: Include tree-vectorizer.h.
4548 (optimize_isl): Use dump_printf_loc to tell when we stopped
4549 optimizing because of an ISL timeout.
4551 2017-09-22 Richard Biener <rguenther@suse.de>
4553 PR tree-optimization/82291
4554 * tree-if-conv.c (predicate_mem_writes): Make sure to
4555 remove writes in blocks predicated with false.
4557 2017-09-22 Richard Biener <rguenther@suse.de>
4559 * sese.c: Include cfganal.h.
4560 (if_region_set_false_region): Remove.
4561 (create_if_region_on_edge): Likewise.
4562 (move_sese_in_condition): Re-implement without destroying
4565 2017-09-22 Richard Biener <rguenther@suse.de>
4567 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
4568 Verify both BBs contain loop PHI nodes before dispatching to
4570 (graphite_regenerate_ast_isl): Do not recompute dominators,
4571 do not verify three times. Restructure for clarity.
4572 * graphite-scop-detection.c (same_close_phi_node,
4573 remove_duplicate_close_phi, make_close_phi_nodes_unique,
4574 defined_in_loop_p, canonicalize_loop_closed_ssa,
4575 canonicalize_loop_closed_ssa_form): Simplify, remove excess
4576 checking and SSA rewrite, move to ...
4577 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
4578 (graphite_initialize): Do not pass in ctx, do not reset the
4579 SCEV cache, compute only dominators.
4580 (graphite_transform_loops): Allocate ISL ctx after
4581 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
4582 Maintain post-dominators only around build_scops.
4583 * sese.c (if_region_set_false_region): Make static. Free
4584 and recompute dominators.
4585 (move_sese_in_condition): Assert we don't get called with
4586 post-dominators computed.
4587 * sese.h (if_region_set_false_region): Remove.
4589 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
4591 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
4592 mode attribute for TARGET_AVX512VL.
4594 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
4596 * config/i386/i386.opt (mprefer-avx256): New option.
4597 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
4599 (ix86_preferred_simd_mode): Return 256-bit AVX modes
4600 for TARGET_PREFER_AVX256.
4601 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
4603 2017-09-21 Jeff Law <law@redhat.com>
4605 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
4606 Fix dump output if the only stack space is for pushed registers.
4608 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4610 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
4613 2017-09-21 Martin Sebor <msebor@redhat.com>
4616 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
4617 code (in C++) or code that triggers warnings.
4619 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
4621 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
4623 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
4625 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
4626 * sched-rgn.c: Ditto.
4627 * sel-sched-ir.c: Ditto.
4629 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
4631 * toplev.h (set_random_seed): Adjust return type.
4632 * toplev.c (init_local_tick): Move eager initialization of random_seed
4633 to get_random_seed. Adjust comment.
4634 (init_random_seed): Inline to get_random_seed, delete.
4635 (get_random_seed): Initialize random_seed lazily.
4636 (set_random_seed): Do not return previous value.
4637 (print_switch_value): Do not call get_random_seed.
4639 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
4641 * cgraph.c (delete_function_version): New, broken out from...
4642 (cgraph_node::delete_function_version): ...here. Rename to
4643 cgraph_node::delete_function_version_by_decl. Update all uses.
4644 (cgraph_node::remove): Call delete_function_version.
4646 2017-09-21 Jakub Jelinek <jakub@redhat.com>
4649 * tree-inline.c (expand_call_inline): Emit clobber stmts for
4650 VAR_DECLs to which addressable non-volatile parameters are mapped
4651 and for id->retvar after the return value assignment. Clear
4652 id->retval and id->retbnd after inlining.
4654 2017-09-21 Richard Biener <rguenther@suse.de>
4656 PR tree-optimization/82276
4657 PR tree-optimization/82244
4658 * tree-vrp.c (build_assert_expr_for): Set
4659 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
4661 (remove_range_assertions): Revert earlier change.
4663 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
4666 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
4668 2017-09-21 Richard Biener <rguenther@suse.de>
4670 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4671 Restore valid IL after code generation errors.
4672 * graphite.c (graphite_transform_loops): Diagnose code
4673 generation issues as MSG_MISSED_OPTIMIZATION and continue
4674 with processing SCOPs.
4676 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4677 Alan Hayward <alan.hayward@arm.com>
4678 David Sherwood <david.sherwood@arm.com>
4680 * calls.c (compute_argument_addresses): Use simplify_gen_binary
4681 rather than choosing between plus_constant and gen_rtx_<CODE>.
4682 * expr.c (emit_push_insn): Likewise.
4683 (expand_expr_real_2): Likewise.
4685 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4686 Alan Hayward <alan.hayward@arm.com>
4687 David Sherwood <david.sherwood@arm.com>
4689 * loop-unroll.c (split_iv): Call copy_rtx on the step.
4691 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4692 Alan Hayward <alan.hayward@arm.com>
4693 David Sherwood <david.sherwood@arm.com>
4695 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
4696 calling tree_to_uhwi.
4698 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4699 Alan Hayward <alan.hayward@arm.com>
4700 David Sherwood <david.sherwood@arm.com>
4702 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
4703 INTEGER_CST rather than a negative test for ADDR_EXPR.
4705 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4706 Alan Hayward <alan.hayward@arm.com>
4707 David Sherwood <david.sherwood@arm.com>
4709 * tree-vrp.c (extract_range_from_binary_expr_1): Check
4710 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
4712 2017-09-21 Richard Biener <rguenther@suse.de>
4714 PR tree-optimization/71351
4715 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
4716 graphite_create_new_loop_guard): Remove, fold remaining parts
4718 (translate_isl_ast_node_for): ... here and simplify.
4720 2017-09-21 Jakub Jelinek <jakub@redhat.com>
4723 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
4724 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
4725 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
4726 alternative always use QI mode, for -Os imov (=R,R) alternative
4727 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
4730 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4731 Jeff Law <law@redhat.com>
4733 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
4734 (allocate_stack_space): New function, partially extracted from
4736 (s390_emit_prologue): Track offset to most recent stack probe.
4737 Code to allocate space moved into allocate_stack_space.
4738 Dump actions when no stack is allocated.
4739 (s390_prologue_plus_offset): New function.
4740 (s390_emit_stack_probe): Likewise.
4742 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
4744 * common.opt (Wa, Wl, Wp, g, gz=): Add
4746 (gno-column-info): Remove.
4747 (gcolumn-info): Drop RejectNegative.
4749 (gno-record-gcc-switches): Remove.
4750 (grecord-gcc-switches): Drop RejectNegative.
4751 (gno-split-dwarf): Remove.
4752 (gsplit-dwarf): Drop RejectNegative.
4753 (gno-strict-dwarf): Remove.
4754 (gstrict-dwarf): Drop RejectNegative.
4755 * config/darwin.opt (gfull, gused): Add RejectNegative.
4756 * dwarf2out.c (gen_producer_string): Drop
4757 gno-record-gcc-switches handler.
4758 * optc-gen.awk: Add g to prefixes with negative forms.
4759 * opts-common.c (remapping_prefix_p): New.
4760 (find_opt): Check it.
4761 (generate_canonical_option): Test g prefix.
4762 (option_map): Add -gno- mapping.
4763 (add_misspelling_candidates): Check remapping_prefix_p.
4765 2017-09-20 Jeff Law <law@redhat.com>
4767 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
4768 thinko in stack clash protection support.
4770 * explow.c (compute_stack_clash_protection_loop_data): Use
4771 CONST_INT_P instead of explicit test. Verify object is a
4772 CONST_INT_P before looking at INTVAL.
4773 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
4774 instead of explicit test.
4776 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
4779 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
4780 address instead of to r1 and r11.
4782 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
4784 * config.gcc: Support "knm".
4785 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
4786 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4788 * config/i386/i386.c (m_KNM): Define.
4789 (processor_target_table): Add "knm".
4791 (ix86_option_override_internal): Add "knm".
4792 (ix86_issue_rate): Add PROCESSOR_KNM.
4793 (ix86_adjust_cost): Ditto.
4794 (ia32_multipass_dfa_lookahead): Ditto.
4795 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
4796 (fold_builtin_cpu): Add M_INTEL_KNM.
4797 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
4798 (processor_type): Add PROCESSOR_KNM.
4799 * config/i386/x86-tune.def: Add m_KNM.
4800 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
4802 2017-09-20 Richard Biener <rguenther@suse.de>
4804 PR tree-optimization/80213
4805 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
4806 are allowed in empty BBs as well.
4807 (canonicalize_loop_closed_ssa): Also look for other complex
4809 (scop_detection::get_sese): Include the loop-closed PHI block
4811 (scop_detection::merge_sese): Remove code adding extra blocks.
4812 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
4813 (build_scops): Assert the final returned scop is invalid.
4815 2017-09-20 Richard Biener <rguenther@suse.de>
4817 PR tree-optimization/82264
4818 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
4820 (vn_phi_lookup): Likewise.
4821 (vn_phi_insert): Likewise.
4823 2017-09-20 Jakub Jelinek <jakub@redhat.com>
4825 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
4826 that fits into uhwi or shwi, add DW_AT_const_value regardless
4827 of early_dwarf without going through RTL, using add_AT_unsigned
4830 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
4831 (DEBUG_LTO_ABBREV_SECTION): Likewise.
4832 (DEBUG_LTO_MACINFO_SECTION): Likewise.
4833 (DEBUG_MACRO_SECTION): Likewise.
4834 (DEBUG_LTO_MACRO_SECTION): Likewise.
4835 (DEBUG_STR_DWO_SECTION): Likewise.
4836 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
4837 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
4838 (DEBUG_LTO_DWO_LINE_SECTION): Define.
4839 (DEBUG_LTO_LINE_STR_SECTION): Define.
4840 (init_sections_and_labels): Initialize debug_line_str_section
4841 variable. Initialize debug_loc_section for -gdwarf-5 to
4842 DEBUG_LOCLISTS_SECTION. Formatting fixes.
4844 2017-09-20 Richard Biener <rguenther@suse.de>
4846 * graphite-sese-to-poly.c (extract_affine): Properly handle
4847 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
4849 2017-09-20 Richard Biener <rguenther@suse.de>
4851 PR tree-optimization/81373
4852 * graphite-scop-detection.c (build_cross_bb_scalars_def):
4853 Force SESE live-out defs to be handled even if they are
4856 2017-09-19 Jeff Law <law@redhat.com>
4858 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
4859 nothing for stack adjustments with REG_STACK_CHECK.
4860 * sched-deps.c (parse_add_or_inc): Reject insns with
4861 REG_STACK_CHECK from dependency breaking.
4862 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
4863 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
4864 * reg-notes.def (STACK_CHECK): New note.
4866 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
4867 (ix86_expand_prologue): Dump stack clash info as needed.
4868 Call ix86_adjust_stack_and_probe_stack_clash as needed.
4870 * function.c (dump_stack_clash_frame_info): New function.
4871 * function.h (dump_stack_clash_frame_info): Prototype.
4872 (enum stack_clash_probes): New enum.
4874 * config/alpha/alpha.c (alpha_expand_prologue): Also check
4875 flag_stack_clash_protection.
4876 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
4877 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
4878 (arm_frame_pointer_required): Likewise.
4879 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
4880 (ia64_expand_prologue): Likewise.
4881 * config/mips/mips.c (mips_expand_prologue): Likewise.
4882 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
4883 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
4884 (sparc_flat_expand_prologue): Likewise.
4885 * config/spu/spu.c (spu_expand_prologue): Likewise.
4887 * explow.c: Include "params.h".
4888 (anti_adjust_stack_and_probe_stack_clash): New function.
4889 (get_stack_check_protect): Likewise.
4890 (compute_stack_clash_protection_loop_data): Likewise.
4891 (emit_stack_clash_protection_loop_start): Likewise.
4892 (emit_stack_clash_protection_loop_end): Likewise.
4893 (allocate_dynamic_stack_space): Use get_stack_check_protect.
4894 Use anti_adjust_stack_and_probe_stack_clash.
4895 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
4896 (emit_stack_clash_protection_loop_start): Likewise.
4897 (emit_stack_clash_protection_loop_end): Likewise.
4898 * rtl.h (get_stack_check_protect): Prototype.
4899 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
4900 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
4901 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
4903 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
4905 * doc/tm.texi: Rebuilt.
4906 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
4907 get_stack_check_protect.
4908 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
4909 * config/arm/arm.c (arm_expand_prologue): Likewise.
4910 (arm_frame_pointer_required): Likewise.
4911 * config/i386/i386.c (ix86_expand_prologue): Likewise.
4912 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
4913 * config/mips/mips.c (mips_expand_prologue): Likewise.
4914 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
4915 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
4916 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
4917 (sparc_flat_expand_prologue): Likewise.
4919 * common.opt (-fstack-clash-protection): New option.
4920 * flag-types.h (enum stack_check_type): Note difference between
4921 -fstack-check= and -fstack-clash-protection.
4922 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
4923 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
4924 * toplev.c (process_options): Issue warnings/errors for cases
4925 not handled with -fstack-clash-protection.
4926 * doc/invoke.texi (-fstack-clash-protection): Document new option.
4927 (-fstack-check): Note additional problem with -fstack-check=generic.
4928 Note that -fstack-check is primarily for Ada and refer users
4929 to -fstack-clash-protection for stack-clash-protection.
4930 Document new params for stack clash protection.
4932 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
4934 * config/i386/i386.c (ix86_split_long_move): Do not handle
4935 address used for LEA in a special way.
4937 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
4939 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
4941 2017-09-19 Martin Sebor <msebor@redhat.com>
4944 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
4945 of incompatible types.
4947 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
4949 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
4950 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
4951 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4952 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
4954 2017-09-19 Richard Biener <rguenther@suse.de>
4956 PR tree-optimization/82244
4957 * tree-vrp.c (remove_range_assertions): Do not propagate
4958 a constant to abnormals but replace the assert with a copy.
4960 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
4962 PR rtl-optimization/57878
4963 PR rtl-optimization/68988
4964 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
4965 avoidance test involving non_reload_pseudos. Move frequency test
4966 below the general fragmentation avoidance test.
4968 2017-09-19 Richard Biener <rguenther@suse.de>
4970 PR tree-optimization/69728
4971 * graphite-sese-to-poly.c (schedule_error): New global.
4972 (add_loop_schedule): Handle empty domain by failing the
4974 (build_original_schedule): Handle schedule_error.
4976 2017-09-19 Richard Biener <rguenther@suse.de>
4978 * graphite-scop-detection.c (scop_detection::can_represent_loop):
4979 Do not iterate to sibling loops but only to siblings of inner
4982 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
4985 * config/m68k/m68k.md (moveq feeding equality comparison): Check
4986 that the registers are different.
4988 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
4990 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
4991 to processor_model and "amdfam17h" to arch_names_table.
4992 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
4994 2017-09-18 Jakub Jelinek <jakub@redhat.com>
4997 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
4999 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5000 Alan Hayward <alan.hayward@arm.com>
5001 David Sherwood <david.sherwood@arm.com>
5003 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
5005 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
5007 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
5008 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
5009 vect_schedule_slp_instance.
5010 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
5011 Update call to vect_slp_analyze_node_operations. Simplify return
5013 (vect_slp_analyze_bb_1): Update call accordingly.
5014 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
5015 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
5016 (vect_schedule_slp): Update call accordingly.
5018 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5019 Alan Hayward <alan.hayward@arm.com>
5020 David Sherwood <david.sherwood@arm.com>
5022 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
5023 with types that aren't in fact scalar.
5025 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5027 * tree-vect-slp.c (vect_record_max_nunits): New function,
5029 (vect_build_slp_tree_1): ...here.
5030 (vect_build_slp_tree_2): Call it for phis too.
5032 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5034 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
5035 to vect_get_vec_def_for_operand when getting the mask operand.
5037 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5038 Alan Hayward <alan.hayward@arm.com>
5039 David Sherwood <david.sherwood@arm.com>
5041 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
5044 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5045 Alan Hayward <alan.hayward@arm.com>
5046 David Sherwood <david.sherwood@arm.com>
5048 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
5049 calculation for vector booleans.
5051 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5052 Alan Hayward <alan.hayward@arm.com>
5053 David Sherwood <david.sherwood@arm.com>
5055 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
5057 (vect_transform_stmt): ...here.
5058 (vect_analyze_stmt): Use it instead of calling
5059 vectorizable_live_operation directly.
5061 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
5063 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
5064 non-SIMT targets in acc vector loops.
5066 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
5068 * configure.ac: Add arc and check if assembler supports gdwarf2.
5069 * configure: Regenerate.
5071 2017-09-18 Richard Biener <rguenther@suse.de>
5073 PR tree-optimization/82220
5074 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
5075 epilogue niters from the min_profitable_iters compute.
5077 2017-09-18 Jakub Jelinek <jakub@redhat.com>
5080 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
5081 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
5082 (ix86_init_pic_reg): Revert 2017-09-01 changes.
5084 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
5087 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
5088 switching to a new text section.
5090 2017-09-18 Richard Biener <rguenther@suse.de>
5092 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
5094 (build_alias_set): Reject aliases with no access function.
5096 2017-09-18 Richard Biener <rguenther@suse.de>
5098 PR tree-optimization/79622
5099 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
5101 (build_cross_bb_scalars_use): Likewise.
5103 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
5105 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
5107 2017-09-18 Alan Modra <amodra@gmail.com>
5110 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
5111 stack_pointer_rtx for count 0. Update comments. Break up
5112 large rtl expression.
5114 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
5116 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
5117 Increase to 20 bytes.
5118 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
5119 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
5120 or avx version of the stub.
5122 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
5125 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
5126 compute the minimum stack alignment. Also update preferred stack
5127 boundary for leaf functions.
5129 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
5131 PR tree-optimization/82228
5132 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
5135 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
5137 * common/config/nds32/nds32-common.c
5138 (nds32_option_optimization_table): Refine formatting.
5139 (nds32_option_optimization_table): Use -fsched-pressure and
5140 -fomit-frame-pointer for specific optimization level.
5142 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
5144 * config/nds32/nds32.c: Refine formatting and comments.
5145 * config/nds32/nds32.h: Likewise.
5146 * config/nds32/nds32.md: Likewise.
5147 * config/nds32/nds32-cost.c: Likewise.
5148 * config/nds32/nds32-isr.c: Likewise.
5149 * config/nds32/nds32-md-auxiliary.c: Likewise.
5150 * config/nds32/nds32-multiple.md: Likewise.
5151 * config/nds32/nds32-predicates.c: Likewise.
5153 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
5154 Jakub Jelinek <jakub@redhat.com>
5156 Add support for -std=c++2a.
5157 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
5159 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
5161 2017-09-15 Steve Ellcey <sellcey@cavium.com>
5164 * doc/extend.texi (Common Function Attributes): Add
5165 references to ARM, AArch64, and S/390 specific attributes.
5166 (Function Specific Option Pragmas): Add AArch64 and S/390
5167 to list of back ends that support the target pragma.
5169 2017-09-15 Nathan Sidwell <nathan@acm.org>
5171 * doc/standards.texi: Fix C++17 description. Update URLs for
5174 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
5176 * common.opt (Wcast-align=strict): New warning option.
5177 * doc/invoke.texi: Document -Wcast-align=strict.
5179 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
5181 * cgraph.h (cgraph_thunk_info): Add comments.
5182 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
5183 assert for VIRTUAL_* arguments stricter.
5185 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
5187 PR tree-optimization/71026
5188 * match.pd: Move RDIV patterns from fold-const.c
5189 * fold-const.c (distribute_real_division): Removed.
5190 (fold_binary_loc): Remove calls to distribute_real_divison.
5192 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5194 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
5195 c++1z and gnu++1z as deprecated. Change other references to
5196 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
5197 Change -Wc++1z-compat to -Wc++17-compat.
5198 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
5199 * dwarf2out.c (highest_c_language): Handle C++17.
5200 (gen_compile_unit_die): Likewise.
5202 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5204 PR rtl-optimization/82192
5205 * combine.c (make_extraction): Don't look through non-paradoxical
5206 SUBREGs or TRUNCATE if pos + len is or might be bigger than
5209 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5210 Alan Hayward <alan.hayward@arm.com>
5211 David Sherwood <david.sherwood@arm.com>
5213 * target.def (function_arg_offset): New hook.
5214 * targhooks.h (default_function_arg_offset): Declare.
5215 * targhooks.c (default_function_arg_offset): New function.
5216 * function.c (locate_and_pad_parm): Use
5217 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
5218 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
5219 (TARGET_FUNCTION_ARG_OFFSET): ...this.
5220 * doc/tm.texi: Regenerate.
5221 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
5222 * config/spu/spu.c (spu_function_arg_offset): New function.
5223 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
5224 * system.h (FUNCTION_ARG_OFFSET): Poison.
5226 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5227 Alan Hayard <alan.hayward@arm.com>
5228 David Sherwood <david.sherwood@arm.com>
5230 * target.def (truly_noop_truncation): New hook.
5231 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
5232 than TRULY_NOOP_TRUNCATION.
5233 * hooks.h (hook_bool_uint_uint_true): Declare.
5234 * hooks.c (hook_bool_uint_uint_true): New function.
5235 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
5236 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
5237 * doc/tm.texi: Regenerate.
5238 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
5239 rather than TRULY_NOOP_TRUNCATION in comments.
5240 (simplify_comparison): Likewise.
5241 (record_truncated_value): Likewise.
5242 * expmed.c (extract_bit_field_1): Likewise.
5243 (extract_split_bit_field): Likewise.
5244 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
5245 instead of TRULY_NOOP_TRUNCATION.
5246 * function.c (assign_parm_setup_block): Likewise.
5247 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
5248 * rtlhooks.c: Include target.h.
5249 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
5250 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
5251 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
5252 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
5253 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
5254 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
5255 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
5256 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
5257 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
5258 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
5259 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
5260 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
5261 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
5262 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
5263 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
5264 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
5265 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
5266 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
5267 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
5268 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
5269 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
5270 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
5271 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
5272 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
5273 * config/mips/mips.c (mips_truly_noop_truncation): New function.
5274 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
5275 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5276 rather than TRULY_NOOP_TRUNCATION in comments.
5277 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
5278 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
5279 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
5280 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
5281 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
5282 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
5283 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
5284 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
5285 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
5286 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
5287 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
5288 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5289 rather than TRULY_NOOP_TRUNCATION in comments.
5290 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
5291 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
5292 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
5293 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
5294 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
5295 TRULY_NOOP_TRUNCATION condition.
5296 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
5297 (TRULY_NOOP_TRUNCATION): Delete.
5298 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
5299 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
5300 * config/spu/spu.c (spu_truly_noop_truncation): New function.
5301 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
5302 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
5303 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
5304 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
5305 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
5306 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5307 rather than TRULY_NOOP_TRUNCATION in comments.
5308 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
5309 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
5310 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
5311 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
5312 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
5313 * system.h (TRULY_NOOP_TRUNCATION): Poison.
5315 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
5318 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
5319 (*cmp_ior): Likewise.
5320 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
5321 (*ior_scc_scc_cmp): Likewise.
5322 (*and_scc_scc): Likewise.
5323 (*and_scc_scc_cmp): Likewise.
5325 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5326 Alan Hayard <alan.hayward@arm.com>
5327 David Sherwood <david.sherwood@arm.com>
5329 * target.def (can_change_mode_class): New hook.
5330 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
5331 (hard_regno_nregs): Likewise.
5332 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
5333 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
5334 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
5335 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
5336 (LOAD_EXTEND_OP): Update accordingly.
5337 * doc/tm.texi: Regenerate.
5338 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
5339 CANNOT_CHANGE_MODE_CLASS.
5340 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
5341 (REG_CAN_CHANGE_MODE_P): ...this new macro.
5342 * combine.c (simplify_set): Update accordingly.
5343 * emit-rtl.c (validate_subreg): Likewise.
5344 * recog.c (general_operand): Likewise.
5345 * regcprop.c (mode_change_ok): Likewise.
5346 * reload1.c (choose_reload_regs): Likewise.
5347 (inherit_piecemeal_p): Likewise.
5348 * rtlanal.c (simplify_subreg_regno): Likewise.
5349 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
5350 instead of CANNOT_CHANGE_MODE_CLASS.
5351 (reload_cse_simplify_operands): Likewise.
5352 * reload.c (push_reload): Use targetm.can_change_mode_class
5353 instead of CANNOT_CHANGE_MODE_CLASS.
5354 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
5355 REG_CANNOT_CHANGE_MODE_P.
5356 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5357 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
5358 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5359 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5360 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5361 (arm_can_change_mode_class): New function.
5362 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
5363 than CANNOT_CHANGE_MODE_CLASS in comments.
5364 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5365 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
5366 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
5367 (ix86_can_change_mode_class): ...this new function, inverting the
5368 sense of the return value.
5369 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5370 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5371 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5372 (ia64_can_change_mode_class): New function.
5373 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5374 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
5375 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
5376 (m32c_can_change_mode_class): ...this new function, inverting the
5377 sense of the return value.
5378 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5379 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5380 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
5381 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
5382 (mips_can_change_mode_class): ...this new function, inverting the
5383 sense of the return value.
5384 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5385 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5386 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5387 (msp430_can_change_mode_class): New function.
5388 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5389 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
5390 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5391 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5392 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5393 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
5394 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5395 (pa_cannot_change_mode_class): Replace with...
5396 (pa_can_change_mode_class): ...this new function, inverting the
5397 sense of the return value.
5398 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
5399 than CANNOT_CHANGE_MODE_CLASS in comments.
5400 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5401 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
5402 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5403 (pdp11_cannot_change_mode_class): Replace with...
5404 (pdp11_can_change_mode_class): ...this new function, inverting the
5405 sense of the return value.
5406 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5407 * config/powerpcspe/powerpcspe-protos.h
5408 (rs6000_cannot_change_mode_class_ptr): Delete.
5409 * config/powerpcspe/powerpcspe.c
5410 (rs6000_cannot_change_mode_class_ptr): Delete.
5411 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5412 (rs6000_option_override_internal): Assign to
5413 targetm.can_change_mode_class instead of
5414 rs6000_cannot_change_mode_class_ptr.
5415 (rs6000_cannot_change_mode_class): Replace with...
5416 (rs6000_can_change_mode_class): ...this new function, inverting the
5417 sense of the return value.
5418 (rs6000_debug_cannot_change_mode_class): Replace with...
5419 (rs6000_debug_can_change_mode_class): ...this new function.
5420 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5421 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
5422 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5423 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5424 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
5426 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
5427 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5428 (rs6000_option_override_internal): Assign to
5429 targetm.can_change_mode_class instead of
5430 rs6000_cannot_change_mode_class_ptr.
5431 (rs6000_cannot_change_mode_class): Replace with...
5432 (rs6000_can_change_mode_class): ...this new function, inverting the
5433 sense of the return value.
5434 (rs6000_debug_cannot_change_mode_class): Replace with...
5435 (rs6000_debug_can_change_mode_class): ...this new function.
5436 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5437 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
5438 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
5439 (s390_can_change_mode_class): ...this new function, inverting the
5440 sense of the return value.
5441 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5442 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5443 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
5444 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5445 (sh_cannot_change_mode_class): Replace with...
5446 (sh_can_change_mode_class): ...this new function, inverting the
5447 sense of the return value.
5448 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5449 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5450 (sparc_can_change_mode_class): New function.
5451 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5452 * config/spu/spu.c (spu_can_change_mode_class): New function.
5453 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5454 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5455 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5456 (visium_can_change_mode_class): New function.
5457 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
5459 2017-09-15 Richard Biener <rguenther@suse.de>
5461 PR tree-optimization/82217
5462 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
5463 but not undefined case.
5465 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5468 * postreload.c (reload_cse_simplify_operands): Skip
5469 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
5471 2017-09-15 Richard Biener <rguenther@suse.de>
5473 PR tree-optimization/68823
5474 * graphite-scop-detection.c (build_alias_set): If we have a
5475 possible dependence check whether we can handle them by just
5476 looking at the DRs DR_ACCESS_FNs.
5477 (build_scops): If build_alias_set fails, fail the SCOP.
5479 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
5481 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
5482 to support float128 built-in functions that require the ISA 3.0
5484 (BU_FLOAT128_3_HW): Likewise.
5485 (SQRTF128): Add support for the IEEE 128-bit square root and fma
5487 (FMAF128): Likewise.
5489 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
5490 support for built-in functions that need the ISA 3.0 IEEE 128-bit
5491 floating point instructions.
5492 (rs6000_invalid_builtin): Likewise.
5493 (rs6000_builtin_mask_names): Likewise.
5494 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
5495 (RS6000_BTM_FLOAT128_HW): Likewise.
5496 (RS6000_BTM_COMMON): Likewise.
5497 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
5499 * doc/extend.texi (RS/6000 built-in functions): Document the
5500 IEEE 128-bit floating point square root and fused multiply-add
5503 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
5505 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
5506 reg (r2) isn't in the set of registers defined in the prologue.
5508 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5509 Alan Hayward <alan.hayward@arm.com>
5510 David Sherwood <david.sherwood@arm.com>
5512 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
5513 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
5514 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
5515 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
5516 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
5518 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
5519 max_vectorization_factor.
5520 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
5522 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5523 Alan Hayward <alan.hayward@arm.com>
5524 David Sherwood <david.sherwood@arm.com>
5526 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
5527 (vect_worthwhile_without_simd_p): Declare.
5528 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
5529 (vectorizable_reduction): Use it.
5530 * tree-vect-stmts.c (vectorizable_shift): Likewise.
5531 (vectorizable_operation): Likewise.
5533 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5534 Alan Hayward <alan.hayward@arm.com>
5535 David Sherwood <david.sherwood@arm.com>
5537 * tree-vectorizer.h (vect_get_num_copies): New function.
5538 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
5539 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5540 (vectorizable_induction): Likewise.
5541 (vectorizable_live_operation): Likewise.
5542 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5543 (vectorizable_bswap): Likewise.
5544 (vectorizable_call): Likewise.
5545 (vectorizable_conversion): Likewise.
5546 (vectorizable_assignment): Likewise.
5547 (vectorizable_shift): Likewise.
5548 (vectorizable_operation): Likewise.
5549 (vectorizable_store): Likewise.
5550 (vectorizable_load): Likewise.
5551 (vectorizable_condition): Likewise.
5552 (vectorizable_comparison): Likewise.
5553 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
5555 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5556 Alan Hayward <alan.hayward@arm.com>
5557 David Sherwood <david.sherwood@arm.com>
5559 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
5560 of vect_init_vector.
5562 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5563 Alan Hayward <alan.hayward@arm.com>
5564 David Sherwood <david.sherwood@arm.com>
5566 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
5567 an inline wrapper that provides a location.
5568 (gimple_build_vector): Likewise.
5569 * gimple-fold.c (gimple_build_vector_from_val): New function.
5570 (gimple_build_vector): Likewise.
5571 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
5572 functions to build the initial value. Always return a gimple value.
5573 (get_initial_defs_for_reduction): Likewise. Only compute
5575 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
5576 vect_init_vector on the results from get_initial_def(s)_for_reduction.
5577 (vectorizable_induction): Use gimple_build_vector rather than
5580 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5581 Alan Hayward <alan.hayward@arm.com>
5582 David Sherwood <david.sherwood@arm.com>
5584 * target.h (vec_perm_indices): New typedef.
5585 (auto_vec_perm_indices): Likewise.
5586 * optabs-query.h: Include target.h
5587 (can_vec_perm_p): Take a vec_perm_indices *.
5588 * optabs-query.c (can_vec_perm_p): Likewise.
5589 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
5590 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
5591 * tree-vect-generic.c (lower_vec_perm): Likewise.
5592 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
5593 (vect_grouped_load_supported): Likewise.
5594 (vect_shift_permute_load_chain): Likewise.
5595 (vect_permute_store_chain): Use auto_vec_perm_indices.
5596 (vect_permute_load_chain): Likewise.
5597 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
5598 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
5599 Update uses of can_vec_perm_p.
5600 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
5601 mode with a number of elements. Take a vec_perm_indices *.
5602 (vect_create_epilog_for_reduction): Update accordingly.
5603 Use auto_vec_perm_indices.
5604 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
5605 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
5606 (vect_transform_slp_perm_load): Likewise.
5607 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
5608 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
5609 (vect_gen_perm_mask_checked): Likewise.
5610 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
5611 (vect_gen_perm_mask_checked): Likewise.
5612 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
5613 (vectorizable_store): Likewise.
5614 (vectorizable_load): Likewise.
5615 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
5616 (vectorizable_bswap): Likewise.
5618 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5619 Alan Hayward <alan.hayward@arm.com>
5620 David Sherwood <david.sherwood@arm.com>
5622 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
5623 * tree.c (build_vector): Likewise.
5624 (build_vector_from_ctor): Update accordingly.
5625 (build_vector_from_val): Likewise.
5626 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
5627 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
5628 * tree-vect-generic.c (add_rshift): Likewise.
5629 (expand_vector_divmod): Likewise.
5630 (optimize_vector_constructor): Likewise.
5631 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
5632 (vect_transform_slp_perm_load): Likewise.
5633 (vect_schedule_slp_instance): Likewise.
5634 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
5635 (vectorizable_call): Likewise.
5636 (vect_gen_perm_mask_any): Likewise. Add elements in order.
5637 * expmed.c (make_tree): Likewise.
5638 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
5639 a vector passed to build_vector.
5640 (fold_convert_const): Likewise.
5641 (exact_inverse): Likewise.
5642 (fold_ternary_loc): Likewise.
5643 (fold_relational_const): Likewise.
5644 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
5645 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
5646 (const_unop): Likewise. Store the reduction accumulator in a
5647 variable rather than an array.
5648 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
5649 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
5650 the new vector, rather than constructing it after the input arrays.
5651 (native_interpret_vector): Use auto_vec<tree> when building
5652 a vector passed to build_vector. Add elements in order.
5653 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
5654 auto_vec<tree> when building a vector passed to build_vector.
5655 (vect_create_epilog_for_reduction): Likewise.
5656 (vectorizable_induction): Likewise.
5657 (get_initial_def_for_reduction): Likewise. Fix indentation of
5659 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
5661 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
5662 passed to build_vector.
5664 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5665 Alan Hayward <alan.hayward@arm.com>
5666 David Sherwood <david.sherwood@arm.com>
5668 * tree-core.h (tree_base::u): Add an "nelts" field.
5669 (tree_vector): Use VECTOR_CST_NELTS as the length.
5670 * tree.c (tree_size): Likewise.
5671 (make_vector): Initialize VECTOR_CST_NELTS.
5672 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
5673 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
5674 TYPE_VECTOR_SUBPARTS.
5675 * expr.c (const_vector_mask_from_tree): Consistently use "units"
5676 as the number of units, setting it from VECTOR_CST_NELTS.
5677 (const_vector_from_tree): Likewise.
5678 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
5679 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
5680 (fold_negate_expr_1): Likewise.
5681 (fold_convert_const): Likewise.
5682 (const_binop): Likewise. Differentiate the number of output and
5684 (const_unop): Likewise.
5685 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
5686 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
5687 in cases that did the opposite.
5689 2017-09-14 Richard Biener <rguenther@suse.de>
5691 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
5694 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
5696 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
5698 2017-09-14 Jakub Jelinek <jakub@redhat.com>
5701 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
5702 if and where to split a bb, except for splitting before debug insn
5703 sequences followed by non-label real insn. Delete debug insns
5704 in between basic blocks.
5706 * combine.c (make_compound_operation_int): Formatting fixes.
5708 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
5709 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5710 * config/netbsd.h (LINK_EH_SPEC): Likewise.
5711 * config/sol2.h (LINK_EH_SPEC): Likewise.
5712 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5713 * config/s390/linux.h (LINK_SPEC): Likewise.
5714 * config/freebsd.h (LINK_EH_SPEC): Likewise.
5715 * config/openbsd.h (LINK_EH_SPEC): Likewise.
5716 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5717 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
5718 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
5719 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5720 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
5721 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
5723 2017-09-13 Jakub Jelinek <jakub@redhat.com>
5725 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
5727 (ENDFILE_LINUX_SPEC): Likewise.
5728 (LINK_EH_SPEC): Likewise.
5729 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
5730 (LINK_OS_LINUX_SPEC32): Likewise.
5731 (LINK_OS_LINUX_SPEC64): Likewise.
5732 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
5733 (LINK_OS_LINUX_SPEC): Likewise.
5735 2017-09-13 Martin Liska <mliska@suse.cz>
5738 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
5739 CASE_HIGH is NULL_TREE.
5741 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
5742 Alan Hayward <alan.hayward@arm.com>
5743 David Sherwood <david.sherwood@arm.com>
5745 * target.def (secondary_memory_needed): New hook.
5746 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
5747 instead of SECONDARY_MEMORY_NEEDED.
5748 (secondary_memory_needed_mode): Likewise.
5749 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
5750 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
5751 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
5752 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
5753 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
5754 * doc/tm.texi: Regenerate.
5755 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
5756 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
5757 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5758 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
5759 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
5760 * config/i386/i386.c (inline_secondary_memory_needed): Put the
5761 mode argument first and change the reg_class arguments to reg_class_t.
5762 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
5763 Make static. Update the call to inline_secondary_memory_needed.
5764 (ix86_register_move_cost): Update the call to
5765 inline_secondary_memory_needed.
5766 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5767 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
5769 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
5770 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5772 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
5773 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
5774 * config/mips/mips.c (mips_secondary_memory_needed): Make static
5775 and match hook interface. Add comment from mips.h.
5776 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5777 * config/mmix/mmix.md (truncdfsf2): Refer to
5778 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5780 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
5781 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
5782 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5783 (pa_secondary_memory_needed): New function.
5784 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
5785 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
5786 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5787 (pdp11_secondary_memory_needed): Make static and match hook interface.
5788 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
5789 * config/powerpcspe/powerpcspe-protos.h
5790 (rs6000_secondary_memory_needed_ptr): Delete.
5791 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
5793 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5794 (rs6000_option_override_internal): Assign to
5795 targetm.secondary_memory_needed rather than
5796 rs6000_secondary_memory_needed_ptr.
5797 (rs6000_secondary_memory_needed): Match hook interface.
5798 (rs6000_debug_secondary_memory_needed): Likewise.
5799 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
5800 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
5801 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
5802 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5803 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
5804 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
5806 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
5807 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5808 (rs6000_option_override_internal): Assign to
5809 targetm.secondary_memory_needed rather than
5810 rs6000_secondary_memory_needed_ptr.
5811 (rs6000_secondary_memory_needed): Match hook interface.
5812 (rs6000_debug_secondary_memory_needed): Likewise.
5813 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
5814 * config/s390/s390.c (s390_secondary_memory_needed): New function.
5815 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5816 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
5817 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5818 (sparc_secondary_memory_needed): New function.
5819 * lra-constraints.c (check_and_process_move): Refer to
5820 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5822 (curr_insn_transform): Likewise.
5823 (process_alt_operands): Use targetm.secondary_memory_needed
5824 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5825 (check_secondary_memory_needed_p): Likewise.
5826 (choose_split_class): Likewise.
5827 * reload.c: Unconditionally include code that was previously
5828 conditional on SECONDARY_MEMORY_NEEDED.
5829 (push_secondary_reload): Use targetm.secondary_memory_needed
5830 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5831 (push_reload): Likewise.
5832 * reload1.c: Unconditionally include code that was previously
5833 conditional on SECONDARY_MEMORY_NEEDED.
5834 (choose_reload_regs): Use targetm.secondary_memory_needed
5835 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5836 (gen_reload): Likewise.
5837 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
5839 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
5840 Alan Hayward <alan.hayward@arm.com>
5841 David Sherwood <david.sherwood@arm.com>
5843 * target.def (secondary_memory_needed_mode): New hook:
5844 * targhooks.c (default_secondary_memory_needed_mode): Declare.
5845 * targhooks.h (default_secondary_memory_needed_mode): New function.
5846 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
5847 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
5848 * doc/tm.texi: Regenerate.
5849 * lra-constraints.c (check_and_process_move): Use
5850 targetm.secondary_memory_needed_mode instead of
5851 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
5852 (curr_insn_transform): Likewise.
5853 * reload.c (get_secondary_mem): Likewise.
5854 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5855 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
5857 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5858 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5859 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
5860 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5861 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
5863 * config/powerpcspe/powerpcspe-protos.h
5864 (rs6000_secondary_memory_needed_mode): Delete.
5865 * config/powerpcspe/powerpcspe.c
5866 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5867 (rs6000_secondary_memory_needed_mode): Make static.
5868 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5869 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
5871 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
5873 (rs6000_secondary_memory_needed_mode): Make static.
5874 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5875 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
5876 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5877 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5878 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
5880 (sparc_secondary_memory_needed_mode): New function.
5881 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
5883 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
5885 * config/aarch64/constraints.md (Umq): New constraint.
5886 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
5888 (mov<mode>): Update condition.
5890 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5892 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
5893 when bitposition is the same.
5895 2017-09-13 Richard Biener <rguenther@suse.de>
5897 * dwarf2out.c (output_die_symbol): Remove.
5898 (output_die): Do not output a DIEs symbol.
5900 2017-09-13 Richard Biener <rguenther@suse.de>
5903 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
5904 default-def to avoid breaking iterator update with the weird
5905 interaction with cgraph_update_edges_for_call_stmt_node.
5907 2017-09-13 Richard Biener <rguenther@suse.de>
5909 * tree-cfg.c (verify_gimple_assign_binary): Add verification
5910 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
5911 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
5912 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
5914 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
5916 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
5917 Disable pc relative literal load irrespective of
5918 TARGET_FIX_ERR_A53_84341 for default.
5920 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
5922 * config/sparc/sparc.c (output_return): Output the source location of
5923 the insn in the delay slot, if any.
5924 (output_sibcall): Likewise.
5926 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
5929 * common.opt (-static-pie): New alias.
5930 (shared): Negate static-pie.
5931 (-no-pie): Update help text.
5933 (static-pie): New option.
5934 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
5935 -static-pie support.
5936 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
5937 (LINK_EH_SPEC): Likewise.
5938 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5939 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
5940 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
5941 * gcc.c (LINK_COMMAND_SPEC): Likewise.
5942 (init_gcc_specs): Likewise.
5943 (init_spec): Likewise.
5944 (display_help): Update help message for -pie.
5945 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
5948 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
5950 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
5951 (movdi_aarch64): Likewise.
5952 (movti_aarch64): Likewise.
5954 2017-09-12 Simon Wright <simon@pushface.org>
5957 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
5958 calculation of the minor version, always output as 0.
5960 2017-09-12 Jakub Jelinek <jakub@redhat.com>
5963 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5964 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
5965 on it early, rather than manual conversion late. For
5966 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
5967 instead of performing manual conversion.
5969 2017-09-12 Carl Love <cel@us.ibm.com>
5971 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
5972 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
5974 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5975 VMULOSW): Add definitions.
5976 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5977 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5978 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5979 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5980 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5982 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
5984 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
5986 (movti_aarch64): Likewise.
5987 (movdf_aarch64): Likewise.
5988 (movtf_aarch64): Likewise.
5989 (load_pairdi): Likewise.
5990 (store_pairdi): Likewise.
5991 (load_pairdf): Likewise.
5992 (store_pairdf): Likewise.
5993 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
5994 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
5995 (ldr_got_small_<mode>): Likewise.
5996 (ldr_got_small_28k_<mode>): Likewise.
5997 (ldr_got_tiny): Likewise.
5998 * config/aarch64/iterators.md (ldst_sz): New.
5999 (ldpstp_sz): Likewise.
6000 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
6002 (thunderx_load): Split load_8 to load_16.
6003 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
6005 (thunderx2t99_storepair_basic): Split store_8 to store_16.
6006 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
6007 (xgene1_store_pair): Split store_8 to store_16.
6008 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
6009 (falkor_st_0_st_sd): Split store_8 to store_16.
6011 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
6013 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
6014 and store1/2/3/4 to store_4/8/12/16.
6015 * config/aarch64/aarch64.md: Update for rename.
6016 * config/arm/arm.md: Likewise.: Likewise.
6017 * config/arm/arm.c: Likewise.
6018 * config/arm/thumb1.md: Likewise.
6019 * config/arm/thumb2.md: Likewise.
6020 * config/arm/vfp.md: Likewise.
6021 * config/arm/arm-generic.md: Likewise.
6022 * config/arm/arm1020e.md: Likewise.
6023 * config/arm/arm1026ejs.md: Likewise.
6024 * config/arm/arm1136jfs.md: Likewise.
6025 * config/arm/arm926ejs.md: Likewise.
6026 * config/arm/cortex-a15.md: Likewise.
6027 * config/arm/cortex-a17.md: Likewise.
6028 * config/arm/cortex-a5.md: Likewise.
6029 * config/arm/cortex-a53.md: Likewise.
6030 * config/arm/cortex-a57.md: Likewise.
6031 * config/arm/cortex-a7.md: Likewise.
6032 * config/arm/cortex-a8.md: Likewise.
6033 * config/arm/cortex-a9.md: Likewise.
6034 * config/arm/cortex-m4.md: Likewise.
6035 * config/arm/cortex-m7.md: Likewise.
6036 * config/arm/cortex-r4.md: Likewise.
6037 * config/arm/exynos-m1.md: Likewise.
6038 * config/arm/fa526.md: Likewise.
6039 * config/arm/fa606te.md: Likewise.
6040 * config/arm/fa626te.md: Likewise.
6041 * config/arm/fa726te.md: Likewise.
6042 * config/arm/fmp626.md: Likewise.
6043 * config/arm/iwmmxt.md: Likewise.
6044 * config/arm/ldmstm.md: Likewise.
6045 * config/arm/marvell-pj4.md: Likewise.
6046 * config/arm/xgene1.md: Likewise.
6047 * config/aarch64/thunderx.md: Likewise.
6048 * config/aarch64/thunderx2t99.md: Likewise.
6049 * config/aarch64/falkor.md: Likewise.
6051 2017-09-12 Martin Liska <mliska@suse.cz>
6053 * attribs.c (private_lookup_attribute): New function.
6054 * attribs.h (private_lookup_attribute): Declared here.
6055 (lookup_attribute): Called from this place.
6057 2017-09-12 Richard Biener <rguenther@suse.de>
6059 PR tree-optimization/82157
6060 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
6061 stmts with side-effects.
6063 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6064 Alan Hayward <alan.hayward@arm.com>
6065 David Sherwood <david.sherwood@arm.com>
6067 * target.def (hard_regno_nregs): New hook.
6068 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
6069 * targhooks.h (default_hard_regno_nregs): Declare.
6070 * targhooks.c (default_hard_regno_nregs): New function.
6071 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
6072 (TARGET_HARD_REGNO_NREGS): ...this hook.
6073 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
6074 (CLASS_MAX_NREGS): Likewise.
6075 * doc/tm.texi: Regenerate.
6076 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
6077 instead of HARD_REGNO_NREGS.
6078 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
6079 HARD_REGNO_NREGS in the comment.
6080 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
6081 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
6082 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
6083 Return an unsigned int.
6084 (TARGET_HARD_REGNO_NREGS): Redefine.
6085 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
6086 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
6087 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
6088 (arc_hard_regno_nregs): New function.
6089 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
6090 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
6091 (arm_hard_regno_nregs): New function.
6092 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
6093 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
6094 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
6095 (TARGET_HARD_REGNO_NREGS): Redefine.
6096 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
6097 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
6098 (HARD_REGNO_NREGS): Delete.
6099 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
6100 (cr16_hard_regno_nregs): New function.
6101 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
6102 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
6103 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
6104 (cris_hard_regno_nregs): New function.
6105 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
6106 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
6107 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
6108 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
6109 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
6110 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
6111 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
6112 (frv_hard_regno_nregs): Make static. Take and return an
6114 (frv_class_max_nregs): Remove outdated copy of documentation.
6115 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
6116 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
6117 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
6118 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
6119 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
6120 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
6121 (TARGET_HARD_REGNO_NREGS): Redefine.
6122 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
6123 (CLASS_MAX_NREGS): Update comment.
6124 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
6125 (ia64_hard_regno_nregs): New function.
6126 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
6127 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
6128 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
6129 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
6130 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
6132 (m32c_hard_regno_nregs): Likewise. Make static.
6133 (TARGET_HARD_REGNO_NREGS): Redefine.
6134 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
6135 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
6136 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
6137 (m68k_hard_regno_nregs): New function.
6138 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
6139 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
6140 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
6141 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
6142 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
6143 Take and return an unsigned int.
6144 (TARGET_HARD_REGNO_NREGS): Redefine.
6145 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
6146 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
6147 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
6148 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
6149 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
6150 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
6151 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
6152 (msp430_hard_regno_nregs): Make static. Take and return an
6154 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
6155 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
6156 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
6157 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
6158 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
6159 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
6160 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
6161 (TARGET_HARD_REGNO_NREGS): Redefine.
6162 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
6163 (PA_HARD_REGNO_NREGS): ...this.
6164 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
6165 (PA_HARD_REGNO_NREGS): ...this.
6166 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
6167 (pa_hard_regno_nregs): New function.
6168 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
6169 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
6170 (pdp11_hard_regno_nregs): New function.
6171 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
6172 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
6173 (rs6000_hard_regno_nregs_hook): New function.
6174 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
6175 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
6176 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
6177 Take and return an unsigned int. Move earlier in file.
6178 (TARGET_HARD_REGNO_NREGS): Redefine.
6179 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
6180 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
6181 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
6182 (rl78_hard_regno_nregs): Make static. Take and return an
6184 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
6185 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
6186 (rs6000_hard_regno_nregs_hook): New function.
6187 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
6188 * config/rx/rx.c (rx_hard_regno_nregs): New function.
6189 (TARGET_HARD_REGNO_NREGS): Redefine.
6190 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
6191 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
6192 instead of HARD_REGNO_NREGS.
6193 (s390_hard_regno_nregs): New function.
6194 (s390_hard_regno_mode_ok): Add comment from s390.h.
6195 (TARGET_HARD_REGNO_NREGS): Redefine.
6196 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
6197 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
6198 (sh_hard_regno_nregs): New function.
6199 (sh_pass_in_reg_p): Use it.
6200 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
6201 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
6202 (sparc_hard_regno_nregs): New function.
6203 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
6204 * config/spu/spu.c (spu_hard_regno_nregs): New function.
6205 (spu_function_arg_advance): Use it, supplying a valid register number.
6206 (TARGET_HARD_REGNO_NREGS): Redefine.
6207 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
6208 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
6209 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
6210 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
6211 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
6212 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
6213 (CLASS_MAX_NREGS): Remove copy of old documentation.
6214 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
6215 (visium_hard_regno_nregs): New function.
6216 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
6217 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
6218 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
6219 xtensa_hard_regno_nregs): New function.
6220 * system.h (HARD_REGNO_NREGS): Poison.
6222 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6224 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
6225 hard_regno_nregs instead of HARD_REGNO_NREGS.
6226 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
6227 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
6228 (c6x_expand_epilogue): Likewise.
6229 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
6230 (frv_read_iacc_argument): Likewise.
6231 * config/sh/sh.c: Include regs.h.
6232 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
6233 (regs_used): Likewise.
6234 (output_stack_adjust): Likewise.
6235 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
6236 * expmed.c: Include regs.h.
6237 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
6238 * ree.c: Include regs.h.
6239 (combine_reaching_defs): Use hard_regno_nregs instead of
6241 (add_removable_extension): Likewise.
6243 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6245 * regs.h (hard_regno_nregs): Turn into a function.
6246 (end_hard_regno): Update accordingly.
6247 * caller-save.c (setup_save_areas): Likewise.
6248 (save_call_clobbered_regs): Likewise.
6249 (replace_reg_with_saved_mem): Likewise.
6250 (insert_restore): Likewise.
6251 (insert_save): Likewise.
6252 * combine.c (can_change_dest_mode): Likewise.
6253 (move_deaths): Likewise.
6254 (distribute_notes): Likewise.
6255 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
6256 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
6257 (rs6000_split_multireg_move): Likewise.
6258 (rs6000_register_move_cost): Likewise.
6259 (rs6000_memory_move_cost): Likewise.
6260 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
6261 (rs6000_split_multireg_move): Likewise.
6262 (rs6000_register_move_cost): Likewise.
6263 (rs6000_memory_move_cost): Likewise.
6264 * cselib.c (cselib_reset_table): Likewise.
6265 (cselib_lookup_1): Likewise.
6266 * emit-rtl.c (set_mode_and_regno): Likewise.
6267 * function.c (aggregate_value_p): Likewise.
6268 * ira-color.c (setup_profitable_hard_regs): Likewise.
6269 (check_hard_reg_p): Likewise.
6270 (calculate_saved_nregs): Likewise.
6271 (assign_hard_reg): Likewise.
6272 (improve_allocation): Likewise.
6273 (calculate_spill_cost): Likewise.
6274 * ira-emit.c (modify_move_list): Likewise.
6275 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
6276 (ira_hard_reg_in_set_p): Likewise.
6277 * ira.c (setup_reg_mode_hard_regset): Likewise.
6278 (clarify_prohibited_class_mode_regs): Likewise.
6279 (check_allocation): Likewise.
6280 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6281 (lra_setup_reg_renumber): Likewise.
6282 (setup_try_hard_regno_pseudos): Likewise.
6283 (spill_for): Likewise.
6284 (assign_hard_regno): Likewise.
6285 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
6286 * lra-constraints.c (in_class_p): Likewise.
6287 (lra_constraint_offset): Likewise.
6288 (simplify_operand_subreg): Likewise.
6289 (lra_constraints): Likewise.
6290 (split_reg): Likewise.
6291 (split_if_necessary): Likewise.
6292 (invariant_p): Likewise.
6293 (inherit_in_ebb): Likewise.
6294 * lra-lives.c (process_bb_lives): Likewise.
6295 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
6296 (get_hard_regs): Likewise.
6297 (do_remat): Likewise.
6298 * lra-spills.c (assign_spill_hard_regs): Likewise.
6299 * mode-switching.c (create_pre_exit): Likewise.
6300 * postreload.c (reload_combine_recognize_pattern): Likewise.
6301 * recog.c (peep2_find_free_register): Likewise.
6302 * regcprop.c (kill_value_regno): Likewise.
6303 (set_value_regno): Likewise.
6304 (copy_value): Likewise.
6305 (maybe_mode_change): Likewise.
6306 (find_oldest_value_reg): Likewise.
6307 (copyprop_hardreg_forward_1): Likewise.
6308 * regrename.c (check_new_reg_p): Likewise.
6309 (regrename_do_replace): Likewise.
6310 * reload.c (push_reload): Likewise.
6311 (combine_reloads): Likewise.
6312 (find_dummy_reload): Likewise.
6313 (operands_match_p): Likewise.
6314 (find_reloads): Likewise.
6315 (find_equiv_reg): Likewise.
6316 (reload_adjust_reg_for_mode): Likewise.
6317 * reload1.c (count_pseudo): Likewise.
6318 (count_spilled_pseudo): Likewise.
6319 (find_reg): Likewise.
6320 (clear_reload_reg_in_use): Likewise.
6321 (free_for_value_p): Likewise.
6322 (allocate_reload_reg): Likewise.
6323 (choose_reload_regs): Likewise.
6324 (reload_adjust_reg_for_temp): Likewise.
6325 (emit_reload_insns): Likewise.
6326 (delete_output_reload): Likewise.
6327 * rtlanal.c (subreg_get_info): Likewise.
6328 * sched-deps.c (sched_analyze_reg): Likewise.
6329 * sel-sched.c (init_regs_for_mode): Likewise.
6330 (mark_unavailable_hard_regs): Likewise.
6331 (choose_best_reg_1): Likewise.
6332 (verify_target_availability): Likewise.
6333 * valtrack.c (dead_debug_insert_temp): Likewise.
6334 * var-tracking.c (track_loc_p): Likewise.
6335 (emit_note_insn_var_location): Likewise.
6336 * varasm.c (make_decl_rtl): Likewise.
6337 * reginfo.c (choose_hard_reg_mode): Likewise.
6338 (init_reg_modes_target): Refer directly to
6339 this_target_regs->x_hard_regno_nregs.
6341 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6343 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
6344 instead of hard_regno_nregs.
6346 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6348 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
6349 end_hard_regno instead of hard_regno_nregs.
6350 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
6351 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
6352 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
6353 * ira-color.c (improve_allocation): Likewise.
6354 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6355 * lra-lives.c (mark_regno_live): Likewise.
6356 (mark_regno_dead): Likewise.
6357 * lra-remat.c (operand_to_remat): Likewise.
6358 * lra.c (collect_non_operand_hard_regs): Likewise.
6359 * postreload.c (reload_combine_note_store): Likewise.
6360 (move2add_valid_value_p): Likewise.
6361 * reload.c (regno_clobbered_p): Likewise.
6363 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6365 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
6367 * config/v850/v850.c (v850_reorg): Likewise.
6368 * reload.c (refers_to_regno_for_reload_p): Likewise.
6369 (find_equiv_reg): Likewise.
6370 * reload1.c (reload_reg_reaches_end_p): Likewise.
6372 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6374 * caller-save.c (add_used_regs): Use REG_NREGS instead of
6376 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
6377 * config/arm/arm.c (output_move_neon): Likewise.
6378 (arm_attr_length_move_neon): Likewise.
6379 (neon_split_vcombine): Likewise.
6380 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
6381 (c6x_mark_reg_written): Likewise.
6382 (c6x_dwarf_register_span): Likewise.
6383 * config/i386/i386.c (ix86_save_reg): Likewise.
6384 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
6385 (rws_access_reg): Likewise.
6386 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
6387 * mode-switching.c (create_pre_exit): Likewise.
6388 * ree.c (combine_reaching_defs): Likewise.
6389 (add_removable_extension): Likewise.
6390 * regcprop.c (find_oldest_value_reg): Likewise.
6391 (copyprop_hardreg_forward_1): Likewise.
6392 * reload.c (reload_inner_reg_of_subreg): Likewise.
6393 (push_reload): Likewise.
6394 (combine_reloads): Likewise.
6395 (find_dummy_reload): Likewise.
6396 (reload_adjust_reg_for_mode): Likewise.
6397 * reload1.c (find_reload_regs): Likewise.
6398 (forget_old_reloads_1): Likewise.
6399 (reload_reg_free_for_value_p): Likewise.
6400 (reload_adjust_reg_for_temp): Likewise.
6401 (emit_reload_insns): Likewise.
6402 (delete_output_reload): Likewise.
6403 * sel-sched.c (choose_best_reg_1): Likewise.
6404 (choose_best_pseudo_reg): Likewise.
6406 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6407 Alan Hayward <alan.hayward@arm.com>
6408 David Sherwood <david.sherwood@arm.com>
6410 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
6411 * target.def (slow_unaligned_access): New hook.
6412 * targhooks.h (default_slow_unaligned_access): Declare.
6413 * targhooks.c (default_slow_unaligned_access): New function.
6414 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
6415 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
6416 * doc/tm.texi: Regenerate.
6417 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
6418 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
6419 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
6421 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
6422 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
6424 (rs6000_slow_unaligned_access): New function.
6425 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
6426 (expand_block_compare): Likewise.
6427 (expand_strn_compare): Likewise.
6428 (rs6000_rtx_costs): Likewise.
6429 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
6430 (riscv_slow_unaligned_access): Likewise.
6431 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
6432 (riscv_slow_unaligned_access_p): ...this and make static.
6433 (riscv_option_override): Update accordingly.
6434 (riscv_slow_unaligned_access): New function.
6435 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
6436 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
6437 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
6438 (rs6000_slow_unaligned_access): New function.
6439 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
6440 (rs6000_rtx_costs): Likewise.
6441 * config/rs6000/rs6000-string.c (expand_block_compare)
6442 (expand_strn_compare): Use targetm.slow_unaligned_access instead
6443 of SLOW_UNALIGNED_ACCESS.
6444 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
6445 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
6446 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
6447 of SLOW_UNALIGNED_ACCESS.
6448 * expmed.c (simple_mem_bitfield_p): Likewise.
6449 * expr.c (alignment_for_piecewise_move): Likewise.
6450 (emit_group_load_1): Likewise.
6451 (emit_group_store): Likewise.
6452 (copy_blkmode_from_reg): Likewise.
6453 (emit_push_insn): Likewise.
6454 (expand_assignment): Likewise.
6455 (store_field): Likewise.
6456 (expand_expr_real_1): Likewise.
6457 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
6458 * lra-constraints.c (simplify_operand_subreg): Likewise.
6459 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
6460 * gimple-ssa-store-merging.c: Likewise in block comment at start
6462 * tree-ssa-strlen.c: Include target.h.
6463 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
6464 of SLOW_UNALIGNED_ACCESS.
6465 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
6467 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6469 PR rtl-optimization/82185
6470 * expmed.c (emit_store_flag_int): Only test tem if it has been
6473 2017-09-12 Richard Biener <rguenther@suse.de>
6476 * match.pd ((FTYPE) N CMP CST): Fix typo.
6478 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
6480 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
6482 (mips_near_type_p): Add 'short_call' attribute as a synonym
6484 * doc/extend.texi (short_call): Document new function attribute.
6486 2017-09-12 Jakub Jelinek <jakub@redhat.com>
6489 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
6490 assertion check that in the condition.
6491 (get_atomic_generic_size): Likewise. Before testing if parameter
6492 has pointer type, if it has array type, call for C++
6493 default_conversion to perform array-to-pointer conversion.
6495 2017-09-12 Richard Biener <rguenther@suse.de>
6497 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
6498 for operations we cannot scalarize.
6500 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
6502 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
6503 vectors heap vectors. Clean up comments.
6504 Make visited_bbs a reference.
6505 (profitable_jump_thread_path): Make GC
6506 vectors heap vectors. Clean up comments.
6508 (convert_and_register_jump_thread_path): Make GC vectors heap
6510 (check_subpath_and_update_thread_path): Same. Clean up comments.
6511 Make visited_bbs a reference.
6512 (handle_phi): Abstract common code to to
6513 register_jump_thread_path_if_profitable.
6514 Rename VAR_BB to DEF_BB.
6516 Make GC vectors heap vectors.
6517 Make visited_bbs a reference.
6518 (handle_assignment): Same.
6519 (register_jump_thread_path_if_profitable): New.
6520 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
6522 Make GC vectors heap vectors. Clean up comments.
6523 Make visited_bbs a reference.
6524 (find_jump_threads_backwards): Make visited_bbs live in the stack.
6525 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
6528 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
6531 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
6532 words of E_DImode object are reachable by xtensa_uimm8x4 access.
6534 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
6536 Revert r251800 and r251799.
6538 2017-09-11 Martin Jambor <mjambor@suse.cz>
6541 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
6542 arguments in advance.
6543 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
6544 use it to find predecessor edges.
6545 (naive_outof_ssa): Collect vector of predecessors.
6547 2017-09-08 Jason Merrill <jason@redhat.com>
6549 PR c++/70029 - ICE with ref-qualifier and -flto
6550 * langhooks.h (struct lang_hooks_for_types): Add
6551 copy_lang_qualifiers.
6552 * attribs.c (build_type_attribute_qual_variant): Use it.
6553 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
6555 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
6556 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
6558 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
6561 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
6562 (*mulsi3_sp64): New instruction.
6563 (mulsi3): New expander.
6565 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
6567 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
6569 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6571 * sancov.c: Include memmodel.h.
6573 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
6576 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
6579 2017-09-07 Carl Love <cel@us.ibm.com>
6581 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
6582 the sldi instruction.
6584 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
6586 * sancov.c: Include tm_p.h.
6588 2017-09-07 Jakub Jelinek <jakub@redhat.com>
6591 * output.h (switch_to_other_text_partition): New declaration.
6592 * varasm.c (switch_to_other_text_partition): New function.
6593 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
6594 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
6595 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
6596 to the other text partition before emitting LCL label and switch back
6597 after emitting the word after it.
6599 2017-09-07 Richard Biener <rguenther@suse.de>
6601 * passes.def (pass_split_crit_edges): Remove instance before PRE.
6602 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
6603 critical edges here, after loop init.
6604 (pass_data_pre): Remove PROP_no_crit_edges flags.
6605 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
6606 for valueization of call args to avoid leaking VN_TOP.
6607 (visit_use): Assert we do not visit default defs.
6608 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
6609 Use error_mark_node to more easily detect leaking VN_TOP.
6610 All default-defs are varying, not VN_TOP. Mark them visited.
6611 (run_scc_vn): Make code match comment.
6613 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6615 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
6616 OPTION_MASK_FLOAT128_KEYWORD.
6617 (POWERPC_MASKS): Likewise.
6618 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
6619 support for the -mfloat128-type option, and make -mfloat128
6620 default on PowerPC Linux systems. Define or undefine
6621 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
6622 Define __float128 to be __ieee128 if IEEE 128-bit support is
6623 enabled, or undefine it.
6624 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
6625 Delete defining __FLOAT128_TYPE__.
6626 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
6627 -mfloat128-type option and make -mfloat128 default on PowerPC
6629 (TARGET_FLOAT128_TYPE): Likewise.
6630 (-mfloat128-type): Likewise.
6631 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6632 Delete the -mfloat128-type option and make -mfloat128 default on
6633 PowerPC Linux systems. Always use __ieee128 to be the keyword for
6634 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
6635 128-bit floating point is enabled. Change tests from using
6636 -mfloat128-type to -mfloat128.
6637 (rs6000_mangle_type): Use the correct mangling for the __float128
6638 type even if normal long double is restricted to 64-bits.
6639 (floatn_mode): Enable the _Float128 type by default on VSX Linux
6641 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
6642 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
6643 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
6645 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
6646 documentation for -mfloat128.
6648 2017-09-06 Olivier Hainque <hainque@adacore.com>
6650 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
6652 2017-09-06 Wish Wu <wishwu007@gmail.com>
6653 Jakub Jelinek <jakub@redhat.com>
6655 * asan.c (initialize_sanitizer_builtins): Add
6656 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
6657 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
6658 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
6659 BT_FN_VOID_UINT64_PTR variables.
6660 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
6661 (BT_FN_VOID_UINT16_UINT16): Likewise.
6662 (BT_FN_VOID_UINT32_UINT32): Likewise.
6663 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
6664 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
6665 (BT_FN_VOID_UINT64_PTR): Likewise.
6666 * common.opt (flag_sanitize_coverage): New variable.
6667 (fsanitize-coverage=trace-pc): Remove.
6668 (fsanitize-coverage=): Add.
6669 * flag-types.h (enum sanitize_coverage_code): New enum.
6670 * fold-const.c (fold_range_test): Disable non-short-circuit
6671 optimization if flag_sanitize_coverage.
6672 (fold_truth_andor): Likewise.
6673 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
6674 * opts.c (COVERAGE_SANITIZER_OPT): Define.
6675 (coverage_sanitizer_opts): New array.
6676 (get_closest_sanitizer_option): Add OPTS argument, handle also
6677 OPT_fsanitize_coverage_.
6678 (parse_sanitizer_options): Adjusted to also handle
6679 OPT_fsanitize_coverage_.
6680 (common_handle_option): Add OPT_fsanitize_coverage_.
6681 * sancov.c (instrument_comparison, instrument_switch): New function.
6682 (sancov_pass): Add trace-cmp support.
6683 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
6684 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
6685 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
6686 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
6687 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
6688 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
6689 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
6690 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
6691 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
6692 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
6694 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
6696 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6697 error. Only quit immediately if parsing is complete.
6698 (BEGIN): Initialize fatal_err and parse_done.
6699 (begin fpu, end fpu): Check number of arguments.
6700 (begin arch, end arch): Likewise.
6701 (begin cpu, end cpu): Likewise.
6702 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6703 (optalias): Likewise.
6705 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
6707 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6708 * config/arm/arm-isa.h: Delete. Move definitions to ...
6709 * arm-cpus.in: ... here. Use new feature and fgroup values.
6710 * config/arm/arm.c (arm_option_override): Use lower case for feature
6712 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6713 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6714 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6716 (print_isa_bits_for): New function.
6717 (gen_isa): New function.
6718 (gen_comm_data): Use print_isa_bits_for.
6719 (define feature): New keyword.
6720 (define fgroup): New keyword.
6721 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
6722 (arm-isa.h): Add rule to generate file.
6723 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6724 case for feature bit names.
6726 2017-09-06 Richard Biener <rguenther@suse.de>
6728 * tree-ssa-pre.c (NECESSARY): Remove.
6729 (create_expression_by_pieces): Do not touch pass-local flags.
6730 (insert_into_preds_of_block): Likewise.
6731 (do_pre_regular_insertion): Likewise.
6732 (eliminate_insert): Likewise.
6733 (eliminate_dom_walker::before_dom_children): Likewise.
6734 (fini_eliminate): Do not look at inserted_exprs.
6735 (mark_operand_necessary): Remove.
6736 (remove_dead_inserted_code): Replace with simple work-list
6737 algorithm based on inserted_exprs and SSA uses.
6738 (pass_pre::execute): Re-order fini_eliminate and
6739 remove_dead_inserted_code.
6741 2017-09-06 Olivier Hainque <hainque@adacore.com>
6743 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
6744 for VxWorks 7. Adjust surrounding comments.
6746 2017-09-06 Richard Biener <rguenther@suse.de>
6748 * gimple-ssa-strength-reduction.c
6749 (find_candidates_dom_walker::before_dom_children): Also allow
6752 2017-09-06 Richard Biener <rguenther@suse.de>
6754 PR tree-optimization/82108
6755 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
6756 for gap in the non-permutation SLP case.
6758 2017-09-06 Martin Jambor <mjambor@suse.cz>
6760 PR tree-optimization/82078
6761 * tree-sra.c (sort_and_splice_var_accesses): Move call to
6762 add_access_to_work_queue...
6763 (build_accesses_from_assign): ...here.
6764 (propagate_all_subaccesses): Make sure racc is the group
6765 representative, if there is one.
6767 2017-09-06 Jakub Jelinek <jakub@redhat.com>
6770 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
6773 2017-09-06 Richard Biener <rguenther@suse.de>
6775 * gimple-ssa-strength-reduction.c
6776 (find_candidates_dom_walker::before_doom_children): Use a
6777 type and not a mode check.
6779 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
6782 * config/arm/predicates.md (arm_general_adddi_operand): Create new
6784 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
6786 2017-09-05 Jeff Law <law@redhat.com>
6788 PR tree-optimization/64910
6789 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
6790 cases where we have 3 or more operands.
6792 2017-09-05 Jakub Jelinek <jakub@redhat.com>
6795 * omp-low.c (lower_omp_for): Recompute tree invariant if
6796 gimple_omp_for_initial/final is ADDR_EXPR.
6799 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
6800 into gimple val before gimplification fo the COND_EXPR.
6802 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
6804 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
6805 REGION_COPY argument.
6806 (thread_through_all_blocks): Remove unused argument to
6807 duplicate_thread_path.
6809 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6810 Alan Hayward <alan.hayward@arm.com>
6811 David Sherwood <david.sherwood@arm.com>
6813 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
6814 Take a scalar_mode rather than a machine_mode.
6815 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6816 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
6817 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6818 (aarch64_gen_adjusted_ldpstp): Likewise.
6819 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
6821 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6822 Alan Hayward <alan.hayward@arm.com>
6823 David Sherwood <david.sherwood@arm.com>
6825 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
6826 Take a scalar_int_mode instead of a machine_mode.
6827 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
6828 (aarch64_output_scalar_simd_mov_immediate): Likewise.
6829 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
6830 (aarch64_simd_attr_length_rglist): Delete.
6831 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
6832 a scalar_int_mode instead of a machine_mode.
6833 (aarch64_add_offset): Likewise.
6834 (aarch64_internal_mov_immediate): Likewise
6835 (aarch64_add_constant_internal): Likewise.
6836 (aarch64_add_constant): Likewise.
6837 (aarch64_movw_imm): Likewise.
6838 (aarch64_rtx_arith_op_extract_p): Likewise.
6839 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
6840 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
6841 Remove assert that the mode isn't a vector.
6842 (aarch64_output_scalar_simd_mov_immediate): Likewise.
6843 (aarch64_expand_mov_immediate): Update calls after above changes.
6844 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
6845 (aarch64_and_bitmask_imm): Check for scalar integer modes.
6846 (aarch64_move_imm): Likewise.
6847 (aarch64_can_const_movi_rtx_p): Likewise.
6848 (aarch64_strip_extend): Likewise.
6849 (aarch64_extr_rtx_p): Likewise.
6850 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
6851 a CONST_INT when the mode parameter is VOIDmode.
6852 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
6854 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6856 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
6857 * stor-layout.c (bitwise_mode_for_mode): Likewise.
6858 (bitwise_type_for_mode): Update accordingly.
6860 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6862 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
6863 * stor-layout.c (mode_for_size_tree): Likewise.
6864 (mode_for_array): Update accordingly.
6865 (layout_decl): Likewise.
6866 (compute_record_mode): Likewise. Only set the mode once.
6868 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6870 * target.def (get_mask_mode): Change return type to opt_mode.
6872 * doc/tm.texi: Regenerate.
6873 * targhooks.h (default_get_mask_mode): Return an opt_mode.
6874 * targhooks.c (default_get_mask_mode): Likewise.
6875 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
6876 * optabs-query.c (can_vec_mask_load_store_p): Update use of
6877 targetm.get_mask_mode.
6878 * tree.c (build_truth_vector_type): Likewise.
6880 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6882 * machmode.h (mode_for_vector): Return an opt_mode.
6883 * stor-layout.c (mode_for_vector): Likewise.
6884 (mode_for_int_vector): Update accordingly.
6885 (layout_type): Likewise.
6886 * config/i386/i386.c (emit_memmov): Likewise.
6887 (ix86_expand_set_or_movmem): Likewise.
6888 (ix86_expand_vector_init): Likewise.
6889 (ix86_get_mask_mode): Likewise.
6890 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
6892 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
6893 * expmed.c (extract_bit_field_1): Likewise.
6894 * expr.c (expand_expr_real_2): Likewise.
6895 * optabs-query.c (can_vec_perm_p): Likewise.
6896 (can_vec_mask_load_store_p): Likewise.
6897 * optabs.c (expand_vec_perm): Likewise.
6898 * targhooks.c (default_get_mask_mode): Likewise.
6899 * tree-vect-stmts.c (vectorizable_store): Likewise.
6900 (vectorizable_load): Likewise.
6901 (get_vectype_for_scalar_type_and_size): Likewise.
6903 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6905 * machmode.h (mode_for_int_vector): New function.
6906 * stor-layout.c (mode_for_int_vector): Likewise.
6907 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
6908 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
6909 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
6910 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
6911 (s390_expand_vcond): Likewise.
6913 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6915 * machmode.h (opt_machine_mode): New type.
6916 (opt_mode<T>): Allow construction from anything that can be
6918 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
6919 (mode_for_size): Return an opt_machine_mode.
6920 * stor-layout.c (mode_for_size): Likewise.
6921 (mode_for_size_tree): Update call accordingly.
6922 (bitwise_mode_for_mode): Likewise.
6923 (make_fract_type): Likewise.
6924 (make_accum_type): Likewise.
6925 * caller-save.c (replace_reg_with_saved_mem): Update call
6927 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6928 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6929 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6930 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6931 * expmed.c (extract_bit_field_1): Likewise.
6932 * reload.c (get_secondary_mem): Likewise.
6933 * varasm.c (assemble_integer): Likewise.
6934 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
6937 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6939 * machmode.h (decimal_float_mode_for_size): New function.
6940 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
6941 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
6942 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
6943 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
6944 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
6945 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
6947 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6949 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
6950 (get_builtin_sync_mode): Likewise.
6951 (expand_ifn_atomic_compare_exchange): Likewise.
6952 (expand_builtin_atomic_clear): Likewise.
6953 (expand_builtin_atomic_test_and_set): Likewise.
6954 (fold_builtin_atomic_always_lock_free): Likewise.
6955 * calls.c (compute_argument_addresses): Likewise.
6956 (emit_library_call_value_1): Likewise.
6957 (store_one_arg): Likewise.
6958 * combine.c (combine_instructions): Likewise.
6959 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
6960 * config/arm/arm.c (arm_function_value): Likewise.
6961 (aapcs_allocate_return_reg): Likewise.
6962 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
6963 * config/i386/i386.c (construct_container): Likewise.
6964 (ix86_gimplify_va_arg): Likewise.
6965 (ix86_expand_sse_cmp): Likewise.
6966 (emit_memmov): Likewise.
6967 (emit_memset): Likewise.
6968 (expand_small_movmem_or_setmem): Likewise.
6969 (ix86_expand_pextr): Likewise.
6970 (ix86_expand_pinsr): Likewise.
6971 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
6972 * config/microblaze/microblaze.c (microblaze_block_move_straight):
6974 * config/mips/mips.c (mips_function_value_1) Likewise.
6975 (mips_block_move_straight): Likewise.
6976 (mips_expand_ins_as_unaligned_store): Likewise.
6977 * config/powerpcspe/powerpcspe.c
6978 (rs6000_darwin64_record_arg_advance_flush): Likewise.
6979 (rs6000_darwin64_record_arg_flush): Likewise.
6980 * config/rs6000/rs6000.c
6981 (rs6000_darwin64_record_arg_advance_flush): Likewise.
6982 (rs6000_darwin64_record_arg_flush): Likewise.
6983 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
6984 (sparc_function_value_1): Likewise.
6985 * config/spu/spu.c (adjust_operand): Likewise.
6986 (spu_emit_branch_or_set): Likewise.
6987 (arith_immediate_p): Likewise.
6988 * emit-rtl.c (gen_lowpart_common): Likewise.
6989 * expr.c (expand_expr_real_1): Likewise.
6990 * function.c (assign_parm_setup_block): Likewise.
6991 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6992 * reload1.c (alter_reg): Likewise.
6993 * stor-layout.c (mode_for_vector): Likewise.
6994 (layout_type): Likewise.
6996 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6998 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
6999 (spu_convert_move): Likewise.
7000 * lower-subreg.c (resolve_simple_move): Likewise.
7002 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7005 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
7006 define_insn to a define_expand.
7007 (altivec_vsum2sws_direct): New define_insn.
7008 (altivec_vsumsws): Convert from a define_insn to a define_expand.
7010 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
7012 * config/arm/arm.c (arm_option_params_internal): Improve setting of
7015 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
7020 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
7021 realign stack if stack alignment needed is less than incoming
7024 2017-09-05 Marek Polacek <polacek@redhat.com>
7027 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
7030 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
7032 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
7034 2017-09-05 Richard Biener <rguenther@suse.de>
7036 PR tree-optimization/82084
7037 * fold-const.c (can_native_encode_string_p): Handle wide characters.
7039 2017-09-05 Richard Biener <rguenther@suse.de>
7041 PR tree-optimization/82102
7042 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
7044 2017-09-05 Martin Liska <mliska@suse.cz>
7046 PR tree-optimization/82032
7047 * tree-cfg.c (generate_range_test): New function.
7048 * tree-cfg.h (generate_range_test): Declared here.
7049 * tree-cfgcleanup.c (convert_single_case_switch): New function.
7050 (cleanup_control_expr_graph): Use it.
7051 * tree-switch-conversion.c (try_switch_expansion): Remove
7053 (emit_case_nodes): Use generate_range_test.
7055 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
7058 * config/i386/i386.md (*<btsc><mode>_mask): Add
7059 TARGET_USE_BT to insn constraint.
7060 (*btr<mode>_mask): Ditto.
7062 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
7064 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
7065 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
7067 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
7070 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
7071 TARGET_NEON and TARGET_IWMMXT.
7072 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
7073 TARGET_NEON and TARGET_IWMMXT.
7074 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
7076 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
7078 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
7079 (ix86_rewrite_tls_address): Ditto.
7080 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
7081 (ix86_rewrite_tls_address_1): Ditto.
7082 (ix86_rewrite_tls_address): Ditto.
7083 * config/i386/predicates.md (tls_address_pattern): New predicate.
7084 * config/i386/i386.md (TLS address splitter): New splitter.
7086 2017-09-04 Richard Biener <rguenther@suse.de>
7088 PR tree-optimization/82084
7089 * fold-const.h (can_native_encode_string_p): Declare.
7090 * fold-const.c (can_native_encode_string_p): Factor out from ...
7091 (native_encode_string): ... here.
7092 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
7093 vectorizing stores from constants we later cannot handle.
7095 2017-09-04 Marek Polacek <polacek@redhat.com>
7098 * doc/invoke.texi: Update -Wtautological-compare documentation.
7100 2017-09-04 Jeff Law <law@redhat.com>
7102 PR tree-optimization/64910
7103 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
7104 swap the first and last operand if the last is a constant.
7106 2017-09-04 Marek Polacek <polacek@redhat.com>
7109 * convert.c (do_narrow): When sanitizing signed integer overflows,
7110 bail out for signed types.
7111 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
7113 2017-09-04 Richard Biener <rguenther@suse.de>
7115 PR tree-optimization/82060
7116 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
7117 Move devirtualization after stmt folding and before EH/AB/noreturn
7118 cleanup to get the stmt refs canonicalized. Use a bool instead
7119 of gimple_modified_p since that doesn't work for NOPs. Schedule
7120 NOPs generated by folding for removal.
7122 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7123 Alan Hayward <alan.hayward@arm.com>
7124 David Sherwood <david.sherwood@arm.com>
7126 * coretypes.h (pad_direction): New enum.
7127 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
7128 (FUNCTION_ARG_PADDING): Likewise.
7129 * target.def (function_arg_padding): New hook.
7130 * targhooks.h (default_function_arg_padding): Declare.
7131 * targhooks.c (default_function_arg_padding): New function.
7132 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
7133 (TARGET_FUNCTION_ARG_PADDING): ...this.
7134 * doc/tm.texi: Regenerate.
7135 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
7136 instead of direction.
7137 (compute_argument_addresses): Likewise.
7138 (load_register_parameters): Likewise.
7139 (emit_library_call_value_1): Likewise.
7140 (store_one_arg): Use targetm.calls.function_arg_padding instead
7141 of FUNCTION_ARG_PADDING.
7142 (must_pass_in_stack_var_size_or_pad): Likewise.
7143 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
7144 (emit_group_store): Likewise.
7145 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
7146 instead of FUNCTION_ARG_PADDING.
7147 (emit_push_insn): Likewise, and propagate enum change throughout
7149 * function.h (direction): Delete.
7150 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
7152 * function.c (assign_parm_find_stack_rtl): Likewise.
7153 (assign_parm_setup_block_p): Likewise.
7154 (assign_parm_setup_block): Likewise.
7155 (gimplify_parameters): Likewise.
7156 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
7157 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
7159 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
7160 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7161 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
7162 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
7163 (aarch64_function_arg_padding): ...this new function.
7164 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
7165 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7166 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
7167 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7168 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
7169 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7170 (arm_pad_arg_upward): Replace with...
7171 (arm_function_arg_padding): ...this new function.
7172 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
7174 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
7175 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
7176 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7177 (ia64_hpux_function_arg_padding): Replace with...
7178 (ia64_function_arg_padding): ...this new function. Use pad_direction
7179 instead of direction. Check for TARGET_HPUX.
7180 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
7181 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7182 (iq2000_function_arg_padding): New function.
7183 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
7184 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
7185 (mips_function_arg_padding): ...this new function.
7186 (mips_pad_reg_upward): Update accordingly.
7187 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7188 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
7189 targetm.calls.function_arg_padding.
7190 (FUNCTION_ARG_PADDING): Delete.
7191 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7192 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
7193 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7194 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
7195 (nios2_block_reg_padding): Return pad_direction instead of direction.
7196 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
7197 instead of direction.
7198 (nios2_function_arg_padding): Likewise. Make static.
7199 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7200 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
7201 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
7202 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
7203 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7204 (pa_function_arg_padding): Make static. Return pad_direction instead
7206 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
7207 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7208 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
7209 instead of direction. Use targetm.calls.function_arg_padding.
7210 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
7211 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
7212 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
7213 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
7214 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
7216 (function_arg_padding): Rename to...
7217 (rs6000_function_arg_padding): ...this. Make static. Return
7218 pad_direction instead of direction.
7219 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
7220 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
7221 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7222 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
7223 instead of direction. Use targetm.calls.function_arg_padding.
7224 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
7225 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
7226 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
7227 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
7228 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7229 (function_arg_padding): Rename to...
7230 (rs6000_function_arg_padding): ...this. Make static. Return
7231 pad_direction instead of direction.
7232 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
7233 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
7234 * config/s390/s390.c (s390_function_arg_padding): New function.
7235 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7236 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
7237 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
7238 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7239 (function_arg_padding): Rename to...
7240 (sparc_function_arg_padding): ...this. Make static. Return
7241 pad_direction instead of direction.
7242 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
7243 * config/spu/spu.c (spu_function_arg_padding): New function.
7244 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7245 * system.h (FUNCTION_ARG_PADDING): Poison.
7247 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7248 Alan Hayward <alan.hayward@arm.com>
7249 David Sherwood <david.sherwood@arm.com>
7251 * target.def (modes_tieable_p): New hook.
7252 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
7253 (TARGET_MODES_TIEABLE_P): ...this.
7254 * doc/tm.texi.in: Regenerate.
7255 * hooks.h (hook_bool_mode_mode_true): Declare.
7256 * hooks.c (hook_bool_mode_mode_true): New function.
7257 * combine.c (subst): Use targetm.modes_tieable_p instead of
7259 * dse.c (find_shift_sequence): Likewise.
7260 * expmed.c (extract_low_bits): Likewise.
7261 * lower-subreg.c: Include target.h.
7262 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
7264 * rtlanal.c (rtx_cost): Likewise.
7265 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
7266 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
7267 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
7268 (TARGET_MODES_TIEABLE_P): Redefine.
7269 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
7270 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
7271 (TARGET_MODES_TIEABLE_P): Redefine.
7272 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
7273 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
7274 (arc_modes_tieable_p): New function.
7275 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
7276 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
7277 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
7278 (arm_modes_tieable_p): Make static.
7279 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
7280 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
7281 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
7282 (TARGET_MODES_TIEABLE_P): Redefine.
7283 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
7284 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
7285 (TARGET_MODES_TIEABLE_P): Redefine.
7286 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
7287 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
7288 (cr16_modes_tieable_p): New function.
7289 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
7290 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
7291 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
7292 (TRULY_NOOP_TRUNCATION): Update comment.
7293 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
7294 (TRULY_NOOP_TRUNCATION): Update comment.
7295 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
7296 (frv_modes_tieable_p): New function.
7297 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
7298 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
7299 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
7300 (TARGET_MODES_TIEABLE_P): Redefine.
7301 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
7302 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
7303 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
7304 (TARGET_MODES_TIEABLE_P): Redefine.
7305 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
7306 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
7307 (ia64_modes_tieable_p): New function.
7308 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
7309 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
7310 (iq2000_modes_tieable_p): New function.
7311 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
7312 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
7313 (lm32_modes_tieable_p): New function.
7314 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
7315 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
7316 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
7317 (TARGET_MODES_TIEABLE_P): Redefine.
7318 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
7319 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
7320 (m32r_modes_tieable_p): New function.
7321 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
7322 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
7323 (m68k_modes_tieable_p): New function.
7324 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
7325 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
7326 (mcore_modes_tieable_p): New function.
7327 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
7328 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
7330 (TARGET_MODES_TIEABLE_P): Redefine.
7331 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
7332 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
7333 * config/mips/mips.c (mips_modes_tieable_p): Make static.
7334 (TARGET_MODES_TIEABLE_P): Redefine.
7335 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
7336 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
7337 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
7338 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
7339 (mn10300_modes_tieable_p): ...this and make static.
7340 (TARGET_MODES_TIEABLE_P): Redefine.
7341 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
7342 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
7343 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
7344 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
7345 (msp430_modes_tieable_p): Make static.
7346 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
7347 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
7348 (TARGET_MODES_TIEABLE_P): Redefine.
7349 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
7350 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
7351 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
7352 (TARGET_MODES_TIEABLE_P): Redefine.
7353 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
7354 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
7355 * config/pa/pa.c (pa_modes_tieable_p): Make static.
7356 (TARGET_MODES_TIEABLE_P): Redefine.
7357 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
7358 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
7359 (pdp11_modes_tieable_p): New function.
7360 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
7361 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
7362 (rs6000_modes_tieable_p): New function.
7363 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
7364 * config/powerpcspe/powerpcspe.md: Update comment.
7365 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
7366 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
7367 (TARGET_MODES_TIEABLE_P): Redefine.
7368 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
7369 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
7370 (rl78_modes_tieable_p): New function.
7371 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
7372 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
7373 (rs6000_modes_tieable_p): New function.
7374 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
7375 * config/rs6000/rs6000.md: Update comment.
7376 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
7377 * config/rx/rx.c (rx_modes_tieable_p): New function.
7378 (TARGET_MODES_TIEABLE_P): Redefine.
7379 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
7380 * config/s390/s390.c (s390_modes_tieable_p): New function.
7381 (TARGET_MODES_TIEABLE_P): Redefine.
7382 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
7383 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
7384 (sh_modes_tieable_p): New function.
7385 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
7386 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
7387 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
7388 (sparc_modes_tieable_p): Make static.
7389 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
7390 * config/spu/spu.c (spu_modes_tieable_p): New function.
7391 (TARGET_MODES_TIEABLE_P): Redefine.
7392 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
7393 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
7394 (TARGET_MODES_TIEABLE_P): Redefine.
7395 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
7396 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
7397 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
7398 * config/v850/v850.c (v850_modes_tieable_p): New function.
7399 (TARGET_MODES_TIEABLE_P): Redefine.
7400 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
7401 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
7402 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
7403 (visium_modes_tieable_p): New function.
7404 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
7405 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
7406 (xtensa_modes_tieable_p): New function.
7407 * system.h (MODES_TIEABLE_P): Poison.
7409 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7410 Alan Hayward <alan.hayward@arm.com>
7411 David Sherwood <david.sherwood@arm.com>
7413 * target.def (hard_regno_mode_ok): New hook.
7414 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
7415 (TARGET_HARD_REGNO_MODE_OK): ...this.
7416 * doc/tm.texi.in: Regenerate.
7417 * hooks.h (hook_bool_uint_mode_true): Declare.
7418 * hooks.c (hook_bool_uint_mode_true): New function.
7419 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
7421 * genpreds.c (write_insn_preds_c): Add an include of target.h.
7422 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
7423 instead of HARD_REGNO_MODE_OK.
7424 * caller-save.c: Include target.h.
7425 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
7427 * combine.c (can_combine_p): Likewise.
7428 (combinable_i3pat): Likewise.
7429 (can_change_dest_mode): Likewise.
7430 * expr.c (init_expr_target): Likewise.
7431 (convert_move): Likewise.
7432 (convert_modes): Likewise.
7433 * ira.c (setup_prohibited_class_mode_regs): Likewise.
7434 (setup_prohibited_mode_move_regs): Likewise.
7435 * ira.h (target_ira): Likewise.
7436 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7437 * lra-constraints.c (process_alt_operands): Likewise.
7438 (split_reg): Likewise.
7439 * recog.c (peep2_find_free_register): Likewise.
7440 * ree.c (combine_reaching_defs): Likewise.
7441 * regcprop.c (maybe_mode_change): Likewise.
7442 * reginfo.c (init_reg_sets_1): Likewise.
7443 (choose_hard_reg_mode): Likewise.
7444 (simplifiable_subregs): Likewise.
7445 * regrename.c (check_new_reg_p): Likewise.
7446 * reload.c (find_valid_class): Likewise.
7447 (find_valid_class_1): Likewise.
7448 (reload_inner_reg_of_subreg): Likewise.
7449 (push_reload): Likewise.
7450 (combine_reloads): Likewise.
7451 (find_dummy_reload): Likewise.
7452 (find_reloads): Likewise.
7453 * reload1.c (find_reg): Likewise.
7454 (set_reload_reg): Likewise.
7455 (allocate_reload_reg): Likewise.
7456 (choose_reload_regs): Likewise.
7457 (reload_adjust_reg_for_temp): Likewise.
7458 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
7459 (simplify_subreg_regno): Likewise.
7460 * sel-sched.c (init_regs_for_mode): Likewise.
7461 * varasm.c (make_decl_rtl): Likewise.
7462 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
7463 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
7465 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
7466 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
7467 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7468 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
7469 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
7470 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7471 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
7472 (arc_mode_class): Delete.
7473 (HARD_REGNO_MODE_OK): Delete.
7474 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7475 (arc_hard_regno_mode_ok): Rename old array to...
7476 (arc_hard_regno_mode_ok_modes): ...this.
7477 (arc_conditional_register_usage): Update accordingly.
7478 (arc_mode_class): Make static.
7479 (arc_hard_regno_mode_ok): New function.
7480 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
7481 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
7482 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7483 (arm_hard_regno_mode_ok): Make static.
7484 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
7486 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
7487 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
7488 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
7490 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7491 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
7492 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
7493 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
7494 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
7495 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7496 * config/bfin/predicates.md (valid_reg_operand): Use
7497 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
7498 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
7499 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
7500 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7501 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
7502 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
7503 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7504 (cr16_hard_regno_mode_ok): Make static and return a bool.
7505 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
7506 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7507 (cris_hard_regno_mode_ok): New function.
7508 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
7509 (epiphany_mode_class): Delete.
7510 (HARD_REGNO_MODE_OK): Delete.
7511 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
7512 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7513 (hard_regno_mode_ok): Rename to...
7514 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
7515 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
7516 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
7518 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
7519 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
7520 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7521 (frv_hard_regno_mode_ok): Make static and return a bool.
7522 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
7524 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
7525 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
7526 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
7527 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
7529 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7530 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
7531 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
7532 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
7534 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7535 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
7536 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7537 (ia64_hard_regno_mode_ok): New function.
7538 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
7539 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7540 (iq2000_hard_regno_mode_ok): New function.
7541 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
7542 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7543 (lm32_hard_regno_mode_ok): New function.
7544 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
7545 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
7546 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
7547 instead of HARD_REGNO_MODE_OK.
7548 (m32c_hard_regno_ok): Rename to...
7549 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
7550 (m32c_cannot_change_mode_class): Update accordingly.
7551 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7552 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
7553 (m32r_mode_class): Delete.
7554 (HARD_REGNO_MODE_OK): Delete.
7555 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7556 (m32r_hard_regno_mode_ok): Rename to...
7557 (m32r_hard_regno_modes): ...this.
7558 (m32r_mode_class): Make static.
7559 (m32r_hard_regno_mode_ok): New function.
7560 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
7561 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
7562 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7563 (m68k_hard_regno_mode_ok): Make static.
7564 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
7565 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7566 (mcore_hard_regno_mode_ok): New function.
7567 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
7568 (HARD_REGNO_MODE_OK): Delete.
7569 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
7571 (microblaze_hard_regno_mode_ok_p): ...this and make static.
7572 (microblaze_hard_regno_mode_ok): New function.
7573 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7574 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
7575 (mips_hard_regno_mode_ok): Delete.
7576 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
7577 (mips_hard_regno_mode_ok_p): ...this and make static.
7578 (mips_hard_regno_mode_ok_p): Rename to...
7579 (mips_hard_regno_mode_ok_uncached): ...this.
7580 (mips_hard_regno_mode_ok): New function.
7581 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
7582 of HARD_REGNO_MODE_OK.
7583 (mips_option_override): Update after above name changes.
7584 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7585 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
7586 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
7587 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
7588 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
7589 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7590 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
7591 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
7592 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
7593 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7594 (msp430_hard_regno_mode_ok): Make static and return a bool.
7595 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
7596 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
7597 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
7599 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7600 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
7601 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
7602 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
7603 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
7604 (PA_HARD_REGNO_MODE_OK): ...this
7605 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
7606 (PA_HARD_REGNO_MODE_OK): ...this.
7607 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7608 (pa_hard_regno_mode_ok): New function.
7609 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
7610 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7611 (pdp11_hard_regno_mode_ok): New function.
7612 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
7613 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
7615 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
7617 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7618 (rs6000_hard_regno_mode_ok): Rename to...
7619 (rs6000_hard_regno_mode_ok_uncached): ...this.
7620 (rs6000_init_hard_regno_mode_ok): Update accordingly.
7621 (rs6000_hard_regno_mode_ok): New function.
7622 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
7623 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
7624 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
7625 (riscv_hard_regno_mode_ok): ...this and make static.
7626 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7627 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
7628 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
7629 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7630 (rl78_hard_regno_mode_ok): Make static and return bool.
7631 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
7632 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
7634 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
7635 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7636 (rs6000_hard_regno_mode_ok): Rename to...
7637 (rs6000_hard_regno_mode_ok_uncached): ...this.
7638 (rs6000_init_hard_regno_mode_ok): Update accordingly.
7639 (rs6000_hard_regno_mode_ok): New function.
7640 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
7641 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
7642 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7643 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
7644 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
7645 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
7646 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7647 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
7648 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
7649 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7650 (sh_hard_regno_mode_ok): Make static.
7651 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
7652 instead of HARD_REGNO_MODE_OK.
7653 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
7654 (sparc_mode_class): Delete.
7655 (HARD_REGNO_MODE_OK): Delete.
7656 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7657 (hard_regno_mode_classes): Make static.
7658 (sparc_mode_class): Likewise.
7659 (sparc_hard_regno_mode_ok): New function.
7660 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
7661 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
7662 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
7664 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7665 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
7666 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
7667 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
7668 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
7669 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7670 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
7671 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
7672 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7673 (visium_hard_regno_mode_ok): New function.
7674 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
7675 instead of HARD_REGNO_MODE_OK.
7676 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
7677 (HARD_REGNO_MODE_OK): Delete.
7678 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
7679 (xtensa_hard_regno_mode_ok_p): ...this and make static.
7680 (xtensa_option_override): Update accordingly.
7681 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7682 (xtensa_hard_regno_mode_ok): New function.
7683 * system.h (HARD_REGNO_MODE_OK): Poison.
7685 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7686 Alan Hayward <alan.hayward@arm.com>
7687 David Sherwood <david.sherwood@arm.com>
7689 * target.def (hard_regno_call_part_clobbered): New hook.
7690 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
7691 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
7692 * doc/tm.texi: Regenerate.
7693 * hooks.h (hook_bool_uint_mode_false): Declare.
7694 * hooks.c (hook_bool_uint_mode_false): New function.
7695 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7696 * cselib.c (cselib_process_insn): Use
7697 targetm.hard_regno_call_part_clobbered instead of
7698 HARD_REGNO_CALL_PART_CLOBBERED.
7699 * ira-conflicts.c (ira_build_conflicts): Likewise.
7700 * ira-costs.c (ira_tune_allocno_costs): Likewise.
7701 * lra-constraints.c (need_for_call_save_p): Likewise.
7702 * lra-lives.c: Include target.h.
7703 (check_pseudos_live_through_calls): Use
7704 targetm.hard_regno_call_part_clobbered instead of
7705 HARD_REGNO_CALL_PART_CLOBBERED.
7706 * regcprop.c: Include target.h.
7707 (copyprop_hardreg_forward_1): Use
7708 targetm.hard_regno_call_part_clobbered instead of
7709 HARD_REGNO_CALL_PART_CLOBBERED.
7710 * reginfo.c (choose_hard_reg_mode): Likewise.
7711 * regrename.c (check_new_reg_p): Likewise.
7712 * reload.c (find_equiv_reg): Likewise.
7713 * reload1.c (emit_reload_insns): Likewise.
7714 * sched-deps.c (deps_analyze_insn): Likewise.
7715 * sel-sched.c (init_regs_for_mode): Likewise.
7716 (mark_unavailable_hard_regs): Likewise.
7717 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
7718 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7719 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
7721 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7722 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7723 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
7725 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
7727 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7728 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7729 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
7731 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7732 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7733 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
7735 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7736 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
7738 * config/powerpcspe/powerpcspe.c
7739 (rs6000_hard_regno_call_part_clobbered): New function.
7740 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7741 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7742 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
7744 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7745 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7746 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
7748 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7749 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7750 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
7752 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7753 Alan Hayward <alan.hayward@arm.com>
7754 David Sherwood <david.sherwood@arm.com>
7756 * rtl.h (subreg_memory_offset): Declare.
7757 * emit-rtl.c (subreg_memory_offset): New function.
7758 * expmed.c (store_bit_field_1): Use it.
7759 * expr.c (undefined_operand_subword_p): Likewise.
7760 * simplify-rtx.c (simplify_subreg): Likewise.
7762 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
7764 PR rtl-optimization/57448
7767 * optabs.c (expand_atomic_load): Place compiler memory barriers if
7768 using atomic_load pattern.
7769 (expand_atomic_store): Likewise.
7771 2017-09-04 Jakub Jelinek <jakub@redhat.com>
7774 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
7775 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
7776 handling. Use replace_call_with_value with NULL instead of
7777 gsi_replace, unlink_stmt_vdef and release_defs.
7779 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
7782 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
7784 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7787 * rtl.h (emit_library_call_value_1): Declare.
7788 (emit_library_call): Replace declaration with a series of overloads.
7789 Remove the parameter count argument.
7790 (emit_library_call_value): Likewise.
7791 * calls.c (emit_library_call_value_1): Make global. Replace varargs
7792 with an "rtx_mode_t *".
7793 (emit_library_call_value): Delete.
7794 (emit_library_call): Likewise.
7795 * asan.c (asan_emit_stack_protection): Update calls accordingly.
7796 (asan_emit_allocas_unpoison): Likewise.
7797 * builtins.c (expand_builtin_powi): Likewise.
7798 (expand_asan_emit_allocas_unpoison): Likewise.
7799 * cfgexpand.c (expand_main_function): Likewise.
7800 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
7801 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
7802 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
7803 * config/arm/arm.c (arm_trampoline_init): Likewise.
7804 (arm_call_tls_get_addr): Likewise.
7805 (arm_expand_divmod_libfunc): Likewise.
7806 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
7807 (smulsi3_highpart): Likewise.
7808 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
7809 (c6x_expand_compare): Likewise.
7810 (c6x_expand_movmem): Likewise.
7811 * config/frv/frv.c (frv_trampoline_init): Likewise.
7812 * config/i386/i386.c (ix86_trampoline_init): Likewise.
7813 (ix86_expand_divmod_libfunc): Likewise.
7814 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
7815 (ia64_expand_compare): Likewise.
7816 (ia64_profile_hook): Likewise.
7817 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
7818 (nonlocal_goto): Likewise.
7819 (restore_stack_nonlocal): Likewise.
7820 * config/m32r/m32r.c (block_move_call): Likewise.
7821 (m32r_trampoline_init): Likewise.
7822 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
7823 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
7824 (m68k_call_m68k_read_tp): Likewise.
7825 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
7826 (microblaze_expand_divide): Likewise.
7827 * config/mips/mips.h (mips_args): Likewise.
7828 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
7829 (MIPS_ICACHE_SYNC): Likewise.
7830 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
7831 (nios2_trampoline_init): Likewise.
7832 * config/pa/pa.c (hppa_tls_call): Likewise.
7833 (pa_trampoline_init): Likewise.
7834 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
7835 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
7836 (expand_strn_compare): Likewise.
7837 (rs6000_generate_compare): Likewise.
7838 (rs6000_expand_float128_convert): Likewise.
7839 (output_profile_hook): Likewise.
7840 (rs6000_trampoline_init): Likewise.
7841 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
7842 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
7843 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
7844 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
7845 (rs6000_generate_compare): Likewise.
7846 (rs6000_expand_float128_convert): Likewise.
7847 (output_profile_hook): Likewise.
7848 (rs6000_trampoline_init): Likewise.
7849 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
7850 * config/sh/sh.c (sh_trampoline_init): Likewise.
7851 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
7852 (sparc_emit_float_lib_cmp): Likewise.
7853 (sparc32_initialize_trampoline): Likewise.
7854 (sparc64_initialize_trampoline): Likewise.
7855 (sparc_profile_hook): Likewise.
7856 * config/spu/spu.c (ea_load_store): Likewise.
7857 * config/spu/spu.md (floatunssidf2): Likewise.
7858 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
7859 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
7860 * config/visium/visium.c (expand_block_move_4): Likewise.
7861 (expand_block_move_2): Likewise.
7862 (expand_block_move_1): Likewise.
7863 (expand_block_set_4): Likewise.
7864 (expand_block_set_2): Likewise.
7865 (expand_block_set_1): Likewise.
7866 (visium_trampoline_init): Likewise.
7867 (visium_profile_hook): Likewise.
7868 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
7869 (xtensa_setup_frame_addresses): Likewise.
7870 (xtensa_trampoline_init): Likewise.
7871 * except.c (sjlj_emit_function_enter): Likewise.
7872 (sjlj_emit_function_exit): Likewise.
7873 * explow.c (allocate_dynamic_stack_space): Likewise.
7874 (probe_stack_range): Likewise.
7875 * expr.c (convert_mode_scalar): Likewise.
7876 * optabs.c (expand_binop): Likewise.
7877 (expand_twoval_binop_libfunc): Likewise.
7878 (expand_unop): Likewise.
7879 (prepare_cmp_insn): Likewise.
7880 (prepare_float_lib_cmp): Likewise.
7881 (expand_float): Likewise.
7882 (expand_fix): Likewise.
7883 (expand_fixed_convert): Likewise.
7884 (maybe_emit_sync_lock_test_and_set): Likewise.
7885 (expand_atomic_compare_and_swap): Likewise.
7886 (expand_mem_thread_fence): Likewise.
7887 (expand_atomic_fetch_op): Likewise.
7889 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
7891 * doc/generic.texi (OpenACC): Adjust URL.
7892 * doc/invoke.texi (C Dialect Options): Ditto.
7894 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
7896 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
7897 predicate for operand 1. Add (m,<S>) constraint.
7898 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
7899 Prevent memory operand 1 with register operand 2.
7901 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
7903 PR rtl-optimization/82024
7904 * combine.c (try_combine): If the combination result is a PARALLEL,
7905 and we only need to retain the SET in there that would be placed
7906 at I2, check that we can place that at I3 instead, before doing so.
7908 2017-09-01 Jakub Jelinek <jakub@redhat.com>
7911 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
7913 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
7914 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
7917 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
7918 Jeff Law <law@redhat.com>
7920 * varasm.c (bss_initializer_p): Do not put constants into .bss
7921 (categorize_decl_for_section): Handle bss_initializer_p returning
7922 false when DECL_INITIAL is NULL.
7924 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7927 * config/s390/s390.c (s390_can_inline_p): New function.
7929 2017-09-01 Jeff Law <law@redhat.com>
7931 PR tree-optimization/82052
7932 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
7933 Always initialize the returned slot after a hash table miss
7934 when INSERT is true.
7936 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
7938 * config/s390/s390.md (mem_signal_fence): Remove.
7939 * doc/md.texi (mem_signal_fence): Remove.
7940 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
7942 * target-insns.def (mem_signal_fence): Remove.
7944 2017-09-01 Jakub Jelinek <jakub@redhat.com>
7947 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
7950 * asan.c (create_odr_indicator): Strip name encoding from assembler
7951 name before appending it after __odr_asan_.
7953 2017-09-01 Martin Liska <mliska@suse.cz>
7955 PR tree-optimization/82059
7956 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
7957 frequency only when an edge is redirected.
7959 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7961 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
7962 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
7963 (arc_conditional_register_usage): Remove ARC600 lp_count
7965 (arc_file_start): Emit Tag_ARC_CPU_variation.
7966 (arc_can_use_doloop_p): New conditions to use ZOLs.
7967 (hwloop_fail): New function.
7968 (hwloop_optimize): Likewise.
7969 (hwloop_pattern_reg): Likewise.
7970 (arc_doloop_hooks): New struct, to be used with reorg_loops.
7971 (arc_reorg_loops): New function, calls reorg_loops.
7972 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
7973 (arc600_corereg_hazard): Remove ZOL checking, case handled by
7975 (arc_loop_hazard): Remove function, functionality moved into
7977 (arc_hazard): Remove arc_loop_hazard call.
7978 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
7979 into hwloop_optimize.
7980 (arc_label_align): Remove ZOL handling.
7981 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
7982 * config/arc/arc.md (doloop_begin): Remove pattern.
7983 (doloop_begin_i): Likewise.
7984 (doloop_end_i): Likewise.
7985 (doloop_fallback): Likewise.
7986 (doloop_fallback_m): Likewise.
7987 (doloop_end): Reimplement expand.
7988 (arc_lp): New pattern for LP instruction.
7989 (loop_end): New pattern.
7990 (loop_fail): Likewise.
7991 (decrement_and_branch_until_zero): Likewise.
7992 * config/arc/arc.opt (mlpc-width): New option.
7993 * doc/invoke.texi (mlpc-width): Document option.
7995 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7997 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
7998 (arc_ccfsm_advance): Fix checking for delay slots.
7999 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
8001 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8003 * config/arc/arc.md (movqi_insn): Add stores to save constant long
8005 (movhi_insn): Update store instruction constraint which are saving
8006 6-bit short immediates.
8007 (movsi_insn): Consider also short scaled load operations.
8008 (zero_extendhisi2_i): Use Usd constraint instead of T.
8009 (extendhisi2_i): Add q constraint.
8010 (arc_clzsi2): Add type and length attributes.
8011 (arc_ctzsi2): Likewise.
8012 * config/arc/constraints.md (Usc): Update constraint, the
8013 assembler can parse two relocations for a single instruction.
8015 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8017 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
8018 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
8020 2017-08-31 Olivier Hainque <hainque@adacore.com>
8022 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
8023 match as powerpc-wrs-vxworks*.
8025 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
8027 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
8028 register constraint for by-element operand.
8029 (aarch64_mls_elt_merge<mode>): Likewise.
8031 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
8033 * config/arc/arc.c (arc_can_follow_jump): Check for short
8036 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
8038 * config.gcc: Use g.opt for arc.
8039 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
8040 functionality moved to ...
8041 (legitimate_scaled_address_p): New function, ...here.
8042 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
8043 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
8044 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
8046 (arc_override_options): Handle G option.
8047 (arc_output_pic_addr_const): Correct function definition.
8048 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
8049 (arc_decl_anon_ns_mem_p): Delete.
8050 (arc_in_small_data_p): Overhaul this function to take into
8051 consideration the value given via G option.
8052 (arc_rewrite_small_data_1): Renamed and corrected old
8053 arc_rewrite_small_data function.
8054 (arc_rewrite_small_data): New function.
8055 (small_data_pattern): Don't use pic_offset_table_rtx.
8056 * config/arc/arc.h (CC1_SPEC): Recognize G option.
8057 * config/arc/simdext.md (movmisalignv2hi): Use
8058 prepare_move_operands function.
8060 (movmisalign*): Likewise.
8061 * doc/invoke.texi (ARC options): Document -G option.
8063 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
8065 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
8067 * config/arc/arc.c (arc_print_operand): Output scalled address for
8068 sdata whenever is possible.
8069 (arc_in_small_data_p): Allow sdata for 64bit datum when double
8070 load/stores are available.
8071 (compact_sda_memory_operand): Check for the alignment required by
8072 code density instructions.
8073 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
8075 * config/arc/constraints.md (Usd): Update constraint.
8076 (Us0): New constraint.
8077 (Usc): Update constraint.
8079 2017-08-31 Richard Biener <rguenther@suse.de>
8082 * dwarf2out.c (dwarf2out_early_global_decl): Process each
8085 2017-08-31 Tamar Christina <tamar.christina@arm.com>
8087 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
8088 Resize type_signature.
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/aarch64.c (aarch64_base_register_rtx_p): Only allow
8095 subregs whose inner modes can be stored in GPRs.
8096 (aarch64_classify_index): Likewise.
8098 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8099 Alan Hayward <alan.hayward@arm.com>
8100 David Sherwood <david.sherwood@arm.com>
8102 * config/aarch64/iterators.md (V_cmp_result): Rename to...
8103 (V_INT_EQUIV): ...this.
8104 (v_cmp_result): Rename to...
8105 (v_int_equiv): ...this.
8106 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
8107 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
8108 (copysign<mode>3): Likewise.
8109 (aarch64_simd_bsl<mode>_internal): Likewise.
8110 (aarch64_simd_bsl<mode>): Likewise.
8111 (vec_cmp<mode><mode>): Likewise.
8112 (vcond<mode><mode>): Likewise.
8113 (vcond<v_cmp_mixed><mode>): Likewise.
8114 (vcondu<mode><v_cmp_mixed>): Likewise.
8115 (aarch64_cm<optab><mode>): Likewise.
8116 (aarch64_cmtst<mode>): Likewise.
8117 (aarch64_fac<optab><mode>): Likewise.
8118 (vec_perm_const<mode>): Likewise.
8119 (vcond_mask_<mode><v_cmp_result>): Rename to...
8120 (vcond_mask_<mode><v_int_equiv>): ...this.
8121 (vec_cmp<mode><v_cmp_result>): Rename to...
8122 (vec_cmp<mode><v_int_equiv>): ...this.
8124 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8125 Alan Hayward <alan.hayward@arm.com>
8126 David Sherwood <david.sherwood@arm.com>
8128 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
8130 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
8131 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
8132 (UNSPEC_LD4_DREG): New unspecs.
8133 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
8134 (aarch64_ld2<mode>_dreg_be): Replace with...
8135 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
8137 (aarch64_ld3<mode>_dreg_le)
8138 (aarch64_ld3<mode>_dreg_be): Replace with...
8139 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
8141 (aarch64_ld4<mode>_dreg_le)
8142 (aarch64_ld4<mode>_dreg_be): Replace with...
8143 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
8146 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8148 PR tree-optimization/81987
8149 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
8150 insert an initializer in a location not dominated by the stride
8153 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
8155 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
8156 on the entire header of the finally block in the fallthru case.
8158 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
8160 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
8162 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
8164 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
8165 rs6000_emit_move_from_cr and call renamed function.
8166 (rs6000_emit_prologue): Call renamed functions.
8167 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
8168 movesi_from_cr, remove volatile CRs.
8170 2017-08-30 Jon Beniston <jon@beniston.com>
8171 Richard Biener <rguenther@suse.de>
8173 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
8174 of VECTOR_MODE_P check.
8175 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
8176 element vector types.
8178 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8180 * df.h (df_read_modify_subreg_p): Remove in favor of...
8181 * rtl.h (read_modify_subreg_p): ...this new function. Take a
8182 const_rtx instead of an rtx.
8183 * cprop.c (local_cprop_find_used_regs): Update accordingly.
8184 * df-problems.c (df_word_lr_mark_ref): Likewise.
8185 * ira-lives.c (mark_pseudo_reg_live): Likewise.
8186 (mark_pseudo_reg_dead): Likewise.
8187 (mark_ref_dead): Likewise.
8188 * reginfo.c (init_subregs_of_mode): Likewise.
8189 * sched-deps.c (sched_analyze_1): Likewise.
8190 * df-scan.c (df_def_record_1): Likewise.
8191 (df_uses_record): Likewise.
8192 (df_read_modify_subreg_p): Remove in favor of...
8193 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
8194 const_rtx instead of an rtx.
8196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8197 Alan Hayward <alan.hayward@arm.com>
8198 David Sherwood <david.sherwood@arm.com>
8200 * rtl.h (partial_subreg_p): New function.
8201 * caller-save.c (save_call_clobbered_regs): Use it.
8202 * calls.c (expand_call): Likewise.
8203 * combine.c (combinable_i3pat): Likewise.
8204 (simplify_set): Likewise.
8205 (make_extraction): Likewise.
8206 (make_compound_operation_int): Likewise.
8207 (gen_lowpart_or_truncate): Likewise.
8208 (force_to_mode): Likewise.
8209 (make_field_assignment): Likewise.
8210 (reg_truncated_to_mode): Likewise.
8211 (record_truncated_value): Likewise.
8212 (move_deaths): Likewise.
8213 * cse.c (record_jump_cond): Likewise.
8214 (cse_insn): Likewise.
8215 * cselib.c (cselib_lookup_1): Likewise.
8216 * expmed.c (extract_bit_field_using_extv): Likewise.
8217 * function.c (assign_parm_setup_reg): Likewise.
8218 * ifcvt.c (noce_convert_multiple_sets): Likewise.
8219 * ira-build.c (create_insn_allocnos): Likewise.
8220 * lra-coalesce.c (merge_pseudos): Likewise.
8221 * lra-constraints.c (match_reload): Likewise.
8222 (simplify_operand_subreg): Likewise.
8223 (curr_insn_transform): Likewise.
8224 * lra-lives.c (process_bb_lives): Likewise.
8225 * lra.c (new_insn_reg): Likewise.
8226 (lra_substitute_pseudo): Likewise.
8227 * regcprop.c (mode_change_ok): Likewise.
8228 (maybe_mode_change): Likewise.
8229 (copyprop_hardreg_forward_1): Likewise.
8230 * reload.c (push_reload): Likewise.
8231 (find_reloads): Likewise.
8232 (find_reloads_subreg_address): Likewise.
8233 * reload1.c (alter_reg): Likewise.
8234 (eliminate_regs_1): Likewise.
8235 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8237 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
8239 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
8240 back to if statements, including unpack.
8242 2017-08-30 Martin Liska <mliska@suse.cz>
8245 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
8247 (func_checker::compare_asm_inputs_outputs): ... this function.
8248 (func_checker::compare_gimple_asm): Use the function to compare
8249 also ASM constrains.
8250 * ipa-icf-gimple.h: Rename the function.
8252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8253 Alan Hayward <alan.hayward@arm.com>
8254 David Sherwood <david.sherwood@arm.com>
8256 * coretypes.h (complex_mode): New type.
8257 * gdbhooks.py (build_pretty_printer): Handle it.
8258 * machmode.h (complex_mode): New class.
8259 (complex_mode::includes_p): New function.
8260 (is_complex_int_mode): Likewise.
8261 (is_complex_float_mode): Likewise.
8262 * genmodes.c (get_mode_class): Handle complex mode classes.
8263 * function.c (expand_function_end): Use is_complex_int_mode.
8265 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8266 Alan Hayward <alan.hayward@arm.com>
8267 David Sherwood <david.sherwood@arm.com>
8269 * coretypes.h (scalar_mode_pod): New typedef.
8270 * gdbhooks.py (build_pretty_printer): Handle it.
8271 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
8272 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
8273 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
8274 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
8277 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8278 Alan Hayward <alan.hayward@arm.com>
8279 David Sherwood <david.sherwood@arm.com>
8281 * machmode.h (mode_for_vector): Take a scalar_mode instead
8283 * stor-layout.c (mode_for_vector): Likewise.
8284 * explow.c (promote_mode): Use as_a <scalar_mode>.
8285 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
8287 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8288 Alan Hayward <alan.hayward@arm.com>
8289 David Sherwood <david.sherwood@arm.com>
8291 * target.def (preferred_simd_mode): Take a scalar_mode
8292 instead of a machine_mode.
8293 * targhooks.h (default_preferred_simd_mode): Likewise.
8294 * targhooks.c (default_preferred_simd_mode): Likewise.
8295 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
8296 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
8297 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
8298 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
8299 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
8300 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
8301 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
8302 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
8304 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
8305 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
8306 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
8307 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
8308 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
8309 * doc/tm.texi: Regenerate.
8310 * optabs-query.c (can_vec_mask_load_store_p): Return false for
8313 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8314 Alan Hayward <alan.hayward@arm.com>
8315 David Sherwood <david.sherwood@arm.com>
8317 * target.def (scalar_mode_supported_p): Take a scalar_mode
8318 instead of a machine_mode.
8319 * targhooks.h (default_scalar_mode_supported_p): Likewise.
8320 * targhooks.c (default_scalar_mode_supported_p): Likewise.
8321 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
8322 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
8323 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
8324 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
8325 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
8326 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
8327 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
8328 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
8329 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
8330 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
8331 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
8332 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
8334 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
8335 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
8336 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
8337 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
8338 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
8340 * doc/tm.texi: Regenerate.
8342 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8343 Alan Hayward <alan.hayward@arm.com>
8344 David Sherwood <david.sherwood@arm.com>
8346 * coretypes.h (opt_scalar_mode): New typedef.
8347 * gdbhooks.py (build_pretty_printers): Handle it.
8348 * machmode.h (mode_iterator::get_2xwider): Add overload for
8350 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
8352 * expr.c (convert_mode_scalar): Likewise.
8353 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8354 * optabs.c (expand_float): Likewise.
8355 (expand_fix): Likewise.
8356 * tree-vect-stmts.c (vectorizable_conversion): 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 * optabs.c (expand_float): Explicitly check for scalars before
8363 using a branching expansion.
8364 (expand_fix): Likewise.
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 * expr.c (convert_mode): Split scalar handling out into...
8371 (convert_mode_scalar): ...this new function. Treat the modes
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 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
8380 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
8382 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8383 Alan Hayward <alan.hayward@arm.com>
8384 David Sherwood <david.sherwood@arm.com>
8386 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
8387 rather than a machine_mode.
8388 (fixed_from_string): Likewise.
8389 (fixed_convert): Likewise.
8390 (fixed_convert_from_int): Likewise.
8391 (fixed_convert_from_real): Likewise.
8392 (real_convert_from_fixed): Likewise.
8393 * fixed-value.c (fixed_from_double_int): Likewise.
8394 (fixed_from_string): Likewise.
8395 (fixed_convert): Likewise.
8396 (fixed_convert_from_int): Likewise.
8397 (fixed_convert_from_real): Likewise.
8398 (real_convert_from_fixed): Likewise.
8399 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
8401 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8402 Alan Hayward <alan.hayward@arm.com>
8403 David Sherwood <david.sherwood@arm.com>
8405 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
8406 of separate mode class checks. Do not allow vector modes here.
8407 (immed_wide_int_const): Use as_a <scalar_mode>.
8408 * explow.c (trunc_int_for_mode): Likewise.
8409 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
8410 (wi::shwi): Likewise.
8411 (wi::min_value): Likewise.
8412 (wi::max_value): Likewise.
8413 * dwarf2out.c (loc_descriptor): Likewise.
8414 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
8417 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8418 Alan Hayward <alan.hayward@arm.com>
8419 David Sherwood <david.sherwood@arm.com>
8421 * tree.h (SCALAR_TYPE_MODE): New macro.
8422 * expr.c (expand_expr_addr_expr_1): Use it.
8423 (expand_expr_real_2): Likewise.
8424 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
8425 (fold_convert_const_fixed_from_int): Likewise.
8426 (fold_convert_const_fixed_from_real): Likewise.
8427 (native_encode_fixed): Likewise
8428 (native_encode_complex): Likewise
8429 (native_encode_vector): Likewise.
8430 (native_interpret_fixed): Likewise.
8431 (native_interpret_real): Likewise.
8432 (native_interpret_complex): Likewise.
8433 (native_interpret_vector): Likewise.
8434 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
8435 (simd_clone_adjust_argument_types): Likewise.
8436 (simd_clone_init_simd_arrays): Likewise.
8437 (simd_clone_adjust): Likewise.
8438 * stor-layout.c (layout_type): Likewise.
8439 * tree.c (build_minus_one_cst): Likewise.
8440 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
8441 * tree-inline.c (estimate_move_cost): Likewise.
8442 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
8443 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
8444 (vectorizable_reduction): Likewise.
8445 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
8446 (vect_recog_mixed_size_cond_pattern): Likewise.
8447 (check_bool_pattern): Likewise.
8448 (adjust_bool_pattern): Likewise.
8449 (search_type_for_mask_1): Likewise.
8450 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
8451 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8452 (vectorizable_load): Likewise.
8453 (vectorizable_store): Likewise.
8454 * ubsan.c (ubsan_encode_value): Likewise.
8455 * varasm.c (output_constant): Likewise.
8457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8458 Alan Hayward <alan.hayward@arm.com>
8459 David Sherwood <david.sherwood@arm.com>
8461 * coretypes.h (scalar_mode): New class.
8462 * machmode.h (scalar_mode): Likewise.
8463 (scalar_mode::includes_p): New function.
8464 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
8465 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
8466 * genmodes.c (get_mode_class): Handle remaining scalar modes.
8467 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
8468 * expmed.c (store_bit_field_1): Likewise.
8469 (extract_bit_field_1): Likewise.
8470 * expr.c (write_complex_part): Likewise.
8471 (read_complex_part): Likewise.
8472 (emit_move_complex_push): Likewise.
8473 (expand_expr_real_2): Likewise.
8474 * function.c (assign_parm_setup_reg): Likewise.
8475 (assign_parms_unsplit_complex): Likewise.
8476 * optabs.c (expand_binop): Likewise.
8477 * rtlanal.c (subreg_get_info): Likewise.
8478 * simplify-rtx.c (simplify_immed_subreg): Likewise.
8479 * varasm.c (output_constant_pool_2): Likewise.
8481 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8482 Alan Hayward <alan.hayward@arm.com>
8483 David Sherwood <david.sherwood@arm.com>
8485 * expmed.c (extract_high_half): Use scalar_int_mode and remove
8487 (expmed_mult_highpart_optab): Likewise.
8488 (expmed_mult_highpart): Likewise.
8490 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8491 Alan Hayward <alan.hayward@arm.com>
8492 David Sherwood <david.sherwood@arm.com>
8494 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
8495 instead of a machine_mode.
8496 (builtin_memset_read_str): Likewise.
8497 * builtins.c (c_readstr): Likewise.
8498 (builtin_memcpy_read_str): Likewise.
8499 (builtin_strncpy_read_str): Likewise.
8500 (builtin_memset_read_str): Likewise.
8501 (builtin_memset_gen_str): Likewise.
8502 (expand_builtin_signbit): Use scalar_int_mode for local variables.
8503 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
8504 instead of a machine_mode.
8505 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
8507 (make_extraction): Likewise.
8508 (try_widen_shift_mode): Take and return scalar_int_modes instead
8510 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
8511 a scalar_int_mode instead of a machine_mode.
8512 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
8513 (avr_addr_space_pointer_mode): Likewise.
8514 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
8515 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
8516 (msp430_unwind_word_mode): Likewise.
8517 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
8518 (spu_addr_space_pointer_mode): Likewise.
8519 (spu_addr_space_address_mode): Likewise.
8520 (spu_libgcc_cmp_return_mode): Likewise.
8521 (spu_libgcc_shift_count_mode): Likewise.
8522 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
8523 (rl78_addr_space_pointer_mode): Likewise.
8524 (fl78_unwind_word_mode): Likewise.
8525 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
8527 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
8528 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
8529 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
8530 (mips_valid_pointer_mode): Likewise.
8531 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
8532 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
8533 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
8535 (ft32_addr_space_address_mode): Likewise.
8536 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
8537 scalar_int_mode instead of a machine_mode.
8538 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
8540 (m32c_addr_space_address_mode): Likewise.
8541 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
8542 (rs6000_eh_return_filter_mode): Likewise.
8543 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
8544 (rs6000_eh_return_filter_mode): Likewise.
8545 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
8546 (s390_libgcc_shift_count_mode): Likewise.
8547 (s390_unwind_word_mode): Likewise.
8548 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
8550 * target.def (mode_rep_extended): Likewise.
8551 (valid_pointer_mode): Likewise.
8552 (addr_space.valid_pointer_mode): Likewise.
8553 (eh_return_filter_mode): Return a scalar_int_mode rather than
8555 (libgcc_cmp_return_mode): Likewise.
8556 (libgcc_shift_count_mode): Likewise.
8557 (unwind_word_mode): Likewise.
8558 (addr_space.pointer_mode): Likewise.
8559 (addr_space.address_mode): Likewise.
8560 * doc/tm.texi: Regenerate.
8561 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
8563 (do_jump): Use scalar_int_mode for local variables.
8564 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
8565 rather than a machine_mode.
8566 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
8567 (scompare_loc_descriptor_wide): Likewise.
8568 (scompare_loc_descriptor_narrow): Likewise.
8569 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
8571 * except.c (sjlj_emit_dispatch_table): Likewise.
8572 (expand_builtin_eh_copy_values): Likewise.
8573 * explow.c (convert_memory_address_addr_space_1): Likewise.
8574 Take a scalar_int_mode rather than a machine_mode.
8575 (convert_memory_address_addr_space): Take a scalar_int_mode rather
8576 than a machine_mode.
8577 (memory_address_addr_space): Use scalar_int_mode for local variables.
8578 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
8579 rather than a machine_mode.
8580 * expmed.c (mask_rtx): Likewise.
8581 (init_expmed_one_conv): Likewise.
8582 (expand_mult_highpart_adjust): Likewise.
8583 (extract_high_half): Likewise.
8584 (expmed_mult_highpart_optab): Likewise.
8585 (expmed_mult_highpart): Likewise.
8586 (expand_smod_pow2): Likewise.
8587 (expand_sdiv_pow2): Likewise.
8588 (emit_store_flag_int): Likewise.
8589 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
8591 (extract_low_bits): Likewise.
8592 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
8594 * expr.c (pieces_addr::adjust): Likewise.
8595 (can_store_by_pieces): Likewise.
8596 (store_by_pieces): Likewise.
8597 (clear_by_pieces_1): Likewise.
8598 (expand_expr_addr_expr_1): Likewise.
8599 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
8600 (expand_expr_real_1): Likewise.
8601 (try_casesi): Likewise.
8602 * final.c (shorten_branches): Likewise.
8603 * fold-const.c (fold_convert_const_int_from_fixed): Change the
8604 type of "mode" to machine_mode.
8605 * internal-fn.c (expand_arith_overflow_result_store): Take a
8606 scalar_int_mode rather than a machine_mode.
8607 (expand_mul_overflow): Use scalar_int_mode for local variables.
8608 * loop-doloop.c (doloop_modify): Likewise.
8609 (doloop_optimize): Likewise.
8610 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
8611 than a machine_mode.
8612 (expand_doubleword_shift_condmove): Likewise.
8613 (expand_doubleword_shift): Likewise.
8614 (expand_doubleword_clz): Likewise.
8615 (expand_doubleword_popcount): Likewise.
8616 (expand_doubleword_parity): Likewise.
8617 (expand_absneg_bit): Use scalar_int_mode for local variables.
8618 (prepare_float_lib_cmp): Likewise.
8619 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
8620 rather than a machine_mode.
8621 (convert_memory_address_addr_space): Likewise.
8622 (get_mode_bounds): Likewise.
8623 (get_address_mode): Return a scalar_int_mode rather than a
8625 * rtlanal.c (get_address_mode): Likewise.
8626 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
8627 than a machine_mode.
8628 * targhooks.c (default_mode_rep_extended): Likewise.
8629 (default_valid_pointer_mode): Likewise.
8630 (default_addr_space_valid_pointer_mode): Likewise.
8631 (default_eh_return_filter_mode): Return a scalar_int_mode rather
8632 than a machine_mode.
8633 (default_libgcc_cmp_return_mode): Likewise.
8634 (default_libgcc_shift_count_mode): Likewise.
8635 (default_unwind_word_mode): Likewise.
8636 (default_addr_space_pointer_mode): Likewise.
8637 (default_addr_space_address_mode): Likewise.
8638 * targhooks.h (default_eh_return_filter_mode): Likewise.
8639 (default_libgcc_cmp_return_mode): Likewise.
8640 (default_libgcc_shift_count_mode): Likewise.
8641 (default_unwind_word_mode): Likewise.
8642 (default_addr_space_pointer_mode): Likewise.
8643 (default_addr_space_address_mode): Likewise.
8644 (default_mode_rep_extended): Take a scalar_int_mode rather than
8646 (default_valid_pointer_mode): Likewise.
8647 (default_addr_space_valid_pointer_mode): Likewise.
8648 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
8650 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
8651 rather than a machine_mode.
8652 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
8653 for local variables.
8654 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8655 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
8656 than a machine_mode.
8658 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8659 Alan Hayward <alan.hayward@arm.com>
8660 David Sherwood <david.sherwood@arm.com>
8662 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
8663 the mode argument to scalar_int_mode.
8664 (do_jump_by_parts_zero_rtx): Likewise.
8665 (do_jump_by_parts_equality_rtx): Likewise.
8666 (do_jump_by_parts_greater): Take a mode argument.
8667 (do_jump_by_parts_equality): Likewise.
8668 (do_jump_1): Update calls accordingly.
8670 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8671 Alan Hayward <alan.hayward@arm.com>
8672 David Sherwood <david.sherwood@arm.com>
8674 * is-a.h (safe_dyn_cast): New function.
8675 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
8676 (jump_table_for_label): Likewise.
8677 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
8678 instead of an rtx_insn *.
8679 (shorten_branches): Use dyn_cast instead of LABEL_P and
8680 JUMP_TABLE_DATA_P. Use jump_table_for_label and
8681 rtx_jump_table_data::get_data_mode.
8682 (final_scan_insn): Likewise.
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 * combine.c (try_combine): Use is_a <scalar_int_mode> when
8689 trying to combine a full-register integer set with a subreg
8692 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8693 Alan Hayward <alan.hayward@arm.com>
8694 David Sherwood <david.sherwood@arm.com>
8696 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
8697 that is always either address_mode or pointer_mode.
8699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8700 Alan Hayward <alan.hayward@arm.com>
8701 David Sherwood <david.sherwood@arm.com>
8703 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
8704 when the two are known to be equal.
8706 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8707 Alan Hayward <alan.hayward@arm.com>
8708 David Sherwood <david.sherwood@arm.com>
8710 * simplify-rtx.c (simplify_const_unary_operation): Use
8711 is_a <scalar_int_mode> instead of checking for a nonzero
8712 precision. Forcibly convert op_mode to a scalar_int_mode
8713 in that case. More clearly differentiate the operand and
8714 result modes and use the former when deciding what the value
8715 of a count-bits operation should be. Use is_int_mode instead
8716 of checking for a MODE_INT. Remove redundant check for whether
8717 this mode has a zero precision.
8719 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8720 Alan Hayward <alan.hayward@arm.com>
8721 David Sherwood <david.sherwood@arm.com>
8723 * optabs.c (widen_leading): Change the type of the mode argument
8724 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
8725 (widen_bswap): Likewise.
8726 (expand_parity): Likewise.
8727 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
8728 (expand_ffs): Likewise.
8729 (epand_unop): Check for scalar integer modes before calling the
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 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
8738 (expand_expr_real_1): Update call accordingly.
8740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8741 Alan Hayward <alan.hayward@arm.com>
8742 David Sherwood <david.sherwood@arm.com>
8744 * expmed.c (store_bit_field_using_insv): Add op0_mode and
8745 value_mode arguments. Use scalar_int_mode internally.
8746 (store_bit_field_1): Rename the new integer mode from imode
8747 to op0_mode and use it instead of GET_MODE (op0). Update calls
8748 to store_split_bit_field, store_bit_field_using_insv and
8749 store_fixed_bit_field.
8750 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
8751 Use scalar_int_mode internally. Use a bit count rather than a mode
8752 when calculating the largest bit size for get_best_mode.
8753 Update calls to store_split_bit_field and store_fixed_bit_field_1.
8754 (store_fixed_bit_field_1): Add mode and value_mode arguments.
8755 Remove assertion that OP0 has a scalar integer mode.
8756 (store_split_bit_field): Add op0_mode and value_mode arguments.
8757 Update calls to extract_fixed_bit_field.
8758 (extract_bit_field_using_extv): Add an op0_mode argument.
8759 Use scalar_int_mode internally.
8760 (extract_bit_field_1): Rename the new integer mode from imode to
8761 op0_mode and use it instead of GET_MODE (op0). Update calls to
8762 extract_split_bit_field, extract_bit_field_using_extv and
8763 extract_fixed_bit_field.
8764 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
8765 to extract_split_bit_field and extract_fixed_bit_field_1.
8766 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
8767 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
8769 (extract_split_bit_field): Add an op0_mode argument. Update call
8770 to extract_fixed_bit_field.
8772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8773 Alan Hayward <alan.hayward@arm.com>
8774 David Sherwood <david.sherwood@arm.com>
8776 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
8777 * explow.c (hard_function_value): Likewise.
8778 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
8779 convert_to_mode call outside the loop.
8780 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
8781 for the mode iterator. Require the mode specified by max_pieces
8783 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
8785 (copy_blkmode_to_reg): Likewise.
8786 (set_storage_via_setmem): Likewise.
8787 * optabs.c (prepare_cmp_insn): Likewise.
8788 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8789 * stor-layout.c (finish_bitfield_representative): Likewise.
8791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8792 Alan Hayward <alan.hayward@arm.com>
8793 David Sherwood <david.sherwood@arm.com>
8795 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
8796 * expr.c (convert_move): Use them.
8797 (convert_modes): Likewise.
8798 (store_expr_with_bounds): Likewise.
8800 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8801 Alan Hayward <alan.hayward@arm.com>
8802 David Sherwood <david.sherwood@arm.com>
8804 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
8805 parameter for the mode of "x". Remove the "known_x", "known_mode"
8806 and "known_ret" arguments. Change the type of the mode argument
8808 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
8809 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
8810 (reg_num_sign_bit_copies_for_combine): Likewise.
8811 * rtlanal.c (nonzero_bits1): Likewise.
8812 (num_sign_bit_copies1): Likewise.
8813 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
8814 (reg_num_sign_bit_copies_general): Likewise.
8815 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
8816 (reg_nonzero_bits_general): Likewise.
8818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8819 Alan Hayward <alan.hayward@arm.com>
8820 David Sherwood <david.sherwood@arm.com>
8822 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
8823 than in subroutines. Return 1 for non-integer modes.
8824 (cached_num_sign_bit_copies): Change the type of the mode parameter
8826 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
8827 classes. Handle CONST_INT_P first and then check whether X also
8828 has a scalar integer mode. Check the same thing for inner registers
8829 of a SUBREG and for values that are being extended or truncated.
8831 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8832 Alan Hayward <alan.hayward@arm.com>
8833 David Sherwood <david.sherwood@arm.com>
8835 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
8836 in subroutines. Return the mode mask for non-integer modes.
8837 (cached_nonzero_bits): Change the type of the mode parameter
8839 (nonzero_bits1): Likewise. Remove early exit for other mode
8840 classes. Handle CONST_INT_P first and then check whether X
8841 also has a scalar integer mode.
8843 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8844 Alan Hayward <alan.hayward@arm.com>
8845 David Sherwood <david.sherwood@arm.com>
8847 * expr.c (widest_int_mode_for_size): Make the comment match the code.
8848 Return a scalar_int_mode and assert that the size is greater than
8850 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
8851 (op_by_pieces_d::op_by_pieces_d): Likewise.
8852 (op_by_pieces_d::run): Likewise.
8853 (can_store_by_pieces): Likewise.
8855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8856 Alan Hayward <alan.hayward@arm.com>
8857 David Sherwood <david.sherwood@arm.com>
8859 * combine.c (extract_left_shift): Add a mode argument and update
8861 (make_compound_operation_int): Change the type of the mode parameter
8862 to scalar_int_mode and update the call to extract_left_shift.
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_and_const_int): Change the type of the mode
8869 parameter to scalar_int_mode.
8870 (simplify_and_const_int_1): Likewise. Update recursive call.
8872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8873 Alan Hayward <alan.hayward@arm.com>
8874 David Sherwood <david.sherwood@arm.com>
8876 * combine.c (simplify_compare_const): Check that the mode is a
8877 scalar_int_mode (rather than VOIDmode) before testing its
8879 (simplify_comparison): Move COMPARISON_P handling out of the
8880 loop and restrict the latter part of the loop to scalar_int_modes.
8881 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
8882 and when considering SUBREG_REGs. Use is_int_mode instead of
8883 checking GET_MODE_CLASS against MODE_INT.
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 (try_widen_shift_mode): Move check for equal modes to...
8890 (simplify_shift_const_1): ...here. Use scalar_int_mode for
8891 shift_unit_mode and for modes involved in scalar shifts.
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 * combine.c (force_int_to_mode): New function, split out from...
8898 (force_to_mode): ...here. Keep xmode up-to-date and use it
8899 instead of GET_MODE (x).
8901 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8902 Alan Hayward <alan.hayward@arm.com>
8903 David Sherwood <david.sherwood@arm.com>
8905 * optabs-query.h (extraction_insn::struct_mode): Change type to
8906 opt_scalar_int_mode and update comment.
8907 (extraction_insn::field_mode): Change type to scalar_int_mode.
8908 (extraction_insn::pos_mode): Likewise.
8909 * combine.c (make_extraction): Update accordingly.
8910 * optabs-query.c (get_traditional_extraction_insn): Likewise.
8911 (get_optab_extraction_insn): Likewise.
8912 * recog.c (simplify_while_replacing): Likewise.
8913 * expmed.c (narrow_bit_field_mem): Change the type of the mode
8914 parameter to opt_scalar_int_mode.
8916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8917 Alan Hayward <alan.hayward@arm.com>
8918 David Sherwood <david.sherwood@arm.com>
8920 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
8921 to a scalar_int_mode instead of a machine_mode.
8922 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
8923 (get_best_mode): Return a boolean and use a pointer argument to store
8924 the selected mode. Replace the limit mode parameter with a bit limit.
8925 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
8926 for the values returned by bit_field_mode_iterator::next_mode.
8927 (store_bit_field): Update call to get_best_mode.
8928 (store_fixed_bit_field): Likewise.
8929 (extract_fixed_bit_field): Likewise.
8930 * expr.c (optimize_bitfield_assignment_op): Likewise.
8931 * fold-const.c (optimize_bit_field_compare): Likewise.
8932 (fold_truth_andor_1): Likewise.
8933 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
8934 Update for new type of m_mode.
8935 (get_best_mode): As above.
8937 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8938 Alan Hayward <alan.hayward@arm.com>
8939 David Sherwood <david.sherwood@arm.com>
8941 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
8942 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
8943 (store_bit_field): Check is_a <scalar_int_mode> before calling
8944 strict_volatile_bitfield_p.
8945 (extract_bit_field): Likewise.
8947 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8948 Alan Hayward <alan.hayward@arm.com>
8949 David Sherwood <david.sherwood@arm.com>
8951 * target.def (cstore_mode): Return a scalar_int_mode.
8952 * doc/tm.texi: Regenerate.
8953 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
8954 * targhooks.h (default_cstore_mode): Likewise.
8955 * targhooks.c (default_cstore_mode): Likewise, using a forced
8957 * expmed.c (emit_cstore): Expect the target of the cstore to be
8960 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8961 Alan Hayward <alan.hayward@arm.com>
8962 David Sherwood <david.sherwood@arm.com>
8964 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
8966 (niter_desc): Likewise mode.
8967 (iv_analyze): Add a mode parameter.
8969 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
8970 and change its type to scalar_int_mode.
8971 * loop-iv.c: Update commentary at head of file.
8972 (iv_constant): Pass the mode paraeter before the rtx it describes
8973 and change its type to scalar_int_mode. Remove VOIDmode handling.
8974 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
8975 (iv_extend): Likewise.
8976 (shorten_into_mode): Likewise.
8977 (iv_add): Use scalar_int_mode.
8978 (iv_mult): Likewise.
8979 (iv_shift): Likewise.
8980 (canonicalize_iv_subregs): Likewise.
8981 (get_biv_step_1): Pass the outer_mode parameter before the rtx
8982 it describes and change its mode to scalar_int_mode. Also change
8983 the type of the returned inner_mode to scalar_int_mode.
8984 (get_biv_step): Likewise, turning outer_mode from a pointer
8985 into a direct parameter. Update call to get_biv_step_1.
8986 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
8987 iv_constant and get_biv_step.
8988 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
8989 and change its type to scalar_int_mode. Don't initialise iv->mode
8990 to VOIDmode and remove later checks for its still being VOIDmode.
8991 Update calls to iv_analyze_op and iv_analyze_expr. Check
8992 is_a <scalar_int_mode> when changing the mode under consideration.
8993 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
8994 Update call to iv_analyze_expr.
8995 (iv_analyze_op): Add a mode parameter. Reject subregs whose
8996 inner register is not also a scalar_int_mode. Update call to
8998 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
8999 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
9000 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
9001 separate mode class checks. Update calls to iv_analyze. Remove
9002 fix-up of VOIDmodes after iv_analyze_biv.
9003 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
9004 don't have a scalar_int_mode. Update call to biv_p.
9006 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9007 Alan Hayward <alan.hayward@arm.com>
9008 David Sherwood <david.sherwood@arm.com>
9010 * cfgexpand.c (convert_debug_memory_address): Use
9011 as_a <scalar_int_mode>.
9012 * combine.c (expand_compound_operation): Likewise.
9013 (make_extraction): Likewise.
9014 (change_zero_ext): Likewise.
9015 (simplify_comparison): Likewise.
9016 * cse.c (cse_insn): Likewise.
9017 * dwarf2out.c (minmax_loc_descriptor): Likewise.
9018 (mem_loc_descriptor): Likewise.
9019 (loc_descriptor): Likewise.
9020 * expmed.c (init_expmed_one_mode): Likewise.
9021 (synth_mult): Likewise.
9022 (emit_store_flag_1): Likewise.
9023 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
9024 of a comparison with size.
9025 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
9026 (reduce_to_bit_field_precision): Likewise.
9027 * function.c (expand_function_end): Likewise.
9028 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
9029 * loop-doloop.c (doloop_modify): Likewise.
9030 * optabs.c (expand_binop): Likewise.
9031 (expand_unop): Likewise.
9032 (expand_copysign_absneg): Likewise.
9033 (prepare_cmp_insn): Likewise.
9034 (maybe_legitimize_operand): Likewise.
9035 * recog.c (const_scalar_int_operand): Likewise.
9036 * rtlanal.c (get_address_mode): Likewise.
9037 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9038 (simplify_cond_clz_ctz): Likewise.
9039 * tree-nested.c (get_nl_goto_field): Likewise.
9040 * tree.c (build_vector_type_for_mode): Likewise.
9041 * var-tracking.c (use_narrower_mode): Likewise.
9043 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9044 Alan Hayward <alan.hayward@arm.com>
9045 David Sherwood <david.sherwood@arm.com>
9047 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
9048 * builtins.c (expand_builtin_signbit): Use it.
9049 * cfgexpand.c (expand_debug_expr): Likewise.
9050 * dojump.c (do_jump): Likewise.
9051 (do_compare_and_jump): Likewise.
9052 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
9053 * expmed.c (make_tree): Likewise.
9054 * expr.c (expand_expr_real_2): Likewise.
9055 (expand_expr_real_1): Likewise.
9056 (try_casesi): Likewise.
9057 * fold-const-call.c (fold_const_call_ss): Likewise.
9058 * fold-const.c (unextend): Likewise.
9059 (extract_muldiv_1): Likewise.
9060 (fold_single_bit_test): Likewise.
9061 (native_encode_int): Likewise.
9062 (native_encode_string): Likewise.
9063 (native_interpret_int): Likewise.
9064 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
9065 * internal-fn.c (expand_addsub_overflow): Likewise.
9066 (expand_neg_overflow): Likewise.
9067 (expand_mul_overflow): Likewise.
9068 (expand_arith_overflow): Likewise.
9069 * match.pd: Likewise.
9070 * stor-layout.c (layout_type): Likewise.
9071 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
9072 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
9073 * tree-ssanames.c (get_range_info): Likewise.
9074 * tree-switch-conversion.c (array_value_type) Likewise.
9075 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
9076 (vect_recog_divmod_pattern): Likewise.
9077 (vect_recog_mixed_size_cond_pattern): Likewise.
9078 * tree-vrp.c (extract_range_basic): Likewise.
9079 (simplify_float_conversion_using_ranges): Likewise.
9080 * tree.c (int_fits_type_p): Likewise.
9081 * ubsan.c (instrument_bool_enum_load): Likewise.
9082 * varasm.c (mergeable_string_section): Likewise.
9083 (narrowing_initializer_constant_valid_p): Likewise.
9084 (output_constant): Likewise.
9086 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9087 Alan Hayward <alan.hayward@arm.com>
9088 David Sherwood <david.sherwood@arm.com>
9090 * machmode.h (NARROWEST_INT_MODE): New macro.
9091 * expr.c (alignment_for_piecewise_move): Use it instead of
9092 GET_CLASS_NARROWEST_MODE (MODE_INT).
9093 (push_block): Likewise.
9094 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
9096 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9098 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9099 Alan Hayward <alan.hayward@arm.com>
9100 David Sherwood <david.sherwood@arm.com>
9102 * postreload.c (move2add_valid_value_p): Change the type of the
9103 mode parameter to scalar_int_mode.
9104 (move2add_use_add2_insn): Add a mode parameter and use it instead
9106 (move2add_use_add3_insn): Likewise.
9107 (reload_cse_move2add): Update accordingly.
9109 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9110 Alan Hayward <alan.hayward@arm.com>
9111 David Sherwood <david.sherwood@arm.com>
9113 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
9115 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
9116 * optabs.c (expand_unop): Likewise.
9118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9119 Alan Hayward <alan.hayward@arm.com>
9120 David Sherwood <david.sherwood@arm.com>
9122 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
9123 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
9124 (popcount_loc_descriptor): Likewise.
9125 (bswap_loc_descriptor): Likewise.
9126 (rotate_loc_descriptor): Likewise.
9127 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
9128 calling the functions above.
9130 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9131 Alan Hayward <alan.hayward@arm.com>
9132 David Sherwood <david.sherwood@arm.com>
9134 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
9136 (try_combine): Likewise.
9137 (simplify_if_then_else): Likewise.
9138 * cse.c (cse_insn): Likewise.
9139 * dwarf2out.c (mem_loc_descriptor): Likewise.
9140 * emit-rtl.c (gen_lowpart_common): Likewise.
9141 * simplify-rtx.c (simplify_truncation): Likewise.
9142 (simplify_binary_operation_1): Likewise.
9143 (simplify_const_relational_operation): Likewise.
9144 (simplify_ternary_operation): Likewise.
9145 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): 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 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
9152 * reload.c (push_reload): Likewise.
9153 (find_reloads): Likewise.
9155 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9156 Alan Hayward <alan.hayward@arm.com>
9157 David Sherwood <david.sherwood@arm.com>
9159 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
9160 (make_compound_operation_int): Likewise.
9161 (change_zero_ext): Likewise.
9162 * expr.c (convert_move): Likewise.
9163 (convert_modes): Likewise.
9164 * fwprop.c (forward_propagate_subreg): Likewise.
9165 * loop-iv.c (get_biv_step_1): Likewise.
9166 * optabs.c (widen_operand): Likewise.
9167 * postreload.c (move2add_valid_value_p): Likewise.
9168 * recog.c (simplify_while_replacing): Likewise.
9169 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9170 (simplify_binary_operation_1): Likewise. Remove redundant
9171 mode equality check.
9173 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9174 Alan Hayward <alan.hayward@arm.com>
9175 David Sherwood <david.sherwood@arm.com>
9177 * combine.c (combine_simplify_rtx): Add checks for
9178 is_a <scalar_int_mode>.
9179 (simplify_if_then_else): Likewise.
9180 (make_field_assignment): Likewise.
9181 (simplify_comparison): Likewise.
9182 * ifcvt.c (noce_try_bitop): Likewise.
9183 * loop-invariant.c (canonicalize_address_mult): Likewise.
9184 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9186 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9187 Alan Hayward <alan.hayward@arm.com>
9188 David Sherwood <david.sherwood@arm.com>
9190 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
9191 is_a <scalar_int_mode> instead of != BLKmode.
9193 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9194 Alan Hayward <alan.hayward@arm.com>
9195 David Sherwood <david.sherwood@arm.com>
9197 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
9198 instead of != VOIDmode.
9199 * combine.c (if_then_else_cond): Likewise.
9200 (change_zero_ext): Likewise.
9201 * dwarf2out.c (mem_loc_descriptor): Likewise.
9202 (loc_descriptor): Likewise.
9203 * rtlanal.c (canonicalize_condition): Likewise.
9204 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
9206 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9207 Alan Hayward <alan.hayward@arm.com>
9208 David Sherwood <david.sherwood@arm.com>
9210 * simplify-rtx.c (simplify_binary_operation_1): Use
9211 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
9213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9214 Alan Hayward <alan.hayward@arm.com>
9215 David Sherwood <david.sherwood@arm.com>
9217 * wide-int.h (int_traits<unsigned char>) New class.
9218 (int_traits<unsigned short>) Likewise.
9219 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
9220 Use GET_MODE_UNIT_PRECISION and remove redundant test for
9222 * combine.c (set_nonzero_bits_and_sign_copies): Use
9223 is_a <scalar_int_mode>.
9224 (find_split_point): Likewise.
9225 (combine_simplify_rtx): Likewise.
9226 (simplify_logical): Likewise.
9227 (expand_compound_operation): Likewise.
9228 (expand_field_assignment): Likewise.
9229 (make_compound_operation): Likewise.
9230 (extended_count): Likewise.
9231 (change_zero_ext): Likewise.
9232 (simplify_comparison): Likewise.
9233 * dwarf2out.c (scompare_loc_descriptor): Likewise.
9234 (ucompare_loc_descriptor): Likewise.
9235 (minmax_loc_descriptor): Likewise.
9236 (mem_loc_descriptor): Likewise.
9237 (loc_descriptor): Likewise.
9238 * expmed.c (init_expmed_one_mode): Likewise.
9239 * lra-constraints.c (lra_constraint_offset): Likewise.
9240 * optabs.c (prepare_libcall_arg): Likewise.
9241 * postreload.c (move2add_note_store): Likewise.
9242 * reload.c (operands_match_p): Likewise.
9243 * rtl.h (load_extend_op): Likewise.
9244 * rtlhooks.c (gen_lowpart_general): Likewise.
9245 * simplify-rtx.c (simplify_truncation): Likewise.
9246 (simplify_unary_operation_1): Likewise.
9247 (simplify_binary_operation_1): Likewise.
9248 (simplify_const_binary_operation): Likewise.
9249 (simplify_const_relational_operation): Likewise.
9250 (simplify_subreg): Likewise.
9251 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9252 * var-tracking.c (adjust_mems): Likewise.
9253 (prepare_call_arguments): Likewise.
9255 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9256 Alan Hayward <alan.hayward@arm.com>
9257 David Sherwood <david.sherwood@arm.com>
9259 * machmode.h (is_int_mode): New fuction.
9260 * combine.c (find_split_point): Use it.
9261 (combine_simplify_rtx): Likewise.
9262 (simplify_if_then_else): Likewise.
9263 (simplify_set): Likewise.
9264 (simplify_shift_const_1): Likewise.
9265 (simplify_comparison): Likewise.
9266 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
9267 * cse.c (notreg_cost): Likewise.
9268 (cse_insn): Likewise.
9269 * cselib.c (cselib_lookup_1): Likewise.
9270 * dojump.c (do_jump_1): Likewise.
9271 (do_compare_rtx_and_jump): Likewise.
9272 * dse.c (get_call_args): Likewise.
9273 * dwarf2out.c (rtl_for_decl_init): Likewise.
9274 (native_encode_initializer): Likewise.
9275 * expmed.c (emit_store_flag_1): Likewise.
9276 (emit_store_flag): Likewise.
9277 * expr.c (convert_modes): Likewise.
9278 (store_field): Likewise.
9279 (expand_expr_real_1): Likewise.
9280 * fold-const.c (fold_read_from_constant_string): Likewise.
9281 * gimple-ssa-sprintf.c (get_format_string): Likewise.
9282 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
9283 * optabs.c (expand_binop): Likewise.
9284 (expand_unop): Likewise.
9285 (expand_abs_nojump): Likewise.
9286 (expand_one_cmpl_abs_nojump): Likewise.
9287 * simplify-rtx.c (mode_signbit_p): Likewise.
9288 (val_signbit_p): Likewise.
9289 (val_signbit_known_set_p): Likewise.
9290 (val_signbit_known_clear_p): Likewise.
9291 (simplify_relational_operation_1): Likewise.
9292 * tree.c (vector_type_mode): Likewise.
9294 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9295 Alan Hayward <alan.hayward@arm.com>
9296 David Sherwood <david.sherwood@arm.com>
9298 * machmode.h (smallest_mode_for_size): Fix formatting.
9299 (smallest_int_mode_for_size): New function.
9300 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
9301 instead of smallest_mode_for_size.
9302 * combine.c (make_extraction): Likewise.
9303 * config/arc/arc.c (arc_expand_movmem): Likewise.
9304 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
9305 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
9306 * config/s390/s390.c (s390_expand_insv): Likewise.
9307 * config/sparc/sparc.c (assign_int_registers): Likewise.
9308 * config/spu/spu.c (spu_function_value): Likewise.
9309 (spu_function_arg): Likewise.
9310 * coverage.c (get_gcov_type): Likewise.
9311 (get_gcov_unsigned_t): Likewise.
9312 * dse.c (find_shift_sequence): Likewise.
9313 * expmed.c (store_bit_field_1): Likewise.
9314 * expr.c (convert_move): Likewise.
9315 (store_field): Likewise.
9316 * internal-fn.c (expand_arith_overflow): Likewise.
9317 * optabs-query.c (get_best_extraction_insn): Likewise.
9318 * optabs.c (expand_twoval_binop_libfunc): Likewise.
9319 * stor-layout.c (layout_type): Likewise.
9320 (initialize_sizetypes): Likewise.
9321 * targhooks.c (default_get_mask_mode): Likewise.
9322 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
9324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9325 Alan Hayward <alan.hayward@arm.com>
9326 David Sherwood <david.sherwood@arm.com>
9328 * machmode.h (opt_mode::else_blk): New function.
9329 (int_mode_for_mode): Declare.
9330 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
9331 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
9333 * cfgexpand.c (expand_debug_expr): Likewise.
9334 * combine.c (gen_lowpart_or_truncate): Likewise.
9335 (gen_lowpart_for_combine): Likewise.
9336 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
9337 * config/avr/avr.c (avr_to_int_mode): Likewise.
9338 (avr_out_plus_1): Likewise.
9339 (avr_out_plus): Likewise.
9340 (avr_out_round): Likewise.
9341 * config/i386/i386.c (ix86_split_to_parts): Likewise.
9342 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
9343 (s390_expand_vcond): Likewise.
9344 * config/spu/spu.c (spu_split_immediate): Likewise.
9345 (spu_expand_mov): Likewise.
9346 * dse.c (get_stored_val): Likewise.
9347 * expmed.c (store_bit_field_1): Likewise.
9348 (convert_extracted_bit_field): Use int_mode_for_mode instead of
9350 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
9351 (extract_low_bits): Likewise.
9352 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
9353 handling rather than repeating the check.
9354 (emit_group_store): Likewise.
9355 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
9356 * optabs.c (expand_absneg_bit): Likewise.
9357 (expand_copysign_absneg): Likewise.
9358 (expand_copysign_bit): Likewise.
9359 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
9360 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
9361 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
9362 * var-tracking.c (prepare_call_arguments): Likewise.
9363 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
9364 int_mode_for_mode instead of mode_for_size.
9365 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
9367 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9368 Alan Hayward <alan.hayward@arm.com>
9369 David Sherwood <david.sherwood@arm.com>
9371 * machmode.h (int_mode_for_size): New function.
9372 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
9373 instead of mode_for_size.
9374 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
9376 * combine.c (expand_field_assignment): Use int_mode_for_size
9377 instead of mode_for_size.
9378 (make_extraction): Likewise.
9379 (simplify_shift_const_1): Likewise.
9380 (simplify_comparison): Likewise.
9381 * dojump.c (do_jump): Likewise.
9382 * dwarf2out.c (mem_loc_descriptor): Likewise.
9383 * emit-rtl.c (init_derived_machine_modes): Likewise.
9384 * expmed.c (flip_storage_order): Likewise.
9385 (convert_extracted_bit_field): Likewise.
9386 * expr.c (copy_blkmode_from_reg): Likewise.
9387 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
9388 * internal-fn.c (expand_mul_overflow): Likewise.
9389 * lower-subreg.c (simple_move): Likewise.
9390 * optabs-libfuncs.c (init_optabs): Likewise.
9391 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9392 * tree.c (vector_type_mode): Likewise.
9393 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
9394 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
9395 * tree-vect-generic.c (expand_vector_parallel): Likewise.
9396 * tree-vect-stmts.c (vectorizable_load): Likewise.
9397 (vectorizable_store): Likewise.
9399 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9400 Alan Hayward <alan.hayward@arm.com>
9401 David Sherwood <david.sherwood@arm.com>
9403 * coretypes.h (pod_mode): New type.
9404 (scalar_int_mode_pod): New typedef.
9405 * machmode.h (pod_mode): New class.
9406 (int_n_data_t::m): Change type to scalar_int_mode_pod.
9407 * genmodes.c (emit_mode_int_n): Update accordingly.
9408 * lower-subreg.h (target_lower_subreg): Change type to
9409 scalar_int_mode_pod.
9410 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
9411 scalar_int_mode_pod.
9413 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9414 Alan Hayward <alan.hayward@arm.com>
9415 David Sherwood <david.sherwood@arm.com>
9417 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
9418 machine_mode to scalar_int_mode.
9419 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
9420 (rs6000_option_override_internal): Remove cast to int.
9421 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
9422 machine_mode to scalar_int_mode.
9423 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
9424 (rs6000_option_override_internal): Remove cast to int.
9425 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
9426 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
9428 * config/s390/s390.c (s390_expand_builtin): Likewise.
9429 * coretypes.h (scalar_int_mode): New type.
9430 (opt_scalar_int_mode): New typedef.
9431 * machmode.h (scalar_int_mode): New class.
9432 (scalar_int_mode::includes_p): New function.
9433 (byte_mode): Change type to scalar_int_mode.
9434 (word_mode): Likewise.
9435 (ptr_mode): Likewise.
9436 * emit-rtl.c (byte_mode): Likewise.
9437 (word_mode): Likewise.
9438 (ptr_mode): Likewise.
9439 (init_derived_machine_modes): Update accordingly.
9440 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
9441 and MODE_PARTIAL_INT.
9442 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
9443 opt_scalar_int_mode.
9445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9446 Alan Hayward <alan.hayward@arm.com>
9447 David Sherwood <david.sherwood@arm.com>
9449 * target.def (libgcc_floating_mode_supported_p): Take a
9451 * doc/tm.texi: Regenerate.
9452 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
9454 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
9455 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
9458 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9459 Alan Hayward <alan.hayward@arm.com>
9460 David Sherwood <david.sherwood@arm.com>
9462 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
9463 * doc/tm.texi: Regenerate.
9464 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
9465 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
9466 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
9467 * targhooks.h (default_floatn_mode): Likewise.
9468 * targhooks.c (default_floatn_mode): Likewise.
9469 * tree.c (build_common_tree_nodes): Update accordingly.
9471 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9472 Alan Hayward <alan.hayward@arm.com>
9473 David Sherwood <david.sherwood@arm.com>
9475 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
9476 (mode_iterator::iterate_p): Likewise.
9477 (mode_iterator::get_wider): Likewise.
9478 * expr.c (init_expr_target): Use opt_scalar_float_mode.
9480 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9481 Alan Hayward <alan.hayward@arm.com>
9482 David Sherwood <david.sherwood@arm.com>
9484 * coretypes.h (opt_scalar_float_mode): New typedef.
9485 * machmode.h (float_mode_for_size): New function.
9486 * emit-rtl.c (double_mode): Delete.
9487 (init_emit_once): Use float_mode_for_size.
9488 * stor-layout.c (layout_type): Likewise.
9489 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
9491 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9492 Alan Hayward <alan.hayward@arm.com>
9493 David Sherwood <david.sherwood@arm.com>
9495 * output.h (assemble_real): Take a scalar_float_mode.
9496 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
9497 * config/arm/arm.md (consttable_4): Likewise.
9498 (consttable_8): Likewise.
9499 (consttable_16): Likewise.
9500 * config/mips/mips.md (consttable_float): Likewise.
9501 * config/s390/s390.c (s390_output_pool_entry): Likewise.
9502 * varasm.c (assemble_real): Take a scalar_float_mode.
9503 (output_constant_pool_2): Update accordingly.
9504 (output_constant): Likewise.
9506 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9507 Alan Hayward <alan.hayward@arm.com>
9508 David Sherwood <david.sherwood@arm.com>
9510 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
9511 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
9512 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
9513 (native_encode_real): Likewise.
9514 (native_interpret_real): Likewise.
9515 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
9516 * tree-vrp.c (simplify_float_conversion_using_ranges): 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 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
9523 <scalar_float_mode>. Simplify.
9524 (gen_extend_conv_libfunc): Likewise.
9526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9527 Alan Hayward <alan.hayward@arm.com>
9528 David Sherwood <david.sherwood@arm.com>
9530 * coretypes.h (scalar_float_mode): New type.
9531 * machmode.h (mode_traits::from_int): Use machine_mode if
9532 USE_ENUM_MODES is defined.
9533 (is_a): New function.
9535 (dyn_cast): Likewise.
9536 (scalar_float_mode): New class.
9537 (scalar_float_mode::includes_p): New function.
9538 (is_float_mode): Likewise.
9539 * gdbhooks.py (MachineModePrinter): New class.
9540 (build_pretty_printer): Use it for scalar_float_mode.
9541 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
9542 (format_helper::format_helper): Turn into a template.
9543 * genmodes.c (get_mode_class): New function.
9544 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
9545 or machine_mode if none.
9546 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
9547 as_a <scalar_float_mode>.
9548 * dwarf2out.c (mem_loc_descriptor): Likewise.
9549 (insert_float): Likewise.
9550 (add_const_value_attribute): Likewise.
9551 * simplify-rtx.c (simplify_immed_subreg): Likewise.
9552 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
9553 (expand_unop): Update accordingly.
9554 (expand_abs_nojump): Likewise.
9555 (expand_copysign_absneg): Take a scalar_float_mode.
9556 (expand_copysign_bit): Likewise.
9557 (expand_copysign): Update accordingly.
9559 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9560 Alan Hayward <alan.hayward@arm.com>
9561 David Sherwood <david.sherwood@arm.com>
9563 * coretypes.h (opt_mode): New class.
9564 * machmode.h (opt_mode): Likewise.
9565 (opt_mode::else_void): New function.
9566 (opt_mode::require): Likewise.
9567 (opt_mode::exists): Likewise.
9568 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
9569 (GET_MODE_2XWIDER_MODE): Likewise.
9570 (mode_iterator::get_wider): Update accordingly.
9571 (mode_iterator::get_2xwider): Likewise.
9572 (mode_iterator::get_known_wider): Likewise, turning into a template.
9573 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
9574 forcing a wider mode to exist.
9575 * config/cr16/cr16.h (LONG_REG_P): Likewise.
9576 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9577 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
9578 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
9579 * lower-subreg.c (init_lower_subreg): Likewise.
9580 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
9581 on the final iteration.
9582 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
9583 a wider mode exists before asking for a move pattern.
9584 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
9585 forcing a wider mode to exist.
9586 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
9587 returning false if no such mode exists.
9588 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
9589 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
9590 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
9591 Avoid checking for a MODE_INT if we already know the mode is not a
9593 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
9594 forcing a wider mode to exist.
9595 (expmed_mult_highpart_optab): Likewise.
9596 (expmed_mult_highpart): Likewise.
9597 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
9599 * lto-streamer-in.c (lto_input_mode_table): Likewise.
9600 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
9601 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9602 * internal-fn.c (expand_mul_overflow): Update use of
9603 GET_MODE_2XWIDER_MODE.
9604 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9605 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
9606 GET_MODE_WIDER_MODE.
9607 (convert_plusminus_to_widen): Likewise.
9608 * tree-switch-conversion.c (array_value_type): Likewise.
9609 * var-tracking.c (emit_note_insn_var_location): Likewise.
9610 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9611 Return false inside rather than outside the loop if no wider mode
9613 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
9614 and GET_MODE_2XWIDER_MODE
9615 (can_compare_p): Use else_void.
9616 * gdbhooks.py (OptMachineModePrinter): New class.
9617 (build_pretty_printer): Use it for opt_mode.
9619 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9620 Alan Hayward <alan.hayward@arm.com>
9621 David Sherwood <david.sherwood@arm.com>
9623 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
9624 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
9626 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9627 Alan Hayward <alan.hayward@arm.com>
9628 David Sherwood <david.sherwood@arm.com>
9630 * machmode.h (mode_traits): New structure.
9631 (get_narrowest_mode): New function.
9632 (mode_iterator::start): Likewise.
9633 (mode_iterator::iterate_p): Likewise.
9634 (mode_iterator::get_wider): Likewise.
9635 (mode_iterator::get_known_wider): Likewise.
9636 (mode_iterator::get_2xwider): Likewise.
9637 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
9638 (FOR_EACH_MODE): Likewise.
9639 (FOR_EACH_MODE_FROM): Likewise.
9640 (FOR_EACH_MODE_UNTIL): Likewise.
9641 (FOR_EACH_WIDER_MODE): Likewise.
9642 (FOR_EACH_2XWIDER_MODE): Likewise.
9643 * builtins.c (expand_builtin_strlen): Use new mode iterators.
9644 * combine.c (simplify_comparison): Likewise
9645 * config/i386/i386.c (type_natural_mode): Likewise.
9646 * cse.c (cse_insn): Likewise.
9647 * dse.c (find_shift_sequence): Likewise.
9648 * emit-rtl.c (init_derived_machine_modes): Likewise.
9649 (init_emit_once): Likewise.
9650 * explow.c (hard_function_value): Likewise.
9651 * expmed.c (extract_fixed_bit_field_1): Likewise.
9652 (extract_bit_field_1): Likewise.
9653 (expand_divmod): Likewise.
9654 (emit_store_flag_1): Likewise.
9655 * expr.c (init_expr_target): Likewise.
9656 (convert_move): Likewise.
9657 (alignment_for_piecewise_move): Likewise.
9658 (widest_int_mode_for_size): Likewise.
9659 (emit_block_move_via_movmem): Likewise.
9660 (copy_blkmode_to_reg): Likewise.
9661 (set_storage_via_setmem): Likewise.
9662 (compress_float_constant): Likewise.
9663 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9664 * optabs-query.c (get_best_extraction_insn): Likewise.
9665 * optabs.c (expand_binop): Likewise.
9666 (expand_twoval_unop): Likewise.
9667 (expand_twoval_binop): Likewise.
9668 (widen_leading): Likewise.
9669 (widen_bswap): Likewise.
9670 (expand_parity): Likewise.
9671 (expand_unop): Likewise.
9672 (prepare_cmp_insn): Likewise.
9673 (prepare_float_lib_cmp): Likewise.
9674 (expand_float): Likewise.
9675 (expand_fix): Likewise.
9676 (expand_sfix_optab): Likewise.
9677 * postreload.c (move2add_use_add2_insn): Likewise.
9678 * reg-stack.c (reg_to_stack): Likewise.
9679 * reginfo.c (choose_hard_reg_mode): Likewise.
9680 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9681 * stor-layout.c (mode_for_size): Likewise.
9682 (smallest_mode_for_size): Likewise.
9683 (mode_for_vector): Likewise.
9684 (finish_bitfield_representative): Likewise.
9685 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
9686 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
9687 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9688 * var-tracking.c (prepare_call_arguments): Likewise.
9690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9691 Alan Hayward <alan.hayward@arm.com>
9692 David Sherwood <david.sherwood@arm.com>
9694 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
9695 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
9696 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
9697 * machmode.h (mode_size): Move earlier in file.
9698 (mode_precision): Likewise.
9699 (mode_inner): Likewise.
9700 (mode_nunits): Likewise.
9701 (mode_unit_size): Likewise.
9702 (unit_unit_precision): Likewise.
9703 (mode_wider): Likewise.
9704 (mode_2xwider): Likewise.
9705 (machine_mode): New class.
9706 (mode_to_bytes): New function.
9707 (mode_to_bits): Likewise.
9708 (mode_to_precision): Likewise.
9709 (mode_to_inner): Likewise.
9710 (mode_to_unit_size): Likewise.
9711 (mode_to_unit_precision): Likewise.
9712 (mode_to_nunits): Likewise.
9713 (GET_MODE_SIZE): Use mode_to_bytes.
9714 (GET_MODE_BITSIZE): Use mode_to_bits.
9715 (GET_MODE_PRECISION): Use mode_to_precision.
9716 (GET_MODE_INNER): Use mode_to_inner.
9717 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
9718 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
9719 (GET_MODE_NUNITS): Use mode_to_nunits.
9720 * system.h (ALWAYS_INLINE): New macro.
9721 * config/powerpcspe/powerpcspe-c.c
9722 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
9723 int for arg1_mode and arg2_mode.
9725 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9726 Alan Hayward <alan.hayward@arm.com>
9727 David Sherwood <david.sherwood@arm.com>
9729 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
9730 Prefix mode names with E_ in case statements.
9731 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9732 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
9733 (aarch64_split_simd_move): Likewise.
9734 (aarch64_gen_storewb_pair): Likewise.
9735 (aarch64_gen_loadwb_pair): Likewise.
9736 (aarch64_gen_store_pair): Likewise.
9737 (aarch64_gen_load_pair): Likewise.
9738 (aarch64_get_condition_code_1): Likewise.
9739 (aarch64_constant_pool_reload_icode): Likewise.
9740 (get_rsqrte_type): Likewise.
9741 (get_rsqrts_type): Likewise.
9742 (get_recpe_type): Likewise.
9743 (get_recps_type): Likewise.
9744 (aarch64_gimplify_va_arg_expr): Likewise.
9745 (aarch64_simd_container_mode): Likewise.
9746 (aarch64_emit_load_exclusive): Likewise.
9747 (aarch64_emit_store_exclusive): Likewise.
9748 (aarch64_expand_compare_and_swap): Likewise.
9749 (aarch64_gen_atomic_cas): Likewise.
9750 (aarch64_emit_bic): Likewise.
9751 (aarch64_emit_atomic_swap): Likewise.
9752 (aarch64_emit_atomic_load_op): Likewise.
9753 (aarch64_evpc_trn): Likewise.
9754 (aarch64_evpc_uzp): Likewise.
9755 (aarch64_evpc_zip): Likewise.
9756 (aarch64_evpc_ext): Likewise.
9757 (aarch64_evpc_rev): Likewise.
9758 (aarch64_evpc_dup): Likewise.
9759 (aarch64_gen_ccmp_first): Likewise.
9760 (aarch64_gen_ccmp_next): Likewise.
9761 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
9762 (alpha_emit_xfloating_libcall): Likewise.
9763 (emit_insxl): Likewise.
9764 (alpha_arg_type): Likewise.
9765 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
9766 (arc_preferred_simd_mode): Likewise.
9767 (arc_secondary_reload): Likewise.
9768 (get_arc_condition_code): Likewise.
9769 (arc_print_operand): Likewise.
9770 (arc_legitimate_constant_p): Likewise.
9771 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9772 * config/arc/arc.md (casesi_load): Likewise.
9773 (casesi_compact_jump): Likewise.
9774 * config/arc/predicates.md (proper_comparison_operator): Likewise.
9775 (cc_use_register): Likewise.
9776 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9777 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
9778 (arm_init_iwmmxt_builtins): Likewise.
9779 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
9780 (neon_expand_vector_init): Likewise.
9781 (arm_attr_length_move_neon): Likewise.
9782 (maybe_get_arm_condition_code): Likewise.
9783 (arm_emit_vector_const): Likewise.
9784 (arm_preferred_simd_mode): Likewise.
9785 (arm_output_iwmmxt_tinsr): Likewise.
9786 (thumb1_output_casesi): Likewise.
9787 (thumb2_output_casesi): Likewise.
9788 (arm_emit_load_exclusive): Likewise.
9789 (arm_emit_store_exclusive): Likewise.
9790 (arm_expand_compare_and_swap): Likewise.
9791 (arm_evpc_neon_vuzp): Likewise.
9792 (arm_evpc_neon_vzip): Likewise.
9793 (arm_evpc_neon_vrev): Likewise.
9794 (arm_evpc_neon_vtrn): Likewise.
9795 (arm_evpc_neon_vext): Likewise.
9796 (arm_validize_comparison): Likewise.
9797 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
9798 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
9799 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
9800 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
9801 (c6x_preferred_simd_mode): Likewise.
9802 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
9803 (epiphany_rtx_costs): Likewise.
9804 * config/epiphany/predicates.md (proper_comparison_operator):
9806 * config/frv/frv.c (condexec_memory_operand): Likewise.
9807 (frv_emit_move): Likewise.
9808 (output_move_single): Likewise.
9809 (output_condmove_single): Likewise.
9810 (frv_hard_regno_mode_ok): Likewise.
9811 (frv_matching_accg_mode): Likewise.
9812 * config/h8300/h8300.c (split_adds_subs): Likewise.
9813 (h8300_rtx_costs): Likewise.
9814 (h8300_print_operand): Likewise.
9815 (compute_mov_length): Likewise.
9816 (output_logical_op): Likewise.
9817 (compute_logical_op_length): Likewise.
9818 (compute_logical_op_cc): Likewise.
9819 (h8300_shift_needs_scratch_p): Likewise.
9820 (output_a_shift): Likewise.
9821 (compute_a_shift_length): Likewise.
9822 (compute_a_shift_cc): Likewise.
9823 (expand_a_rotate): Likewise.
9824 (output_a_rotate): Likewise.
9825 * config/i386/i386.c (classify_argument): Likewise.
9826 (function_arg_advance_32): Likewise.
9827 (function_arg_32): Likewise.
9828 (function_arg_64): Likewise.
9829 (function_value_64): Likewise.
9830 (ix86_gimplify_va_arg): Likewise.
9831 (ix86_legitimate_constant_p): Likewise.
9832 (put_condition_code): Likewise.
9833 (split_double_mode): Likewise.
9834 (ix86_avx256_split_vector_move_misalign): Likewise.
9835 (ix86_expand_vector_logical_operator): Likewise.
9836 (ix86_split_idivmod): Likewise.
9837 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
9838 (ix86_build_const_vector): Likewise.
9839 (ix86_build_signbit_mask): Likewise.
9840 (ix86_match_ccmode): Likewise.
9841 (ix86_cc_modes_compatible): Likewise.
9842 (ix86_expand_branch): Likewise.
9843 (ix86_expand_sse_cmp): Likewise.
9844 (ix86_expand_sse_movcc): Likewise.
9845 (ix86_expand_int_sse_cmp): Likewise.
9846 (ix86_expand_vec_perm_vpermi2): Likewise.
9847 (ix86_expand_vec_perm): Likewise.
9848 (ix86_expand_sse_unpack): Likewise.
9849 (ix86_expand_int_addcc): Likewise.
9850 (ix86_split_to_parts): Likewise.
9851 (ix86_vectorize_builtin_gather): Likewise.
9852 (ix86_vectorize_builtin_scatter): Likewise.
9853 (avx_vpermilp_parallel): Likewise.
9854 (inline_memory_move_cost): Likewise.
9855 (ix86_tieable_integer_mode_p): Likewise.
9856 (x86_maybe_negate_const_int): Likewise.
9857 (ix86_expand_vector_init_duplicate): Likewise.
9858 (ix86_expand_vector_init_one_nonzero): Likewise.
9859 (ix86_expand_vector_init_one_var): Likewise.
9860 (ix86_expand_vector_init_concat): Likewise.
9861 (ix86_expand_vector_init_interleave): Likewise.
9862 (ix86_expand_vector_init_general): Likewise.
9863 (ix86_expand_vector_set): Likewise.
9864 (ix86_expand_vector_extract): Likewise.
9865 (emit_reduc_half): Likewise.
9866 (ix86_emit_i387_round): Likewise.
9867 (ix86_mangle_type): Likewise.
9868 (ix86_expand_round_sse4): Likewise.
9869 (expand_vec_perm_blend): Likewise.
9870 (canonicalize_vector_int_perm): Likewise.
9871 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
9872 (expand_vec_perm_1): Likewise.
9873 (expand_vec_perm_interleave3): Likewise.
9874 (expand_vec_perm_even_odd_pack): Likewise.
9875 (expand_vec_perm_even_odd_1): Likewise.
9876 (expand_vec_perm_broadcast_1): Likewise.
9877 (ix86_vectorize_vec_perm_const_ok): Likewise.
9878 (ix86_expand_vecop_qihi): Likewise.
9879 (ix86_expand_mul_widen_hilo): Likewise.
9880 (ix86_expand_sse2_abs): Likewise.
9881 (ix86_expand_pextr): Likewise.
9882 (ix86_expand_pinsr): Likewise.
9883 (ix86_preferred_simd_mode): Likewise.
9884 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
9885 * config/i386/sse.md (*andnot<mode>3): Likewise.
9886 (<mask_codefor><code><mode>3<mask_name>): Likewise.
9887 (*<code><mode>3): Likewise.
9888 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
9889 (ia64_expand_atomic_op): Likewise.
9890 (ia64_arg_type): Likewise.
9891 (ia64_mode_to_int): Likewise.
9892 (ia64_scalar_mode_supported_p): Likewise.
9893 (ia64_vector_mode_supported_p): Likewise.
9894 (expand_vec_perm_broadcast): Likewise.
9895 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
9896 (iq2000_function_arg_advance): Likewise.
9897 (iq2000_function_arg): Likewise.
9898 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
9899 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
9900 (m68k_libcall_value): Likewise.
9901 (m68k_function_value): Likewise.
9902 (sched_attr_op_type): Likewise.
9903 * config/mcore/mcore.c (mcore_output_move): Likewise.
9904 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
9906 (microblaze_function_arg): Likewise.
9907 * config/mips/mips.c (mips16_build_call_stub): Likewise.
9908 (mips_print_operand): Likewise.
9909 (mips_mode_ok_for_mov_fmt_p): Likewise.
9910 (mips_vector_mode_supported_p): Likewise.
9911 (mips_preferred_simd_mode): Likewise.
9912 (mips_expand_vpc_loongson_even_odd): Likewise.
9913 (mips_expand_vec_unpack): Likewise.
9914 (mips_expand_vi_broadcast): Likewise.
9915 (mips_expand_vector_init): Likewise.
9916 (mips_expand_vec_reduc): Likewise.
9917 (mips_expand_msa_cmp): Likewise.
9918 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
9919 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
9920 (cc_flags_for_mode): Likewise.
9921 * config/msp430/msp430.c (msp430_print_operand): Likewise.
9922 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
9923 (nds32_output_casesi_pc_relative): Likewise.
9924 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9925 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
9926 (nvptx_gen_unpack): Likewise.
9927 (nvptx_gen_pack): Likewise.
9928 (nvptx_gen_shuffle): Likewise.
9929 (nvptx_gen_wcast): Likewise.
9930 (nvptx_preferred_simd_mode): Likewise.
9931 * config/pa/pa.c (pa_secondary_reload): Likewise.
9932 * config/pa/predicates.md (base14_operand): Likewise.
9933 * config/powerpcspe/powerpcspe-c.c
9934 (altivec_resolve_overloaded_builtin): Likewise.
9935 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
9937 (rs6000_preferred_simd_mode): Likewise.
9938 (output_vec_const_move): Likewise.
9939 (rs6000_expand_vector_extract): Likewise.
9940 (rs6000_split_vec_extract_var): Likewise.
9941 (reg_offset_addressing_ok_p): Likewise.
9942 (rs6000_legitimate_offset_address_p): Likewise.
9943 (rs6000_legitimize_address): Likewise.
9944 (rs6000_emit_set_const): Likewise.
9945 (rs6000_const_vec): Likewise.
9946 (rs6000_emit_move): Likewise.
9947 (spe_build_register_parallel): Likewise.
9948 (rs6000_darwin64_record_arg_recurse): Likewise.
9949 (swap_selector_for_mode): Likewise.
9950 (spe_init_builtins): Likewise.
9951 (paired_init_builtins): Likewise.
9952 (altivec_init_builtins): Likewise.
9953 (do_load_for_compare): Likewise.
9954 (rs6000_generate_compare): Likewise.
9955 (rs6000_expand_float128_convert): Likewise.
9956 (emit_load_locked): Likewise.
9957 (emit_store_conditional): Likewise.
9958 (rs6000_output_function_epilogue): Likewise.
9959 (rs6000_handle_altivec_attribute): Likewise.
9960 (rs6000_function_value): Likewise.
9961 (emit_fusion_gpr_load): Likewise.
9962 (emit_fusion_p9_load): Likewise.
9963 (emit_fusion_p9_store): Likewise.
9964 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
9965 (fusion_gpr_mem_load): Likewise.
9966 (fusion_addis_mem_combo_load): Likewise.
9967 (fusion_addis_mem_combo_store): Likewise.
9968 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
9969 (fusion_gpr_mem_load): Likewise.
9970 (fusion_addis_mem_combo_load): Likewise.
9971 (fusion_addis_mem_combo_store): Likewise.
9972 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9974 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
9975 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
9976 (rs6000_preferred_simd_mode): Likewise.
9977 (output_vec_const_move): Likewise.
9978 (rs6000_expand_vector_extract): Likewise.
9979 (rs6000_split_vec_extract_var): Likewise.
9980 (reg_offset_addressing_ok_p): Likewise.
9981 (rs6000_legitimate_offset_address_p): Likewise.
9982 (rs6000_legitimize_address): Likewise.
9983 (rs6000_emit_set_const): Likewise.
9984 (rs6000_const_vec): Likewise.
9985 (rs6000_emit_move): Likewise.
9986 (rs6000_darwin64_record_arg_recurse): Likewise.
9987 (swap_selector_for_mode): Likewise.
9988 (paired_init_builtins): Likewise.
9989 (altivec_init_builtins): Likewise.
9990 (rs6000_expand_float128_convert): Likewise.
9991 (emit_load_locked): Likewise.
9992 (emit_store_conditional): Likewise.
9993 (rs6000_output_function_epilogue): Likewise.
9994 (rs6000_handle_altivec_attribute): Likewise.
9995 (rs6000_function_value): Likewise.
9996 (emit_fusion_gpr_load): Likewise.
9997 (emit_fusion_p9_load): Likewise.
9998 (emit_fusion_p9_store): Likewise.
9999 * config/rx/rx.c (rx_gen_move_template): Likewise.
10000 (flags_from_mode): Likewise.
10001 * config/s390/predicates.md (s390_alc_comparison): Likewise.
10002 (s390_slb_comparison): Likewise.
10003 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
10004 (s390_vector_mode_supported_p): Likewise.
10005 (s390_cc_modes_compatible): Likewise.
10006 (s390_match_ccmode_set): Likewise.
10007 (s390_canonicalize_comparison): Likewise.
10008 (s390_emit_compare_and_swap): Likewise.
10009 (s390_branch_condition_mask): Likewise.
10010 (s390_rtx_costs): Likewise.
10011 (s390_secondary_reload): Likewise.
10012 (__SECONDARY_RELOAD_CASE): Likewise.
10013 (s390_expand_cs): Likewise.
10014 (s390_preferred_simd_mode): Likewise.
10015 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
10016 * config/sh/sh.c (sh_print_operand): Likewise.
10017 (dump_table): Likewise.
10018 (sh_secondary_reload): Likewise.
10019 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10020 * config/sh/sh.md (casesi_worker_1): Likewise.
10021 (casesi_worker_2): Likewise.
10022 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
10023 (fcc_comparison_operator): Likewise.
10024 * config/sparc/sparc.c (sparc_expand_move): Likewise.
10025 (emit_soft_tfmode_cvt): Likewise.
10026 (sparc_preferred_simd_mode): Likewise.
10027 (output_cbranch): Likewise.
10028 (sparc_print_operand): Likewise.
10029 (sparc_expand_vec_perm_bmask): Likewise.
10030 (vector_init_bshuffle): Likewise.
10031 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10032 (spu_vector_mode_supported_p): Likewise.
10033 (spu_expand_insv): Likewise.
10034 (spu_emit_branch_or_set): Likewise.
10035 (spu_handle_vector_attribute): Likewise.
10036 (spu_builtin_splats): Likewise.
10037 (spu_builtin_extract): Likewise.
10038 (spu_builtin_promote): Likewise.
10039 (spu_expand_sign_extend): Likewise.
10040 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10041 (tilegx_simd_int): Likewise.
10042 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
10043 (tilepro_simd_int): Likewise.
10044 * config/v850/v850.c (const_double_split): Likewise.
10045 (v850_print_operand): Likewise.
10046 (ep_memory_offset): Likewise.
10047 * config/vax/vax.c (vax_rtx_costs): Likewise.
10048 (vax_output_int_move): Likewise.
10049 (vax_output_int_add): Likewise.
10050 (vax_output_int_subtract): Likewise.
10051 * config/visium/predicates.md (visium_branch_operator): Likewise.
10052 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
10053 (visium_print_operand_address): Likewise.
10054 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10055 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
10056 (xtensa_expand_conditional_branch): Likewise.
10057 (xtensa_copy_incoming_a7): Likewise.
10058 (xtensa_output_literal): Likewise.
10059 * dfp.c (decimal_real_maxval): Likewise.
10060 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
10062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10063 Alan Hayward <alan.hayward@arm.com>
10064 David Sherwood <david.sherwood@arm.com>
10066 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
10067 (mode_nunits_inline): Likewise.
10068 (mode_inner_inline): Likewise.
10069 (mode_unit_size_inline): Likewise.
10070 (mode_unit_precision_inline): Likewise.
10071 (emit_insn_modes_h): Likewise. Also emit a #define of the
10073 (emit_mode_wider): Add an E_ prefix to mode names.
10074 (emit_mode_complex): Likewise.
10075 (emit_mode_inner): Likewise.
10076 (emit_mode_adjustments): Likewise.
10077 (emit_mode_int_n): Likewise.
10078 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
10079 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
10080 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
10081 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
10082 (CRC32_BUILTIN, ENTRY): Likewise.
10083 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
10084 (aarch64_pop_regs): Likewise.
10085 (aarch64_process_components): Likewise.
10086 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
10087 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
10088 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
10089 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
10090 * config/arm/arm.c (arm_init_libfuncs): Likewise.
10091 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
10093 * config/i386/i386-builtin.def (pcmpestr): Likewise.
10094 (pcmpistr): Likewise.
10095 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
10096 * config/mmix/mmix.c (mmix_output_condition): Likewise.
10097 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
10099 * config/rl78/rl78.c (mduc_regs): Likewise.
10100 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10101 (htm_expand_builtin): Likewise.
10102 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
10103 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
10104 * config/xtensa/xtensa.c (print_operand): Likewise.
10105 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
10106 (NUM_MODE_VECTOR_INT): Likewise.
10107 * genoutput.c (null_operand): Likewise.
10108 (output_operand_data): Likewise.
10109 * genrecog.c (print_parameter_value): Likewise.
10110 * lra.c (debug_operand_data): Likewise.
10112 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10114 * dwarf2out.c (scompare_loc_descriptor_wide)
10115 (scompare_loc_descriptor_narrow): New functions, split out from...
10116 (scompare_loc_descriptor): ...here.
10117 * expmed.c (emit_store_flag_int): New function, split out from...
10118 (emit_store_flag): ...here.
10120 2017-08-30 Richard Biener <rguenther@suse.de>
10122 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
10123 (dwarf2out_early_finish): Move setting of AT_pubnames from
10124 early debug output to early finish.
10126 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10128 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
10129 and -mdata-region to the assembler.
10131 2017-08-30 Richard Biener <rguenther@suse.de>
10133 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
10135 (gen_subprogram_die): Add DW_AT_object_pointer only early.
10136 (dwarf2out_early_global_decl): Only generate a DIE for the
10137 abstract origin if it doesn't already exist or is a declaration DIE.
10138 (resolve_addr): Do not add the linkage name twice when
10139 generating a stub DIE for the DW_TAG_GNU_call_site target.
10141 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10143 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10144 Use machine_mode rather than int for arg1_mode.
10146 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
10149 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
10150 that the second argument of the built-in functions to unpack
10151 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
10152 switch statement instead a lot of if statements.
10153 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
10154 Allow 64-bit values to be in Altivec registers as well as
10155 traditional floating point registers.
10156 (pack<mode>, FMOVE128_VSX iterator): Likewise.
10158 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
10160 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
10161 MAX_REGS_PER_ADDRESS == 1.
10163 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
10165 * config/i386/i386.opt (flag_fentry): Do not init to -1.
10166 * config/i386/i386.c (ix86_option_override_internal): Simplify
10167 setting of opts->x_flag_entry.
10169 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10170 Jakub Jelinek <jakub@redhat.com>
10171 Richard Biener <rguenther@suse.de>
10173 PR tree-optimization/81503
10174 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
10175 folded constant fits in the target type; reorder tests for clarity.
10177 2017-08-29 Martin Liska <mliska@suse.cz>
10179 * passes.def: Include pass_lower_switch.
10180 * stmt.c (dump_case_nodes): Remove and move to
10181 tree-switch-conversion.
10182 (case_values_threshold): Likewise.
10183 (expand_switch_as_decision_tree_p): Likewise.
10184 (emit_case_decision_tree): Likewise.
10185 (expand_case): Likewise.
10186 (balance_case_nodes): Likewise.
10187 (node_has_low_bound): Likewise.
10188 (node_has_high_bound): Likewise.
10189 (node_is_bounded): Likewise.
10190 (emit_case_nodes): Likewise.
10191 (struct simple_case_node): New struct.
10192 (add_case_node): Remove.
10193 (emit_case_dispatch_table): Use vector instead of case_list.
10194 (reset_out_edges_aux): Remove.
10195 (compute_cases_per_edge): Likewise.
10196 (expand_case): Build list of simple_case_node.
10197 (expand_sjlj_dispatch_table): Use it.
10198 * tree-switch-conversion.c (struct case_node): Moved from
10199 stmt.c and adjusted.
10200 (emit_case_nodes): Likewise.
10201 (node_has_low_bound): Likewise.
10202 (node_has_high_bound): Likewise.
10203 (node_is_bounded): Likewise.
10204 (case_values_threshold): Likewise.
10205 (reset_out_edges_aux): Likewise.
10206 (compute_cases_per_edge): Likewise.
10207 (add_case_node): Likewise.
10208 (dump_case_nodes): Likewise.
10209 (balance_case_nodes): Likewise.
10210 (expand_switch_as_decision_tree_p): Likewise.
10211 (emit_jump): Likewise.
10212 (emit_case_decision_tree): Likewise.
10213 (try_switch_expansion): Likewise.
10214 (do_jump_if_equal): Likewise.
10215 (emit_cmp_and_jump_insns): Likewise.
10216 (fix_phi_operands_for_edge): New function.
10217 (record_phi_operand_mapping): Likewise.
10218 (class pass_lower_switch): New pass.
10219 (pass_lower_switch::execute): New function.
10220 (make_pass_lower_switch): Likewise.
10221 (conditional_probability):
10222 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
10223 * tree-pass.h: Add make_pass_lower_switch.
10225 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10228 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
10231 2017-08-29 Richard Biener <rguenther@suse.de>
10233 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
10234 we do not add a DW_AT_inline attribute twice.
10235 (gen_subprogram_die): Remove code setting DW_AT_inline on
10236 DECL_ABSTRACT_P nodes.
10238 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
10240 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
10241 calls to internal functions.
10242 (gimplify_modify_expr): Likewise.
10243 * tree-call-cdce.c (use_internal_fn): Likewise.
10244 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
10245 (convert_to_divmod): Set the nothrow flag.
10246 * tree-if-conv.c (predicate_mem_writes): Likewise.
10247 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
10248 (vectorizable_call): Likewise.
10249 (vectorizable_store): Likewise.
10250 (vectorizable_load): Likewise.
10251 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
10252 (vect_recog_mask_conversion_pattern): Likewise.
10254 2017-08-29 Martin Liska <mliska@suse.cz>
10257 * gcc.c (driver_handle_option): Add new argument.
10258 * opts-common.c (handle_option): Pass
10259 target_option_override_hook.
10260 * opts-global.c (lang_handle_option): Add new option.
10261 (set_default_handlers): Add new argument.
10262 (decode_options): Likewise.
10263 * opts.c (target_handle_option): Likewise.
10264 (common_handle_option): Call target_option_override_hook.
10265 * opts.h (struct cl_option_handler_func): Add hook for
10266 target option override.
10267 (struct cl_option_handlers): Likewise.
10268 (set_default_handlers): Add new argument.
10269 (decode_options): Likewise.
10270 (common_handle_option): Likewise.
10271 (target_handle_option): Likewise.
10272 * toplev.c (toplev::main): Pass targetm.target_option.override
10275 2017-08-29 Richard Biener <rguenther@suse.de>
10276 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
10278 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
10279 life to the active subtree.
10281 2017-08-28 Jeff Law <law@redhat.com>
10283 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
10284 derive_equivalences.
10285 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
10286 Code moved into....
10287 (edge_info::derive_equivalences): New private member function
10289 * tree-ssa-dom.c (class edge_info): Changed from a struct
10290 to a class. Add ctor/dtor, methods and data members.
10291 (edge_info::edge_info): Renamed from allocate_edge_info.
10292 Initialize additional members.
10293 (edge_info::~edge_info): New.
10294 (free_dom_edge_info): Delete the edge info.
10295 (record_edge_info): Use new class & associated member functions.
10296 Tighten forms for testing for edge equivalences.
10297 (record_temporary_equivalences): Iterate over the simple
10298 equivalences rather than assuming there's only one per edge.
10299 (cprop_into_successor_phis): Iterate over the simple
10300 equivalences rather than assuming there's only one per edge.
10301 (optimize_stmt): Use operand_equal_p rather than pointer
10302 equality for mini-DSE code.
10304 2017-08-28 Nathan Sidwell <nathan@acm.org>
10306 * gcc.c (execute): Fold SIGPIPE handling into switch
10307 statement. Adjust internal error message.
10309 2017-08-28 Richard Biener <rguenther@suse.de>
10312 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
10313 Do nothing for removed DIEs.
10315 2017-08-28 Richard Biener <rguenther@suse.de>
10317 PR tree-optimization/81977
10318 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
10321 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
10324 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
10325 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
10326 using targetm.gen_mem_thread_fence.
10328 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
10331 * config/i386/i386.md (*<btsc><mode>): Change operand 2
10332 predicate to register_operand. Reorder operands.
10333 (*btr<mode>): Ditto.
10334 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
10335 (*btr<mode>_mask): Ditto.
10337 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
10339 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
10340 * config/rs6000/xmmintrin.h: New file.
10341 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
10343 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10346 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
10347 parameter and_insn and return it.
10348 (recombine_lvx_pattern): Insert a copy to ensure availability of
10349 the base register of the copied masking operation at the point of
10350 the instruction replacement.
10351 (recombine_stvx_pattern): Likewise.
10353 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
10355 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
10356 undocumented switches.
10357 (-mpower9-dform-vector): Likewise.
10358 (-mpower9-dform): Likewise.
10359 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
10360 comments to delete references to -mpower9-dform* switches.
10361 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
10362 Delete reference to -mpower9-dform* switches, test for
10363 -mpower9-vector instead.
10364 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
10365 (OTHER_P9_VECTOR_MASKS): Likewise.
10366 (POWERPC_MASKS): Likewise.
10367 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
10368 tests against -mpower9-dform* to -mpower9-vector. Delete code
10369 that checked for -mpower9-dform* consistancy with other options.
10370 Add test for -mpower9-misc to enable other power9 switches.
10371 (rs6000_init_hard_regno_mode_ok): Likewise.
10372 (rs6000_option_override_internal): Likewise.
10373 (rs6000_emit_prologue): Likewise.
10374 (rs6000_emit_epilogue): Likewise.
10375 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
10376 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
10377 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
10379 (emit_fusion_p9_store): Likewise.
10380 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
10381 resetting these macros if the assembler does not support ISA 3.0
10383 (TARGET_P9_DFORM_VECTOR): Likewise.
10384 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
10385 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
10387 2017-08-25 Alan Modra <amodra@gmail.com>
10389 PR rtl-optimization/81747
10390 * cse.c (cse_extended_basic_block): Don't attempt to record
10391 equivalences for degenerate conditional branches that branch
10392 to their fall-through.
10394 2017-08-24 Martin Sebor <msebor@redhat.com>
10396 PR middle-end/81908
10397 * gimple-fold.c (size_must_be_zero_p): New function.
10398 (gimple_fold_builtin_memory_op): Call it.
10400 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
10402 * config/rs6000/mm_malloc.h: New file.
10404 2017-08-24 Bin Cheng <bin.cheng@arm.com>
10406 PR tree-optimization/81913
10407 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
10408 analysis when either IVs in condition can wrap.
10410 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
10412 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
10413 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
10415 2017-08-24 Richard Biener <rguenther@suse.de>
10418 * targhooks.c (default_target_can_inline_p): Properly
10419 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
10420 is present and always compare.
10421 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
10422 infer -mfpmath=sse from TARGET_SSE_P.
10423 (ix86_can_inline_p): Properly use target_option_default_node when
10424 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
10426 2017-08-24 Richard Biener <rguenther@suse.de>
10429 * dwarf2out.c (output_die): Handle flag_generate_offload like
10431 (output_comp_unit): Likewise.
10432 (gen_array_type_die): Likewise.
10433 (dwarf2out_early_finish): Likewise.
10434 (note_variable_value_in_expr): Likewise.
10435 (dwarf2out_finish): Likewise. Adjust assert.
10436 * cgraphunit.c (symbol_table::compile): Move setting of
10437 flag_generate_offload earlier ...
10438 (symbol_table::finalize_compilation_unit): ... here, before
10439 early debug finalization.
10441 2017-08-24 Richard Biener <rguenther@suse.de>
10443 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
10444 and ipa-fnsummary.h.
10445 (ix86_can_inline_p): When ix86_fpmath flags do not match
10446 check whether the callee uses FP math at all.
10448 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
10450 PR middle-end/81931
10451 * tree-ssanames.c (get_nonzero_bits): Use element_precision
10452 instead of TYPE_PRECISION.
10454 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
10455 Alan Hayward <alan.hayward@arm.com>
10456 David Sherwood <david.sherwood@arm.com>
10458 * combine.c (make_extraction): Use subreg_offset_from_lsb.
10460 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
10462 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
10464 (ix86_frame::stack_realign_allocate): New field.
10465 (struct machine_frame_state): Modify comments.
10466 (machine_frame_state::sp_realigned_fp_end): New field.
10467 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
10468 layout calculation.
10469 (sp_valid_at): Add assertion to assure no attempt to access invalid
10470 offset of a realigned stack.
10471 (fp_valid_at): Likewise.
10472 (choose_baseaddr): Modify comments.
10473 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
10474 ix86_expand_prologue.
10475 (ix86_expand_prologue): Modify stack realignment and allocation.
10476 (ix86_expand_epilogue): Modify comments.
10477 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
10478 avx2_runtime, avx512f, and avx512f_runtime.
10480 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
10482 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
10483 (mstackrealign): Do not init to -1.
10484 * config/i386/i386.c (ix86_option_override_internal):
10485 Check opts_set, not opts when setting default value of
10486 opts->x_ix86_force_align_arg_pointer.
10488 2017-08-23 Richard Biener <rguenther@suse.de>
10490 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
10491 lang_hooks.decl_printable_name.
10492 * print-rtl-function.c (print_rtx_function): Likewise.
10493 * tree-pretty-print.c (dump_function_header): Likewise.
10495 2017-08-23 Richard Biener <rguenther@suse.de>
10498 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
10499 -g0 at compile-time.
10501 2017-08-23 Tamar Christina <tamar.christina@arm.com>
10503 PR middle-end/19706
10504 * doc/sourcebuild.texi (Other hardware attributes):
10507 2017-08-23 Tamar Christina <tamar.christina@arm.com>
10509 PR middle-end/19706
10510 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
10511 Fix single-use check.
10513 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
10515 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
10517 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
10519 * doc/install.texi: Modify to add more details on running selected
10522 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
10524 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
10525 is combined with -mabi=ms.
10526 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
10529 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10531 PR tree-optimization/81488
10532 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
10533 and cached_basis fields.
10534 (MAX_SPREAD): New constant.
10535 (alloc_cand_and_find_basis): Initialize new fields.
10536 (clear_visited): New function.
10537 (create_phi_basis_1): Rename from create_phi_basis, set visited
10538 and cached_basis fields.
10539 (create_phi_basis): New wrapper function.
10540 (phi_add_costs_1): Rename from phi_add_costs, add spread
10541 parameter, set visited field, short-circuit when limits reached.
10542 (phi_add_costs): New wrapper function.
10543 (record_phi_increments_1): Rename from record_phi_increments, set
10545 (record_phi_increments): New wrapper function.
10546 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
10547 (phi_incr_cost): New wrapper function.
10548 (all_phi_incrs_profitable_1): Rename from
10549 all_phi_incrs_profitable, set visited field.
10550 (all_phi_incrs_profitable): New wrapper function.
10552 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
10553 Alan Hayward <alan.hayward@arm.com>
10554 David Sherwood <david.sherwood@arm.com>
10556 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
10557 that takes the outer and inner modes.
10558 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
10559 comparison as the canonical test for a paradoxical subreg.
10560 * combine.c (simplify_set): Use paradoxical_subreg_p.
10561 (make_extraction): Likewise.
10562 (force_to_mode): Likewise.
10563 (rtx_equal_for_field_assignment_p): Likewise.
10564 (gen_lowpart_for_combine): Likewise.
10565 (simplify_comparison): Likewise.
10566 * cse.c (equiv_constant): Likewise.
10567 * expmed.c (store_bit_field_1): Likewise.
10568 * final.c (alter_subreg): Likewise.
10569 * fwprop.c (propagate_rtx): Likewise.
10570 (forward_propagate_subreg): Likewise.
10571 * ira-conflicts.c (ira_build_conflicts): Likewise.
10572 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
10573 * lra-constraints.c (curr_insn_transform): Likewise.
10574 (split_reg): Likewise.
10575 * lra-eliminations.c (move_plus_up): Likewise.
10576 (lra_eliminate_regs_1): Likewise.
10577 * recog.c (general_operand): Likewise.
10578 * ree.c (combine_reaching_defs): Likewise.
10579 * reload.c (push_reload): Likewise.
10580 (find_reloads): Likewise.
10581 * reload1.c (elimination_effects): Likewise.
10582 (compute_reload_subreg_offset): Likewise.
10583 (choose_reload_regs): Likewise.
10584 * rtlanal.c (subreg_lsb_1): Likewise.
10585 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10586 (simplify_subreg): Likewise.
10587 * var-tracking.c (track_loc_p): Likewise.
10588 * emit-rtl.c (byte_lowpart_offset): Likewise.
10589 (paradoxical_subreg_p): Delete out-of-line definition.
10591 2017-08-22 Jeff Law <law@redhat.com>
10593 PR tree-optimization/81741
10594 PR tree-optimization/71947
10595 * tree-ssa-dom.c: Include tree-inline.h.
10596 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
10597 equivalences if one is more expensive to compute than the other.
10598 * tree-ssa-scopedtables.h (class const_or_copies): Make
10599 record_const_or_copy_raw method private.
10600 (class avail_exprs_stack): New method simplify_binary_operation.
10601 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
10602 avail_exprs_stack::simplify_binary_operation as needed.
10603 (avail_exprs_stack::simplify_binary_operation): New function.
10605 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
10607 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
10608 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
10609 (DOT_SYMBOLS): Likewise.
10610 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
10611 (RELOCATABLE_NEEDS_FIXUP): Likewise.
10612 (RS6000_ABI_NAME): Likewise.
10613 (TARGET_CMODEL): Likewise.
10614 (TOC_SECTION_ASM_OP): Likewise.
10615 (SET_CMODEL): New macro.
10616 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
10618 2017-08-22 Richard Biener <rguenther@suse.de>
10620 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
10621 to allow for free-lang-data replacements similar to verify_type_variant.
10623 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
10624 Alan Hayward <alan.hayward@arm.com>
10625 David Sherwood <david.sherwood@arm.com>
10627 * config/aarch64/aarch64.md (casesi): Use DImode rather than
10628 VOIDmode for the LABEL_REF.
10630 2017-08-22 Richard Biener <rguenther@suse.de>
10632 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
10634 2017-08-22 Richard Biener <rguenther@suse.de>
10636 * common.opt (feliminate-dwarf2-dups): Ignore.
10637 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
10638 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
10639 same_die_p_wrap, compute_section_prefix,
10640 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
10641 (comdat_symbol_id, comdat_symbol_number): Likewise.
10642 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
10644 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
10645 (output_die): Mark unreachable path unreachable.
10646 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
10647 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
10648 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
10649 (dwarf2out_early_finish): Likewise.
10651 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
10653 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
10655 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
10658 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
10659 not VAR_P. Filter attribute warnings with OPT_Wattributes.
10660 (avr_attribute_table) <io, io_low, address>: Initialize
10661 .decl_required with true.
10663 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
10665 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
10666 undocumented debugging options.
10667 (-mvsx-scalar-double): Likewise.
10668 (-mallow-df-permute): Likewise.
10669 (-mvectorize-builtins): Likewise.
10670 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10671 (rs6000_builtin_vectorized_function): Likewise.
10672 (rs6000_builtin_md_vectorized_function): Likewise.
10673 (rs6000_opt_vars): Likewise.
10675 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
10678 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
10679 (*btrq_imm): Rename from *btrq.
10680 (*btcq_imm): Rename from *btcq.
10681 (btsc): New code attribute.
10682 (*<btsc><mode>): New insn pattern.
10683 (*btr<mode>): Ditto.
10684 (*<btsc><mode>_mask): New insn_and_split pattern.
10685 (*btr<mode>_mask): Ditto.
10687 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10688 Alan Hayward <alan.hayward@arm.com>
10689 David Sherwood <david.sherwood@arm.com>
10691 * function.c (pad_below): Simplify padding calculation.
10693 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10694 Alan Hayward <alan.hayward@arm.com>
10695 David Sherwood <david.sherwood@arm.com>
10697 * target.def (function_prologue): Remove frame size argument.
10698 (function_epilogue): Likewise.
10699 * doc/tm.texi: Regenerate.
10700 * final.c (final_start_function): Update call to function_prologue.
10701 (final_end_function): Update call to function_epilogue.
10702 (default_function_pro_epilogue): Remove frame size argument.
10703 * output.h (default_function_pro_epilogue): Likewise.
10704 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
10705 (arm_output_function_prologue): Likewise.
10706 * config/frv/frv.c (frv_function_prologue): Likewise.
10707 (frv_function_epilogue): Likewise.
10708 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
10709 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
10710 (ia64_output_function_epilogue): Likewise.
10711 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
10712 (m32r_output_function_epilogue): Likewise.
10713 * config/microblaze/microblaze.c (microblaze_function_prologue)
10714 (microblaze_function_epilogue): Likewise.
10715 * config/mips/mips.c (mips_output_function_prologue): Likewise.
10716 (mips_output_function_epilogue): Likewise.
10717 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
10718 (mmix_target_asm_function_epilogue): Likewise.
10719 * config/msp430/msp430.c (msp430_start_function): Likewise.
10720 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
10721 (nds32_asm_function_epilogue): Likewise.
10722 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
10723 * config/pa/pa.c (pa_output_function_prologue): Likewise.
10724 (pa_output_function_epilogue): Likewise.
10725 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
10726 (rs6000_output_function_epilogue): Likewise.
10727 * config/rl78/rl78.c (rl78_start_function): Likewise.
10728 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
10729 (rs6000_output_function_epilogue): Likewise.
10730 * config/rx/rx.c (rx_output_function_prologue): Likewise.
10731 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
10732 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
10733 (sparc_asm_function_epilogue): Likewise.
10735 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10737 * tree.h (type_has_mode_precision_p): New function.
10738 * convert.c (convert_to_integer_1): Use it.
10739 * expr.c (expand_expr_real_2): Likewise.
10740 (expand_expr_real_1): Likewise.
10741 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
10742 * match.pd: Likewise.
10743 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
10744 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
10745 * tree-tailcall.c (process_assignment): Likewise.
10746 * tree-vect-loop.c (vectorizable_reduction): Likewise.
10747 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
10748 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
10749 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10750 (vectorizable_assignment): Likewise.
10751 (vectorizable_shift): Likewise.
10752 (vectorizable_operation): Likewise.
10753 * tree-vrp.c (register_edge_assert_for_2): Likewise.
10755 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
10757 * match.pd: Add pow (C, x) simplification.
10759 2017-08-21 Richard Biener <rguenther@suse.de>
10761 PR tree-optimization/81900
10762 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
10763 for blocks with abnormal predecessors.
10764 (compute_antic): Do not set visited flag prematurely.
10766 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
10769 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
10771 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10773 * stor-layout.h (vector_type_mode): Move to...
10774 * tree.h (vector_type_mode): ...here.
10775 * stor-layout.c (vector_type_mode): Move to...
10776 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
10778 2017-08-21 Richard Biener <rguenther@suse.de>
10780 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
10781 register_external_die hooks.
10782 (debug_false_tree_charstarstar_uhwistar): Declare.
10783 (debug_nothing_tree_charstar_uhwi): Likewise.
10784 * debug.c (do_nothing_debug_hooks): Adjust.
10785 (debug_false_tree_charstarstar_uhwistar): New do nothing.
10786 (debug_nothing_tree_charstar_uhwi): Likewise.
10787 * dbxout.c (dbx_debug_hooks): Adjust.
10788 (xcoff_debug_hooks): Likewise.
10789 * sdbout.c (sdb_debug_hooks): Likewise.
10790 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
10791 * dwarf2out.c (macinfo_label_base): New global.
10792 (dwarf2out_register_external_die): New function for the
10793 register_external_die hook.
10794 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
10795 (dwarf2_debug_hooks): Use them.
10796 (dwarf2_lineno_debug_hooks): Adjust.
10797 (struct die_struct): Add with_offset flag.
10798 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
10799 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
10800 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
10801 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
10802 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
10803 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
10804 defining section names for the early LTO debug variants.
10805 (reset_indirect_string): New helper.
10806 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
10807 (print_dw_val): Add support for offsetted symbol references.
10808 (get_ultimate_context): Split out from is_cxx.
10809 (is_cxx): Use get_ultimate_context.
10810 (is_fortran): Add decl overload.
10811 (compute_comp_unit_symbol): Split out worker from
10812 compute_section_prefix.
10813 (compute_section_prefix): Call compute_comp_unit_symbol and
10814 set comdat_type_p here.
10815 (output_die): Skip DIE symbol output for the LTO added one.
10816 Handle DIE symbol references with offset.
10817 (output_comp_unit): Guard section name mangling properly.
10818 For LTO debug sections emit a symbol at the section beginning
10819 which we use to refer to its DIEs.
10820 (add_abstract_origin_attribute): For DIEs registered via
10821 dwarf2out_register_external_die directly refer to the early
10822 DIE rather than indirectly through the shadow one we created.
10823 Remove obsolete call to dwarf2out_abstract_function for
10824 non-function/block origins.
10825 (gen_array_type_die): When generating early LTO debug do
10826 not emit DW_AT_string_length.
10827 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
10828 late when in LTO. As suggested place a gcc_unreachable for
10829 the DECL_ABSTRACT_P case.
10830 (gen_subprogram_die): Avoid another specification DIE
10831 for early built declarations/definitions for the late LTO case.
10832 (gen_variable_die): Add type references for late duplicated VLA dies
10834 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
10835 we have the abstract instance already.
10836 (process_scope_var): Adjust decl DIE contexts in LTO which
10837 first puts them in limbo.
10838 (gen_decl_die): Do not generate type DIEs late apart from
10839 types for VLAs or for decls we do not yet have a DIE. Do not
10840 call dwarf2out_abstract_function late.
10841 (dwarf2out_early_global_decl): Make sure to create DIEs
10842 for abstract instances of a decl first.
10843 (dwarf2out_late_global_decl): Adjust comment.
10844 (output_macinfo_op): With multiple macro sections use
10845 macinfo_label_base to distinguish labels.
10846 (output_macinfo): Likewise. Update macinfo_label_base.
10847 Pass in the line info label.
10848 (note_variable_value_in_expr): When generating LTO resolve
10849 all variable values here by generating DIEs as needed.
10850 (init_sections_and_labels): Add early LTO debug flag parameter
10851 and generate different sections and names if set. Add generation
10852 counter for the labels so we can have multiple of them.
10853 (reset_dies): Helper to allow DIEs to be output multiple times.
10854 (dwarf2out_finish): When outputting DIEs to the fat part of an
10855 LTO object first reset DIEs.
10856 (dwarf2out_early_finish): Output early DIEs when generating LTO.
10857 (modified_type_die): Check for decl_ultimate_origin being self
10859 (gen_type_die_with_usage): Likewise.
10860 (gen_typedef_die): Allow decl_ultimate_origin being self.
10861 (set_decl_abstract_flags): Remove.
10862 (set_block_abstract_flags): Likewise.
10863 (dwarf2out_abstract_function): Treat the early generated DIEs
10864 as the abstract copy and only add DW_AT_inline and
10865 DW_AT_artificial here and call set_decl_origin_self.
10866 If the DIE has an abstract origin don't do anything.
10867 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
10868 if we have none yet (Go fails to build one, PR78628).
10869 (variably_modified_type_p): Prevent endless recursion for Ada
10870 cyclic pointer types.
10871 * lto-streamer-in.c: Include debug.h.
10872 (dref_queue): New global.
10873 (lto_read_tree_1): Stream in DIE references.
10874 (lto_input_tree): Register DIE references.
10875 (input_function): Stream DECL_DEBUG_ARGS.
10876 * lto-streamer-out.c: Include debug.h.
10877 (lto_write_tree_1): Output DIE references.
10878 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
10879 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
10880 (output_function): Stream DECL_DEBUG_ARGS.
10881 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
10882 Stream DECL_ABSTRACT_ORIGIN.
10883 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
10884 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
10885 DECL_CONTEXT for file-scope decls.
10886 * lto-streamer.h (struct dref_entry): Declare.
10887 (dref_queue): Likewise.
10888 * cfgexpand.c (pass_expand::execute): Do not call the
10889 outlining_inline_function hook here.
10890 * lto-wrapper.c (debug_obj): New global.
10891 (tool_cleanup): Unlink it if required.
10892 (debug_objcopy): New function.
10893 (run_gcc): Handle early debug sections in the IL files by
10894 extracting them to separate files, partially linkin them and
10895 feeding the result back as result to the linker.
10896 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
10897 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
10898 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
10899 sections into a separate segment.
10900 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
10902 (darwin_asm_dwarf_section): Likewise.
10903 (darwin_asm_output_dwarf_offset): Likewise.
10904 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
10906 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10907 Alan Hayward <alan.hayward@arm.com>
10908 David Sherwood <david.sherwood@arm.com>
10910 * read-md.h (md_reader::record_potential_iterator_use): Replace
10911 pointer argument with an rtx and an index.
10912 * read-rtl.c (iterator_group::apply_iterator): Likewise.
10913 (apply_mode_iterator): Likewise.
10914 (apply_code_iterator): Likewise.
10915 (apply_int_iterator): Likewise.
10916 (apply_subst_iterator): Likewise.
10917 (record_iterator_use): Likewise.
10918 (record_attribute_use): Likewise.
10919 (md_reader::record_potential_iterator_use): Likewise. Update calls
10920 to record_iterator_use and apply_iterator.
10921 (iterator_use): Replace ptr with x and index.
10922 (attribute_use): Likewise.
10923 (apply_attribute_uses): Update calls to apply_iterator.
10924 (apply_iterators): Likewise. Update initialization of iterator_use.
10925 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
10926 and record_potential_iterator_use.
10927 (rtx_reader::read_rtx_operand): Likewise.
10929 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10930 Alan Hayward <alan.hayward@arm.com>
10931 David Sherwood <david.sherwood@arm.com>
10933 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
10936 2017-08-21 Richard Biener <rguenther@suse.de>
10938 PR middle-end/81884
10939 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
10940 at struct end conservatively when comparing common bases.
10942 2017-08-21 Richard Biener <rguenther@suse.de>
10944 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
10945 (mem_ref_in_stmt): Remove.
10946 (determine_max_movement): Use ref index to get at the reference.
10947 (invariantness_dom_walker::before_dom_children): Deal with
10948 lim data already initialized.
10949 (gather_mem_refs_stmt): Initialize lim data and record ref index.
10951 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
10953 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
10954 (TARGET_ISA_ROUND): Ditto.
10955 (TARGET_ROUND): Ditto.
10956 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
10957 * config/i386/i386.md: Ditto.
10958 * config/i386/sse.md: Ditto.
10959 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
10960 with OPTION_MASK_ISA_SSE4_1.
10962 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
10965 * doc/extend.texi (x86 Built-in Functions): Correct the name of
10966 __builtin_ia32_lzcnt_u16.
10968 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
10971 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
10972 (rs6000_set_current_function): Rewrite function to use it.
10974 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
10977 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
10978 and TYPE_WARN_IF_NOT_ALIGN.
10979 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
10980 (handle_warn_if_not_align): New.
10981 (place_union_field): Call handle_warn_if_not_align.
10982 (place_field): Call handle_warn_if_not_align.
10983 Copy TYPE_WARN_IF_NOT_ALIGN.
10984 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
10985 (layout_type): Likewise.
10986 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
10988 (tree_decl_common): Add warn_if_not_align.
10989 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
10990 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
10991 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
10992 (DECL_WARN_IF_NOT_ALIGN): Likewise.
10993 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
10994 * doc/extend.texi: Document warn_if_not_aligned attribute.
10995 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
10997 2017-08-17 Martin Liska <mliska@suse.cz>
11000 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
11001 (get_data_dependence): Use it as pointer type.
11002 (distribute_loop): Likewise.
11004 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11006 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
11007 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
11008 (p8_vmrgow_<mode>_direct): New define_insn.
11009 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
11010 handle endianness for vmrgew and vmrgow permute patterns.
11012 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
11014 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
11015 * config/rs6000/rs6000-cpus.def: Remove comment.
11016 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
11017 (POWERPC_MASKS): Likewise.
11018 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
11019 use of TARGET_VSX_TIMODE.
11020 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
11021 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
11022 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
11023 (rs6000_option_override_internal): Remove dead code.
11024 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
11025 (rs6000_legitimize_reload_address): Likewise.
11026 (rs6000_legitimate_address_p): Likewise.
11027 (rs6000_opt_masks): Delete "vsx-timode".
11028 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
11029 from function comment.
11030 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
11031 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
11032 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
11034 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
11035 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
11036 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
11037 (VSX_TI): Likewise.
11039 (define_peephole2): Likewise.
11041 2017-08-17 Martin Sebor <msebor@redhat.com>
11044 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
11045 past the end of an array.
11046 (test_pp_format): Add test cases.
11048 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
11050 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
11051 missing ECF_NOTHROW flags.
11053 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
11056 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
11057 operands residing in integer registers.
11058 (*vsx_le_perm_load_<mode>): Likewise.
11059 (*vsx_le_perm_store_<mode>): Likewise.
11060 (define_peephole2): Add peepholes to optimize the above.
11062 2017-08-17 Marek Polacek <polacek@redhat.com>
11064 PR middle-end/81814
11065 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
11066 to mimic what shorten_compare did. Change the return type to bool.
11067 (fold_cond_expr_with_comparison): Update call to
11068 operand_equal_for_comparison_p.
11069 (fold_ternary_loc): Likewise.
11071 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
11073 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
11075 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
11077 2017-08-17 Richard Biener <rguenther@suse.de>
11079 * tree-ssa-structalias.c (solve_graph): When propagating
11080 to successors update the graphs succ edges and avoid duplicate work.
11082 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
11085 * config/i386/i386.c (ix86_option_override_internal): Save target
11086 specific options after ix86_stack_protector_guard_reg was changed.
11088 2017-08-17 Richard Biener <rguenther@suse.de>
11090 PR tree-optimization/81827
11091 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
11092 (new_var_info): Initialize it conservatively.
11093 (get_call_vi): Mark register vars.
11094 (new_scalar_tmp_constraint_exp): Likewise.
11095 (handle_rhs_call): Likewise.
11096 (handle_const_call): Likewise.
11097 (create_function_info_for): Likewise.
11098 (solve_constraints): Sort varinfos to separate register from
11099 non-register vars to pack points-to solution bitmaps during
11102 2017-08-17 Marek Polacek <polacek@redhat.com>
11104 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
11106 2017-08-17 Richard Biener <rguenther@suse.de>
11108 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
11109 to true when overflow is undefined and we saturated the result.
11111 2017-08-17 Alan Modra <amodra@gmail.com>
11114 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
11115 Don't use store multiple if only one reg needs saving.
11116 (interesting_frame_related_regno): New function.
11117 (rs6000_frame_related): Don't emit frame info for regs that
11119 (rs6000_emit_epilogue): Likewise.
11121 2017-08-16 Nathan Sidwell <nathan@acm.org>
11123 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
11124 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
11125 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
11126 * tree.c (free_lang_data_in_type): Use else-if chain. Always
11127 clear TYPE_LANG_1. Remove obsolete member-function stripping.
11128 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
11129 (verify_type): Adjust for TYPE_BINFO move.
11130 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
11131 process TYPE_BINFO directly.
11132 (hash_tree): Likewise.
11133 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
11135 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
11138 2017-08-16 David Malcolm <dmalcolm@redhat.com>
11140 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
11142 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
11145 * config/i386/i386.md (*anddi_1_btr): Change predicates of
11146 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
11147 Add ix86_binary_operator_ok to insn constraint.
11148 (*iordi_1_bts): Ditto.
11149 (*xordi_1_btc): Ditto.
11150 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
11151 Update corresponding peephole2 pattern.
11155 2017-08-16 Bin Cheng <bin.cheng@arm.com>
11157 PR tree-optimization/81832
11158 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
11159 copy loop header which has IFN_LOOP_DIST_ALIAS call.
11161 2017-08-16 Marek Polacek <polacek@redhat.com>
11164 * fold-const.c (fold_indirect_ref_1): Restore original behavior
11165 regarding size_zero_node.
11167 2017-08-16 Martin Liska <mliska@suse.cz>
11170 * config.gcc: Respect previously set extra_objs in case
11173 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
11175 PR tree-optimization/81835
11176 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
11177 the phi SSA_NAME. Check that the condition in a COND_EXPR does
11178 not depend on the phi.
11180 2017-08-16 Alan Modra <amodra@gmail.com>
11182 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
11185 2017-08-16 Alan Modra <amodra@gmail.com>
11187 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
11188 (save_reg_p): ..into this. Update all callers.
11189 (first_reg_to_save): Simplify.
11191 2017-08-16 Alan Modra <amodra@gmail.com>
11193 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
11196 2017-08-15 Joseph Myers <joseph@codesourcery.com>
11200 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
11201 constant count if that count is less than 32.
11203 2017-08-15 Nathan Sidwell <nathan@acm.org>
11205 * gcc.c (execute): Emit friendlier message if inferior is killed
11206 by an external cause.
11208 2017-08-15 Richard Biener <rguenther@suse.de>
11210 PR tree-optimization/81790
11211 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
11212 CONSTRUCTORs from simplifying and VN.
11214 2017-08-14 Martin Sebor <msebor@redhat.com>
11216 * builtin-attrs.def: Add comments.
11218 2017-08-14 Martin Sebor <msebor@redhat.com>
11221 * doc/extend.texi (attribute nonstring): Document new attribute.
11223 2017-08-14 Martin Sebor <msebor@redhat.com>
11226 * tree-diagnostic.c (default_tree_printer): Handle %G.
11227 * gimple-pretty-print.h (percent_G_format): Declare new function.
11228 * gimple-pretty-print.c (percent_G_format): Define.
11229 * tree-pretty-print.c (percent_K_format): Add argument.
11231 2017-08-14 Martin Sebor <msebor@redhat.com>
11233 PR translation/79998
11234 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
11235 Remove a stray space.
11237 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
11240 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
11241 (*iordi_1_bts): Ditto.
11242 (*xordi_1_btc): Ditto.
11244 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11247 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
11248 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11250 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
11251 quoted strings, and make more translator-friendly.
11252 (darwin_rs6000_override_options): Likewise.
11253 (rs6000_option_override_internal): Likewise.
11254 (rs6000_return_in_memory): Fix overlong line.
11255 (init_cmulative_args): Use quoted strings, and make more
11256 translator-friendly.
11257 (rs6000_pass_by_reference): Fix overlong line.
11258 (def_builtin): Use quoted strings.
11259 (altivec_expand_predicate_builtin): Use quoted strings, and make
11260 more translator-friendly.
11261 (htm_expand_builtin): Use quoted strings.
11262 (cpu_expand_builtin): Use quoted strings, and make more
11263 translator-friendly.
11264 (altivec_expand_builtin): Likewise.
11265 (paired_expand_predicate_builtin): Likewise.
11266 (rs6000_invalid_builtin): Likewise.
11267 (builtin_function_type): Use quoted strings.
11268 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
11269 more translator-friendly.
11270 (rs6000_trampoline_init): Likewise.
11271 (rs6000_handle_altivec_attribute): Likewise.
11272 (rs6000_inner_target_options): Use quoted strings.
11273 (rs6000_disable_incompatible_switches): Likewise.
11274 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
11275 strings, and make more translator-friendly.
11276 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
11278 2017-08-14 Bin Cheng <bin.cheng@arm.com>
11280 PR tree-optimization/81799
11281 * tree-loop-distribution.c (version_loop_by_alias_check): Force
11282 cond_expr to simple gimple operand.
11284 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
11286 PR middle-end/46932
11287 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
11289 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
11293 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
11294 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
11295 TARGET_GASISR_PROLOGUES.
11296 * config/avr/avr.c (avr_option_override): Same.
11297 (avr_pass_pre_proep::execute): Same.
11299 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
11302 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
11303 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
11305 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
11307 * config/i386/i386.md (*load_tp_<mode>): Redefine as
11308 define_insn_and_split. Split to a memory load from 0 in
11309 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
11310 using PTR mode iterator.
11311 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
11312 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
11313 (*add_tp_<mode>): Redefine as define_insn_and_split.
11314 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
11315 address space. Merge with *add_tp_x32 using PTR mode iterator.
11316 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
11317 Split to an add with a memory load from 0 in
11318 DEFAULT_TLS_SEG_REG address space.
11320 2017-08-12 Andrew Pinski <apinski@cavium.com>
11322 * config/aarch64/aarch64-option-extensions.def (rdma):
11323 Fix feature string to what Linux prints out in /proc/cpuinfo.
11325 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
11328 * dwarf2out.c (modified_type_die): For C typedef types that have
11329 an ultimate origin, process the ultimate origin instead of the
11331 (gen_typedef_die): Assert that input DECLs have no ultimate
11333 (gen_type_die_with_usage): For typedef variants that have an
11334 ultimate origin, just call gen_decl_die on the original DECL.
11335 (process_scope_var): Avoid creating DIEs for local typedefs and
11336 concrete static variables.
11338 2017-08-12 Alan Modra <amodra@gmail.com>
11342 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
11343 match gnu-user.h startfile.
11344 (ENDFILE_LINUX_SPEC): Similarly.
11346 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
11349 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
11351 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
11353 2017-08-11 Tamar Christina <tamar.christina@arm.com>
11354 * config/aarch64/aarch64.md (mov<mode>): Change.
11355 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
11356 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
11357 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
11359 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
11361 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
11362 for storage order barriers.
11364 2017-08-11 Martin Liska <mliska@suse.cz>
11367 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
11368 variables of void type.
11370 2017-08-11 Martin Liska <mliska@suse.cz>
11372 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
11373 TARGET_SUPPORTS_ALIASES.
11374 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
11375 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
11376 (optimize_weakref): Likewise.
11377 * symtab.c (symtab_node::noninterposable_alias): Likewise.
11378 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
11379 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
11381 2017-08-11 Martin Liska <mliska@suse.cz>
11384 * config/i386/i386.c (make_resolver_func): Do complete
11385 refactoring of the function.
11387 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
11390 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
11391 * config/i386/i386.c (ix86_stack_protect_guard): Use
11392 ix86_stack_protect_guard_symbol_str to generate varible declaration.
11393 * doc/invoke.texi (x86 Options): Document
11394 -mstack-protector-guard-symbol= option.
11396 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
11398 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
11399 * config/i386/i386.c (ix86_split_stack_guard): New function.
11400 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
11401 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
11402 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
11403 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
11404 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
11405 (split_stack_space_check): Call ix86_split_stack_guard.
11407 2017-08-10 Martin Sebor <msebor@redhat.com>
11409 * print-tree.c (print_node): Print location using the established
11411 Replace spaces with colons.
11412 (debug_raw, debug): Ditto.
11414 2017-08-10 Martin Sebor <msebor@redhat.com>
11417 * pretty-print.c (pp_format): Correct the handling of %s precision.
11419 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
11422 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
11424 (ix86_finalize_stack_frame_flags): This. Also clear
11425 frame_pointer_needed if -fno-omit-frame-pointer is used without
11427 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
11428 with ix86_finalize_stack_frame_flags.
11429 (ix86_expand_epilogue): Likewise.
11430 (ix86_expand_split_stack_prologue): Likewise.
11431 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
11433 2017-08-10 Martin Liska <mliska@suse.cz>
11436 * c-attribs.c (handle_target_attribute):
11437 Report warning for an empty string argument of target attribute.
11439 2017-08-09 Jakub Jelinek <jakub@redhat.com>
11442 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
11444 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
11445 or DECL_NONLOCAL labels.
11446 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
11447 or DECL_NONLOCAL labels here.
11449 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
11451 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
11452 to indicate when early gimple folding has been disabled.
11453 (rs6000_gimple_fold_builtin): Add debug content.
11454 (rs6000_invalid_builtin): Fix whitespace.
11455 (rs6000_expand_builtin): Fix whitespace.
11456 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
11458 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
11461 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
11462 SAVE_MULTIPLE if not all the registers that saves, should be saved.
11464 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
11466 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
11467 (qdf24xx): Likewise.
11468 * config/aarch64/aarch64.md: Include falkor.md.
11469 * config/aarch64/falkor.md: New.
11471 2017-08-09 Marek Polacek <polacek@redhat.com>
11474 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
11475 * diagnostic.c (emit_diagnostic): Add a comment.
11476 (emit_diagnostic_valist): New function.
11478 2017-08-09 Marek Polacek <polacek@redhat.com>
11481 * input.c (make_location): New overload.
11482 * input.h (make_location): Declare.
11484 2017-08-08 Alan Modra <amodra@gmail.com>
11485 H.J. Lu <hongjiu.lu@intel.com>
11488 * gcc.c (NO_PIE_SPEC): Delete.
11489 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
11491 (LINK_PIE_SPEC): ..to here.
11492 (LINK_COMMAND_SPEC): Support -no-pie.
11493 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
11494 chain of crtbegin*.o selection, update for PIE_SPEC changes and
11496 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
11497 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
11498 (ENDFILE_CRTEND_SPEC): Similarly.
11500 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
11503 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
11504 (mstack-protector-guard-offset=): Ditto.
11505 * config/i386/i386.c (ix86_option_override): Handle
11506 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
11508 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
11509 ix86_stack_protect_guard_offset variables.
11510 (TARGET_STACK_PROTECT_GUARD): Always define.
11511 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
11512 and -mstack-protector-guard-offset= options.
11514 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11516 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
11517 boundary case for the last candidate.
11519 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11521 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
11522 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
11524 2017-08-08 Tamar Christina <tamar.christina@arm.com>
11526 PR middle-end/19706
11527 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
11528 * config/aarch64/aarch64-builtins.c
11529 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
11530 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
11531 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
11533 2017-08-08 Tamar Christina <tamar.christina@arm.com>
11534 Andrew Pinski <pinskia@gmail.com>
11536 PR middle-end/19706
11537 * internal-fn.def (XORSIGN): New.
11538 * optabs.def (xorsign_optab): New.
11539 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
11540 (convert_expand_mult_copysign): New.
11541 (pass_optimize_widening_mul::execute): Call
11542 convert_expand_mult_copysign.
11544 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11546 PR tree-optimization/81354
11547 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
11548 Insert on edges rather than explicitly creating landing pads.
11549 (analyze_candidates_and_replace): Commit edge inserts.
11551 2017-08-08 Richard Biener <rguenther@suse.de>
11553 PR middle-end/81719
11554 * tree-ssa-loop-niter.c: Include tree-dfa.h.
11555 (expand_simple_operations): Also look through ADDR_EXPRs with
11556 MEM_REF bases treating them as POINTER_PLUS_EXPR.
11558 2017-08-08 Richard Biener <rguenther@suse.de>
11560 PR tree-optimization/81723
11561 * tree-vect-slp.c (struct bst_traits): New hash traits.
11562 (bst_fail): New global.
11563 (vect_build_slp_tree_2): New worker, split out from ...
11564 (vect_build_slp_tree): ... this now wrapping it with using
11565 bst_fail set to cache SLP tree build fails. Properly handle
11567 (vect_analyze_slp_instance): Allocate and free bst_fail.
11569 2017-08-08 Martin Liska <mliska@suse.cz>
11572 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
11573 LABEL_DECLs that can be from a different function.
11575 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11577 PR tree-optimization/81744
11578 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
11579 loop's number of iterations.
11581 2017-08-08 Martin Liska <mliska@suse.cz>
11583 * asan.c: Include header files.
11584 * attribs.c (build_decl_attribute_variant): New function moved
11586 (build_type_attribute_qual_variant): Likewise.
11587 (cmp_attrib_identifiers): Likewise.
11588 (simple_cst_list_equal): Likewise.
11589 (omp_declare_simd_clauses_equal): Likewise.
11590 (attribute_value_equal): Likewise.
11591 (comp_type_attributes): Likewise.
11592 (build_type_attribute_variant): Likewise.
11593 (lookup_ident_attribute): Likewise.
11594 (remove_attribute): Likewise.
11595 (merge_attributes): Likewise.
11596 (merge_type_attributes): Likewise.
11597 (merge_decl_attributes): Likewise.
11598 (merge_dllimport_decl_attributes): Likewise.
11599 (handle_dll_attribute): Likewise.
11600 (attribute_list_equal): Likewise.
11601 (attribute_list_contained): Likewise.
11602 * attribs.h (lookup_attribute): New function moved from tree.[ch].
11603 (lookup_attribute_by_prefix): Likewise.
11604 * bb-reorder.c: Include header files.
11605 * builtins.c: Likewise.
11606 * calls.c: Likewise.
11607 * cfgexpand.c: Likewise.
11608 * cgraph.c: Likewise.
11609 * cgraphunit.c: Likewise.
11610 * convert.c: Likewise.
11611 * dwarf2out.c: Likewise.
11612 * final.c: Likewise.
11613 * fold-const.c: Likewise.
11614 * function.c: Likewise.
11615 * gimple-expr.c: Likewise.
11616 * gimple-fold.c: Likewise.
11617 * gimple-pretty-print.c: Likewise.
11618 * gimple.c: Likewise.
11619 * gimplify.c: Likewise.
11620 * hsa-common.c: Likewise.
11621 * hsa-gen.c: Likewise.
11622 * internal-fn.c: Likewise.
11623 * ipa-chkp.c: Likewise.
11624 * ipa-cp.c: Likewise.
11625 * ipa-devirt.c: Likewise.
11626 * ipa-fnsummary.c: Likewise.
11627 * ipa-inline.c: Likewise.
11628 * ipa-visibility.c: Likewise.
11630 * lto-cgraph.c: Likewise.
11631 * omp-expand.c: Likewise.
11632 * omp-general.c: Likewise.
11633 * omp-low.c: Likewise.
11634 * omp-offload.c: Likewise.
11635 * omp-simd-clone.c: Likewise.
11636 * opts-global.c: Likewise.
11637 * passes.c: Likewise.
11638 * predict.c: Likewise.
11639 * sancov.c: Likewise.
11640 * sanopt.c: Likewise.
11641 * symtab.c: Likewise.
11642 * toplev.c: Likewise.
11643 * trans-mem.c: Likewise.
11644 * tree-chkp.c: Likewise.
11645 * tree-eh.c: Likewise.
11646 * tree-into-ssa.c: Likewise.
11647 * tree-object-size.c: Likewise.
11648 * tree-parloops.c: Likewise.
11649 * tree-profile.c: Likewise.
11650 * tree-ssa-ccp.c: Likewise.
11651 * tree-ssa-live.c: Likewise.
11652 * tree-ssa-loop.c: Likewise.
11653 * tree-ssa-sccvn.c: Likewise.
11654 * tree-ssa-structalias.c: Likewise.
11655 * tree-ssa.c: Likewise.
11656 * tree-streamer-in.c: Likewise.
11657 * tree-vectorizer.c: Likewise.
11658 * tree-vrp.c: Likewise.
11659 * tsan.c: Likewise.
11660 * ubsan.c: Likewise.
11661 * varasm.c: Likewise.
11662 * varpool.c: Likewise.
11663 * tree.c: Remove functions moved to attribs.[ch].
11664 * tree.h: Likewise.
11665 * config/aarch64/aarch64.c: Add attrs.h header file.
11666 * config/alpha/alpha.c: Likewise.
11667 * config/arc/arc.c: Likewise.
11668 * config/arm/arm.c: Likewise.
11669 * config/avr/avr.c: Likewise.
11670 * config/bfin/bfin.c: Likewise.
11671 * config/c6x/c6x.c: Likewise.
11672 * config/cr16/cr16.c: Likewise.
11673 * config/cris/cris.c: Likewise.
11674 * config/darwin.c: Likewise.
11675 * config/epiphany/epiphany.c: Likewise.
11676 * config/fr30/fr30.c: Likewise.
11677 * config/frv/frv.c: Likewise.
11678 * config/ft32/ft32.c: Likewise.
11679 * config/h8300/h8300.c: Likewise.
11680 * config/i386/winnt.c: Likewise.
11681 * config/ia64/ia64.c: Likewise.
11682 * config/iq2000/iq2000.c: Likewise.
11683 * config/lm32/lm32.c: Likewise.
11684 * config/m32c/m32c.c: Likewise.
11685 * config/m32r/m32r.c: Likewise.
11686 * config/m68k/m68k.c: Likewise.
11687 * config/mcore/mcore.c: Likewise.
11688 * config/microblaze/microblaze.c: Likewise.
11689 * config/mips/mips.c: Likewise.
11690 * config/mmix/mmix.c: Likewise.
11691 * config/mn10300/mn10300.c: Likewise.
11692 * config/moxie/moxie.c: Likewise.
11693 * config/msp430/msp430.c: Likewise.
11694 * config/nds32/nds32-isr.c: Likewise.
11695 * config/nds32/nds32.c: Likewise.
11696 * config/nios2/nios2.c: Likewise.
11697 * config/nvptx/nvptx.c: Likewise.
11698 * config/pa/pa.c: Likewise.
11699 * config/pdp11/pdp11.c: Likewise.
11700 * config/powerpcspe/powerpcspe.c: Likewise.
11701 * config/riscv/riscv.c: Likewise.
11702 * config/rl78/rl78.c: Likewise.
11703 * config/rx/rx.c: Likewise.
11704 * config/s390/s390.c: Likewise.
11705 * config/sh/sh.c: Likewise.
11706 * config/sol2.c: Likewise.
11707 * config/sparc/sparc.c: Likewise.
11708 * config/spu/spu.c: Likewise.
11709 * config/stormy16/stormy16.c: Likewise.
11710 * config/tilegx/tilegx.c: Likewise.
11711 * config/tilepro/tilepro.c: Likewise.
11712 * config/v850/v850.c: Likewise.
11713 * config/vax/vax.c: Likewise.
11714 * config/visium/visium.c: Likewise.
11715 * config/xtensa/xtensa.c: Likewise.
11717 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11720 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
11721 constraints since the -mupper-regs-* switches have been
11723 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
11724 into a vector from a double word element that was extracted from
11725 another vector, and eliminate extra XXPERMDI instructions.
11726 (vsx_concat_<mode>_2): Likewise.
11727 (vsx_concat_<mode>_3): Likewise.
11728 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
11729 concat to allow optimizing inserts from previous extracts.
11731 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
11733 * config/i386/i386.c (ix86_stack_protect_guard): Generate
11734 memory reference to a SSP offset in TLS address space.
11735 (ix86_print_operand) <case '@'>: Remove.
11736 (ix86_print_operand_punct_valid_p): Remove '@' code.
11737 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
11738 UNSPEC_SP_TLS_TEST.
11739 (stack_tls_protect_set_<mode>): Remove.
11740 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
11741 (stack_tls_protect_test_<mode>): Remove.
11742 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
11744 2017-08-07 Olivier Hainque <hainque@adacore.com>
11747 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
11749 2017-08-07 Douglas Rupp <rupp@adacore.com>
11751 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
11752 variable was referenced as multidir in command.
11754 2017-08-07 Jakub Jelinek <jakub@redhat.com>
11757 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
11760 2017-08-07 Martin Liska <mliska@suse.cz>
11762 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
11763 * config/rl78/rl78.c: Add include of attribs.h.
11764 * config/sh/sh.c: Likewise.
11765 * config/v850/v850.c: Likewise.
11767 2017-08-07 Tom de Vries <tom@codesourcery.com>
11769 PR middle-end/78266
11770 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
11772 2017-08-07 Martin Liska <mliska@suse.cz>
11774 * config/mips/mips.c: Include attribs.h.
11776 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
11779 * doc/invoke.texi: Document change in behvaior for -Ofast for
11782 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
11784 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
11786 (aarch64_pop_regs): Likewise.
11787 (aarch64_gen_load_pair): Likewise.
11788 (aarch64_save_callee_saves): Likewise.
11789 (aarch64_restore_callee_saves): Likewise.
11791 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11793 * config/i386/i386.c: Revert the last change.
11795 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11798 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
11800 (ix86_finalize_stack_frame_flags): This. Also clear
11801 frame_pointer_needed if -fno-omit-frame-pointer is used without
11803 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
11804 with ix86_finalize_stack_frame_flags.
11805 (ix86_expand_epilogue): Likewise.
11806 (ix86_expand_split_stack_prologue): Likewise.
11808 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11811 * config/i386/i386.c (get_builtin_code_for_version): Set priority
11812 to P_AES for Westmere.
11814 2017-08-07 Jonathan Yong <10walls@gmail.com>
11816 * config/i386/mingw.opt (fset-stack-executable): Removed.
11817 * config/i386/cygming.opt (fset-stack-executable): Moved
11819 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
11821 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
11823 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
11825 2017-08-07 Marek Polacek <polacek@redhat.com>
11827 PR middle-end/81737
11828 * fold-const.c (fold_indirect_ref_1): Check type_domain.
11830 2017-08-07 Martin Liska <mliska@suse.cz>
11832 * attribs.h (canonicalize_attr_name): New function.
11833 (cmp_attribs): Move from c-format.c and adjusted.
11834 (is_attribute_p): Moved from tree.h.
11835 * tree-inline.c: Add new includes.
11836 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
11837 (private_is_attribute_p): Remove.
11838 (private_lookup_attribute): Likewise.
11839 (private_lookup_attribute_by_prefix): Simplify.
11840 (remove_attribute): Use is_attribute_p.
11841 * tree.h: Remove removed declarations.
11843 2017-08-07 Jakub Jelinek <jakub@redhat.com>
11845 PR middle-end/81698
11846 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
11847 instead of computing it in the function. Formatting fix.
11848 (expand_case): Don't rely on default_edge being the first edge,
11849 clear it if removing it, pass default_edge to
11850 emit_case_dispatch_table.
11851 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
11854 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
11856 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
11857 insn in the function, emit NOP after the insn.
11859 2017-08-06 Tom de Vries <tom@codesourcery.com>
11861 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
11864 2017-08-06 Tom de Vries <tom@codesourcery.com>
11866 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
11869 2017-08-04 Yury Gribov <tetra2005@gmail.com>
11871 PR tree-optimization/57371
11872 * match.pd: New pattern.
11874 2017-08-04 Marek Polacek <polacek@redhat.com>
11876 PR middle-end/81695
11877 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
11878 perform the computation in offset_int.
11880 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11882 PR tree-optimization/81136
11883 * tree-vectorizer.h: Include tree-hash-traits.h.
11884 (vec_base_alignments): New typedef.
11885 (vec_info): Add a base_alignments field.
11886 (vect_record_base_alignments): Declare.
11887 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
11889 (DR_IS_CONDITIONAL_IN_STMT): New macro.
11890 (create_data_ref): Add an is_conditional_in_stmt argument.
11891 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
11892 the is_conditional_in_stmt field.
11893 (data_ref_loc): Add an is_conditional_in_stmt field.
11894 (get_references_in_stmt): Set the is_conditional_in_stmt field.
11895 (find_data_references_in_stmt): Update call to create_data_ref.
11896 (graphite_find_data_references_in_stmt): Likewise.
11897 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
11898 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
11899 (vect_record_base_alignment): New function.
11900 (vect_record_base_alignments): Likewise.
11901 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
11902 for nested statements even if we fail to compute a misalignment.
11903 Use pooled base alignments for unconditional references.
11904 (vect_find_same_alignment_drs): Compare base addresses instead
11906 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
11907 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
11909 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11911 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
11912 Add an explicit name for the enum. Use auto_vec for slp_instances
11913 and grouped_stores.
11914 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
11916 (_bb_vec_info): Add a constructor and destructor.
11917 (vinfo_for_stmt): Return NULL for uids of -1 as well.
11918 (destroy_loop_vec_info): Delete.
11919 (vect_destroy_datarefs): Likewise.
11920 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
11921 (vec_info::vec_info): New function.
11922 (vec_info::~vec_info): Likewise.
11923 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
11924 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
11925 destroy_loop_vec_info.
11926 * tree-vect-loop.c (new_loop_vec_info): Replace with...
11927 (_loop_vec_info::_loop_vec_info): ...this.
11928 (destroy_loop_vec_info): Replace with...
11929 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
11930 the stmt_vec_infos. Leave handling of vec_info information to its
11931 destructor. Remove explicit vector releases.
11932 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
11933 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
11934 * tree-vect-slp.c (new_bb_vec_info): Replace with...
11935 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
11936 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
11937 (destroy_bb_vec_info): Replace with...
11938 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
11939 information to its destructor.
11940 (vect_slp_analyze_bb_1): Use new and delete instead of
11941 new_bb_vec_info and destroy_bb_vec_info.
11942 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
11945 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11947 * tree-data-ref.h (subscript): Add access_fn field.
11948 (data_dependence_relation): Add could_be_independent_p.
11949 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
11950 (same_access_functions): Move to tree-data-ref.c.
11951 * tree-data-ref.c (ref_contains_union_access_p): New function.
11952 (access_fn_component_p): Likewise.
11953 (access_fn_components_comparable_p): Likewise.
11954 (dr_analyze_indices): Add a reference to access_fn_component_p.
11955 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
11957 (constant_access_functions): Likewise.
11958 (add_other_self_distances): Likewise.
11959 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
11960 (initialize_data_dependence_relation): Use XCNEW and remove
11961 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
11962 of access functions that have the same type. Allow the
11963 subsequence to end with different bases in some circumstances.
11964 Record the chosen access functions in SUB_ACCESS_FN.
11965 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
11966 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
11967 (subscript_dependence_tester_1): Likewise dra and drb.
11968 (build_classic_dist_vector): Update calls accordingly.
11969 (subscript_dependence_tester): Likewise.
11970 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
11971 DDR_COULD_BE_INDEPENDENT_P.
11972 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
11973 comp_alias_ddrs instead of may_alias_ddrs.
11974 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
11976 (vect_analyze_data_ref_dependence): Use it if
11977 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
11978 distance vectors if that fails.
11979 (dependence_distance_ge_vf): New function.
11980 (vect_prune_runtime_alias_test_list): Use it. Don't clear
11981 LOOP_VINFO_MAY_ALIAS_DDRS.
11983 2017-08-04 Richard Biener <rguenther@suse.de>
11985 PR middle-end/81705
11986 * fold-const.c (fold_binary_loc): Properly restrict
11987 minus_var0 && minus_var1 case when associating undefined overflow
11990 2017-08-04 Tom de Vries <tom@codesourcery.com>
11992 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
11994 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11996 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11997 Don't start diagnostic messages with a capital letter.
11998 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12000 (rs6000_invalid_builtin): Likewise.
12001 (rs6000_trampoline_init): Likewise.
12003 2017-08-03 Jakub Jelinek <jakub@redhat.com>
12006 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
12007 after setting changeable df flags.
12009 2017-08-03 Richard Biener <rguenther@suse.de>
12011 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
12012 up if the use is in USE - X.
12014 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
12016 * toplev.c (dumpfile.h): New include.
12017 (internal_error_reentered): New static function. Use it...
12018 (internal_error_function): ...here to handle reentered internal_error.
12020 2017-08-03 Richard Biener <rguenther@suse.de>
12022 PR middle-end/81148
12023 * fold-const.c (split_tree): Add minus_var and minus_con
12024 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
12025 here but always use minus_*.
12026 (associate_trees): Assert we never associate with MINUS_EXPR
12027 and NULL first operand. Do not recurse for PLUS_EXPR operands
12028 when associating as MINUS_EXPR either.
12029 (fold_binary_loc): Track minus_var and minus_con.
12031 2017-08-03 Tom de Vries <tom@codesourcery.com>
12034 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
12035 ACCEL_COMPILER, apply finish_options on
12036 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
12038 2017-08-03 Tom de Vries <tom@codesourcery.com>
12041 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
12042 function_entry_patch_area_size > 0.
12044 2017-08-03 Jakub Jelinek <jakub@redhat.com>
12047 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
12048 instead of 10??LU, perform unit multiplication in wide_int,
12049 don't change alloc_object_size_limit if the limit is larger
12052 PR tree-optimization/81655
12053 PR tree-optimization/81588
12054 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
12055 the case when ranges[i].low and high are 1 for unsigned type with
12058 PR middle-end/81052
12059 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
12060 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
12062 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12064 * tree-vrp.h: Add include guard.
12066 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
12069 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
12070 (ud2): New insn pattern.
12071 * config/i386/i386.c (ix86_expand_epilogue):
12072 For naked functions, generate ud2 instead of trap insn.
12074 2017-08-02 Marek Polacek <polacek@redhat.com>
12077 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
12079 2017-08-02 Tom de Vries <tom@codesourcery.com>
12080 Cesar Philippidis <cesar@codesourcery.com>
12082 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
12083 Add missing edge probabilities.
12085 2017-08-02 Tamar Christina <tamar.christina@arm.com>
12087 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
12088 Correct endianness.
12090 2017-08-02 Jakub Jelinek <jakub@redhat.com>
12092 PR middle-end/79499
12093 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
12094 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
12095 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
12096 prologue_seq sequences - if any.
12098 2017-08-02 Richard Biener <rguenther@suse.de>
12100 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
12101 via vectors if supported, integer extracts via punning if supported
12102 or otherwise vector extracts.
12104 2017-08-02 Richard Biener <rguenther@suse.de>
12106 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
12108 (bitmap_insert_into_set): ... this.
12110 2017-08-02 Richard Biener <rguenther@suse.de>
12112 PR tree-optimization/81633
12114 2015-08-17 Alan Hayward <alan.hayward@arm.com>
12116 PR tree-optimization/71752
12117 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
12119 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
12121 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
12122 (machine_function::call_ms2sysv_pad_out): Remove field.
12123 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
12124 (ix86_compute_frame_layout): Likewise.
12126 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
12129 * config/i386/i386.c (ix86_set_func_type): Disallow naked
12130 attribute with interrupt attribute.
12132 2017-08-01 Andrew Pinski <apinski@cavium.com>
12134 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
12135 BIT_INSERT_EXPR's operand 1
12136 to see if the types precision matches.
12138 2017-08-01 Martin Liska <mliska@suse.cz>
12140 PR middle-end/70140
12141 * builtins.c (expand_builtin_memcpy_args): Remove.
12142 (expand_builtin_memcpy): Call newly added function
12143 expand_builtin_memory_copy_args.
12144 (expand_builtin_memcpy_with_bounds): Likewise.
12145 (expand_builtin_mempcpy): Remove last argument.
12146 (expand_builtin_mempcpy_with_bounds): Likewise.
12147 (expand_builtin_memory_copy_args): New function created from
12148 expand_builtin_mempcpy_args with small modifications.
12149 (expand_builtin_mempcpy_args): Remove.
12150 (expand_builtin_stpcpy): Remove unused argument.
12151 (expand_builtin): Likewise.
12152 (expand_builtin_with_bounds): Likewise.
12154 2017-08-01 Martin Liska <mliska@suse.cz>
12157 Make mempcpy more optimal (PR middle-end/70140).
12159 2017-08-01 Jakub Jelinek <jakub@redhat.com>
12162 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12163 __builtin_vec_cmpne verify both arguments are compatible vectors
12164 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
12165 verify arg1_type is a pointer or array type. For __builtin_vec_st,
12166 move computation of aligned to after checking the argument types.
12170 * config/rs6000/vsx.md (vextract_fp_from_shorth,
12171 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
12174 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
12175 Doug Rupp <rupp@adacore.com>
12176 Olivier Hainque <hainque@adacore.com>
12178 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
12179 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
12181 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
12182 for TARGET_OS_CPP_BUILTIN.
12183 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
12184 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
12186 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
12187 passing required abi options to the assembler for EABI configurations.
12188 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
12189 of .text.hot and .text.unlikely sections for kernel modules when
12190 using ARM style exceptions.
12191 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
12192 options. Add EXTRA_CC1_SPEC.
12193 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
12195 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
12197 (ARM_TARGET2_DWARF_FORMAT): Define.
12198 * config/arm/t-vxworks: Adjust multilib control to removal of the
12199 Diab command line options.
12201 2017-08-01 Martin Liska <mliska@suse.cz>
12203 PR gcov-profile/81561
12204 * gcov.c (unblock): Make unblocking safe as we need to preserve
12205 index correspondence of blocks and block_lists.
12207 2017-08-01 Richard Biener <rguenther@suse.de>
12209 PR tree-optimization/81181
12210 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
12211 (compute_antic): ... end of iteration here.
12213 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
12215 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
12216 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
12217 (ftree-slp-vectorize): Likewise.
12218 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
12219 can no longer be set independent of flag_tree_loop_vectorize.
12220 * omp-general.c (emp_max_vf): Likewise.
12221 * opts.c (enable_fdo_optimizations): Remove references to
12222 flag_tree_vectorize, these are now implicit.
12223 (common_handle_option): Remove handling for OPT_ftree_vectorize,
12224 and leave it for the options machinery.
12226 2017-08-01 Martin Liska <mliska@suse.cz>
12228 PR middle-end/70140
12229 * builtins.c (expand_builtin_memcpy_args): Remove.
12230 (expand_builtin_memcpy): Call newly added function
12231 expand_builtin_memory_copy_args.
12232 (expand_builtin_memcpy_with_bounds): Likewise.
12233 (expand_builtin_mempcpy): Remove last argument.
12234 (expand_builtin_mempcpy_with_bounds): Likewise.
12235 (expand_builtin_memory_copy_args): New function created from
12236 expand_builtin_mempcpy_args with small modifications.
12237 (expand_builtin_mempcpy_args): Remove.
12238 (expand_builtin_stpcpy): Remove unused argument.
12239 (expand_builtin): Likewise.
12240 (expand_builtin_with_bounds): Likewise.
12242 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
12245 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
12246 print "ds:" only for immediates in generic address space.
12248 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
12251 * config/i386/i386.c (ix86_funciton_naked): New prototype.
12252 (ix86_function_ok_for_sibcall): Return false for naked functions.
12254 2017-08-01 Richard Biener <rguenther@suse.de>
12256 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
12257 (compute_antic): Seed worklist with exit block predecessors.
12258 * cfganal.c (dfs_find_deadend): For a cycle return the source
12259 of the edge closing it.
12261 2017-08-01 Tamar Christina <tamar.christina@arm.com>
12263 * config/aarch64/aarch64.c
12264 (aarch64_can_const_movi_rtx_p): Move 0 check.
12266 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12268 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
12269 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
12271 * match.pd: Ditto in address comparison pattern.
12273 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12275 PR tree-optimization/81627
12276 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
12277 closed ssa form for store-store chain.
12279 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12281 PR tree-optimization/81620
12282 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
12283 for store-store chain.
12285 2017-08-01 Jakub Jelinek <jakub@redhat.com>
12287 PR tree-optimization/81588
12288 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
12289 ranges[i].in_p, invert comparison code ccode. For >/>=,
12290 swap rhs1 and rhs2 and comparison code unconditionally,
12291 for </<= don't do that. Don't swap rhs1/rhs2 again if
12292 ranges[i].in_p, instead invert comparison code ccode if
12293 opcode or oe->rank is BIT_IOR_EXPR.
12296 * optabs.def (vec_extract_optab, vec_init_optab): Change from
12297 a direct optab to conversion optab.
12298 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
12299 with GET_MODE_INNER as last argument instead of optab_handler.
12300 * expmed.c (extract_bit_field_1): Likewise. Use vector from
12301 vector extraction if possible and optab is available.
12302 * expr.c (store_constructor): Use convert_optab_handler instead
12303 of optab_handler. Use vector initialization from smaller
12304 vectors if possible and optab is available.
12305 * tree-vect-stmts.c (vectorizable_load): Likewise.
12306 * doc/md.texi (vec_extract, vec_init): Document that the optabs
12307 now have two modes.
12308 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
12309 of vec_init from half-sized vectors with the same element mode.
12310 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
12311 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
12312 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
12313 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
12314 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
12315 after mode in gen_vec_extract* calls.
12316 (vec_extract<mode>): Renamed to ...
12317 (vec_extract<mode><ssescalarmodelower>): ... this.
12318 (vec_extract<mode><ssehalfvecmodelower>): New expander.
12319 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
12320 element mode after mode in gen_vec_init* calls.
12321 (VEC_INIT_HALF_MODE): New mode iterator.
12322 (vec_init<mode>): Renamed to ...
12323 (vec_init<mode><ssescalarmodelower>): ... this.
12324 (vec_init<mode><ssehalfvecmodelower>): New expander.
12325 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
12326 (vec_extractv2sfsf): ... this.
12327 (vec_initv2sf): Renamed to ...
12328 (vec_initv2sfsf): ... this.
12329 (vec_extractv2si): Renamed to ...
12330 (vec_extractv2sisi): ... this.
12331 (vec_initv2si): Renamed to ...
12332 (vec_initv2sisi): ... this.
12333 (vec_extractv4hi): Renamed to ...
12334 (vec_extractv4hihi): ... this.
12335 (vec_initv4hi): Renamed to ...
12336 (vec_initv4hihi): ... this.
12337 (vec_extractv8qi): Renamed to ...
12338 (vec_extractv8qiqi): ... this.
12339 (vec_initv8qi): Renamed to ...
12340 (vec_initv8qiqi): ... this.
12341 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
12342 (vec_init<mode>): Renamed to ...
12343 (vec_init<mode><VEC_base_l>): ... this.
12344 (vec_extract<mode>): Renamed to ...
12345 (vec_extract<mode><VEC_base_l>): ... this.
12346 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
12347 (vec_initv2sfsf): ... this.
12348 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
12349 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
12350 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
12351 element mode after mode in gen_vec_init* calls.
12352 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
12353 (vec_init<mode><Vel>): ... this.
12354 (vec_extract<mode>): Renamed to ...
12355 (vec_extract<mode><Vel>): ... this.
12356 * config/aarch64/iterators.md (Vel): New mode attribute.
12357 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
12358 Add element mode after mode in gen_vec_extract* calls.
12359 * config/s390/vector.md (non_vec_l): New mode attribute.
12360 (vec_extract<mode>): Renamed to ...
12361 (vec_extract<mode><non_vec_l>): ... this.
12362 (vec_init<mode>): Renamed to ...
12363 (vec_init<mode><non_vec_l>): ... this.
12364 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
12365 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
12367 * config/arm/iterators.md (V_elem_l): New mode attribute.
12368 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
12369 (vec_extract<mode><V_elem_l>): ... this.
12370 (vec_extractv2di): Renamed to ...
12371 (vec_extractv2didi): ... this.
12372 (vec_init<mode>): Renamed to ...
12373 (vec_init<mode><V_elem_l>): ... this.
12374 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
12375 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
12376 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
12377 Add element mode after gen_vec_extract* calls.
12378 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
12379 (vec_init<mode><unitmode>): ... this.
12380 (vec_extract<mode>): Renamed to ...
12381 (vec_extract<mode><unitmode>): ... this.
12382 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
12383 (vec_init<mode><unitmode>): ... this.
12384 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
12385 (vec_initv2sfsf): ... this.
12386 (vec_extractv2sf): Renamed to ...
12387 (vec_extractv2sfsf): ... this.
12388 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
12389 Add element mode after gen_vec_extract* calls.
12390 * config/mips/mips.md (unitmode): New mode iterator.
12391 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
12392 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
12393 * config/spu/spu.md (inner_l): New mode attribute.
12394 (vec_init<mode>): Renamed to ...
12395 (vec_init<mode><inner_l>): ... this.
12396 (vec_extract<mode>): Renamed to ...
12397 (vec_extract<mode><inner_l>): ... this.
12398 * config/sparc/sparc.md (veltmode): New mode iterator.
12399 (vec_init<VMALL:mode>): Renamed to ...
12400 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
12401 * config/ia64/vect.md (vec_initv2si): Renamed to ...
12402 (vec_initv2sisi): ... this.
12403 (vec_initv2sf): Renamed to ...
12404 (vec_initv2sfsf): ... this.
12405 (vec_extractv2sf): Renamed to ...
12406 (vec_extractv2sfsf): ... this.
12407 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
12408 (vec_init<mode>): Renamed to ...
12409 (vec_init<mode><VEC_base_l>): ... this.
12410 (vec_extract<mode>): Renamed to ...
12411 (vec_extract<mode><VEC_base_l>): ... this.
12412 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
12413 (vec_initv2sfsf): ... this.
12414 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
12415 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
12416 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
12417 gen_vec_init* calls.
12419 2017-08-01 Richard Biener <rguenther@suse.de>
12421 PR tree-optimization/81297
12422 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
12423 TREE_OVERFLOW from INTEGER_CSTs.
12425 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
12427 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
12429 2017-07-31 Carl Love <cel@us.ibm.com>
12431 * config/rs6000/rs6000-c: Add support for built-in functions
12432 vector signed char vec_xl_be (signed long long, signed char *);
12433 vector unsigned char vec_xl_be (signed long long, unsigned char *);
12434 vector signed int vec_xl_be (signed long long, signed int *);
12435 vector unsigned int vec_xl_be (signed long long, unsigned int *);
12436 vector signed long long vec_xl_be (signed long long, signed long long *);
12437 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
12438 vector signed short vec_xl_be (signed long long, signed short *);
12439 vector unsigned short vec_xl_be (signed long long, unsigned short *);
12440 vector double vec_xl_be (signed long long, double *);
12441 vector float vec_xl_be (signed long long, float *);
12442 * config/rs6000/altivec.h (vec_xl_be): Add #define.
12443 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
12444 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
12446 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
12447 (altivec_expand_builtin): Add switch statement to call
12448 altivec_expand_xl_be for each builtin.
12449 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
12450 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
12451 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
12452 __builtin_vsx_le_be_v16qi.
12453 * doc/extend.texi: Update the built-in documentation file for the
12454 new built-in functions.
12456 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
12459 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
12461 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
12463 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12465 * config.gcc: Add z14.
12466 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
12467 CPU model numbers for z13s and z14.
12468 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
12470 * config/s390/s390-opts.h (enum processor_type): Rename
12471 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
12472 * config/s390/s390.c (processor_table): Add field for CPU name to
12473 be passed to Binutils.
12474 (s390_asm_output_machine_for_arch): Use the new field in
12475 processor_table for Binutils.
12476 (s390_expand_builtin): Replace arch12 with z14.
12477 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
12478 (s390_get_sched_attrmask): Likewise.
12479 (s390_get_unit_mask): Likewise.
12480 * config/s390/s390.opt: Add z14 to processor_type enum.
12482 2017-07-31 Martin Jambor <mjambor@suse.cz>
12485 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
12486 regardless of optimization level.
12488 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
12489 Martin Liska <mliska@suse.cz>
12491 * predict.def: Remove old comment and adjust probability.
12492 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
12493 PREDICT statements.
12495 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
12498 * config/i386/i386.c (ix86_function_naked): New function.
12499 (ix86_can_use_return_insn_p): Return false for naked functions.
12500 (ix86_expand_prologue): Skip prologue for naked functions.
12501 (ix86_expand_epilogue): Skip epilogue for naked functions
12502 and emit trap instruction.
12503 (ix86_warn_func_return): New function.
12504 (ix86_attribute_table): Add "naked" attribute specification.
12505 (TARGET_WARN_FUNC_RETURN): Define.
12506 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
12508 2017-07-31 Martin Liska <mliska@suse.cz>
12510 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
12511 (dump_gimple_bb_header): Always dump BB info.
12512 (pp_cfg_jump): Do not append info about BB when dumping a jump.
12514 2017-07-31 Martin Liska <mliska@suse.cz>
12517 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
12518 also with current_function_decl non-null equality.
12520 2017-07-31 Jakub Jelinek <jakub@redhat.com>
12523 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
12524 change type to the element type, instead add eltype variable and
12525 use it where we are interested in the element type.
12527 PR tree-optimization/81603
12528 * ipa-polymorphic-call.c
12529 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
12530 offset arithmetic in offset_int, bail out if the resulting bit offset
12531 doesn't fit into shwi.
12533 2017-07-31 Martin Liska <mliska@suse.cz>
12535 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
12536 (gimplify_save_expr): Fix comment.
12538 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
12541 * config/i386/i386.c (ix86_function_arg): Update arguments for
12543 (ix86_compute_frame_layout): Set the initial stack offset to
12544 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
12545 INCOMING_FRAME_SP_OFFSET.
12546 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
12547 stack before exception handler returns.
12548 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
12549 the 'ERROR_CODE' for exception handler.
12551 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
12553 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
12554 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
12555 (ASM_OUTPUT_REG_POP): Ditto.
12556 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
12557 instead of asm_fprintf to output pure string.
12559 2017-07-29 Jakub Jelinek <jakub@redhat.com>
12561 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
12562 to imported_module_or_decl hook.
12563 (debug_nothing_tree_tree_tree_bool): Remove.
12564 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
12565 * debug.c (do_nothing_debug_hooks): Use
12566 debug_nothing_tree_tree_tree_bool_bool instead of
12567 debug_nothing_tree_tree_tree_bool.
12568 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12569 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
12570 * sdbout.c (sdb_debug_hooks): Likewise.
12571 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
12572 (gen_namespace_die): Add DW_AT_export_symbols attribute if
12574 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
12575 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
12576 attribute, don't add anything.
12578 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12580 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
12581 (fold_build2_stat_loc): Likewise.
12582 (fold_build3_stat_loc): Likewise.
12583 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
12584 (fold_build1_loc): Remove macro.
12585 (fold_build2_loc): Likewise.
12586 (fold_build3_loc): Likewise.
12588 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12590 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
12591 (gimple_build_debug_bind_source_stat): Likewise.
12592 * gimple.h (gimple_build_debug_bind): Remove macro.
12593 (gimple_build_debug_bind_source): Likewise.
12595 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12597 * bitmap.c (bitmap_alloc): Adjust.
12598 (bitmap_gc_alloc): Likewise.
12599 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
12601 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12603 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
12604 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
12605 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
12606 (bitmap_gc_alloc_stat): Likewise.
12607 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
12609 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12611 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
12612 * rtl.h (shallow_copy_rtx): Remove macro.
12614 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12616 * emit-rtl.c (gen_raw_REG): Adjust.
12617 * gengenrtl.c (gendef): Likewise.
12618 * rtl.c (rtx_alloc_stat): Remove _stat from name.
12619 * rtl.h (rtx_alloc): Remove macro.
12621 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12623 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
12624 (build_tree_list_stat): Likewise.
12625 * tree.h (build_tree_list): Remove macro.
12626 (build_tree_list_vec): Likewise.
12628 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12630 * tree.c (make_vector_stat): Remove _stat from name.
12631 (build_vector_stat): Likewise.
12632 * tree.h (make_vector_stat): Remove macro.
12633 (build_vector_stat): Likewise.
12635 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12637 * tree.h (build_var_debug_value): Remove prototype.
12639 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12641 * tree.c (tree_cons_stat): Remove _stat from name.
12642 * tree.h (tree_cons): Remove macro.
12644 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12646 * tree.c (build_vl_exp_stat): Remove _stat from name.
12647 * tree.h (build_vl_exp): Remove macro.
12649 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12651 * tree.c (build_decl_stat): Remove _stat from name.
12652 * tree.h (build_decl): Remove macro.
12654 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12656 * gimple.c (gimple_build_with_ops_stat): Adjust.
12657 (gimple_alloc_stat): Remove _stat from name.
12658 * gimple.h (gimple_alloc): Remove macro.
12660 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12662 * tree.c (make_tree_vec_stat): Remove _stat from name.
12663 (grow_tree_vec_stat): Likewise.
12664 * tree.h (make_tree_vec_stat): Adjust prototype.
12665 (grow_tree_vec_stat): Likewise.
12666 (make_tree_vec): Remove macro.
12667 (grow_tree_vec): Likewise.
12669 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12671 * fold-const.c (fold_build1_stat_loc): Adjust.
12672 (fold_build2_stat_loc): Likewise.
12673 (fold_build3_stat_loc): Likewise.
12674 * tree.c (build0_stat): Remove _stat from name.
12675 (build1_stat): Likewise.
12676 (build2_stat): Likewise.
12677 (build3_stat): Likewise.
12678 (build4_stat): Likewise.
12679 (build5_stat): Likewise.
12680 * tree.h (build1_loc): Remove macro, and rename _stat function
12682 (build2_loc): Likewise.
12683 (build3_loc): Likewise.
12684 (build4_loc): Likewise.
12685 (build5_loc): Likewise.
12687 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12689 * tree.c (make_int_cst_stat): Remove _stat from name.
12690 * tree.h (make_int_cst_stat): Adjust prototype.
12691 (make_int_cst): Remove macro.
12693 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12695 * tree.c (make_tre_binfo_stat): Remove _stat from name.
12696 * tree.h (make_tree_binfo_stat): Adjust prototype.
12697 (make_tree_binfo): Remove.
12699 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12701 * tree.c (copy_node_stat): Rename to copy_node.
12702 (build_distinct_type_copy): Adjust.
12703 * tree.h (copy_node_stat): Adjust prototype.
12704 (copy_node): Remove macro.
12706 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12708 * tree.c (make_node_stat): rename to make_node.
12709 (build_tree_list_stat): Adjust.
12710 (build0_stat): Likewise.
12711 (build2_stat): Likewise.
12712 (build3_stat): Likewise.
12713 (build4_stat): Likewise.
12714 (build5_stat): Likewise.
12715 (build_decl_stat): Likewise.
12716 * tree.h (make_node_stat): Adjust prototype.
12717 (make_node): remove macro.
12719 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
12721 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
12722 (PPC_FEATURE2_SCV): Likewise.
12723 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
12725 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12727 * config/aarch64/aarch64.c
12728 (aarch64_internal_mov_immediate): Add new special pattern.
12729 * config/aarch64/aarch64.md (*movdi_aarch64):
12730 Add reg/32bit const mov case.
12732 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12733 Richard Sandiford <richard.sandiford@linaro.org>
12735 * config/aarch64/aarch64.md (mov<mode>): Generalize.
12736 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
12737 Add integer and movi cases.
12738 (movi-split-hf-df-sf split, fp16): New.
12739 (enabled): Added TARGET_FP_F16INST.
12740 * config/aarch64/iterators.md (GPF_HF): New.
12741 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
12743 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12745 * config/aarch64/aarch64.c
12746 (aarch64_simd_container_mode): Add prototype.
12747 (aarch64_expand_mov_immediate): Add HI support.
12748 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
12749 (aarch64_can_const_movi_rtx_p): New.
12750 (aarch64_preferred_reload_class):
12751 Remove restrictions of using FP registers for certain SIMD operations.
12752 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
12753 (aarch64_valid_floating_const): Add integer move validation.
12754 (aarch64_simd_imm_scalar_p): Remove.
12755 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
12756 (aarch64_legitimate_constant_p): Expand list of supported cases.
12757 * config/aarch64/aarch64-protos.h
12758 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
12759 (aarch64_reinterpret_float_as_int): New.
12760 (aarch64_simd_imm_scalar_p): Remove.
12761 * config/aarch64/constraints.md (Uvi): New.
12762 (Dd): Split into Ds and new Dd.
12763 * config/aarch64/aarch64.md (*movsi_aarch64):
12765 (*movdi_aarch64): Add SIMD mov case.
12767 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12769 * tree-predcom.c: (struct chain): Handle store-store chain in which
12770 stores for elimination only store loop invariant values.
12771 (execute_pred_commoning_chain): Ditto.
12772 (prepare_initializers_chain_store_elim): Ditto.
12773 (prepare_finalizers): Ditto.
12774 (is_inv_store_elimination_chain): New function.
12775 (initialize_root_vars_store_elim_1): New function.
12777 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12779 * tree-predcom.c: Revise general description of the pass.
12780 (enum chain_type): New enum type for store elimination.
12781 (struct chain): New field supporting store elimination.
12782 (struct component): Ditto.
12783 (dump_chain): Dump store-stores chain.
12784 (release_chain): Release resources.
12785 (split_data_refs_to_components): Compute and create component
12786 contains only stores for elimination.
12787 (get_chain_last_ref_at): New function.
12788 (make_invariant_chain): Initialization.
12789 (make_rooted_chain): Specify chain type in parameter and record it.
12790 (add_looparound_copies): Skip for store-stores chain.
12791 (determine_roots_comp): Compute type of chain and pass it to
12793 (initialize_root_vars_store_elim_2): New function.
12794 (finalize_eliminated_stores): New function.
12795 (remove_stmt): Handle store for elimination.
12796 (execute_pred_commoning_chain): Execute predictive commoning on
12797 store-store chains.
12798 (determine_unroll_factor): Skip unroll for store-stores chain.
12799 (prepare_initializers_chain_store_elim): New function.
12800 (prepare_initializers_chain): Hanlde store-store chain.
12801 (prepare_finalizers_chain, prepare_finalizers): New function.
12802 (tree_predictive_commoning_loop): Return integer value indicating
12803 if loop is unrolled or lcssa form is corrupted.
12804 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
12806 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12808 * tree-predcom.c (initialize_root): Delete.
12809 (execute_pred_commoning_chain): Initialize root vars and replace
12810 reference of non-combined chain directly, rather than call above
12813 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12815 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
12816 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
12818 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12820 * tree-predcom.c (struct chain): New field init_seq.
12821 (release_chain): Release init_seq.
12822 (prepare_initializers_chain): Record intialization stmts in above
12824 (insert_init_seqs): New function.
12825 (tree_predictive_commoning_loop): Call insert_init_seqs.
12827 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12829 * tree-predcom.c (determine_roots_comp): Skip trivial components.
12831 2017-07-28 Richard Biener <rguenther@suse.de>
12833 * match.pd: Remove superfluous :c.
12834 * genmatch.c (simplify::id): Add member.
12835 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
12837 (current_id): New global.
12838 (dt_node::parent): Move from ...
12839 (dt_operand::parent): ... here. Add for_id member.
12840 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
12841 (decision_tree::find_node): Relax order requirement when
12842 merging DT_TRUE nodes to ones inbetween the current simplify
12843 and the one we try to merge with. Add diagnostic whenever
12844 we need to enforce pattern order by not merging.
12845 (decision_tree::insert): Set current_id.
12846 (decision_tree::print_node): Dump parent node and for_id.
12847 (parser::last_id): Add member.
12848 (parser::push_simplify): Assign unique id.
12849 (parser::parser): Initialize last_id.
12851 2017-07-28 Martin Liska <mliska@suse.cz>
12854 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
12855 gimple_build_debug_bind.
12857 2017-07-28 Richard Biener <rguenther@suse.de>
12859 PR tree-optimization/81502
12860 * match.pd: Add pattern combining BIT_INSERT_EXPR with
12862 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
12864 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
12865 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
12866 for BIT_FIELD_REF args.
12867 * fold-const.c (make_bit_field_ref): Likewise.
12868 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
12870 2017-07-28 Jakub Jelinek <jakub@redhat.com>
12873 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
12874 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
12875 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
12876 Or it into SANITIZER_UNDEFINED.
12877 * ubsan.c: Include gimple-fold.h and varasm.h.
12878 (ubsan_expand_ptr_ifn): New function.
12879 (instrument_pointer_overflow): New function.
12880 (maybe_instrument_pointer_overflow): New function.
12881 (instrument_object_size): Formatting fix.
12882 (pass_ubsan::execute): Call instrument_pointer_overflow
12883 and maybe_instrument_pointer_overflow.
12884 * internal-fn.c (expand_UBSAN_PTR): New function.
12885 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
12886 * sanitizer.def (__ubsan_handle_pointer_overflow,
12887 __ubsan_handle_pointer_overflow_abort): New builtins.
12888 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
12889 * internal-fn.def (UBSAN_PTR): New internal function.
12890 * opts.c (sanitizer_opts): Add pointer-overflow.
12891 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
12892 * fold-const.c (build_range_check): Compute pointer range check in
12893 integral type if pointer arithmetics would be needed. Formatting
12896 2017-07-28 Martin Liska <mliska@suse.cz>
12899 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
12900 parameters that are of a variable-length.
12902 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
12904 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
12906 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
12907 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
12908 (CRT_CALL_STATIC_FUNCTION): Likewise.
12909 (ASM_DEFAULT_SPEC): New define.
12910 (ASM_SPEC32): Likewise.
12911 (ASM_SPEC64): Likewise.
12912 (ASM_SPEC_COMMON): Likewise.
12913 (ASM_SPEC): Likewise.
12914 (INVALID_64BIT): Likewise.
12915 (LINK_OS_DEFAULT_SPEC): Likewise.
12916 (LINK_OS_SPEC32): Likewise.
12917 (LINK_OS_SPEC64): Likewise.
12918 (POWERPC_LINUX): Likewise.
12919 (PTRDIFF_TYPE): Likewise.
12920 (RESTORE_FP_PREFIX): Likewise.
12921 (RESTORE_FP_SUFFIX): Likewise.
12922 (SAVE_FP_PREFIX): Likewise.
12923 (SAVE_FP_SUFFIX): Likewise.
12924 (SIZE_TYPE): Likewise.
12925 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
12926 (TARGET_64BIT): Likewise.
12927 (TARGET_64BIT): Likewise.
12928 (TARGET_AIX): Likewise.
12929 (WCHAR_TYPE_SIZE): Likewise.
12930 (WCHAR_TYPE): Undefine.
12931 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
12932 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
12933 (CPP_OS_RTEMS_SPEC): Delete.
12934 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
12935 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
12937 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
12939 2017-07-28 Jakub Jelinek <jakub@redhat.com>
12941 PR tree-optimization/81578
12942 * tree-parloops.c (build_new_reduction): Bail out if
12943 reduction_code isn't one of the standard OpenMP reductions.
12944 Move the details printing after that decision.
12946 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12948 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
12949 related to reload_in_progress.
12950 (splat_input_operand): Likewise.
12951 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
12953 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
12955 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
12956 (TARGET_INSTANTIATE_DECLS): Likewise.
12957 (legitimate_indexed_address_p): Delete reload_in_progress code.
12958 (rs6000_debug_legitimate_address_p): Likewise.
12959 (rs6000_eliminate_indexed_memrefs): Likewise.
12960 (rs6000_emit_le_vsx_store): Likewise.
12961 (rs6000_emit_move_si_sf_subreg): Likewise.
12962 (rs6000_emit_move): Likewise.
12963 (register_to_reg_type): Likewise.
12964 (rs6000_pre_atomic_barrier): Likewise.
12965 (rs6000_machopic_legitimize_pic_address): Likewise.
12966 (rs6000_allocate_stack_temp): Likewise.
12967 (rs6000_address_for_fpconvert): Likewise.
12968 (rs6000_address_for_altivec): Likewise.
12969 (rs6000_secondary_memory_needed_rtx): Delete function.
12970 (rs6000_check_sdmode): Likewise.
12971 (rs6000_alloc_sdmode_stack_slot): Likewise.
12972 (rs6000_instantiate_decls): Likewise.
12973 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
12974 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
12975 Delete reload_in_progress.
12976 (*vec_reload_and_plus_<mptrsize>): Likewise.
12977 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
12978 (vsx_div_v2di): Likewise.
12979 (vsx_udiv_v2di): Likewise.
12981 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12983 * config/rs6000/rs6000.opt (mlra): Replace with stub.
12984 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
12985 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
12986 (rs6000_debug_reg_global): Delete print of LRA status.
12987 (rs6000_option_override_internal): Delete dead LRA related code.
12988 (rs6000_lra_p): Delete function.
12989 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
12991 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
12993 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
12994 * config/riscv/rtems.h: New file.
12996 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12997 Sudakshina Das <sudi.das@arm.com>
12999 * config/aarch64/aarch64.md
13000 (define_split for and<mode>3nr_compare): Move
13001 non aarch64_logical_operand to a register.
13002 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
13003 register immediate operand to a register.
13004 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
13006 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
13008 PR middle-end/81564
13009 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
13011 2017-07-27 Richard Biener <rguenther@suse.de>
13013 PR tree-optimization/81573
13014 PR tree-optimization/81494
13015 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
13016 multi defuse cycle case.
13018 2017-07-27 Richard Biener <rguenther@suse.de>
13020 PR tree-optimization/81571
13021 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
13024 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
13026 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
13027 earlier and only if MASK_FPU is set. Adjust formatting.
13029 2017-07-27 Martin Liska <mliska@suse.cz>
13031 * opt-functions.awk: Add validation of value of Init.
13032 * optc-gen.awk: Pass new argument.
13034 2017-07-27 Martin Liska <mliska@suse.cz>
13036 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
13037 Fix wrong condition.
13039 2017-07-27 Martin Liska <mliska@suse.cz>
13041 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
13042 BBs and edges seen by autoFDO.
13044 2017-07-27 Richard Biener <rguenther@suse.de>
13046 PR tree-optimization/81502
13047 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
13048 with incompatible but same sized type.
13049 (execute_update_addresses_taken): Likewise.
13051 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
13053 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
13054 flag_tree_loop_vectorize rather than flag_tree_vectorize.
13056 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13059 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
13060 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
13061 Change s_operand to memory_operand.
13063 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
13065 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
13066 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
13067 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
13068 Emit instructions rather than returning an expression. Handle TFmode
13069 and KFmode by casting to TImode.
13070 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
13071 (rs6000_emit_le_vsx_store): Likewise.
13072 * config/rs6000/vsx.md (VSX_TI): New iterator.
13073 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
13074 (*vsx_le_undo_permute_<mode>): Likewise.
13075 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
13076 emit the split sequence.
13077 (*vsx_le_perm_store_<mode>): Likewise.
13079 2017-07-27 Jakub Jelinek <jakub@redhat.com>
13081 PR tree-optimization/81555
13082 PR tree-optimization/81556
13083 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
13084 if true, force CHANGED for the recursive invocation.
13085 (reassociate_bb): Remember original length of ops array, pass
13086 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
13088 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
13089 attributes for noipa attribute. For naked attribute use
13090 lookup_attribute first before lookup_attribute_spec.
13091 * final.c (rest_of_handle_final): Disable IPA RA for functions with
13093 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
13094 for functions with noipa attribute.
13095 (cgraph_externally_visible_p): Return true for functions with noipa
13097 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
13098 for functions with noipa attribute.
13099 * doc/extend.texi: Document noipa function attribute.
13100 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
13101 also for functions with noipa attribute.
13102 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
13104 2017-07-26 Andrew Pinski <apinski@cavium.com>
13106 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
13107 vec_unalign_load_cost and vec_unalign_store_cost.
13109 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13111 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
13112 -mvsx-small-integer option.
13113 (ISA_3_0_MASKS_IEEE): Likewise.
13114 (OTHER_VSX_VECTOR_MASKS): Likewise.
13115 (POWERPC_MASKS): Likewise.
13116 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
13117 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
13118 code, only testing for DImode being allowed in non-VSX floating
13120 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
13121 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
13123 (rs6000_option_override_internal): Delete -mvsx-small-integer.
13124 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
13125 TARGET_P8_VECTOR test.
13126 (rs6000_secondary_reload_simple_move): Likewise.
13127 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
13128 since TARGET_P9_VECTOR was already tested.
13129 (rs6000_opt_masks): Remove -mvsx-small-integer.
13130 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
13131 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13133 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
13134 test for TARGET_VEXTRACTUB was used, and that uses
13136 (p9 extract splitter): Likewise.
13137 (vsx_extract_<mode>_di_p9): Likewise.
13138 (vsx_extract_<mode>_store_p9): Likewise.
13139 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
13140 for TARGET_P9_VECTOR was used. Delete code that is now dead with
13141 the elimination of TARGET_VSX_SMALL_INTEGER.
13142 (vsx_extract_<mode>_p8): Likewise.
13143 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
13144 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
13145 (vsx_set_<mode>_p9): Likewise.
13146 (vsx_set_v4sf_p9): Likewise.
13147 (vsx_set_v4sf_p9_zero): Likewise.
13148 (vsx_insert_extract_v4sf_p9): Likewise.
13149 (vsx_insert_extract_v4sf_p9_2): Likewise.
13150 * config/rs6000/rs6000.md (sign extend splitter): Change
13151 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
13152 (floatsi<mode>2_lfiwax_mem): Likewise.
13153 (floatunssi<mode>2_lfiwzx_mem): Likewise.
13154 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
13155 since a test for TARGET_P9_VECTOR was used.
13156 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13157 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
13158 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13159 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
13160 TARGET_P8_VECTOR test.
13161 (fix_trunc<mode>si2_stfiwx): Likewise.
13162 (fix_trunc<mode>si2_internal): Likewise.
13163 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
13164 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13166 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
13167 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
13168 TARGET_P8_VECTOR test.
13169 (fixuns_trunc<mode>si2_stfiwx): Likewise.
13170 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
13171 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13173 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
13174 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
13175 since a test for TARGET_P9_VECTOR was used.
13176 (splitter for loading small constants): Likewise.
13178 2017-07-26 Andrew Pinski <apinski@cavium.com>
13180 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
13183 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
13186 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
13187 (fp_valid_at): Likewise.
13189 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
13191 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
13192 (qdf24xx_addrcost_table): Likewise.
13193 (cortexa57_tunings): Update to use generic_branch_cost.
13194 (cortexa72_tunings): Likewise.
13195 (cortexa73_tunings): Likewise.
13196 (qdf24xx_tunings): Likewise.
13198 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
13200 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
13201 (thunderx2t99_branch_cost): Likewise.
13202 (cortexa35_tunings): Update to use generic_branch_cost.
13203 (cortexa53_tunings): Likewise.
13204 (cortexa57_tunings): Likewise.
13205 (cortexa72_tunings): Likewise.
13206 (cortexa73_tunings): Likewise.
13207 (thunderx2t99_tunings): Likewise.
13209 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
13211 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
13212 (sparc_option_override): Honour MASK_FSMULD.
13213 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
13214 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
13215 * config/sparc/sparc.opt (mfsmuld): New option.
13216 * doc/invoke.texi (mfsmuld): Document option.
13218 2017-07-26 Marek Polacek <polacek@redhat.com>
13220 PR middle-end/70992
13221 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
13223 2017-07-26 Richard Biener <rguenther@suse.de>
13225 * gimple-match-head.c (do_valueize): Return OP if valueize
13227 (get_def): New helper to get at the def stmt of a SSA name
13228 if valueize allows.
13229 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
13230 do_valueize to get at the def stmt.
13231 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
13233 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
13235 PR middle-end/46932
13236 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
13238 2017-07-26 Martin Liska <mliska@suse.cz>
13241 * function.c (expand_function_start): Make expansion of
13242 nonlocal_goto_save_area after parm_birth_insn.
13244 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
13246 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
13247 from all CPU target flags enable members.
13249 2017-07-26 Richard Biener <rguenther@suse.de>
13251 * genmatch.c (dt_simplify::gen): Make iterator vars const.
13252 (decision_tree::gen): Make 'type' const.
13253 (write_predicate): Likewise.
13255 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
13257 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
13258 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
13259 (rs6000_option_override_internal): Likewise.
13260 (rs6000_expand_vector_set): Likewise.
13261 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
13262 (TARGET_UPPER_REGS_SF): Likewise.
13263 (TARGET_UPPER_REGS_DI): Likewise.
13264 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
13265 (TARGET_DIRECT_MOVE_64BIT): Likewise.
13266 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
13267 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13268 (Splitters for DI constants in Altivec registers): Likewise.
13269 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
13270 (vsx_set_v4sf_p9): Likewise.
13271 (vsx_set_v4sf_p9_zero): Likewise.
13272 (vsx_insert_extract_v4sf_p9): Likewise.
13273 (vsx_insert_extract_v4sf_p9_2): Likewise.
13275 2017-07-25 Carl Love <cel@us.ibm.com>
13277 * doc/extend.texi: Update the built-in documentation file for the
13278 existing built-in functions
13279 vector signed char vec_cnttz (vector signed char);
13280 vector unsigned char vec_cnttz (vector unsigned char);
13281 vector signed short vec_cnttz (vector signed short);
13282 vector unsigned short vec_cnttz (vector unsigned short);
13283 vector signed int vec_cnttz (vector signed int);
13284 vector unsigned int vec_cnttz (vector unsigned int);
13285 vector signed long long vec_cnttz (vector signed long long);
13286 vector unsigned long long vec_cnttz (vector unsigned long long);
13288 2017-07-25 Andrew Pinski <apinski@cavium.com>
13290 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
13291 accesses where the use is for the first operand of a BIT_INSERT.
13293 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
13296 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
13297 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
13299 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
13301 * config/i386/gstabs.h: Delete.
13302 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
13304 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
13306 * config/i386/i386.c (ix86_decompose_address): Do not check for
13307 register RTX when looking at index_reg or base_reg.
13308 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
13310 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
13312 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
13313 to update EH info here.
13315 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
13317 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
13319 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
13321 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
13323 2017-07-25 Torsten Duwe <duwe@suse.de>
13325 * common.opt: Introduce -fpatchable-function-entry
13326 command line option, and its variables function_entry_patch_area_size
13327 and function_entry_patch_area_start.
13328 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
13329 including a two-value parser.
13330 * target.def (print_patchable_function_entry): New target hook.
13331 * targhooks.h (default_print_patchable_function_entry): New function.
13332 * targhooks.c (default_print_patchable_function_entry): Likewise.
13333 * toplev.c (process_options): Switch off IPA-RA if
13334 patchable function entries are being generated.
13335 * varasm.c (assemble_start_function): Look at the
13336 patchable-function-entry command line switch and current
13337 function attributes and maybe generate NOP instructions by
13338 calling the print_patchable_function_entry hook.
13339 * doc/extend.texi: Document patchable_function_entry attribute.
13340 * doc/invoke.texi: Document -fpatchable_function_entry
13341 command line option.
13342 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
13344 * doc/tm.texi: Re-generate.
13346 2017-07-25 Jakub Jelinek <jakub@redhat.com>
13349 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
13350 TARGET_AVX512DQ rather than TARGET_AVX512BW.
13352 2017-07-25 Tamar Christina <tamar.christina@arm.com>
13354 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
13356 2017-07-25 Richard Biener <rguenther@suse.de>
13358 PR tree-optimization/81455
13359 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
13360 not walk in cycles when looking for guards.
13362 2017-07-25 Richard Biener <rguenther@suse.de>
13364 PR tree-optimization/81529
13365 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
13366 when optimizing backedge uses.
13368 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
13370 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
13372 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
13373 to dl_section_ref. On AIX, append an expression to subtract
13374 the size of the section length to dl_section_ref.
13376 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
13378 * configure.ac: If any of the config.* scripts fail, exit 1.
13379 * configure: Regenerate.
13381 2017-07-25 Richard Biener <rguenther@suse.de>
13383 PR middle-end/81546
13384 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
13385 of immediate uses, be more verbose on errors.
13387 2017-07-25 Richard Biener <rguenther@suse.de>
13389 PR tree-optimization/81510
13390 * tree-vect-loop.c (vect_is_simple_reduction): When the
13391 reduction stmt is not inside the loop bail out.
13393 2017-07-25 Richard Biener <rguenther@suse.de>
13395 PR tree-optimization/81303
13396 * tree-vect-loop-manip.c (vect_loop_versioning): Build
13397 profitability check against LOOP_VINFO_NITERSM1.
13399 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
13401 * domwalk.c (cmp_bb_postorder): Simplify.
13402 (sort_bbs_postorder): New function. Use it...
13403 (dom_walker::walk): ...here to optimize common cases.
13405 2017-07-25 Martin Liska <mliska@suse.cz>
13408 * ipa-visibility.c (function_and_variable_visibility): Make the
13409 redirection just on target that supports aliasing.
13410 Fix GNU coding style.
13412 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
13415 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
13417 * config/arm/rtems.h: Likewise.
13418 * config/bfin/rtems.h: Likewise.
13419 * config/i386/rtemself.h: Likewise.
13420 * config/lm32/rtems.h: Likewise.
13421 * config/m32c/rtems.h: Likewise.
13422 * config/m68k/rtemself.h: Likewise.
13423 * config/microblaze/rtems.h: Likewise.
13424 * config/mips/rtems.h: Likewise.
13425 * config/moxie/rtems.h: Likewise.
13426 * config/nios2/rtems.h: Likewise.
13427 * config/powerpcspe/rtems.h: Likewise.
13428 * config/rs6000/rtems.h: Likewise.
13429 * config/rtems.h: Likewise.
13430 * config/sh/rtems.h: Likewise.
13431 * config/sh/rtemself.h: Likewise.
13432 * config/sparc/rtemself.h: Likewise.
13434 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
13437 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
13438 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
13439 * tree-ssa-structalias.c (alias_get_name): Same.
13441 2017-07-25 Bin Cheng <bin.cheng@arm.com>
13444 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
13445 instructions if no du chain is found.
13447 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
13449 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
13451 2017-07-25 Richard Biener <rguenther@suse.de>
13453 PR middle-end/81505
13454 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
13457 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
13459 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
13460 upper-regs options.
13461 (ISA_2_7_MASKS_SERVER): Likewise.
13462 (ISA_3_0_MASKS_IEEE): Likewise.
13463 (OTHER_P8_VECTOR_MASKS): Likewise.
13464 (OTHER_VSX_VECTOR_MASKS): Likewise.
13465 (POWERPC_MASKS): Likewise.
13466 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
13467 duplicate list of options.
13468 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
13469 explicit -mupper-regs options.
13470 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
13471 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
13472 alias for -mupper-regs-df.
13473 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13474 (rs6000_init_hard_regno_mode_ok): Likewise.
13475 (rs6000_option_override_internal): Likewise.
13476 (rs6000_opt_masks): Likewise.
13477 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
13478 options in terms of whether -mvsx or -mpower8-vector was used.
13479 (TARGET_UPPER_REGS_DI): Likewise.
13480 (TARGET_UPPER_REGS_SF): Likewise.
13481 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
13482 -mupper-regs-* options.
13484 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
13486 * passes.c (emergency_dump_function): Print some empty lines and a
13487 header before the RTL dump.
13489 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
13491 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
13493 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
13496 * config/aarch64/aarch64.c (aarch64_classify_symbol):
13497 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
13499 2017-07-24 Carl Love <cel@us.ibm.com>
13501 * config/rs6000/rs6000-c.c: Add support for built-in functions
13502 vector float vec_extract_fp32_from_shorth (vector unsigned short);
13503 vector float vec_extract_fp32_from_shortl (vector unsigned short);
13504 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
13505 vec_extract_fp_from_shortl): Add defines for the two builtins.
13506 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
13507 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
13509 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
13510 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
13511 * doc/extend.texi: Update the built-in documentation file for the
13512 new built-in function.
13514 2017-07-24 Jakub Jelinek <jakub@redhat.com>
13517 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
13519 * doc/generic.texi: Likewise.
13520 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
13521 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
13523 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
13525 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
13526 (aarch64_mls_elt_merge<mode>): Likewise.
13528 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
13530 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
13531 having __cxa_atexit.
13533 2017-07-23 Michael Collison <michael.collison@arm.com>
13535 * config/arm/arm.c (arm_option_override): Deprecate
13536 use of -mstructure-size-boundary.
13537 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
13538 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
13540 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13543 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
13544 Reduce cost estimate for direct moves.
13546 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
13549 * config/i386/i386.c (ix86_option_override_internal): Disable
13550 BMI, BMI2 and TBM instructions for -m16.
13552 2017-07-21 Carl Love <cel@us.ibm.com>
13554 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13555 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13556 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13557 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13558 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13559 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13560 VMULOSW): New enum "unspec" values.
13561 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13562 altivec_vmulosw): New patterns.
13563 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13564 VMULOSW): Add definitions.
13566 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
13568 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
13569 (qdf24xx): Likewise.
13570 * config/aarch64/aarch64-options-extensions.def (rdma); New.
13571 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
13572 (AARCH64_FL_V8_1): Renumber.
13573 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
13574 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
13575 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
13576 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
13577 rdma to feature modifiers list.
13579 2017-07-21 Yury Gribov <tetra2005@gmail.com>
13581 PR middle-end/56727
13582 * ipa-visibility (function_and_variable_visibility): Convert
13583 recursive PLT call to direct call if appropriate.
13585 2017-07-21 Andrew Pinski <apinski@cavium.com>
13587 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
13588 operand 1 to see if the types precision matches.
13589 * fold-const.c (operand_equal_p): Likewise.
13591 2017-07-21 Richard Biener <rguenther@suse.de>
13593 PR tree-optimization/81303
13594 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
13595 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
13596 (vect_peeling_hash_get_lowest_cost): Adjust.
13597 (vect_enhance_data_refs_alignment): Likewise. Use
13598 vect_get_peeling_costs_all_drs to compute the penalty for no
13599 peeling to match up costs.
13601 2017-07-21 Richard Biener <rguenther@suse.de>
13603 PR tree-optimization/81500
13604 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
13605 we didn't identify a reduction path.
13607 2017-07-21 Tom de Vries <tom@codesourcery.com>
13608 Cesar Philippidis <cesar@codesourcery.com>
13610 PR gcov-profile/81442
13611 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
13614 2017-07-21 Tom de Vries <tom@codesourcery.com>
13617 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
13619 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
13620 nvptx_override_options_after_change.
13622 2017-07-21 Ulrich Drepper <drepper@redhat.com>
13624 * dwarf2out.c (output_file_names): Avoid double testing for
13625 dwarf_version >= 5.
13627 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
13629 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
13631 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
13633 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
13635 (try_crossjump_to_edge): Do not punt on partitioned functions.
13637 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
13639 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
13640 Put all BBs reachable only via paths crossing cold region to cold
13642 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
13644 2016-07-21 Richard Biener <rguenther@suse.de>
13646 PR tree-optimization/81303
13647 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
13648 into account prologue and epilogue iterations when raising
13649 min_profitable_iters to sth at least covering one vector iteration.
13651 2017-07-21 Tamar Christina <tamar.christina@arm.com>
13653 * config/arm/arm.c (arm_test_cpu_arch_dat):
13656 2017-07-20 Nathan Sidwell <nathan@acm.org>
13658 Remove TYPE_METHODS.
13659 * tree.h (TYPE_METHODS): Delete.
13660 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
13661 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
13662 (dbxout_type_methods): Scan TYPE_FIELDS.
13663 (dbxout_type): Don't check TYPE_METHODS here.
13664 * function.c (use_register_for_decl): Always ignore register for
13665 class types when not optimizing.
13666 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
13667 * tree.c (free_lang_data_in_type): Stitch out member functions and
13668 templates from TYPE_FIELDS.
13669 (build_distinct_type_copy, verify_type_variant,
13670 verify_type): Member fns are on TYPE_FIELDS.
13671 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
13672 * tree-pretty-print.c (dump_generic_node): Likewise.
13674 2017-07-20 Jakub Jelinek <jakub@redhat.com>
13677 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
13678 V2TImode and V4TImode.
13679 (ix86_expand_vector_extract): Likewise.
13680 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
13681 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
13682 (ssescalarmode): Handle V4TImode and V2TImode.
13683 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
13684 (*vec_extractv2ti, *vec_extractv4ti): New insns.
13685 (VEXTRACTI128_MODE): New mode iterator.
13686 (splitter for *vec_extractv?ti first element): New.
13687 (VEC_INIT_MODE): New mode iterator.
13688 (vec_init<mode>): Consolidate 3 expanders into one using
13689 VEC_INIT_MODE mode iterator.
13691 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
13693 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
13694 non_spilled_static_chain_regno_p.
13696 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
13698 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
13700 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
13702 * bb-reorder.c (connect_traces): Allow copying of blocks within
13705 2017-07-20 Richard Biener <rguenther@suse.de>
13707 * gimple.h (gimple_phi_result): Add gphi * overload.
13708 (gimple_phi_result_ptr): Likewise.
13709 (gimple_phi_arg): Likewise. Adjust index assert to only
13710 allow actual argument accesses rather than all slots available
13712 (gimple_phi_arg_def): Add gphi * overload.
13713 * tree-phinodes.c (make_phi_node): Initialize only actual
13715 (resize_phi_node): Clear memory not covered by old node,
13716 do not initialize excess argument slots.
13717 (reserve_phi_args_for_new_edge): Initialize new argument slot
13720 2017-07-20 Bin Cheng <bin.cheng@arm.com>
13722 PR tree-optimization/81388
13724 2016-07-21 Bin Cheng <bin.cheng@arm.com>
13726 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
13727 by removing computation of may_be_zero.
13729 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13730 Tom de Vries <tom@codesourcery.com>
13732 PR middle-end/81030
13733 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
13734 when gimple level profile disagrees with what RTL expander did.
13736 2017-07-20 Richard Biener <rguenther@suse.de>
13738 PR tree-optimization/61171
13739 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
13740 (vect_analyze_stmt): Add slp instance parameter.
13741 (vectorizable_reduction): Likewise.
13742 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
13743 (vect_is_simple_reduction): Deal with chains not detected
13744 as SLP reduction chain, specifically not properly associated
13745 chains containing a mix of plus/minus.
13746 (get_reduction_op): Remove.
13747 (get_initial_defs_for_reduction): Simplify, pass in whether
13748 this is a reduction chain, pass in the SLP node for the PHIs.
13749 (vect_create_epilog_for_reduction): Get the SLP instance as
13751 (vectorizable_reduction): Get the SLP instance as arg.
13752 During analysis remember the SLP node with the PHIs in the
13753 instance. Simplify getting at the vectorized reduction PHIs.
13754 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
13755 through SLP instance.
13756 (vect_slp_analyze_operations): Likewise.
13757 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
13758 (vect_transform_stmt): Likewise.
13760 2017-07-20 Tom de Vries <tom@codesourcery.com>
13762 PR tree-optimization/81489
13763 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
13764 read of phi arg location to before loop that modifies phi.
13766 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13768 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
13771 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13773 PR middle-end/81331
13774 * except.c (execute): Fix ordering issue.
13776 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
13778 PR rtl-optimization/81423
13779 * combine.c (make_compound_operation_int): Don't try to optimize
13780 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
13782 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
13784 PR rtl-optimization/81423
13785 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
13786 with a constant that is -1 in the truncated to mode.
13788 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13790 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
13791 (determine_unlikely_bbs): ... here.
13792 * predict.h (propagate_unlikely_bbs_forward): Declare.
13793 * cfgexpand.c (pass_expand::execute): Use it.
13794 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
13796 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
13797 propagate_unlikely_bbs_forward.
13799 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13801 PR middle-end/81331
13802 * except.c (maybe_add_nop_after_section_switch): New function.
13805 2017-07-19 Tom de Vries <tom@codesourcery.com>
13807 * gimple.h (gimple_phi_set_arg): Make assert more strict.
13809 2017-07-19 Tom de Vries <tom@codesourcery.com>
13811 * gimple.h (gimple_phi_arg): Make assert more strict.
13813 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
13815 * config.gcc (powerpc*-*-*): Add mmintrin.h.
13816 * config/rs6000/mmintrin.h: New file.
13817 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
13819 2017-07-19 Jakub Jelinek <jakub@redhat.com>
13821 PR tree-optimization/81346
13822 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
13824 2017-07-19 Tom de Vries <tom@codesourcery.com>
13826 * config/nvptx/nvptx.md (VECIM): Add V2DI.
13828 2017-07-19 Tom de Vries <tom@codesourcery.com>
13830 * config/nvptx/nvptx-modes.def: Add V2DImode.
13831 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
13832 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
13833 (nvptx_output_mov_insn): Handle lack of mov.b128.
13834 (nvptx_print_operand): Handle 'H' and 'L' codes.
13835 (nvptx_vector_mode_supported): Allow V2DImode.
13836 (nvptx_preferred_simd_mode): New function.
13837 (nvptx_data_alignment): New function.
13838 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
13839 nvptx_preferred_simd_mode.
13840 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
13842 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
13844 2017-07-19 Tom de Vries <tom@codesourcery.com>
13846 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
13847 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
13848 (nvptx_vector_mode_supported): New function. Allow V2SImode.
13849 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
13850 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
13851 (mov<VECIM>_insn): New define_insn.
13852 (define_expand "mov<VECIM>): New define_expand.
13854 2017-07-19 Tom de Vries <tom@codesourcery.com>
13856 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
13858 2017-07-19 Jakub Jelinek <jakub@redhat.com>
13860 PR tree-optimization/81346
13861 * fold-const.h (fold_div_compare, range_check_type): Declare.
13862 * fold-const.c (range_check_type): New function.
13863 (build_range_check): Use range_check_type.
13864 (fold_div_compare): No longer static, rewritten into
13865 a match.pd helper function.
13866 (fold_comparison): Don't call fold_div_compare here.
13867 * match.pd (X / C1 op C2): New optimization using fold_div_compare
13868 as helper function.
13870 2017-07-19 Nathan Sidwell <nathan@acm.org>
13872 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
13873 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
13874 * tree.c (find_decls_types_r, verify_type): Use
13875 TYPE_{MIN,MAX}_VALUE_RAW.
13876 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
13877 (hash_tree): Likewise.
13878 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13880 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13883 2017-07-18 Tom de Vries <tom@codesourcery.com>
13885 PR middle-end/81464
13886 * omp-expand.c (expand_omp_for_static_chunk): Handle
13887 equal-argument loop exit phi.
13889 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
13892 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
13893 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
13894 operand 2 predicate.
13895 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
13896 operand 2 predicate.
13897 (ror,rol -> rorx splitters): Use const_int_operand as
13898 operand 2 predicate.
13900 2017-06-18 Richard Biener <rguenther@suse.de>
13902 PR tree-optimization/81410
13903 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
13904 the gap in the ! slp_perm SLP case after each group.
13906 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13908 PR middle-end/81463
13909 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
13912 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13914 PR middle-end/81462
13915 * predict.c (set_even_probabilities): Cleanup; do not affect
13916 probabilities that are already known.
13917 (combine_predictions_for_bb): Call even when count is set.
13919 2017-07-18 Nathan Sidwell <nathan@acm.org>
13921 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
13924 2017-07-18 Bin Cheng <bin.cheng@arm.com>
13927 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
13928 optimization for loop niter analysis.
13930 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
13933 * config/avr/avr.c (avr_optimize_casesi): Don't use
13934 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
13936 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
13938 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
13939 body_cost_vec from _vect_peel_extended_info.
13940 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
13941 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
13944 2017-07-18 Bin Cheng <bin.cheng@arm.com>
13946 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
13948 2017-07-18 Richard Biener <rguenther@suse.de>
13950 PR tree-optimization/80620
13951 PR tree-optimization/81403
13952 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
13953 info when re-using a VN table entry.
13955 2017-07-18 Richard Biener <rguenther@suse.de>
13957 PR tree-optimization/81418
13958 * tree-vect-loop.c (vectorizable_reduction): Properly compute
13959 vectype_in. Verify that with lane-reducing reduction operations
13960 we have a single def-use cycle.
13962 2017-07-17 Carl Love <cel@us.ibm.com>
13964 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
13966 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13967 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13968 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13969 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13970 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13971 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13972 VMULOSW): New enum "unspec" values.
13973 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13974 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
13975 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13976 altivec_vmulosw): New patterns.
13977 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13978 VMULOSW): Add definitions.
13979 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
13981 * config/alpha/alpha.c: Include predict.h.
13983 2017-07-17 Yury Gribov <tetra2005@gmail.com>
13985 * tree-vrp.c (compare_assert_loc): Fix comparison function
13986 to return predictable results.
13988 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13990 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
13992 (subdi3): Likewise.
13993 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
13994 * doc/invoke.texi (mexpand-adddi): Update text.
13996 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13998 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
13999 that also clobbers the CC register. The old expand code is moved
14001 (*arc_clzsi2): ... here.
14002 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
14003 the CC register. The old expand code is moved to ...
14004 (arc_ctzsi2): ... here.
14006 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14008 * config/arc/arc.opt (mindexed-loads): Use initial value
14009 TARGET_INDEXED_LOADS_DEFAULT.
14010 (mauto-modify-reg): Use initial value
14011 TARGET_AUTO_MODIFY_REG_DEFAULT.
14012 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
14013 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
14014 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
14015 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
14017 2017-07-17 Martin Liska <mliska@suse.cz>
14020 * opts.c (finish_options): Do not allow -fgnu-tm
14021 w/ -fsanitize={kernel-,}address. Say sorry.
14023 2017-07-17 Bin Cheng <bin.cheng@arm.com>
14026 * tree-loop-distribution.c (classify_partition): Only assert on
14027 numer of iterations.
14028 (merge_dep_scc_partitions): Delete prameter. Update function call.
14029 (distribute_loop): Remove code handling loop with unknown niters.
14030 (pass_loop_distribution::execute): Skip loop with unknown niters.
14032 2017-07-17 Bin Cheng <bin.cheng@arm.com>
14035 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
14036 function sort_partitions_by_post_order.
14038 2017-07-17 Bin Cheng <bin.cheng@arm.com>
14040 PR tree-optimization/81374
14041 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
14042 the max index of basic blocks, rather than number of basic blocks.
14044 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14046 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
14048 (arc_legitimate_pic_operand_p): Likewise.
14049 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
14051 (arc_needs_pcl_p): Likewise.
14052 (arc_legitimate_pc_offset_p): Likewise.
14053 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
14054 function is also used in constrains.md.
14055 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
14056 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
14057 PLUS. Only return true/false in known cases, otherwise assert.
14058 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
14059 is already called in arc_legitimate_constant_p.
14060 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
14062 (LEGITIMATE_PIC_OPERAND_P): Use
14063 arc_raw_symbolic_reference_mentioned_p function.
14064 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
14069 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14070 Andrew Burgess <andrew.burgess@embecosm.com>
14072 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
14073 (arc_return_address_register): New function.
14074 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
14075 (arc_handle_fndecl_attribute): Add naked attribute.
14076 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14077 (TARGET_WARN_FUNC_RETURN): Likewise.
14078 (arc_allocate_stack_slots_for_args): New function.
14079 (arc_warn_func_return): Likewise.
14080 (machine_function): Change type fn_type.
14081 (arc_compute_function_type): Consider new naked function type,
14082 change function return type.
14083 (arc_must_save_register): Adapt to handle new
14084 arc_compute_function_type's return type.
14085 (arc_expand_prologue): Likewise.
14086 (arc_expand_epilogue): Likewise.
14087 (arc_return_address_regs): Delete.
14088 (arc_return_address_register): New function.
14089 (arc_epilogue_uses): Use above function.
14090 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
14091 (arc_function_type): Change encoding, add naked type.
14092 (ARC_INTERRUPT_P): Change to handle the new encoding.
14093 (ARC_FAST_INTERRUPT_P): Likewise.
14094 (ARC_NORMAL_P): Define.
14095 (ARC_NAKED_P): Likewise.
14096 (arc_compute_function_type): Delete prototype.
14097 * config/arc/arc.md (in_ret_delay_slot): Use
14098 arc_return_address_register function.
14099 (simple_return): Likewise.
14100 (p_return_i): Likewise.
14102 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14104 PR tree-optimization/81428
14105 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
14106 can't be built for those types.
14108 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
14110 Remove stuff dead since r239246.
14112 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
14113 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
14114 (avr_inform_devices): Remove dead stuff.
14116 2017-07-17 Tamar Christina <tamar.christina@arm.com>
14118 * config/arm/arm_neon.h: Fix softp typo.
14120 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14122 PR tree-optimization/81365
14123 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
14124 aggregate moves onto bb predecessor edges, make sure there are no
14125 loads that could alias the lhs in between the start of bb and the
14128 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
14131 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
14132 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
14133 [LSHIFTRT, outer_code = TRUNCATE]: Same.
14135 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14137 PR tree-optimization/81396
14138 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
14139 (init_symbolic_number): Initialize it to 1.
14140 (perform_symbolic_merge): Add n_ops from both operands into the new
14142 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
14143 without base_addr as useless if they need more than one operation.
14144 (bswap_replace): Handle !bswap case for NULL base_addr.
14146 2017-07-17 Tom de Vries <tom@codesourcery.com>
14149 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
14152 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
14154 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
14155 conditional builtin define __FIX_LEON3FT_B2BST.
14157 2017-07-17 Daniel Cederman <cederman@gaisler.com>
14159 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
14160 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
14163 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
14165 PR rtl-optimization/81424
14166 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
14167 to remove potential trapping from operands if -fnon-call-exceptions.
14169 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14171 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
14172 profile_proability for scalling.
14173 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
14175 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14177 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
14179 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14181 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
14182 fixpoint arithmetics.
14184 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14186 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
14187 fixpoint arithmetics.
14189 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14191 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
14192 fixpoint arithmetics.
14194 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14196 * profile-count.h (profile_probability::from_reg_br_prob_note,
14197 profile_probability::to_reg_br_prob_note): New functions.
14198 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
14199 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
14200 * predict.c (probability_reliable_p): Update.
14201 (edge_probability_reliable_p): Update.
14202 (br_prob_note_reliable_p): Update.
14203 (invert_br_probabilities): Update.
14204 (add_reg_br_prob_note): New function.
14205 (combine_predictions_for_insn): Update.
14206 * asan.c (asan_clear_shadow): Update.
14207 * cfgbuild.c (compute_outgoing_frequencies): Update.
14208 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
14209 (update_br_prob_note): Update.
14210 (rtl_verify_edges): Update.
14211 (purge_dead_edges): Update.
14212 (fixup_reorder_chain): Update.
14213 * emit-rtl.c (try_split): Update.
14214 * ifcvt.c (cond_exec_process_insns): Update.
14215 (cond_exec_process_if_block): Update.
14216 (dead_or_predicable): Update.
14217 * internal-fn.c (expand_addsub_overflow): Update.
14218 (expand_neg_overflow): Update.
14219 (expand_mul_overflow): Update.
14220 * loop-doloop.c (doloop_modify): Update.
14221 * loop-unroll.c (compare_and_jump_seq): Update.
14222 * optabs.c (emit_cmp_and_jump_insn_1): Update.
14223 * predict.h: Update.
14224 * reorg.c (mostly_true_jump): Update.
14226 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
14227 * config/alpha/alpha.c (emit_unlikely_jump): Update.
14228 * config/arc/arc.c: (emit_unlikely_jump): Update.
14229 * config/arm/arm.c: (emit_unlikely_jump): Update.
14230 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
14231 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
14232 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
14233 (ix86_print_operand): Update.
14234 (ix86_split_fp_branch): Update.
14235 (predict_jump): Update.
14236 * config/ia64/ia64.c (ia64_print_operand): Update.
14237 * config/mmix/mmix.c (mmix_print_operand): Update.
14238 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
14239 (rs6000_expand_split_stack_prologue): Update.
14240 * config/rs6000/rs6000.c: Update.
14241 * config/s390/s390.c (s390_expand_vec_strlen): Update.
14242 (s390_expand_vec_movstr): Update.
14243 (s390_expand_cs_tdsi): Update.
14244 (s390_expand_split_stack_prologue): Update.
14245 * config/sh/sh.c (sh_print_operand): Update.
14246 (expand_cbranchsi4): Update.
14247 (expand_cbranchdi4): Update.
14248 * config/sparc/sparc.c (output_v9branch): Update.
14249 * config/spu/spu.c (get_branch_target): Update.
14250 (ea_load_store_inline): Update.
14251 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
14252 * config/tilepro/tilepro.c: Update.
14254 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
14256 * gimplify.c (mostly_copy_tree_r): Revert latest change.
14257 (gimplify_save_expr): Likewise.
14259 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14261 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
14263 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14265 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
14267 * timevar.def (TV_IPA_FNSUMMARY): Define.
14269 2017-07-16 Daniel Cederman <cederman@gaisler.com>
14271 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
14272 to back store errata sensitive sequence from being generated.
14273 (sqrtdf2_fix): Likewise.
14275 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14277 * tree-ssa-threadupdate.c (compute_path_counts,
14278 update_joiner_offpath_counts): Use profile_probability.
14280 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
14283 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14285 * config/arm/arm-c.c (arm_cpu_builtins): Define
14286 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
14288 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
14290 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14291 array entries to represent __ieee128 versions of the
14292 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
14293 scalar_extract_sig, and scalar_insert_exp built-in functions.
14294 (altivec_resolve_overloaded_builtin): Add special case handling
14295 for the __builtin_scalar_insert_exp function, as represented by
14296 the P9V_BUILTIN_VEC_VSIEDP constant.
14297 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
14298 exponent support for __ieee128 argument.
14299 (VSESQP): Add scalar extract signature support for __ieee128
14301 (VSTDCNQP): Add scalar test negative support for __ieee128
14303 (VSIEQP): Add scalar insert exponent support for __int128 argument
14304 with __ieee128 result.
14305 (VSIEQPF): Add scalar insert exponent support for __ieee128
14306 argument with __ieee128 result.
14307 (VSTDCQP): Add scalar test data class support for __ieee128
14309 (VSTDCNQP): Add overload support for scalar test negative with
14310 __ieee128 argument.
14311 (VSTDCQP): Add overload support for scalar test data class
14312 __ieee128 argument.
14313 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
14314 UNSPEC_VSX_SXSIGDP.
14315 (UNSPEC_VSX_SIEXPQP): New constant.
14316 (xsxexpqp): New insn for VSX scalar extract exponent quad
14318 (xsxsigqp): New insn for VSX scalar extract significand quad
14320 (xsiexpqpf): New insn for VSX scalar insert exponent quad
14321 precision with floating point argument.
14322 (xststdcqp): New expand for VSX scalar test data class quad
14324 (xststdcnegqp): New expand for VSX scalar test negative quad
14326 (xststdcqp): New insn to match expansions for VSX scalar test data
14327 class quad precision and VSX scalar test negative quad precision.
14328 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
14329 special case operand checking to enforce that second operand of
14330 VSX scalar test data class with quad precision argument is a 7-bit
14332 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
14333 prototypes and descriptions of __ieee128 versions of
14334 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
14335 scalar_test_data_class, and scalar_test_neg built-in functions.
14337 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14339 PR tree-optimization/81162
14340 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
14341 replace a negate with an add.
14343 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
14345 * doc/invoke.texi (arm/-mcpu): Document +crypto.
14347 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14349 * config/arm/arm-c.c (arm_cpu_builtins): Define
14350 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
14352 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14354 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
14355 (armv8-r): Set ARM Cortex-R52 as default CPU.
14356 * config/arm/arm-tables.opt: Regenerate.
14357 * config/arm/arm-tune.md: Regenerate.
14358 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
14360 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
14361 extension for -mcpu=cortex-r52.
14363 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14365 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
14366 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
14367 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
14368 (fp-armv8): Define it as FP_ARMv8 only.
14369 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
14370 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
14372 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
14373 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
14374 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
14375 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
14376 than TARGET_FPU_ARMV8.
14377 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
14378 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
14379 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
14381 * config/arm/neon.md (neon_vrint): Likewise.
14382 (neon_vcvt): Likewise.
14383 (neon_<fmaxmin_op><mode>): Likewise.
14384 (<fmaxmin><mode>3): Likewise.
14385 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
14386 * config/arm/predicates.md (arm_cond_move_operator): Check against
14387 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
14389 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
14391 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
14392 to top of function.
14394 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14396 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
14397 loop in comment with memset.
14399 2017-07-14 Martin Liska <mliska@suse.cz>
14401 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
14402 * dwarf2out.c (is_java): Remove the function.
14403 (output_pubname): Remove usage of the function.
14404 (lower_bound_default): Remove usage of DW_LANG_Java.
14405 (gen_compile_unit_die): Likewise.
14406 * gcc.c: Remove compiler defaults for .java and .zip files.
14407 * gimple-expr.c (remove_suffix): Change as there's no longer
14408 extension than 4-letter one.
14409 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
14410 (gimplify_save_expr): Likewise.
14411 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
14412 as it's possible even for other languages than Java.
14413 * langhooks.h (struct lang_hooks): Remove Java from a comment.
14414 * lto-opts.c (lto_write_options): Remove reference to Java.
14415 * opts.c (strip_off_ending): Update file extension handling.
14416 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
14417 * tree-eh.c (lower_resx): Likewise.
14418 * tree.c (free_lang_data_in_type): Remove dead code.
14419 (find_decls_types_r): Likewise.
14420 (build_common_builtin_nodes): Remove Java from a comment.
14421 (verify_type): Remove dead code.
14422 * varasm.c (assemble_external): Remove Java from a comment.
14424 2017-07-14 Martin Liska <mliska@suse.cz>
14426 * opts.c (finish_options): Add quotes.
14427 (common_handle_option): Likewise.
14429 2017-07-14 Martin Liska <mliska@suse.cz>
14431 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
14432 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
14433 Remove N_SO_PASCAL.
14434 * dwarf2out.c (lower_bound_default): Do not handle
14436 (gen_compile_unit_die): Likewise.
14437 * gcc.c: Remove default extension binding for GNU Pascal.
14438 * stmt.c: Remove Pascal language from a comment.
14439 * xcoffout.c: Likewise.
14441 2017-07-13 David Malcolm <dmalcolm@redhat.com>
14444 * diagnostic-show-locus.c (fixit_cmp): New function.
14445 (layout::layout): Sort m_fixit_hints.
14446 (column_range::column_range): Assert that the values are valid.
14447 (struct char_span): New struct.
14448 (correction::overwrite): New method.
14449 (struct source_line): New struct.
14450 (line_corrections::add_hint): Add assertions. Reimplement memcpy
14451 calls in terms of classes source_line and char_span, and
14452 correction::overwrite.
14453 (selftest::test_overlapped_fixit_printing_2): New function.
14454 (selftest::diagnostic_show_locus_c_tests): Call it.
14456 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
14458 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
14459 early if there is no lhs.
14461 2017-07-13 Martin Liska <mliska@suse.cz>
14463 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
14464 (gen_reference_type_die): Likewise.
14465 * stor-layout.c: Remove Pascal-related comment.
14467 2017-07-13 Martin Liska <mliska@suse.cz>
14469 * opts.c (finish_options): Add quotes to error messages.
14470 (parse_sanitizer_options): Likewise.
14472 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
14474 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
14476 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
14478 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
14480 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
14482 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
14484 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
14486 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
14489 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
14490 provides the hardware capability bits, define the macro
14491 __BUILTIN_CPU_SUPPORTS__.
14492 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
14493 if GLIBC does not provide the hardware capability bits. Add a
14494 gcc_unreachable call if the built-in cpu function is neither
14495 __builtin_cpu_is nor __builtin_cpu_supports.
14496 (rs6000_get_function_versions_dispatcher): Change the warning
14497 that an old GLIBC is used which does not export the capability
14498 bits to be an error.
14499 * doc/extend.texi (target_clones attribute): Document the
14500 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
14501 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
14502 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
14503 the macros defined by GCC if the newer GLIBC is available.
14505 2017-07-12 Jeff Law <law@redhat.com>
14507 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
14508 remaining includes slightly.
14509 * config/riscv/riscv-builtins.c: Include profile-count.h.
14511 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
14514 * config/avr/avr.c (avr_set_current_function): In diagnostic
14515 messages: Quote keywords and (parts of) identifiers.
14516 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
14519 2017-07-12 Carl Love <cel@us.ibm.com>
14521 * config/rs6000/rs6000-c.c: Add support for built-in functions
14522 vector bool char vec_revb (vector bool char);
14523 vector bool short vec_revb (vector short char);
14524 vector bool int vec_revb (vector bool int);
14525 vector bool long long vec_revb (vector bool long long);
14526 * doc/extend.texi: Update the built-in documentation file for the
14527 new built-in functions.
14529 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14531 * config/s390/s390.md: Remove movcc splitter.
14533 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14535 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
14536 load/store on condition.
14538 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
14541 * config/avr/avr.c (avr_encode_section_info)
14542 [progmem && !TREE_READONLY]: Error if progmem object needs
14545 2017-07-11 Michael Collison <michael.collison@arm.com>
14547 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
14550 2017-07-11 Carl Love <cel@us.ibm.com>
14552 * config/rs6000/rs6000-c.c: Add support for builtins
14553 vector unsigned int vec_parity_lsbb (vector signed int);
14554 vector unsigned int vec_parity_lsbb (vector unsigned int);
14555 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
14556 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
14557 vector unsigned long long vec_parity_lsbb (vector signed long long);
14558 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
14559 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
14560 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
14561 * doc/extend.texi: Update the built-in documentation file for the
14562 new built-in functions.
14564 2017-07-11 David Malcolm <dmalcolm@redhat.com>
14566 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
14567 (layout::m_primary_loc): New field.
14568 (layout::layout): Initialize new field. Move location filtering
14569 logic from here to...
14570 (layout::maybe_add_location_range): ...this new method. Add
14571 support for filtering to just the lines already specified by other
14573 (layout::will_show_line_p): New method.
14574 (gcc_rich_location::add_location_if_nearby): New method.
14575 (selftest::test_add_location_if_nearby): New test function.
14576 (selftest::diagnostic_show_locus_c_tests): Call it.
14577 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
14580 2017-07-11 Tom de Vries <tom@codesourcery.com>
14582 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
14583 (bb_first_real_insn): New function.
14584 (nvptx_single): Add extra initialization of broadcasted condition
14587 2017-07-11 Nathan Sidwell <nathan@acm.org>
14589 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
14591 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
14593 * doc/extend.texi (AVR Function Attributes): Remove weblink to
14594 Binutils doc as TEXI will mess them up.
14595 * doc/invoke.texi (AVR Options): Same here.
14597 2017-07-11 Daniel Cederman <cederman@gaisler.com>
14599 * config/sparc/sparc.opt (mfix-ut700): New option.
14600 (mfix-gr712rc): Likewise.
14601 (sparc_fix_b2bst): New variable.
14602 * doc/invoke.texi (SPARC options): Document them.
14603 (ARM options): Fix warnings.
14604 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
14605 instructions to prevent sequences that can trigger the store-store
14606 errata for certain LEON3FT processors.
14607 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
14608 (sparc_option_override): Set sparc_fix_b2bst appropriately.
14609 * config/sparc/sparc.md (fix_b2bst): New attribute.
14610 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
14612 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
14615 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
14617 (*rsqrtsf2_sse): Ditto.
14619 (div<mode>3): Macroize insn from divdf3 and divsf3
14620 using MODEF mode iterator.
14622 2017-07-10 Martin Sebor <msebor@redhat.com>
14624 PR tree-optimization/80397
14625 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
14626 instead of testing for equality to INTEGER_TYPE.
14628 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
14630 * config.gcc: Remove uclibc from arc target spec.
14632 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
14634 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
14636 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14639 * lto-wrapper.c (remove_option): New function.
14640 (merge_and_complain): Merge PIC/PIE options more realistically.
14642 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
14644 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
14648 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
14649 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
14650 * config.in: Regenerate.
14651 * configure: Regenerate.
14652 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
14653 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
14654 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
14655 (TARGET_GASISR_PROLOGUES): ...target mask.
14656 * common/config/avr/avr-common.c
14657 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
14658 Set -mgas-isr-prologues.
14659 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
14660 INSERT_PASS_BEFORE for it.
14661 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
14662 * config/avr/avr.c (avr_option_override)
14663 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
14664 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
14665 (avr_attribute_table) <no_gccisr>: Add new function attribute.
14666 (avr_set_current_function) <is_no_gccisr>: Init machine field.
14667 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
14669 (make_avr_pass_pre_proep): New function.
14670 (emit_push_sfr) <treg>: Add argument to function and use it
14671 instead of TMP_REG.
14672 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
14673 and set machine->gasisr.yes.
14674 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
14675 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
14676 __gcc_isr.n_pushed to .L__stack_usage.
14677 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
14678 (avr_asm_final_postscan_insn): ...this new static function.
14679 * config/avr/avr.h (machine_function)
14680 <is_no_gccisr, use_L__stack_usage>: New fields.
14681 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
14682 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
14683 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
14684 (gasisr, *gasisr): New expander and insn.
14685 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
14686 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
14687 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
14689 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
14691 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
14694 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
14696 Move jump-tables out of .text again.
14699 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
14700 (ASM_OUTPUT_ADDR_VEC): New function.
14701 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
14702 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
14703 INSN_ADDRESSes as asm comment.
14704 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
14705 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
14706 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
14707 * config/avr/avr.md (*tablejump): Adjust comment.
14708 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
14709 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
14711 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
14712 (avr_output_addr_vec): New proto.
14713 (avr_log_t) <insn_addresses>: New field.
14715 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
14718 * config/i386/i386.c (ix86_function_arg_advance): Set
14719 outgoing_args_on_stack to true if there are outgoing arguments
14721 (ix86_function_arg): Likewise.
14722 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
14723 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
14724 * config/i386/i386.h (machine_function): Add
14725 outgoing_args_on_stack.
14727 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
14729 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
14730 supporting pthreds.
14731 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
14733 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
14735 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
14736 (REAL_H): Remove $(MACHMODE_H).
14737 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
14739 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
14740 $(MACHMODE_H) and double-int.h.
14741 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
14743 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
14746 2017-07-07 Andrew Pinski <apinski@cavium.com>
14748 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
14749 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
14751 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14753 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
14754 Add warning if GCC was not configured to link against a GLIBC that
14755 exports the hardware capability bits.
14756 (make_resolver_func): Make resolver function private and not a
14757 COMDAT function. Create the name with clone_function_name instead
14758 of make_unique_name.
14761 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
14762 correct operand in doing the split.
14764 2017-07-07 Carl Love <cel@us.ibm.com>
14766 * config/rs6000/rs6000-c: Add support for built-in function
14767 vector unsigned short vec_pack_to_short_fp32 (vector float,
14769 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
14770 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
14771 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
14772 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
14773 (convert_4f32_8i16): Add define_expand.
14774 * doc/extend.texi: Update the built-in documentation file for the
14775 new built-in function.
14777 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14779 * config/sparc/m8.md: New file.
14780 * config/sparc/sparc.md: Include m8.md.
14782 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14784 * config/sparc/sparc.opt: New option -mvis4b.
14785 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
14786 (sparc_option_override): Handle VIS4B.
14787 (enum sparc_builtins): Define
14788 SPARC_BUILTIN_DICTUNPACK{8,16,32},
14789 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
14790 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
14791 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
14792 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
14793 (check_constant_argument): New function.
14794 (sparc_vis_init_builtins): Define builtins
14795 __builtin_vis_dictunpack{8,16,32},
14796 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
14797 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
14798 __builtin_vis_fpcmpde{8,16,32}shl and
14799 __builtin_vis_fpcmpur{8,16,32}shl.
14800 (sparc_expand_builtin): Check that the constant operands to
14801 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
14802 constant and in range.
14803 * config/sparc/sparc-c.c (sparc_target_macros): Handle
14805 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
14806 (SPARC_IMM5_P): Likewise.
14807 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
14808 (enabled): Handle vis4b.
14809 (UNSPEC_DICTUNPACK): New unspec.
14810 (UNSPEC_FPCMPSHL): Likewise.
14811 (UNSPEC_FPUCMPSHL): Likewise.
14812 (UNSPEC_FPCMPDESHL): Likewise.
14813 (UNSPEC_FPCMPURSHL): Likewise.
14814 (cpu_feature): New CPU feature `vis4b'.
14815 (dictunpack{8,16,32}): New insns.
14816 (FPCSMODE): New mode iterator.
14817 (fpcscond): New code iterator.
14818 (fpcsucond): Likewise.
14819 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
14820 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
14821 (fpcmpde{8,16,32}{si,di}shl): Likewise.
14822 (fpcmpur{8,16,32}{si,di}shl): Likewise.
14823 * config/sparc/constraints.md: Define constraints `q' for unsigned
14824 2-bit integer constants and `t' for unsigned 5-bit integer
14826 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
14828 (imm5_operand_dictunpack16): Likewise.
14829 (imm5_operand_dictunpack32): Likewise.
14830 (imm2_operand): Likewise.
14831 * doc/invoke.texi (SPARC Options): Document -mvis4b.
14832 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
14833 ditunpack* and fpcmp*shl builtins.
14835 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14837 * config.gcc: Handle m8 in --with-{cpu,tune} options.
14838 * config.in: Add HAVE_AS_SPARC6 define.
14839 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
14841 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
14843 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
14844 (CPP_CPU_SPEC): Handle m8.
14845 (ASM_CPU_SPEC): Likewise.
14846 * config/sparc/sparc-opts.h (enum processor_type): Add
14848 * config/sparc/sparc.c (m8_costs): New struct.
14849 (sparc_option_override): Handle TARGET_CPU_m8.
14850 (sparc32_initialize_trampoline): Likewise.
14851 (sparc64_initialize_trampoline): Likewise.
14852 (sparc_issue_rate): Likewise.
14853 (sparc_register_move_cost): Likewise.
14854 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
14855 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
14856 (ASM_CPU64_DEFAULT_SPEC): Likewise.
14857 (CPP_CPU_SPEC): Handle M8.
14858 (ASM_CPU_SPEC): Likewise.
14859 (AS_M8_FLAG): Define.
14860 * config/sparc/sparc.md: Add m8 to the cpu attribute.
14861 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
14862 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
14864 * configure: Regenerate.
14865 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
14868 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14870 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
14872 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
14873 ("*movdi_insn_sp32"): Do not set v3pipe.
14874 ("*movsi_insn"): Likewise.
14875 ("*movdi_insn_sp64"): Likewise.
14876 ("*movsf_insn"): Likewise.
14877 ("*movdf_insn_sp32"): Likewise.
14878 ("*movdf_insn_sp64"): Likewise.
14879 ("*zero_extendsidi2_insn_sp64"): Likewise.
14880 ("*sign_extendsidi2_insn"): Likewise.
14881 ("*mov<VM32:mode>_insn"): Likewise.
14882 ("*mov<VM64:mode>_insn_sp64"): Likewise.
14883 ("*mov<VM64:mode>_insn_sp32"): Likewise.
14884 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
14885 ("<vlop:code><VL:mode>3"): Likewise.
14886 ("*not_<vlop:code><VL:mode>3"): Likewise.
14887 ("*nand<VL:mode>_vis"): Likewise.
14888 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
14889 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
14890 ("one_cmpl<VL:mode>2"): Likewise.
14891 ("faligndata<VM64:mode>_vis"): Likewise.
14892 ("alignaddrsi_vis"): Likewise.
14893 ("alignaddrdi_vis"): Likweise.
14894 ("alignaddrlsi_vis"): Likewise.
14895 ("alignaddrldi_vis"): Likewise.
14896 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
14897 ("bmaskdi_vis"): Likewise.
14898 ("bmasksi_vis"): Likewise.
14899 ("bshuffle<VM64:mode>_vis"): Likewise.
14900 ("cmask8<P:mode>_vis"): Likewise.
14901 ("cmask16<P:mode>_vis"): Likewise.
14902 ("cmask32<P:mode>_vis"): Likewise.
14903 ("pdistn<P:mode>_vis"): Likewise.
14904 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
14906 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14908 * config/sparc/sparc.md ("subtype"): New insn attribute.
14909 ("*wrgsr_sp64"): Set insn subtype.
14910 ("*rdgsr_sp64"): Likewise.
14911 ("alignaddrsi_vis"): Likewise.
14912 ("alignaddrdi_vis"): Likewise.
14913 ("alignaddrlsi_vis"): Likewise.
14914 ("alignaddrldi_vis"): Likewise.
14915 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
14916 ("fexpand_vis"): Likewise.
14917 ("fpmerge_vis"): Likewise.
14918 ("faligndata<VM64:mode>_vis"): Likewise.
14919 ("bshuffle<VM64:mode>_vis"): Likewise.
14920 ("cmask8<P:mode>_vis"): Likewise.
14921 ("cmask16<P:mode>_vis"): Likewise.
14922 ("cmask32<P:mode>_vis"): Likewise.
14923 ("fchksm16_vis"): Likewise.
14924 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
14925 ("fmean16_vis"): Likewise.
14926 ("fp<plusminus_insn>64_vis"): Likewise.
14927 ("<plusminus_insn>v8qi3"): Likewise.
14928 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
14929 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
14930 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
14931 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
14932 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
14933 ("*movqi_insn"): Likewise.
14934 ("*movhi_insn"): Likewise.
14935 ("*movsi_insn"): Likewise.
14936 ("movsi_pic_gotdata_op"): Likewise.
14937 ("*movdi_insn_sp32"): Likewise.
14938 ("*movdi_insn_sp64"): Likewise.
14939 ("movdi_pic_gotdata_op"): Likewise.
14940 ("*movsf_insn"): Likewise.
14941 ("*movdf_insn_sp32"): Likewise.
14942 ("*movdf_insn_sp64"): Likewise.
14943 ("*zero_extendhisi2_insn"): Likewise.
14944 ("*zero_extendqihi2_insn"): Likewise.
14945 ("*zero_extendqisi2_insn"): Likewise.
14946 ("*zero_extendqidi2_insn"): Likewise.
14947 ("*zero_extendhidi2_insn"): Likewise.
14948 ("*zero_extendsidi2_insn_sp64"): Likewise.
14949 ("ldfsr"): Likewise.
14950 ("prefetch_64"): Likewise.
14951 ("prefetch_32"): Likewise.
14952 ("tie_ld32"): Likewise.
14953 ("tie_ld64"): Likewise.
14954 ("*tldo_ldub_sp32"): Likewise.
14955 ("*tldo_ldub1_sp32"): Likewise.
14956 ("*tldo_ldub2_sp32"): Likewise.
14957 ("*tldo_ldub_sp64"): Likewise.
14958 ("*tldo_ldub1_sp64"): Likewise.
14959 ("*tldo_ldub2_sp64"): Likewise.
14960 ("*tldo_ldub3_sp64"): Likewise.
14961 ("*tldo_lduh_sp32"): Likewise.
14962 ("*tldo_lduh1_sp32"): Likewise.
14963 ("*tldo_lduh_sp64"): Likewise.
14964 ("*tldo_lduh1_sp64"): Likewise.
14965 ("*tldo_lduh2_sp64"): Likewise.
14966 ("*tldo_lduw_sp32"): Likewise.
14967 ("*tldo_lduw_sp64"): Likewise.
14968 ("*tldo_lduw1_sp64"): Likewise.
14969 ("*tldo_ldx_sp64"): Likewise.
14970 ("*mov<VM32:mode>_insn"): Likewise.
14971 ("*mov<VM64:mode>_insn_sp64"): Likewise.
14972 ("*mov<VM64:mode>_insn_sp32"): Likewise.
14974 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14976 * config/sparc/sparc.md ("type"): New insn type viscmp.
14977 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
14979 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
14980 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
14981 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
14982 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
14984 ("n7_vis_logical_11cycle"): Likewise.
14985 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
14986 * config/sparc/niagara2.md ("niag3_vis": Likewise.
14987 * config/sparc/niagara.md ("niag_vis"): Likewise.
14988 * config/sparc/ultra3.md ("us3_fga"): Likewise.
14989 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
14991 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14993 * config/sparc/sparc.md: New instruction type `bmask'.
14994 (bmaskdi_vis): Use the `bmask' type.
14995 (bmasksi_vis): Likewise.
14996 * config/sparc/ultra3.md (us3_array): Likewise.
14997 * config/sparc/niagara7.md (n7_array): Likewise.
14998 * config/sparc/niagara4.md (n4_array): Likewise.
14999 * config/sparc/niagara2.md (niag2_vis): Likewise.
15000 (niag3_vis): Likewise.
15001 * config/sparc/niagara.md (niag_vis): Likewise.
15003 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15005 * ipa-comdats.c: Remove optimize check from gate.
15006 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
15007 for functions not optimized.
15008 (ipa_fn_summary_read): Skip optimize check.
15009 (ipa_fn_summary_write): Likewise.
15010 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
15012 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
15014 (can_inline_edge_p): Check flag_pcc_struct_return for match.
15015 (check_callers): Give up on caller which is not optimized.
15016 (inline_small_functions): Likewise.
15017 (ipa_inline): Do not give up when not optimizing.
15018 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
15019 away unoptimizes cdtors.
15020 (whole_program_function_and_variable_visibility): Do
15021 ipa_discover_readonly_nonaddressable_vars in LTO mode.
15022 * ipa.c (process_references): Do not check optimize.
15023 (symbol_table::remove_unreachable_nodes): Update optimize check.
15024 (set_writeonly_bit): Update optimize check.
15025 (pass_ipa_cdtor_merge::gate): Do not check optimize.
15026 (pass_ipa_single_use::gate): Remove.
15028 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15030 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
15031 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
15032 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
15033 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
15034 permute_load, permute_store, adjust_extract, adjust_splat,
15035 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
15036 replace_swap_with_copy, dump_swap_insn_table,
15037 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
15038 recombine_lvx_pattern, recombine_stvx_pattern,
15039 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
15040 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
15041 to file rs6000-p8swap.c.
15042 * config/rs6000/rs6000-p8swap.c: New file.
15043 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
15044 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
15045 and rs6000*-*-* targets.
15047 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15049 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
15051 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15053 * lto-wrapper.c (merge_and_complain): Do not merge
15054 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
15055 fsigned_zeros, ftrapping_math, fwrapv.
15056 (append_compiler_options): Do not track these options.
15057 (append_linker_options): Likewie
15059 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15061 * cgraphunit.c (cgraph_node::finalize_function): When
15062 !flag_toplevel_reorde set no_reorder flag.
15063 (varpool_node::finalize_decl): Likewise.
15064 (symbol_table::compile): Drop no toplevel reorder path.
15066 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15068 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
15069 edges; zero probability is not better than uninitialized.
15071 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
15073 * asan.h (asan_sanitize_allocas_p): Declare.
15074 * asan.c (asan_sanitize_allocas_p): New function.
15075 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
15076 (handle_builtin_alloca): Likewise.
15077 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
15078 if !asan_sanitize_allocas_p.
15079 * params.def (asan-instrument-allocas): Add new option.
15080 * params.h (ASAN_PROTECT_ALLOCAS): Define.
15081 * opts.c (common_handle_option): Disable allocas sanitization for
15084 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
15086 * asan.c: Include gimple-fold.h.
15087 (get_last_alloca_addr): New function.
15088 (handle_builtin_stackrestore): Likewise.
15089 (handle_builtin_alloca): Likewise.
15090 (asan_emit_allocas_unpoison): Likewise.
15091 (get_mem_refs_of_builtin_call): Add new parameter, remove const
15092 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
15093 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
15094 (instrument_builtin_call): Pass gimple iterator to
15095 get_mem_refs_of_builtin_call.
15096 (last_alloca_addr): New global.
15097 * asan.h (asan_emit_allocas_unpoison): Declare.
15098 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
15099 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
15100 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
15101 if function calls alloca.
15102 * gimple-fold.c (replace_call_with_value): Remove static keyword.
15103 * gimple-fold.h (replace_call_with_value): Declare.
15104 * internal-fn.c: Include asan.h.
15105 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
15106 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
15108 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15110 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
15111 (C_SELFTEST_FLAGS): New.
15112 (CPP_SELFTEST_FLAGS): New.
15113 (SELFTEST_DEPS): New, from deps of s-selftest.
15114 (C_SELFTEST_DEPS): New, from deps of s-selftest.
15115 (CPP_SELFTEST_DEPS): New.
15116 (selftest): Add dependency on s-selftest-c++.
15117 (s-selftest): Rename to...
15118 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
15119 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
15120 than SELFTEST_FLAGS.
15121 (selftest-gdb): Rename to...
15122 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
15124 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
15125 (selftest-valgrind): Rename to...
15126 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
15128 (selftest-valgrind): Reintroduce as an alias for
15129 selftest-c-valgrind.
15130 (s-selftest-c++): New.
15131 (selftest-c++-gdb): New.
15132 (selftest-c++-valgrind): New.
15134 2017-07-06 Olivier Hainque <hainque@adacore.com>
15136 * gcc.c (process_command): When deciding if undefined variables
15137 should be ignored when processing specs, accept "gcc -v" as well.
15139 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15141 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
15142 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
15144 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
15146 * config/arm/arm-cpus.in (armv8-r): Add new entry.
15147 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
15148 * config/arm/arm-tables.opt: Regenerate.
15149 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
15151 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
15153 2017-07-06 Carl Love <cel@us.ibm.com>
15155 * ChangeLog: Clean up from mid air collision
15157 2017-07-06 Carl Love <cel@us.ibm.com>
15159 * config/rs6000/rs6000-c.c: Add support for built-in functions
15160 vector signed int vec_subc (vector signed int, vector signed int);
15161 vector signed __int128 vec_subc (vector signed __int128,
15162 vector signed __int128);
15163 vector unsigned __int128 vec_subc (vector unsigned __int128,
15164 vector unsigned __int128);
15165 vector signed int vec_sube (vector signed int, vector signed int,
15166 vector signed int);
15167 vector unsigned int vec_sube (vector unsigned int,
15168 vector unsigned int,
15169 vector unsigned int);
15170 vector signed __int128 vec_sube (vector signed __int128,
15171 vector signed __int128,
15172 vector signed__int128);
15173 vector unsigned __int128 vec_sube (vector unsigned __int128,
15174 vector unsigned __int128,
15175 vector unsigned __int128);
15176 vector signed int vec_subec (vector signed int, vector signed int,
15177 vector signed int);
15178 vector unsigned int vec_subec (vector unsigned int,
15179 vector unsigned int,
15180 vector unsigned int);
15181 vector signed __int128 vec_subec (vector signed __int128,
15182 vector signed __int128,
15183 vector signed__int128);
15184 vector unsigned __int128 vec_subec (vector unsigned __int128,
15185 vector unsigned __int128,
15186 vector unsigned __int128);
15187 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
15188 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
15189 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
15190 BU_ALTIVEC_OVERLOAD_X definitions.
15191 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
15192 * doc/extend.texi: Update the built-in documentation file for the new
15193 built-in functions.
15195 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15198 * diagnostic-show-locus.c (layout::layout): Use start and finish
15199 spelling location for the start and finish of each range.
15200 * genmatch.c (linemap_client_expand_location_to_spelling_point):
15201 Add unused aspect param.
15202 * input.c (expand_location_1): Add "aspect" param, and use it
15203 to access the correct part of the location.
15204 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
15206 (expand_location_to_spelling_point): Likewise.
15207 (linemap_client_expand_location_to_spelling_point): Add "aspect"
15208 param, and pass it to expand_location_1.
15210 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
15212 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
15213 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
15214 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
15215 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
15216 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
15217 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
15218 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
15219 _mm_maskz_getmant_ss): New intrinsics.
15220 (__builtin_ia32_getexpss128_mask): Changed to ...
15221 __builtin_ia32_getexpss128_round ... this.
15222 (__builtin_ia32_getexpsd128_mask): Changed to ...
15223 __builtin_ia32_getexpsd128_round ... this.
15224 * config/i386/i386-builtin-types.def
15225 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
15226 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
15227 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
15228 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
15229 __builtin_ia32_getmantss_mask_round): New builtins.
15230 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
15231 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
15232 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
15233 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
15234 * config/i386/sse.md
15235 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
15236 avx512f_sgetexp<mode><mask_scalar_name>
15237 <round_saeonly_scalar_name> ... this.
15238 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
15239 %0, %1, %2<round_saeonly_op3>}): Changed to ...
15240 vgetexp<ssescalarmodesuffix>
15241 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15242 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
15243 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
15244 avx512f_vgetmant<mode><mask_scalar_name>
15245 <round_saeonly_scalar_name> ... this.
15246 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
15247 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
15248 vgetmant<ssescalarmodesuffix>
15249 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
15250 %0<mask_scalar_operand4>, %1, %2
15251 <round_saeonly_scalar_mask_op4>, %3} ... this.
15252 * config/i386/subst.md (mask_scalar_operand4,
15253 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
15254 round_saeonly_scalar_nimm_predicate): New subst attributes.
15256 2017-07-06 Julia Koval <julia.koval@intel.com>
15258 * config/i386/i386.c (ix86_erase_embedded_rounding):
15259 Remove code for old rounding pattern.
15261 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
15263 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
15265 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
15267 * doc/sourcebuild.texi (Test Directives, Variants of
15268 dg-require-support): Add documentation for dg-require-stack-check.
15270 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
15272 * config/i386/subst.md (mask_scalar, round_scalar,
15273 round_saeonly_scalar): New meta-templates.
15274 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
15275 round_scalar_mask_operand3, round_scalar_mask_op3,
15276 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
15277 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
15278 round_saeonly_scalar_constraint,
15279 round_saeonly_scalar_prefix): New subst attribute.
15280 * config/i386/sse.md
15281 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
15282 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
15283 <round_scalar_name> ... this.
15284 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
15285 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
15286 <round_scalar_name> ... this.
15287 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
15288 <sse>_vm<code><mode>3<mask_scalar_name>
15289 <round_saeonly_scalar_name> ... this.
15290 (v<plusminus_mnemonic><ssescalarmodesuffix>
15291 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
15292 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
15293 v<plusminus_mnemonic><ssescalarmodesuffix>
15294 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15295 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
15296 (v<multdiv_mnemonic><ssescalarmodesuffix>
15297 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
15298 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
15299 v<multdiv_mnemonic><ssescalarmodesuffix>
15300 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15301 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
15302 (v<maxmin_float><ssescalarmodesuffix>
15303 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
15304 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
15305 v<maxmin_float><ssescalarmodesuffix>
15306 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15307 %0<mask_scalar_operand3>, %1, %<iptr>2
15308 <round_saeonly_scalar_mask_op3>} ... this.
15310 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
15312 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
15313 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
15315 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
15316 Alan Hayward <alan.hayward@arm.com>
15317 David Sherwood <david.sherwood@arm.com>
15319 * combine.c (simplify_if_then_else): Remove "enum" before
15321 * compare-elim.c (can_eliminate_compare): Likewise.
15322 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
15324 (aarch64_lookup_simd_builtin_type): Likewise.
15325 (aarch64_simd_builtin_type): Likewise.
15326 (aarch64_init_simd_builtin_types): Likewise.
15327 (aarch64_simd_expand_args): Likewise.
15328 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
15330 (aarch64_reverse_mask): Likewise.
15331 (aarch64_simd_emit_reg_reg_move): Likewise.
15332 (aarch64_gen_adjusted_ldpstp): Likewise.
15333 (aarch64_ccmp_mode_to_code): Likewise.
15334 (aarch64_operands_ok_for_ldpstp): Likewise.
15335 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15336 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
15338 (aarch64_min_divisions_for_recip_mul): Likewise.
15339 (aarch64_reassociation_width): Likewise.
15340 (aarch64_get_condition_code_1): Likewise.
15341 (aarch64_simd_emit_reg_reg_move): Likewise.
15342 (aarch64_simd_attr_length_rglist): Likewise.
15343 (aarch64_reverse_mask): Likewise.
15344 (aarch64_operands_ok_for_ldpstp): Likewise.
15345 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15346 (aarch64_gen_adjusted_ldpstp): Likewise.
15347 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
15349 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
15350 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
15351 (arm_lookup_simd_builtin_type): Likewise.
15352 (arm_simd_builtin_type): Likewise.
15353 (arm_init_simd_builtin_types): Likewise.
15354 (arm_expand_builtin_args): Likewise.
15355 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
15356 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
15357 (ft32_setup_incoming_varargs): Likewise.
15358 (ft32_function_arg): Likewise.
15359 (ft32_function_arg_advance): Likewise.
15360 (ft32_pass_by_reference): Likewise.
15361 (ft32_arg_partial_bytes): Likewise.
15362 (ft32_valid_pointer_mode): Likewise.
15363 (ft32_addr_space_pointer_mode): Likewise.
15364 (ft32_addr_space_legitimate_address_p): Likewise.
15365 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
15367 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
15368 (ix86_emit_outlined_ms2sysv_restore): Likewise.
15369 (iamcu_alignment): Likewise.
15370 (canonicalize_vector_int_perm): Likewise.
15371 (ix86_noce_conversion_profitable_p): Likewise.
15372 (ix86_mpx_bound_mode): Likewise.
15373 (ix86_operands_ok_for_move_multiple): Likewise.
15374 * config/microblaze/microblaze-protos.h
15375 (microblaze_expand_conditional_branch_reg): Likewise.
15376 * config/microblaze/microblaze.c
15377 (microblaze_expand_conditional_branch_reg): Likewise.
15378 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
15380 (rs6000_reassociation_width): Likewise.
15381 (rs6000_invalid_binary_op): Likewise.
15382 (fusion_p9_p): Likewise.
15383 (emit_fusion_p9_load): Likewise.
15384 (emit_fusion_p9_store): Likewise.
15385 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
15387 (riscv_hard_regno_mode_ok_p): Likewise.
15388 (riscv_address_insns): Likewise.
15389 (riscv_split_symbol): Likewise.
15390 (riscv_legitimize_move): Likewise.
15391 (riscv_function_value): Likewise.
15392 (riscv_hard_regno_nregs): Likewise.
15393 (riscv_expand_builtin): Likewise.
15394 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
15395 (riscv_build_integer): Likewise.
15396 (riscv_split_integer): Likewise.
15397 (riscv_legitimate_constant_p): Likewise.
15398 (riscv_cannot_force_const_mem): Likewise.
15399 (riscv_regno_mode_ok_for_base_p): Likewise.
15400 (riscv_valid_base_register_p): Likewise.
15401 (riscv_valid_offset_p): Likewise.
15402 (riscv_valid_lo_sum_p): Likewise.
15403 (riscv_classify_address): Likewise.
15404 (riscv_legitimate_address_p): Likewise.
15405 (riscv_address_insns): Likewise.
15406 (riscv_load_store_insns): Likewise.
15407 (riscv_force_binary): Likewise.
15408 (riscv_split_symbol): Likewise.
15409 (riscv_force_address): Likewise.
15410 (riscv_legitimize_address): Likewise.
15411 (riscv_move_integer): Likewise.
15412 (riscv_legitimize_const_move): Likewise.
15413 (riscv_legitimize_move): Likewise.
15414 (riscv_address_cost): Likewise.
15415 (riscv_subword): Likewise.
15416 (riscv_output_move): Likewise.
15417 (riscv_canonicalize_int_order_test): Likewise.
15418 (riscv_emit_int_order_test): Likewise.
15419 (riscv_function_arg_boundary): Likewise.
15420 (riscv_pass_mode_in_fpr_p): Likewise.
15421 (riscv_pass_fpr_single): Likewise.
15422 (riscv_pass_fpr_pair): Likewise.
15423 (riscv_get_arg_info): Likewise.
15424 (riscv_function_arg): Likewise.
15425 (riscv_function_arg_advance): Likewise.
15426 (riscv_arg_partial_bytes): Likewise.
15427 (riscv_function_value): Likewise.
15428 (riscv_pass_by_reference): Likewise.
15429 (riscv_setup_incoming_varargs): Likewise.
15430 (riscv_print_operand): Likewise.
15431 (riscv_elf_select_rtx_section): Likewise.
15432 (riscv_save_restore_reg): Likewise.
15433 (riscv_for_each_saved_reg): Likewise.
15434 (riscv_register_move_cost): Likewise.
15435 (riscv_hard_regno_mode_ok_p): Likewise.
15436 (riscv_hard_regno_nregs): Likewise.
15437 (riscv_class_max_nregs): Likewise.
15438 (riscv_memory_move_cost): Likewise.
15439 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
15440 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
15441 (rl78_addr_space_address_mode): Likewise.
15442 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15444 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15445 (rs6000_reassociation_width): Likewise.
15446 (rs6000_invalid_binary_op): Likewise.
15447 (fusion_p9_p): Likewise.
15448 (emit_fusion_p9_load): Likewise.
15449 (emit_fusion_p9_store): Likewise.
15450 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
15451 (ok_for_simple_move_operands): Likewise.
15452 (ok_for_simple_move_strict_operands): Likewise.
15453 (ok_for_simple_arith_logic_operands): Likewise.
15454 (visium_legitimize_reload_address): Likewise.
15455 (visium_select_cc_mode): Likewise.
15456 (output_cbranch): Likewise.
15457 (visium_split_double_move): Likewise.
15458 (visium_expand_copysign): Likewise.
15459 (visium_expand_int_cstore): Likewise.
15460 (visium_expand_fp_cstore): Likewise.
15461 * config/visium/visium.c (visium_pass_by_reference): Likewise.
15462 (visium_function_arg): Likewise.
15463 (visium_function_arg_advance): Likewise.
15464 (visium_libcall_value): Likewise.
15465 (visium_setup_incoming_varargs): Likewise.
15466 (visium_legitimate_constant_p): Likewise.
15467 (visium_legitimate_address_p): Likewise.
15468 (visium_legitimize_address): Likewise.
15469 (visium_secondary_reload): Likewise.
15470 (visium_register_move_cost): Likewise.
15471 (visium_memory_move_cost): Likewise.
15472 (prepare_move_operands): Likewise.
15473 (ok_for_simple_move_operands): Likewise.
15474 (ok_for_simple_move_strict_operands): Likewise.
15475 (ok_for_simple_arith_logic_operands): Likewise.
15476 (visium_function_value_1): Likewise.
15477 (rtx_ok_for_offset_p): Likewise.
15478 (visium_legitimize_reload_address): Likewise.
15479 (visium_split_double_move): Likewise.
15480 (visium_expand_copysign): Likewise.
15481 (visium_expand_int_cstore): Likewise.
15482 (visium_expand_fp_cstore): Likewise.
15483 (visium_split_cstore): Likewise.
15484 (visium_select_cc_mode): Likewise.
15485 (visium_split_cbranch): Likewise.
15486 (output_cbranch): Likewise.
15487 (visium_print_operand_address): Likewise.
15488 * expmed.c (flip_storage_order): Likewise.
15489 * expmed.h (emit_cstore): Likewise.
15490 (flip_storage_order): Likewise.
15491 * genrecog.c (validate_pattern): Likewise.
15492 * hsa-gen.c (gen_hsa_addr): Likewise.
15493 * internal-fn.c (expand_arith_overflow): Likewise.
15494 * ira-color.c (allocno_copy_cost_saving): Likewise.
15495 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15496 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
15497 (process_invariant_for_inheritance): Likewise.
15498 * lra-eliminations.c (move_plus_up): Likewise.
15499 * omp-low.c (lower_oacc_reductions): Likewise.
15500 * simplify-rtx.c (simplify_subreg): Likewise.
15501 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
15502 (TARGET_CHKP_BOUND_MODE): Likewise..
15503 * targhooks.c (default_chkp_bound_mode): Likewise.
15504 (default_setup_incoming_vararg_bounds): Likewise.
15505 * targhooks.h (default_chkp_bound_mode): Likewise.
15506 (default_setup_incoming_vararg_bounds): Likewise.
15507 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
15508 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
15509 (have_whole_vector_shift): Likewise.
15510 * tree-vect-stmts.c (vectorizable_load): Likewise.
15511 * doc/tm.texi: Regenerate.
15513 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
15515 Graceful degrade if Binutils PR21472 is not available.
15518 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
15519 .rodata in flash test fails.
15520 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
15521 * confgure: Regenerate.
15522 * config.in: Regenerate.
15523 * config/avr/avr.c (avr_asm_named_section)
15524 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
15525 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
15526 (avr_asm_init_sections): Same.
15528 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15530 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
15531 (fma<VH:mode>4_intrinsic): Likewise.
15532 (*fmsub<VCVTF:mode>4): Likewise.
15533 (*fmsub<VH:mode>4_intrinsic): Likewise.
15535 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
15538 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
15539 Don't depend on "optimize > 0".
15540 (out_movhi_r_mr, out_movqi_mr_r): Same.
15541 (out_movhi_mr_r, out_movqi_r_mr): Same.
15542 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
15543 io_address_operand on "optimize > 0".
15545 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15547 * tree-loop-distribution.c: Add general explanantion on the pass.
15548 (generate_loops_for_partition): Mark distributed loop.
15549 (pg_add_dependence_edges): New parameter. Handle alias data
15550 dependence specially and record it in the parameter if asked.
15551 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
15552 (init_partition_graph_vertices, add_partition_graph_edge): New.
15553 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
15554 (free_partition_graph_vdata, build_partition_graph): New.
15555 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
15556 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
15557 (data_ref_segment_size, latch_dominated_by_data_ref): New.
15558 (compute_alias_check_pairs, version_loop_by_alias_check): New.
15559 (version_for_distribution_p, finalize_partitions): New.
15560 (distribute_loop): Handle alias data dependence specially. Factor
15561 out loop fusion code as functions and call these functions.
15563 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15565 * tree-loop-distribution.c (classify_partition): New parameter and
15566 better handle reduction statement.
15567 (rdg_build_partitions): Revise comment.
15568 (distribute_loop): Compute statements in all partitions and pass it
15569 to classify_partition.
15571 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15573 * tree-loop-distribution.c (enum partition_type): New.
15574 (struct partition): New field type.
15575 (partition_merge_into): Add parameter. Update partition type.
15576 (data_dep_in_cycle_p, update_type_for_merge): New functions.
15577 (build_rdg_partition_for_vertex): Compute partition type.
15578 (rdg_build_partitions): Dump partition type.
15579 (distribute_loop): Update calls to partition_merge_into.
15581 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15583 * tree-loop-distribution.c (struct ddr_hasher): New.
15584 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
15586 (classify_partition): Call get_data_dependence.
15587 (pg_add_dependence_edges): Ditto.
15588 (distribute_loop): Release data dependence hash table.
15590 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15592 * tree-loop-distribution.c (ref_base_address): Delete.
15593 (similar_memory_accesses): Rename ...
15594 (share_memory_accesses): ... to this. Check if partitions access
15595 the same memory reference.
15596 (distribute_loop): Call share_memory_accesses.
15598 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15600 * tree-loop-distribution.c (struct partition): New field recording
15601 its data reference.
15602 (partition_alloc, partition_free): Init and release data refs.
15603 (partition_merge_into): Merge data refs.
15604 (build_rdg_partition_for_vertex): Collect data refs for partition.
15605 (pg_add_dependence_edges): Change parameters from vector to bitmap.
15607 (distribute_loop): Remve data refs from vertice data of partition
15610 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15612 * tree-loop-distribution.c (params.h): Include header file.
15613 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
15614 (datarefs_vec): New global var.
15615 (create_rdg_vertices): Use datarefs_vec directly.
15616 (free_rdg): Don't free data references.
15617 (build_rdg): Update use. Don't free data references.
15618 (distribute_loop): Compute global variable for data references.
15619 Bail out if there are too many data references.
15621 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15623 * tree-loop-distribution.c (loop_nest): New global var.
15624 (build_rdg): Use loop directly, rather than loop nest.
15625 (pg_add_dependence_edges): Remove loop nest parameter. Use global
15627 (distribute_loop): Compute global variable loop nest. Update use.
15629 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15631 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
15632 (partition_merge_into): New parameter. Dump reason for fusion.
15633 (distribute_loop): Update use of partition_merge_into.
15635 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15637 * tree-loop-distribution.c (bb_top_order_index): New.
15638 (bb_top_order_index_size, bb_top_order_cmp): New.
15639 (stmts_from_loop): Use topological order.
15640 (pass_loop_distribution::execute): Compute and release topological
15641 order for basic blocks.
15643 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15645 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
15648 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15650 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
15651 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
15652 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
15653 * internal-fn.def (LOOP_DIST_ALIAS): New.
15654 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
15655 (fold_loop_internal_call): ... this.
15656 (vect_loop_dist_alias_call): New function.
15657 (set_uid_loop_bbs): Call fold_loop_internal_call.
15658 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
15661 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
15664 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
15665 Require dead FLAGS_REG at the beginning of a peephole.
15667 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
15670 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
15671 arguments in the call to __builtin_ia32_sbb_u32.
15672 (_subborrow_u64): Swap _X and _Y arguments in the call to
15673 __builtin_ia32_sbb_u64.
15675 2017-07-04 Jakub Jelinek <jakub@redhat.com>
15678 * tree-vrp.c (compare_assert_loc): Turn into a function template
15679 with stable template parameter. Only test if a->e is NULL,
15680 !a->e == !b->e has been verified already. Use e == NULL or
15681 e != NULL instead of e or ! e tests. If stable is true, don't use
15682 iterative_hash_expr, on the other side allow a or b or both NULL
15683 and sort the NULLs last.
15684 (process_assert_insertions): Sort using compare_assert_loc<false>
15685 instead of compare_assert_loc, later sort using
15686 compare_assert_loc<true> before calling process_assert_insertions_for
15687 in a loop. Use break instead of continue once seen NULL pointer.
15689 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15691 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
15692 Cortex-R7 and Cortex-R8 processors.
15694 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15696 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
15697 uninitialized while src is not.
15699 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
15701 * common/config/arm/arm-common.c: Adjust include path for
15703 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
15704 (arm-cpu.h): Create in build directory. Adjust dependency rules.
15705 (arm-cpu-data.h): Likewise.
15706 (arm-cpu-cdata.h): Likewise.
15707 * config/arm/arm-cpu.h: Delete.
15708 * config/arm/arm-cpu-cdata.h: Delete.
15709 * config/arm/arm-cpu-data.h: Delete.
15711 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
15713 * config/arm/arm-cpus.in (cortex-a55): New.
15714 (cortex-a75): Likewise.
15715 (cortex-a75.cortex-a55): Likewise.
15716 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
15718 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
15719 * config/arm/arm-cpu-cdata.h: Regenerate.
15720 * config/arm/arm-cpu-data.h: Regenerate.
15721 * config/arm/arm-cpu.h: Regenerate.
15722 * config/arm/arm-tables.opt: Regenerate.
15723 * config/arm/arm-tune.md: Regenerate.
15725 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15727 * haifa-sched.c (sched_create_recovery_edges): Update profile.
15729 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15731 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
15734 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
15736 PR tree-optimization/81292
15737 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
15738 full_string_p, also call adjust_related_strinfos if the adjustment
15739 is simple, otherwise invalidate related strinfos.
15741 2017-07-04 Martin Liska <mliska@suse.cz>
15744 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
15745 newly created variable as DECL_IGNORED_P.
15747 2017-07-04 Martin Liska <mliska@suse.cz>
15750 * ipa-inline.c (inline_small_functions):
15751 Use xstrdup_for_dump.
15753 2017-07-04 Tom de Vries <tom@codesourcery.com>
15755 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
15757 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
15760 * config/darwin.c (darwin_function_switched_text_sections):
15763 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
15765 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
15767 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
15769 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
15771 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15773 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
15774 min_profitable_iters, and th as inclusive lower bounds.
15775 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
15776 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
15777 for min_profitable_iters and min_profitable_estimate.
15778 (vect_transform_loop): Treat th as an inclusive lower bound.
15779 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
15781 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
15784 * config/darwin.c (darwin_function_switched_text_sections):
15785 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
15786 in two pieces, and suppress the use of buf.
15788 2017-07-03 Nathan Sidwell <nathan@acm.org>
15790 * hash-table.h (hash_table_mod1): Fix indentation.
15792 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15794 PR middle-end/81290
15795 * predict.c (force_edge_cold): Be more careful about propagation
15797 * profile-count.h (profile_probability::guessed,
15798 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
15800 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
15802 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
15804 * doc/invoke.texi (rcpc architecture extension): Document it.
15806 2017-07-03 Richard Biener <rguenther@suse.de>
15808 PR tree-optimization/60510
15809 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
15810 the scalar reduction PHI and use it.
15811 (vectorizable_reduction): Properly guard the single_defuse_cycle
15812 path for non-SLP reduction chains where we cannot use it.
15813 Rework reduc_def/index and vector type deduction. Rework
15814 vector operand gathering during reduction op code-gen.
15815 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
15816 chains dissolve the chain and leave it to non-SLP reduction
15819 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15821 * tree-data-ref.h (dr_alignment): Declare.
15822 * tree-data-ref.c (dr_alignment): New function.
15823 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
15824 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
15826 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
15828 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15830 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
15831 and base_misalignment fields.
15832 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
15833 * tree-data-ref.c: Include builtins.h.
15834 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
15835 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
15836 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
15837 * tree-vect-data-refs.c: Include tree-cfg.h.
15838 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
15839 fields instead of calculating an alignment here.
15840 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
15841 innermost_loop_behavior fields.
15843 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15845 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
15847 (DR_STEP_ALIGNMENT): New macro.
15848 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
15849 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
15850 (create_data_ref): Print it.
15851 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
15852 to tell whether the step preserves vector (mis)alignment.
15853 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
15854 Move the check for an integer step and generalise to all INTEGER_CST.
15855 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
15856 Print the outer step alignment.
15858 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15860 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
15861 with offset_alignment.
15862 (DR_ALIGNED_TO): Delete.
15863 (DR_OFFSET_ALIGNMENT): New macro.
15864 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
15865 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
15866 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
15867 (create_data_ref): Likewise.
15868 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
15869 (vect_analyze_data_refs): Likewise.
15870 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
15871 creating dummy innermost behavior.
15873 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15875 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
15876 with a "innermost_loop_behavior *" and refeence tree.
15877 * tree-data-ref.c (dr_analyze_innermost): Likewise.
15878 (create_data_ref): Update call accordingly.
15879 * tree-predcom.c (find_looparound_phi): Likewise.
15881 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15883 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
15884 fields with dr_wrt_vec_loop.
15885 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
15886 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
15887 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
15888 (vect_dr_behavior): New function.
15889 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
15890 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
15891 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
15892 track whether the step preserves the misalignment.
15893 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
15894 Use vect_dr_behavior.
15895 (vect_setup_realignment): Update call accordingly.
15896 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
15897 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
15898 call to vect_create_addr_base_for_vector_ref.
15899 (vect_create_cond_for_align_checks): Likewise.
15900 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
15901 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
15902 (vect_recog_mask_conversion_pattern): Likewise.
15903 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
15904 (new_stmt_vec_info): Remove redundant zeroing.
15906 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
15908 * common/config/arm/arm-common.c (arm_be8_option): New function.
15909 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
15910 (ISA_ARMv6): Add isa_bit_be8.
15911 * config/arm/arm.h (arm_be8_option): Add prototype.
15912 (BE8_SPEC_FUNCTION): New define.
15913 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
15914 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
15915 (mlittle-endian): Similarly.
15916 (mbe8, mbe32): New options.
15917 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
15918 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
15920 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15922 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
15924 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15926 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
15927 (cleanup_tree_cfg_bb): Use it.
15928 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
15930 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
15932 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15935 * loop-doloop.c (add_test): Update profile.
15937 2017-07-03 Martin Liska <mliska@suse.cz>
15940 * sanopt.c (rewrite_usage_of_param): New function.
15941 (sanitize_rewrite_addressable_params): Likewise.
15942 (pass_sanopt::execute): Call rewrite_usage_of_param.
15944 2017-07-03 Richard Biener <rguenther@suse.de>
15946 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
15947 back to using VIEW_CONVERT_EXPR.
15949 2017-07-03 Martin Liska <mliska@suse.cz>
15952 * doc/extend.texi: Document when a resolver function is
15953 generated for target_clones.
15955 2017-07-03 Martin Liska <mliska@suse.cz>
15957 * asan.c (asan_emit_stack_protection): Unpoison just red zones
15958 and shadow memory of auto variables which are subject of
15959 use-after-scope sanitization.
15960 (asan_expand_mark_ifn): Add do set only when is_poison.
15962 2016-07-03 Richard Biener <rguenther@suse.de>
15964 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
15966 (vect_force_simple_reduction): Record reduction def -> phi mapping.
15967 (vectorizable_reduction): Perform reduction PHI creation when
15968 visiting a reduction PHI and adjust and simplify code generation
15969 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
15970 (vect_transform_loop): Visit reduction PHIs.
15971 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
15972 defs into the SLP tree.
15973 (vect_build_slp_tree): Reduction defs terminate the recursion.
15974 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
15976 (vect_get_vec_defs_for_stmt_copy): Export.
15977 (vect_get_vec_defs): Likewise.
15978 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
15980 (vect_get_vec_defs_for_stmt_copy): Declare.
15981 (vect_get_vec_defs): Likewise.
15983 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15985 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
15986 parameter with a "loop" parameter and use it instead of the
15987 loop containing DR_STMT. Don't check simple_iv when doing
15988 BB analysis. Describe the two analysis modes in the comment.
15990 2017-07-03 Tom de Vries <tom@codesourcery.com>
15992 PR tree-optimization/69468
15993 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
15994 (find_same_succ_bb): Handle ignore_edge_flags.
15996 2017-07-03 Tom de Vries <tom@codesourcery.com>
15998 PR tree-optimization/81192
15999 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
16001 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
16003 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
16005 2017-07-03 Tom de Vries <tom@codesourcery.com>
16007 PR tree-optimization/81192
16008 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
16009 BB_SAME_SUCC (bb) == NULL.
16011 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16013 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
16016 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16018 * dumpfile.c: Include profile-count.h
16019 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
16021 (insert_cond_bb): Update profile.
16022 * tree-cfg.h (insert_cond_bb): Update prototype.
16023 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
16024 * tree-dump.c: Do not include tree-cfg.
16026 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16028 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
16030 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16032 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
16035 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16037 * tree-complex.c (expand_complex_div_wide): update profile.
16039 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16040 Alan Hayward <alan.hayward@arm.com>
16041 David Sherwood <david.sherwood@arm.com>
16043 * Makefile.in (MACHMODE_H): Remove insn-modes.h
16044 (CORETYPES_H): New define.
16045 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
16046 (insn-modes-inline.h, s-modes-inline-h): New rules.
16047 (generated_files): Add insn-modes-inline.h.
16048 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
16049 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
16050 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
16051 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
16052 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
16053 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
16054 (build/gencodes.o, build/genconditions.o): Likewise.
16055 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
16056 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
16057 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
16058 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
16059 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
16060 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
16061 * coretypes.h: Include everything up to real.h for generators.
16062 Include insn-modes.h first. Include wide-int-print.h after
16063 wide-int.h. Include insn-modes-inline.h and then machmode.h.
16064 * machmode.h: Don't include insn-modes.h here.
16065 * function-tests.c: Remove includes of signop.h, machmode.h,
16066 double-int.h and wide-int.h.
16068 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
16070 * optc-save-gen.awk: Likewise.
16071 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
16072 * godump.c: Remove include of wide-int-print.h.
16073 * pretty-print.h: Likewise.
16074 * wide-int-print.cc: Likewise.
16075 * wide-int.cc: Likewise.
16076 * hash-map-tests.c: Remove include of signop.h.
16077 * hash-set-tests.c: Likewise.
16078 * rtl-tests.c: Likewise.
16079 * mkconfig.sh: Remove include of machmode.h.
16080 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
16082 (emit_insn_modes_inline_h): ...this new function. Emit the code
16083 into an insn-modes-inline.h header file, adding appropriate
16084 include guards and end comments.
16085 (emit_insn_modes_c_header): Remove include of machmode.h.
16086 (emit_min_insn_modes_c_header): Include coretypes.h rather than
16088 (main): Handle -i flag and call emit_insn_modes_inline_h when
16091 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16093 * tree-ssa-strlen.c (strinfo): Rename the length field to
16094 nonzero_chars. Add a full_string_p field.
16095 (compare_nonzero_chars, zero_length_string_p): New functions.
16096 (get_addr_stridx): Add an offset_out parameter.
16097 Use compare_nonzero_chars.
16098 (get_stridx): Update accordingly. Use compare_nonzero_chars.
16099 (new_strinfo): Update after above changes to strinfo.
16100 (set_endptr_and_length): Set full_string_p.
16101 (get_string_length): Update after above changes to strinfo.
16102 (unshare_strinfo): Update call to new_strinfo.
16103 (maybe_invalidate): Likewise.
16104 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
16105 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
16106 as a uhwi instead of an shwi. Update after above changes to
16107 strinfo and new_strinfo.
16108 (zero_length_string): Assert that chainsi contains full strings.
16109 Use zero_length_string_p. Update call to new_strinfo.
16110 (adjust_related_strinfos): Update after above changes to strinfo.
16111 Copy full_string_p from origsi.
16112 (adjust_last_stmt): Use zero_length_string_p.
16113 (handle_builtin_strlen): Update after above changes to strinfo and
16114 new_strinfo. Install the lhs as the string length if the previous
16115 entry didn't describe a full string.
16116 (handle_builtin_strchr): Update after above changes to strinfo
16118 (handle_builtin_strcpy): Likewise.
16119 (handle_builtin_strcat): Likewise.
16120 (handle_builtin_malloc): Likewise.
16121 (handle_pointer_plus): Likewise.
16122 (handle_builtin_memcpy): Likewise. Track nonzero characters
16123 that aren't necessarily followed by a nul terminator.
16124 (handle_char_store): Likewise.
16126 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16128 PR tree-optimization/80769
16129 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
16130 for malloc and calloc. Document the new invariant that all related
16131 strinfos have delayed lengths or none do.
16132 (verify_related_strinfos): Move earlier in file.
16133 (set_endptr_and_length): New function, split out from...
16134 (get_string_length): ...here. Also set the lengths of related
16136 (zero_length_string): Assert that chainsi has known (rather than
16138 (adjust_related_strinfos): Likewise.
16140 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16142 PR tree-optimization/81136
16143 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
16144 assert that two references with the same misalignment have the same
16145 compile-time misalignment if those compile-time misalignments
16148 2017-07-01 Andi Kleen <ak@linux.intel.com>
16150 * print-tree.c (print_node): Print all attributes.
16152 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16154 * cfg.c (scale_bbs_frequencies): New function.
16155 * cfg.h (scale_bbs_frequencies): Declare it.
16156 * cfgloopanal.c (single_likely_exit): Cleanup.
16157 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
16159 (scale_loop_profile): Likewise.
16160 (loop_version): Likewise.
16161 (create_empty_loop_on_edge): Update.
16162 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
16163 scale_loop_frequencies, scale_loop_profile, loopify,
16164 loop_version): Update prototypes.
16165 * modulo-sched.c (sms_schedule): Update.
16166 * predict.c (unlikely_executed_edge_p): Also check probability.
16167 (probably_never_executed_edge_p): Fix typo.
16168 * tree-if-conv.c (version_loop_for_if_conversion): Update.
16169 * tree-parloops.c (gen_parallel_loop): Update.
16170 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
16171 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16172 * tree-ssa-loop-split.c (split_loop): Update.
16173 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
16174 * tree-vect-loop-manip.c (vect_do_peeling): Update.
16175 (vect_loop_versioning): Update.
16176 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16178 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16180 * trans-mem.c (split_bb_make_tm_edge): Update profile.
16182 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16184 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
16185 to keep profile consistent.
16187 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16189 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
16190 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
16191 * profile-count.h (max_safe_multiplier): Make unsigned.
16192 (profile_count::guessed_zero): New.
16194 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16196 * bb-reorder.c (fix_up_crossing_landing_pad,
16197 fix_crossing_conditional_branches): Use make_single_succ_edge
16198 to keep profile consistent.
16200 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16202 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
16205 2017-07-01 Jakub Jelinek <jakub@redhat.com>
16208 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
16209 the right scopes, make sure cond_jump isn't preserved between multiple
16210 iterations. Search for fallthru edge whenever there are 3+ edges and
16211 use find_fallthru_edge for it.
16213 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16215 Patch by Alexander Monakov <amonakov@ispras.ru>
16216 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
16217 probabilities consistently.
16219 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16221 * pa.c (pa_expand_compare_and_swap_loop): Update call of
16222 emit_cmp_and_jump_insns.
16224 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16227 * tree-inline.c (expand_call_inline): Combine profile statuses.
16229 2017-06-30 Andrew Pinski <apinski@cavium.com>
16231 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
16232 fold_stmt returned true.
16234 2017-06-30 Nathan Sidwell <nathan@acm.org>
16236 * ggc.h (empty_string): Delete.
16237 * cfgexpand.c (expand_asm_stmt): Use plain "".
16238 * optabs.c (expand_asm_memory_barrier): Likewise.
16239 * stringpool.c (empty_string): Delete.
16240 (digit_vector, digit_string): Delete.
16241 (ggc_alloc_string): Use plain "", don't optimize single digit
16242 strings. Use ggc_alloc_atomic.
16244 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
16246 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
16247 comparison set and one other set, use the cost of the non-comparison
16250 2017-06-30 Nathan Sidwell <nathan@acm.org>
16252 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
16255 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
16257 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
16258 loops. Remove now unneeded calls to gimple_switch_set_label() that
16259 just set removed labels to NULL_TREE.
16261 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
16263 * tree-ssanames.c (set_range_info_raw): Abstract from ...
16264 (set_range_info): ...here. Only call set_range_info_raw if domain
16266 (set_nonzero_bits): Call set_range_info_raw.
16267 * tree-ssanames.h (set_range_info_raw): New.
16269 2017-06-30 Jakub Jelinek <jakub@redhat.com>
16272 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
16273 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
16274 of nonimmediate_operand and <store_mask_constraint> instead of m
16275 for the input operand. For V8FI iterator, always split if input
16276 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
16277 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
16278 <store_mask_predicate> instead of register_operand and
16279 <store_mask_constraint> instead of v for the input operand. Make
16280 sure both operands aren't MEMs for if not <mask_applied>.
16282 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
16284 * lto-wrapper.c (copy_file) Close both file descriptors before
16287 2017-06-30 Martin Liska <mliska@suse.cz>
16290 * multiple_target.c (create_dispatcher_calls): Make ifunc
16291 also for function that don't have calls or are not referenced.
16293 2017-06-30 Richard Biener <rguenther@suse.de>
16295 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
16296 analyze the first scalar stmt. Move vector type computation
16297 for the BB case here from ...
16298 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
16299 live operation processing in the SLP case properly.
16301 2017-06-30 Richard Biener <rguenther@suse.de>
16303 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
16305 2017-06-30 Martin Liska <mliska@suse.cz>
16308 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
16309 before BUILT_IN_UNWIND_RESUME when ASAN is used.
16311 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
16313 * doc/invoke.texi (AArch64): Add missing options and remove redundant
16316 2017-06-30 Richard Biener <rguenther@suse.de>
16318 PR tree-optimization/81249
16319 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
16320 condition reduction result to original scalar type.
16322 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16324 * profile-count.h (enum profile_quality): Fix typos and whitespace
16327 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16329 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
16330 type for branch probabilities.
16332 2017-06-29 Julian Brown <julian@codesourcery.com>
16333 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16335 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
16336 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
16337 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
16338 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
16340 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16342 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
16343 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
16344 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
16345 CC usage from generic code to here.
16346 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
16347 CC usage into the target macros.
16349 2017-06-29 Maya Rashish <coypu@sdf.org>
16351 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
16354 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16356 * arm/arm-builtins.c: Include profile-count.h
16357 * except.c (sjlj_emit_function_enter): Use
16358 profile_probability::unlikely.
16360 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16362 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
16363 and tocrel_offset be pointer args rather than implicitly using
16365 (legitimate_constant_pool_address_p, rs6000_emit_move,
16366 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
16367 tocrel_offset and use in toc_relative_expr_p call.
16368 (print_operand, print_operand_address): Use static tocrel_base_oac
16369 and tocrel_offset_oac.
16370 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
16373 2017-06-29 Maya Rashish <coypu@sdf.org>
16375 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
16377 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
16379 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
16380 objects, take into account only the alignment of 'op0' and 'mode1' if
16383 2017-06-29 Steve Ellcey <sellcey@cavium.com>
16385 * ccmp.c (ccmp_tree_comparison_p): New function.
16386 (ccmp_candidate_p): Update to use above function.
16387 (get_compare_parts): New function.
16388 (expand_ccmp_next): Update to use new functions.
16389 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
16391 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
16392 take mode as argument.
16393 * ccmp.h (expand_ccmp_expr): Add mode as argument.
16394 * expr.c (expand_expr_real_1): Pass mode as argument.
16396 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
16398 * combine.c (combine_instructions): Print insns to dump_file, together
16401 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16403 * asan.c (asan_emit_stack_protection): Update.
16404 (create_cond_insert_point): Update.
16405 * auto-profile.c (afdo_propagate_circuit): Update.
16406 * basic-block.h (struct edge_def): Turn probability to
16407 profile_probability.
16408 (EDGE_FREQUENCY): Update.
16409 * bb-reorder.c (find_traces_1_round): Update.
16410 (better_edge_p): Update.
16411 (sanitize_hot_paths): Update.
16412 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
16413 (make_single_succ_edge): Update.
16414 (check_bb_profile): Update.
16415 (dump_edge_info): Update.
16416 (update_bb_profile_for_threading): Update.
16417 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
16418 probabilitycount to 0.
16419 * cfgbuild.c (compute_outgoing_frequencies): Update.
16420 * cfgcleanup.c (try_forward_edges): Update.
16421 (outgoing_edges_match): Update.
16422 (try_crossjump_to_edge): Update.
16423 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
16424 (expand_gimple_tailcall): Update.
16425 (construct_init_block): Use make_single_succ_edge.
16426 (construct_exit_block): Use make_single_succ_edge.
16427 * cfghooks.c (verify_flow_info): Update.
16428 (redirect_edge_succ_nodup): Update.
16429 (split_edge): Update.
16430 (account_profile_record): Update.
16431 * cfgloopanal.c (single_likely_exit): Update.
16432 * cfgloopmanip.c (scale_loop_profile): Update.
16433 (set_zero_probability): Remove.
16434 (duplicate_loop_to_header_edge): Update.
16435 * cfgloopmanip.h (loop_version): Update prototype.
16436 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
16437 (force_nonfallthru_and_redirect): Update.
16438 (update_br_prob_note): Update.
16439 (rtl_verify_edges): Update.
16440 (purge_dead_edges): Update.
16441 (rtl_lv_add_condition_to_bb): Update.
16442 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
16443 * cgraphunit.c (init_lowered_empty_function): Update.
16444 (cgraph_node::expand_thunk): Update.
16445 * cilk-common.c: Include profile-count.h
16446 * dojump.c (inv): Remove.
16447 (jumpifnot): Update.
16448 (jumpifnot_1): Update.
16449 (do_jump_1): Update.
16451 (do_jump_by_parts_greater_rtx): Update.
16452 (do_compare_rtx_and_jump): Update.
16453 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
16454 do_jump_1. do_compare_rtx_and_jump): Update prototype.
16455 * dwarf2cfi.c: Include profile-count.h
16456 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
16457 (sjlj_emit_dispatch_table): Likewise.
16458 * explow.c: Include profile-count.h
16459 * expmed.c (emit_store_flag_force): Update.
16460 (do_cmp_and_jump): Update.
16461 * expr.c (compare_by_pieces_d::generate): Update.
16462 (compare_by_pieces_d::finish_mode): Update.
16463 (emit_block_move_via_loop): Update.
16464 (store_expr_with_bounds): Update.
16465 (store_constructor): Update.
16466 (expand_expr_real_2): Update.
16467 (expand_expr_real_1): Update.
16468 * expr.h (try_casesi, try_tablejump): Update prototypes.
16469 * gimple-pretty-print.c (dump_probability): Update.
16470 (dump_profile): New.
16471 (dump_gimple_label): Update.
16472 (dump_gimple_bb_header): Update.
16473 * graph.c (draw_cfg_node_succ_edges): Update.
16474 * hsa-gen.c (convert_switch_statements): Update.
16475 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
16476 (find_if_case_1): Update.
16477 (find_if_case_2): Update.
16478 * internal-fn.c (expand_arith_overflow_result_store): Update.
16479 (expand_addsub_overflow): Update.
16480 (expand_neg_overflow): Update.
16481 (expand_mul_overflow): Update.
16482 (expand_vector_ubsan_overflow): Update.
16483 * ipa-cp.c (good_cloning_opportunity_p): Update.
16484 * ipa-split.c (split_function): Use make_single_succ_edge.
16485 * ipa-utils.c (ipa_merge_profiles): Update.
16486 * loop-doloop.c (add_test): Update.
16487 (doloop_modify): Update.
16488 * loop-unroll.c (compare_and_jump_seq): Update.
16489 (unroll_loop_runtime_iterations): Update.
16490 * lra-constraints.c (lra_inheritance): Update.
16491 * lto-streamer-in.c (input_cfg): Update.
16492 * lto-streamer-out.c (output_cfg): Update.
16493 * mcf.c (adjust_cfg_counts): Update.
16494 * modulo-sched.c (sms_schedule): Update.
16495 * omp-expand.c (expand_omp_for_init_counts): Update.
16496 (extract_omp_for_update_vars): Update.
16497 (expand_omp_ordered_sink): Update.
16498 (expand_omp_for_ordered_loops): Update.
16499 (expand_omp_for_generic): Update.
16500 (expand_omp_for_static_nochunk): Update.
16501 (expand_omp_for_static_chunk): Update.
16502 (expand_cilk_for): Update.
16503 (expand_omp_simd): Update.
16504 (expand_omp_taskloop_for_outer): Update.
16505 (expand_omp_taskloop_for_inner): Update.
16506 * omp-simd-clone.c (simd_clone_adjust): Update.
16507 * optabs.c (expand_doubleword_shift): Update.
16508 (expand_abs): Update.
16509 (emit_cmp_and_jump_insn_1): Update.
16510 (expand_compare_and_swap_loop): Update.
16511 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
16512 * predict.c (predictable_edge_p): Update.
16513 (edge_probability_reliable_p): Update.
16514 (set_even_probabilities): Update.
16515 (combine_predictions_for_insn): Update.
16516 (combine_predictions_for_bb): Update.
16517 (propagate_freq): Update.
16518 (estimate_bb_frequencies): Update.
16519 (force_edge_cold): Update.
16520 * profile-count.c (profile_count::dump): Add missing space into dump.
16521 (profile_count::debug): Add newline.
16522 (profile_count::differs_from_p): Explicitly convert to unsigned.
16523 (profile_count::stream_in): Update.
16524 (profile_probability::dump): New member function.
16525 (profile_probability::debug): New member function.
16526 (profile_probability::differs_from_p): New member function.
16527 (profile_probability::differs_lot_from_p): New member function.
16528 (profile_probability::stream_in): New member function.
16529 (profile_probability::stream_out): New member function.
16530 * profile-count.h (profile_count_quality): Rename to ...
16531 (profile_quality): ... this one.
16532 (profile_probability): New.
16533 (profile_count): Update.
16534 * profile.c (compute_branch_probabilities): Update.
16535 * recog.c (peep2_attempt): Update.
16536 * sched-ebb.c (schedule_ebbs): Update.
16537 * sched-rgn.c (find_single_block_region): Update.
16538 (compute_dom_prob_ps): Update.
16539 (schedule_region): Update.
16540 * sel-sched-ir.c (compute_succs_info): Update.
16541 * stmt.c (struct case_node): Update.
16542 (do_jump_if_equal): Update.
16543 (get_outgoing_edge_probs): Update.
16544 (conditional_probability): Update.
16545 (emit_case_dispatch_table): Update.
16546 (expand_case): Update.
16547 (expand_sjlj_dispatch_table): Update.
16548 (emit_case_nodes): Update.
16549 * targhooks.c: Update.
16550 * tracer.c (better_p): Update.
16551 (find_best_successor): Update.
16552 * trans-mem.c (expand_transaction): Update.
16553 * tree-call-cdce.c: Update.
16554 * tree-cfg.c (gimple_split_edge): Upate.
16555 (move_sese_region_to_fn): Upate.
16556 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
16557 * tree-eh.c (lower_resx): Upate.
16558 (cleanup_empty_eh_move_lp): Upate.
16559 * tree-if-conv.c (version_loop_for_if_conversion): Update.
16560 * tree-inline.c (copy_edges_for_bb): Update.
16561 (copy_cfg_body): Update.
16562 * tree-parloops.c (gen_parallel_loop): Update.
16563 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
16564 (gimple_gen_time_profiler): Update.
16565 * tree-ssa-dce.c (remove_dead_stmt): Update.
16566 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
16567 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
16568 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
16569 (unloop_loops): Update.
16570 (try_peel_loop): Update.
16571 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16572 * tree-ssa-loop-split.c (connect_loops): Update.
16573 (split_loop): Update.
16574 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
16575 (hoist_guard): Update.
16576 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
16577 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
16578 (value_replacement): Update.
16579 * tree-ssa-reassoc.c (branch_fixup): Update.
16580 * tree-ssa-tail-merge.c (replace_block_by): Update.
16581 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
16582 (create_edge_and_update_destination_phis): Update.
16583 (compute_path_counts): Update.
16584 (recompute_probabilities): Update.
16585 (update_joiner_offpath_counts): Update.
16586 (freqs_to_counts_path): Update.
16587 (duplicate_thread_path): Update.
16588 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
16589 (struct switch_conv_info): Update.
16590 (gen_inbound_check): Update.
16591 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
16592 (vect_do_peeling): Update.
16593 (vect_loop_versioning): Update.
16594 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16595 (optimize_mask_stores): Update.
16596 * ubsan.c (ubsan_expand_null_ifn): Update.
16597 * value-prof.c (gimple_divmod_fixed_value): Update.
16598 (gimple_divmod_fixed_value_transform): Update.
16599 (gimple_mod_pow2): Update.
16600 (gimple_mod_pow2_value_transform): Update.
16601 (gimple_mod_subtract): Update.
16602 (gimple_mod_subtract_transform): Update.
16603 (gimple_ic): Update.
16604 (gimple_stringop_fixed_value): Update.
16605 (gimple_stringops_transform): Update.
16606 * value-prof.h: Update.
16608 2017-06-29 Carl Love <cel@us.ibm.com>
16610 * config/rs6000/rs6000-c.c: Add support for built-in functions
16611 vector signed int vec_signed (vector float);
16612 vector signed long long vec_signed (vector double);
16613 vector signed int vec_signed2 (vector double, vector double);
16614 vector signed int vec_signede (vector double);
16615 vector signed int vec_signedo (vector double);
16616 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
16617 instruction generator.
16618 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
16619 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
16620 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
16622 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
16623 vunsignede_v2df): Add define_expands.
16624 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
16625 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
16626 VEC_UNSIGNEDO): Add definitions.
16627 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
16628 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
16629 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
16630 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
16631 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
16632 * config/rs6000/altivec.h (vec_signed, vec_signed2,
16633 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
16634 vec_unsignede, vec_unsignedo): Add builtin defines.
16635 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
16637 * doc/extend.texi: Update the built-in documentation file for the
16638 new built-in functions.
16640 2017-06-29 Richard Biener <rguenther@suse.de>
16642 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
16643 reduction chains to LOOP_VINFO_REDUCTIONS.
16644 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
16645 SLP reductions after processing reduction chains.
16647 2017-06-29 Nathan Sidwell <nathan@acm.org>
16649 * builtins.c (fold_builtin_FUNCTION): Use
16650 lang_hooks.decl_printable_name.
16652 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
16654 PR middle-end/81194
16655 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
16656 with only one label.
16657 * stmt.c (expand_case): Assert NCASES is greater than one.
16659 2017-06-29 Richard Biener <rguenther@suse.de>
16661 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
16663 (group_case_labels): Likewise.
16664 (find_taken_edge): Push sanity checking on val to workers...
16665 (find_taken_edge_cond_expr): ... here
16666 (find_taken_edge_switch_expr): ... and here, handle cases
16667 with just a default label.
16668 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
16669 (group_case_labels): Likewise.
16670 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
16671 group_case_labels does anything cleanup the CFG again.
16673 2017-06-29 Bin Cheng <bin.cheng@arm.com>
16675 PR tree-optimization/81196
16676 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
16677 exit condition comparing two IVs.
16679 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
16681 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
16682 profile to the dummy entry at the end of the list of architectures.
16683 * config/arm/arm-cpu-cdata.h: Regenerated.
16685 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16686 Michael Collison <michael.collison@arm.com>
16689 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
16691 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
16692 (*aarch64_reg_<mode>3_minus_mask): New pattern.
16693 (*aarch64_<optab>_reg_di3_mask2): New pattern.
16694 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
16695 of shift when the shift amount is masked with constant equal to
16696 the size of the mode.
16697 * config/aarch64/predicates.md (subreg_lowpart_operator): New
16700 2017-06-29 Martin Liska <mliska@suse.cz>
16702 * config/i386/i386.opt: Change range from [1,5] to [0,5].
16704 2017-06-29 Yury Gribov <tetra2005@gmail.com>
16707 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
16710 (edge_growth_cache_entry::edge_growth_cache_entry): New
16712 (reset_edge_growth_cache): Update to use constructor.
16714 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16716 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
16717 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
16718 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
16720 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
16722 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
16723 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
16725 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
16727 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
16728 (*-linux-uclibc*): Add t-uclibc tmake_file.
16729 * config/t-musl: New.
16730 * config/t-uclibc: New.
16732 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
16734 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
16736 (gen_comm_data): Emit architectural setting of arch_prof.
16737 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
16739 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
16740 (armv8-m.base, armv8-m.main): Likewise.
16741 * arm-protos.h (arm_build_target): Add profile field.
16742 (arch_option): Likewise.
16743 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
16745 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
16746 arm_active_target.profile.
16748 2017-06-28 Richard Biener <rguenther@suse.de>
16750 PR middle-end/81227
16751 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
16752 TYPE_OVERFLOW_WRAPS.
16753 * match.pd (negate_expr_p): Likewise.
16754 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
16755 fold_build2, not fold_binary.
16757 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
16759 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
16760 Convert memory address to Pmode.
16761 (aarch64_print_operand): Assert MEM operands are always Pmode.
16763 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
16766 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
16767 Remove redundant if.
16768 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
16769 * config/arm/aarch-common-protos.h
16770 (aarch_forward_to_shift_is_not_shifted_re): Remove.
16771 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
16773 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
16776 * multiple_target.c (create_dispatcher_calls): Set the default
16777 clone to be static, not public.
16779 2017-06-28 Richard Biener <rguenther@suse.de>
16781 * tree-vect-loop.c (vectorizable_reduction): Move special
16782 cond reduction IV var creation ...
16783 (vect_create_epilog_for_reduction): ... here. Remove induction_index
16784 parameter. Use STMT_VINFO_VECTYPE.
16785 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
16788 2017-06-28 Martin Liska <mliska@suse.cz>
16791 * ipa-visibility.c (non_local_p): Handle visibility.
16793 2017-06-28 Martin Liska <mliska@suse.cz>
16796 * common.opt: Add IntegerRange to various options.
16797 * opt-functions.awk (integer_range_info): New function.
16798 * optc-gen.awk: Add integer_range_info to cl_options struct.
16799 * opts-common.c (decode_cmdline_option): Handle
16800 CL_ERR_INT_RANGE_ARG.
16801 (cmdline_handle_error): Likewise.
16802 * opts.c (print_filtered_help): Show valid interval in
16803 when --help is provided.
16804 * opts.h (struct cl_option): Add range_min and range_max fields.
16805 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
16807 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
16809 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
16810 (x * C EQ/NE y * C): New transformation.
16812 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
16814 * genmultilib (combination_space): Accept '+' in option names.
16816 2017-06-28 Martin Liska <mliska@suse.cz>
16819 * asan.c (instrument_derefs): Bail out inner references
16820 that are hard register variables.
16822 2017-06-28 Jakub Jelinek <jakub@redhat.com>
16825 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
16826 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
16828 2017-06-28 Richard Biener <rguenther@suse.de>
16830 * tree-vectorizer.h (vect_get_vec_defs): Remove.
16831 (vect_get_slp_defs): Adjust.
16832 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
16834 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
16836 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
16837 get_initial_defs_for_reduction instead of vect_get_vec_defs.
16838 (vectorizable_reduction): Adjust.
16839 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
16841 (vect_get_slp_defs): Likewise.
16842 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
16843 (vectorizable_bswap): Adjust.
16844 (vectorizable_call): Likewise.
16845 (vectorizable_conversion): Likewise.
16846 (vectorizable_assignment): Likewise.
16847 (vectorizable_shift): Likewise.
16848 (vectorizable_operation): Likewise.
16849 (vectorizable_store): Likewise.
16850 (vectorizable_condition): Likewise.
16851 (vectorizable_comparison): Likewise.
16853 2017-06-28 Michael Collison <michael.collison@arm.com>
16856 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
16858 (arm_gen_movmemqi): Removed unused variable 'i'.
16859 Convert 'for' loop into 'while' loop.
16860 (arm_expand_prologue): Remove last unnecessary set of insn.
16861 (thumb_pop): Remove unused variable 'pushed_words'.
16862 (thumb_exit): Remove last unnecessary set of regs_to_pop.
16864 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16866 * config/s390/predicates.md: Use s390_rel_address_ok_p.
16867 * config/s390/s390-protos.h: Add prototype of
16868 s390_rel_address_ok_p.
16869 * config/s390/s390.c (s390_got_symbol): New function.
16870 (s390_rel_address_ok_p): New function.
16871 (legitimize_pic_address): Use s390_rel_address_ok_p.
16872 (s390_load_got): Use s390_got_symbol.
16873 (s390_option_override): Issue error if
16874 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
16875 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
16877 * config/s390/s390.opt: New option mpic-data-is-text-relative.
16879 2017-06-27 Andrew Pinski <apinski@cavium.com>
16881 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
16882 (X * copysign (1.0, X)): New pattern.
16883 (X * copysign (1.0, -X)): New pattern.
16884 (copysign (-1.0, CST)): New pattern.
16886 2017-06-27 Joseph Myers <joseph@codesourcery.com>
16888 * genmultilib (combination_space): Remove variable.
16889 Validate reuse rules against regular expression for any sequence
16890 of multilib options in any order.
16892 2017-06-27 Michael Collison <michael.collison@arm.com>
16894 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
16895 call aarch64_split_simd_combine.
16896 * (aarch64_combine_internal<mode>): Delete pattern.
16897 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
16898 Allow register and subreg operands.
16900 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16902 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
16903 specific need, just fallback on defaults.
16904 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
16906 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16907 Olivier Hainque <hainque@adacore.com>
16909 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
16911 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
16912 targets. Pick a default if no particular attempt applied.
16913 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
16916 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16918 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
16919 (x86_64-wrs-vxworks7): Likewise.
16921 2017-06-27 Marek Polacek <polacek@redhat.com>
16924 * ubsan.c (instrument_null): Check get_base_address's result for null.
16926 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
16928 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
16930 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
16932 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
16933 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
16934 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
16935 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
16936 New function types.
16937 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
16938 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
16939 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
16940 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
16941 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
16942 BUILT_IN_FEUPDATEENV): New builtins.
16943 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
16944 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
16945 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
16946 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
16948 (builtin_structptr_types): Adjust size.
16949 * tree.c (builtin_structptr_types): Add four entries.
16951 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16952 Olivier Hainque <hainque@adacore.com>
16954 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
16955 (TLS_SYM): New local macro, forcing reference to __tls__ on
16956 link command lines for VxWorks 7 RTPs, triggering initialization
16958 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
16959 OS features TLS support, true for RTPs on VxWorks 7.
16960 * config/vxworks.c (vxworks_override_options): Setup emutls
16963 2017-06-27 Jakub Jelinek <jakub@redhat.com>
16965 * predict.c (test_prediction_value_range): Use -1U instead of -1
16966 to avoid narrowing conversion warning.
16967 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
16968 to avoid narrowing conversion warning.
16969 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
16971 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
16973 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16975 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
16976 64bit configurations.
16977 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
16978 (SIZE_TYPE): Likewise.
16979 * config/vxworks.c (vxworks_emutls_var_fields): Use
16980 long_unsigned_type_node instead of unsigned_type_node as the offset
16981 field type, which is "pointer" mode in emutls.c.
16983 2017-06-27 Jakub Jelinek <jakub@redhat.com>
16986 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
16988 PR middle-end/81207
16989 * gimple-fold.c (replace_call_with_call_and_fold): Handle
16990 gimple_vuse copying separately from gimple_vdef copying.
16992 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16994 * value-prof.c (free_hist): Remove call to memset and the enclosing if
16997 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
16998 Olivier Hainque <hainque@adacore.com>
17000 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
17001 for all vxworks7 targets.
17002 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
17003 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
17004 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
17005 variations for VX6/VX7 and 32/64bits later on in ...
17006 (VXWORKS_LIB_SPEC): Leverage new macros.
17007 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
17008 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
17010 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
17012 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
17013 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
17015 2017-06-26 Carl Love <cel@us.ibm.com>
17017 * config/rs6000/rs6000-c.c: Add support for built-in functions
17018 vector bool char vec_reve (vector bool char);
17019 vector signed char vec_reve (vector signed char);
17020 vector unsigned char vec_reve (vector unsigned char);
17021 vector bool int vec_reve (vector bool int);
17022 vector signed int vec_reve (vector signed int);
17023 vector unsigned int vec_reve (vector unsigned int);
17024 vector bool long long vec_reve (vector bool long long);
17025 vector signed long long vec_reve (vector signed long long);
17026 vector unsigned long long vec_reve (vector unsigned long long);
17027 vector bool short vec_reve (vector bool short);
17028 vector signed short vec_reve (vector signed short);
17029 vector double vec_reve (vector double);
17030 vector float vec_reve (vector float);
17031 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
17032 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
17033 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
17034 (altivec_vreve): New pattern.
17035 * config/rs6000/altivec.h (vec_reve): New define.
17036 * doc/extend.texi (vec_rev): Update the built-in documentation file
17037 for the new built-in functions.
17039 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17041 PR tree-optimization/71815
17042 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
17044 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
17046 (slsr_process_phi): Likewise.
17047 (replace_uncond_cands_and_profitable_phis): Don't replace a
17048 multiply candidate with a stride of 1 (copy or cast).
17049 (phi_incr_cost): Call uses_consumed_by_stmt rather than
17051 (lowest_cost_path): Likewise.
17052 (total_savings): Likewise.
17054 2017-06-26 Richard Biener <rguenther@suse.de>
17057 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
17058 Use def_builtin_pure for all gather builtins.
17060 2017-06-26 Richard Biener <rguenther@suse.de>
17062 PR tree-optimization/81203
17063 * tree-tailcall.c (find_tail_calls): Do not move stmts into
17064 non-dominating BBs.
17066 2017-06-26 Marek Polacek <polacek@redhat.com>
17069 * doc/invoke.texi: Document -Wmultistatement-macros.
17071 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
17073 * doc/sourcebuild.texi (ARM-specific attributes): Document new
17074 arm_neon_ok_no_float_abi effective target.
17076 2017-06-26 Richard Biener <rguenther@suse.de>
17078 PR tree-optimization/80928
17079 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
17080 (copy_bbs): Set BB_DUPLICATED flag early.
17081 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
17083 (execute_on_shrinking_pred): Likewise.
17084 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
17085 BB_DUPLICATED blocks.
17086 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
17087 iterate over all PHIs considering removal of *gsi.
17089 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
17091 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
17094 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17096 * config/rs6000/rs6000-string.c: (expand_block_clear,
17097 do_load_for_compare, select_block_compare_mode,
17098 compute_current_alignment, expand_block_compare,
17099 expand_strncmp_align_check, expand_strn_compare,
17100 expand_block_move, rs6000_output_load_multiple)
17101 Move functions related to string/block move/compare
17102 to a separate file.
17103 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
17104 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
17105 for this function which is now used in two files.
17106 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
17107 * config.gcc: Add rs6000-string.o to extra_objs for
17108 targets powerpc*-*-* and rs6000*-*-*.
17110 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
17113 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
17114 32-bit, since indexed is not valid for DImode.
17115 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
17116 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
17117 (define_peephole2 for Altivec d-form load): Add 32-bit support.
17118 (define_peephole2 for Altivec d-form store): Likewise.
17121 * multiple_target.c (create_dispatcher_calls): Only create the
17122 dispatcher call if the function is the default clone of a
17123 versioned function.
17125 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
17127 PR middle-end/80902
17128 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
17129 a call, force the call to not be a tail call.
17131 2017-06-23 Jeff Law <law@redhat.com>
17133 * doc/contrib.texi: Add entry for Steven Pemberton's work on
17136 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
17138 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
17139 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
17140 handling for early expansion of vector shifts (sl,sr,sra,rl).
17141 (builtin_function_type): Add vector shift right instructions
17142 to the unsigned argument list.
17144 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
17146 rtl-optimizatoin/79286
17147 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
17148 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
17149 trap. PIC register plus a const unspec without offset can never trap.
17151 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
17153 * tree.h (builtin_structptr_type): New type.
17154 (builtin_structptr_types): Declare new array.
17155 * tree.c (builtin_structptr_types): New array.
17156 (free_lang_data, build_common_tree_nodes): Use it.
17158 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
17161 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
17164 2017-06-22 Matt Turner <mattst88@gmail.com>
17166 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
17167 Lake models to skylake case. Assume skylake for unknown
17168 models with clflushopt.
17170 2017-06-22 Jeff Law <law@redhat.com>
17172 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
17173 frame sizes that do not satisfy aarch64_uimm12_shift.
17175 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
17177 * profile-count.h (apply_probability,
17178 apply_scale, probability_in): Fix checks for zero.
17180 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17182 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
17183 * doc/cppdiropts.texi (-I @var{dir}): Document it.
17185 2016-06-22 Richard Biener <rguenther@suse.de>
17187 * tree-vect-loop.c (vect_model_reduction_cost): Handle
17188 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
17189 REDUC_MAX_EXPR support.
17190 (vectorizable_reduction): Likewise.
17191 (vect_create_epilog_for_reduction): Likewise.
17193 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
17195 * match.pd (A / (1 << B) -> A >> B): New.
17196 * generic-match-head.c: Include optabs-tree.h.
17197 * gimple-match-head.c: Likewise.
17198 * optabs-tree.h (target_supports_op_p): New.
17199 * optabs-tree.c (target_supports_op_p): New.
17201 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17203 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
17204 $gcc_cv_ld --help output.
17205 (gcc_cv_ld_demangle): Likewise.
17206 (gcc_cv_ld_eh_frame_hdr): Likewise.
17207 (gcc_cv_ld_pie): Likewise.
17208 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
17209 (gcc_cv_ld_buildid): Likewise.
17210 (gcc_cv_ld_sysroot): Likewise.
17211 (ld_bndplt_support): Likewise.
17212 (ld_pushpopstate_support): Likewise.
17213 * configure: Regenerate.
17214 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
17216 2017-06-21 Jakub Jelinek <jakub@redhat.com>
17219 * config/i386/sse.md (round<mode>2): Renumber match_dup and
17220 operands indexes to avoid gap between operands and match_dups.
17222 2017-06-21 Andrew Pinski <apinski@cavium.com>
17224 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
17225 Increment Arith_shift and Arith_shift_reg by 1.
17226 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
17228 * config/aarch64/aarch64.c (thunderx_tunings): Enable
17229 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
17230 (aarch64_strip_extend): Add new argument and test for it.
17231 (aarch64_cheap_mult_shift_p): New function.
17232 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
17233 add a cost if it is true.
17234 Update calls to aarch64_strip_extend.
17235 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
17237 2017-06-21 Andrew Pinski <apinski@cavium.com>
17239 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
17241 (thunderxt88): Likewise.
17242 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
17243 (thunderx_prefetch_tune): New variable.
17244 (thunderx2t99_prefetch_tune): Update for the correct values.
17245 (thunderxt88_tunings): New variable.
17246 (thunderx_tunings): Use thunderx_prefetch_tune instead of
17247 generic_prefetch_tune.
17248 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
17250 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17252 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
17253 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
17254 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
17255 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
17256 (aarch64_atomic_cas<mode>, GPI): Likewise.
17258 2017-06-21 Martin Liska <mliska@suse.cz>
17260 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
17261 statements on cold and hot labels.
17262 * predict.c (tree_estimate_probability_bb): Remove the
17263 prediction from this place.
17265 2017-06-21 Martin Liska <mliska@suse.cz>
17267 PR tree-optimization/79489
17268 * gimplify.c (maybe_add_early_return_predict_stmt): New
17270 (gimplify_return_expr): Call the function.
17271 * predict.c (tree_estimate_probability_bb): Remove handling
17273 * predict.def: Update comment about early return predictor.
17274 * gimple-predict.h (is_gimple_predict): New function.
17275 * predict.def: Change default value of early return to 66.
17276 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
17278 * passes.def: Put pass_strip_predict_hints to the beginning of
17281 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
17283 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
17284 FUNCTION_DECL declarations.
17285 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
17287 (dwaf2out_decl): Likewise.
17288 * godump.c (go_early_global_decl): Skip call to the real debug hook
17289 for FUNCTION_DECL declarations.
17290 * passes.c (rest_of_decl_compilation): Skip call to the
17291 early_global_decl debug hook for FUNCTION_DECL declarations, unless
17292 -fdump-go-spec is passed.
17294 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
17296 * config/i386/i386.c (struct builtin_isa): New field pure_p.
17297 Reorder for compactness.
17298 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
17299 (def_builtin_pure, def_builtin_pure2): New functions.
17300 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
17302 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
17304 * match.pd (nop_convert): New predicate.
17305 ((A +- CST1) +- CST2): Allow some NOP conversions.
17307 2017-06-21 Jakub Jelinek <jakub@redhat.com>
17310 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
17311 with ctors/dtors if GOVD_SHARED is set.
17313 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
17315 * config/aarch64/aarch64.md (movti_aarch64):
17316 Emit mov rather than orr.
17317 (movtf_aarch64): Likewise.
17318 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
17319 Emit mov rather than orr.
17321 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
17323 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
17324 Swap alternatives, make integer dup more expensive.
17326 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
17328 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
17329 Return true for non-tls symbols.
17331 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
17333 * config/aarch64/aarch64-cores.def (cortex-a55): New.
17334 (cortex-a75): Likewise.
17335 (cortex-a75.cortex-a55): Likewise.
17336 * config/aarch64/aarch64-tune.md: Regenerate.
17337 * doc/invoke.texi (-mtune): Document new values for -mtune.
17339 2017-06-21 Tom de Vries <tom@codesourcery.com>
17341 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
17342 stack_size feature.
17343 (Effective-Target Keywords, Other attributes): Suggest using
17344 dg-add-options stack_size feature to get stack limit in stack_size
17345 effective target documentation.
17347 2017-06-21 Julian Brown <julian@codesourcery.com>
17348 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17350 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
17351 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
17352 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
17354 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
17355 attribute type list for neon_multiply.
17356 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
17357 attribute type list for neon_multiply.
17358 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
17359 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
17360 attribute type list for neon_multiply.
17361 * config/arm/types.md (crypto_pmull): Add.
17362 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
17363 attribute type list.
17365 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
17367 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
17370 2017-06-20 Jakub Jelinek <jakub@redhat.com>
17372 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
17373 to make sure not to dereference a NULL cost_classes_ptr pointer.
17375 2017-06-20 Carl Love <cel@us.ibm.com>
17377 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17378 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17379 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17380 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17381 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17382 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17383 VMULOSW): New enum "unspec" values.
17384 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17385 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17386 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17387 altivec_vmulosw): New patterns.
17388 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17389 VMULOSW): Add definitions.
17391 2017-06-20 Julia Koval <julia.koval@intel.com>
17393 * config/i386/i386.c: Fix rounding expand for new pattern.
17394 * config/i386/subst.md: Fix pattern (parallel -> unspec).
17396 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17398 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
17399 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
17401 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17403 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
17406 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17408 * config/aarch64/aarch64-cores.def: Rearrange to sort by
17409 architecture, then by implementer ID.
17410 * config/aarch64/aarch64-tune.md: Regenerate.
17412 2017-06-20 Richard Biener <rguenther@suse.de>
17414 PR middle-end/81097
17415 * fold-const.c (split_tree): Fold to type before negating.
17417 2017-06-20 David Malcolm <dmalcolm@redhat.com>
17419 * diagnostic-show-locus.c
17420 (selftest::test_fixit_deletion_affecting_newline): New function.
17421 (selftest::diagnostic_show_locus_c_tests): Call it.
17423 2017-06-20 Andreas Schwab <schwab@suse.de>
17426 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
17429 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17431 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
17432 __ARM_FEATURE_COPROC according to support.
17434 2017-06-20 Jakub Jelinek <jakub@redhat.com>
17436 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
17437 Rewritten to avoid overflow for > 32-bit pointers.
17440 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
17441 by removing enum keyword.
17442 (ubsan_type_descriptor): Likewise. Formatting fix.
17445 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
17446 splitter): Require TARGET_SSE2 in the condition.
17448 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
17451 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
17452 for doing vector set of SFmode on ISA 3.0.
17453 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
17454 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
17456 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
17457 SFmode value into a V4SF variable that was extracted from another
17458 V4SF variable without converting the element to double precision
17459 and back to single precision vector format.
17460 (vsx_insert_extract_v4sf_p9_2): Likewise.
17462 2017-06-19 Jakub Jelinek <jakub@redhat.com>
17464 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
17465 in UWHI to avoid undefined overflow.
17468 * ubsan.h (enum ubsan_encode_value_phase): New.
17469 (ubsan_encode_value): Change second argument to
17470 enum ubsan_encode_value_phase with default value of
17471 UBSAN_ENCODE_VALUE_GENERIC.
17472 * ubsan.c (ubsan_encode_value): Change second argument to
17473 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
17474 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
17475 create_tmp_var_raw instead of create_tmp_var and use a
17477 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
17478 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
17479 ubsan_encode_value callers.
17482 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
17483 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
17484 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
17486 2017-06-19 Richard Biener <rguenther@suse.de>
17488 PR middle-end/81118
17489 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
17490 estimates if we changed anything.
17492 2017-06-19 Richard Biener <rguenther@suse.de>
17494 PR tree-optimization/80887
17495 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
17496 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
17497 simplified lookups, then reset mprts_hook.
17498 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
17500 (try_to_simplify): Likewise.
17502 2017-06-19 Martin Liska <mliska@suse.cz>
17505 * gimplify.c (gimplify_switch_expr):
17506 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
17508 2017-06-19 Martin Liska <mliska@suse.cz>
17510 * doc/install.texi: Document that PGO runs in 4 stages.
17512 2017-06-19 Martin Liska <mliska@suse.cz>
17515 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
17516 to dispatcher function name.
17517 * multiple_target.c (replace_function_decl): New function.
17518 (create_dispatcher_calls): Redirect both edges and references.
17520 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
17522 * profile-count.c (profile_count::dump): Dump quality.
17523 (profile_count::differs_from_p): Update for unsigned val.
17524 * profile-count.h (profile_count_quality): New enum.
17525 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
17527 2017-06-19 Richard Biener <rguenther@suse.de>
17529 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
17530 struct function as arg.
17531 (estimate_numbers_of_iterations): Export overload with loop arg.
17532 (free_numbers_of_iterations_estimates_loop): Use an overload of
17533 free_numbers_of_iterations_estimates instead.
17534 * tree-cfg.c (remove_bb): Adjust.
17535 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
17536 * tree-parloops.c (gen_parallel_loop): Likewise.
17537 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
17539 (tree_unroll_loops_completely): Likewise.
17540 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
17541 Use an overload instead and export.
17542 (estimated_loop_iterations): Adjust.
17543 (max_loop_iterations): Likewise.
17544 (likely_max_loop_iterations): Likewise.
17545 (estimate_numbers_of_iterations): Take struct function as arg
17547 (loop_exits_before_overflow): Adjust.
17548 (free_numbers_of_iterations_estimates_loop): Use an overload.
17549 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
17550 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
17552 2017-06-19 Richard Biener <rguenther@suse.de>
17555 * ipa-prop.c (find_constructor_constant_at_offset): Handle
17556 RANGE_EXPR conservatively.
17558 2017-06-16 Carl Love <cel@us.ibm.com>
17560 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17561 definitions for vec_float, vec_float2, vec_floato,
17562 vec_floate built-ins.
17563 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
17564 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
17566 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
17567 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
17568 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
17569 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
17570 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
17571 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
17572 vec_floato): Add builtin defines.
17573 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
17574 Update the built-in documentation file for the new built-in
17577 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17579 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
17580 (mthumb): Mark as the negative of -marm.
17582 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17584 * doc/invoke.texi (ARM Options, -mcpu): Document supported
17586 (ARM Options, -mtune): Document that this accepts the same
17587 extension options as -mcpu.
17588 (ARM Options, -mfpu): Document addition of -mfpu=auto.
17590 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17592 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
17593 permitted extensions.
17595 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17597 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
17598 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
17599 (armv8-m.main): Add option +nodsp.
17600 * config/arm/arm-cpu-cdata.h: Regenerated.
17602 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17604 * config/arm/t-fuchsia: New file.
17605 * config.gcc (arm*-*-fuchsia*): Use it.
17607 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17609 * config/arm/t-symbian: Rewrite for new option infrastructure.
17611 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17613 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
17614 (MULTILIB_REQUIRED): Likewise.
17616 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17618 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
17619 (MULTILIB_RESUE): Likewise.
17620 (MULTILIB_MATCHES): Likewise.
17621 (MULTLIB_REQUIRED): Likewise.
17623 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17625 * config/arm/t-rtems: Rewrite for new option framework.
17627 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17629 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
17630 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
17631 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
17632 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
17633 * config/arm/t-multilib: ... here.
17634 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
17635 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
17636 armv7-a and armv8*-a when A-profile libraries have not been built.
17637 * config/arm/t-rmprofile: Rewrite.
17639 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17641 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
17642 with a backslash. Remove the backslash after substituting unescaped
17644 * doc/fragments.texi (MULTILIB_REUSE): Document it.
17646 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17648 * config.gcc: (arm*-*-*): When building a-profile libraries, force
17649 the driver to pass through the default setting of -mfloat-abi.
17650 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
17652 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
17653 (all_feat_combs): New rule.
17654 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
17656 * config/arm/t-aprofile: Rewrite.
17658 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17660 * config/arm/arm.h (FPUTYPE_AUTO): Define.
17661 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
17662 fpu is not specified by the user/command-line.
17663 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
17664 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
17665 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
17666 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
17667 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
17668 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
17670 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17672 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
17673 * config/arm/t-arm-elf: Rewritten.
17675 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17677 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
17678 have some floating-point instructions.
17679 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
17680 (TARGET_MAYBE_HARD_FLOAT): New macro.
17681 * config/arm/arm-builtins.c (arm_init_builtins): Use
17682 TARGET_MAYBE_HARD_FLOAT.
17683 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
17685 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17687 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
17688 (configargs.h): Include it.
17689 (arm_print_hint_for_fpu_option): New function.
17690 (arm_parse_fpu_option): New function.
17691 (candidate_extension): New class.
17692 (arm_canon_for_multilib): New function.
17693 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
17694 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
17695 (ARCH_CANONICAL_SPECS): New macro.
17696 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
17698 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17700 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
17701 are set after handling multilib fragments. Set target_cpu_default2
17704 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17706 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
17708 (arm*-*-*): Set target_cpu_default2 to a quoted string.
17709 * config/arm/parsecpu.awk (check_cpu): Validate any extension
17711 (check_arch): Likewise.
17712 * config/arm/arm.c (arm_configure_build_target): Handle
17713 TARGET_CPU_DEFAULT being a string constant. Scan any feature
17714 options in the default.
17716 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17718 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
17719 when an option is an alias of another.
17720 * config/arm/parsecpu.awk (optalias): New parser token.
17721 (gen_comm_data): Mark non-alias options as such. Emit entries
17722 for extension aliases.
17723 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
17724 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
17725 (armv6kz, armv6zk, armv6t2): Likewise.
17726 (armv7): Make vfpv3-d16 an alias.
17727 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
17729 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
17730 Sort in canonical order.
17731 (armv8-a): Sort in canonical order.
17732 (armv8.1-a, armv8.2-a): Likewise.
17733 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
17735 (cortex-a9): Sort in canonical order.
17736 * config/arm/arm.c (selftests.h): Include it.
17737 (arm_test_cpu_arch_data): New function.
17738 (arm_run_self_tests): New function.
17739 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
17740 (targetm): Move declaration to the end of the file.
17741 * arm-cpu-cdata.h: Regenerated.
17743 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17745 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
17746 call to target_mode_check describing the type of option passed.
17747 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
17748 (arm_target_thumb_only): Use arm_parse_arch_option_name or
17749 arm_parse_cpu_option_name to match parameters against list of
17751 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
17752 arm_arch_core_flags data structure.
17753 * config/arm/arm-cpu_cdata.h: Regenerated.
17755 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17757 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
17759 (arm_print_hint_for_cpu_option): Likewise.
17760 (arm_print_hint_for_arch_option): Likewise.
17761 (arm_parse_cpu_option_name): Likewise.
17762 (arm_parse_arch_option_name): Likewise.
17763 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
17764 of entries in the all_fpus list.
17765 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
17766 (arm_parse_cpu_option_name): Declare.
17767 (arm_parse_arch_option_name): Declare.
17768 (arm_parse_option_features): Declare.
17769 (arm_intialize_isa): Declare.
17770 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
17772 (gen_comm_data): ... here. Make definitions non-static.
17773 * config/arm/arm-cpu-data.h: Regenerated.
17774 * config/arm/arm-cpu-cdata.h: Regenerated.
17776 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17778 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
17779 (cpu_arch_extension): New structure.
17780 (cpu_arch_option, arch_option, cpu_option): New structures.
17781 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
17782 architecture types.
17783 (gen_data): Generate new format data tables.
17784 * config/arm/arm.c (cpu_tune): New structure.
17785 (cpu_option, processors): Delete.
17786 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
17787 (arm_print_hint_for_cpu_option): ... this and ...
17788 (arm_print_hint_for_arch_option): ... this.
17789 (arm_parse_arch_cpu_name): Delete. Replace with ...
17790 (arm_parse_cpu_option_name): ... this and ...
17791 (arm_parse_arch_option_name): ... this.
17792 (arm_unrecognized_feature): Change type of target parameter to
17794 (arm_parse_arch_cpu_features): Delete. Replace with ...
17795 (arm_parse_option_features): ... this.
17796 (arm_configure_build_target): Rework to use new configuration data
17798 (arm_print_tune_info): Rework for new configuration data tables.
17799 * config/arm/arm-cpu-data.h: Regenerated.
17800 * config/arm/arm-cpu.h: Regenerated.
17802 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17804 * Makefile.in (OBJS): Move sbitmap.o from here ...
17805 (OBJS-libcommon): ... to here.
17807 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17809 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
17810 (ISA_ALL_CRYPTO): New macro.
17811 (ISA_ALL_SIMD): New macro
17812 (ISA_ALL_FP): New macro.
17813 * config/arm/arm.c (fpu_bitlist): Update initializer.
17814 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
17816 (arm9e): Add fpu. Add option for nofp
17817 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
17818 (arm926ej-s, arm1026ej-s): Likewise.
17819 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
17820 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
17821 neon-fp16, neon-vfpv4, nofp and nosimd.
17822 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
17823 (cortex-a8): Add fpu. Add option for nofp.
17824 (cortex-a9): Add fpu. Add options for nosimd and nofp.
17825 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
17826 (cortex-r4f): Add fpu.
17827 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
17828 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
17830 (cortex-r8): Likewise.
17831 (cortex-m4): Add fpu. Add option for nofp.
17832 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
17833 (cortex-a17.cortex-a7): Likewise.
17834 (cortex-a32): Add fpu. Add options for crypto and nofp.
17835 (cortex-a35, cortex-a53): Likewise.
17836 (cortex-a57): Add fpu. Add option for crypto.
17837 (cortex-a72, cortex-a73): Likewise.
17838 (exynos-m1): Likewise.
17839 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
17840 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
17841 (cortex-m33): Add fpu. Add option for nofp.
17842 * config/arm/arm-cpu-cdata.h: Regenerated
17843 * config/arm/arm-cpu-data.h: Regenerated.
17845 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17847 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
17848 (armv5te, armv5tej): Likewise.
17849 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
17850 (armv7): Add options fp and vfpv3-d16.
17851 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
17852 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
17854 (armv7ve): Likewise.
17855 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
17856 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
17857 (armv8-a): Add nocrypto option.
17858 (armv8.1-a, armv8.2-a): Likewise.
17859 (armv8-m.main): add options fp, fp.dp and nofp.
17861 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17863 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
17865 (armv8-a+crc): Delete.
17866 (armv8.1-a): Add options simd, crypto and nofp.
17867 (armv8.2-a): Add options fp16, simd, crypto and nofp.
17868 (armv8.2-a+fp16): Delete.
17869 (armv8-m.main): Add option dsp.
17870 (armv8-m.main+dsp): Delete.
17871 (cortex-a8): Add fpu. Add nofp option.
17872 (cortex-a9): Add fpu. Add nofp and nosimd options.
17873 * config/arm/parsecpu.awk (gen_data): Generate option tables and
17874 link to main cpu and architecture data structures.
17875 (gen_comm_data): Only put isa attributes from the main architecture
17877 (option): New statement for architecture and CPU entries.
17878 * arm.c (struct cpu_option): New structure.
17879 (struct processors): Add entry for options.
17880 (arm_unrecognized_feature): New function.
17881 (arm_parse_arch_cpu_name): Ignore any characters after the first
17883 (arm_parse_arch_cpu_feature): New function.
17884 (arm_configure_build_target): Separate out any CPU and architecture
17885 features and parse separately. Don't error out if -mfpu=auto is
17886 used with only an architecture string.
17887 (arm_print_asm_arch_directives): New function.
17888 (arm_file_start): Call it.
17889 * config/arm/arm-cpu-cdata.h: Regenerated.
17890 * config/arm/arm-cpu-data.h: Likewise.
17891 * config/arm/arm-tables.opt: Likewise.
17893 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17895 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
17896 assembler when it is not -mfpu=auto.
17898 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17900 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
17901 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
17902 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
17903 (ASM_CPU_SPEC): Rewrite.
17904 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
17905 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
17906 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
17907 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
17908 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
17909 copied string is NUL-terminated. Also strip any characters prefixed
17911 (arm_rewrite_selected_arch): New function.
17912 (arm_rewrite_march): New function.
17914 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17916 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
17917 (x_arm_cpu_string, x_arm_tune_string): Likewise.
17918 (march, mcpu, mtune): Convert to string-based options.
17919 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
17920 (arm_parse_arch_cpu_name): New function.
17921 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
17922 identify selected architecture or CPU.
17923 (arm_option_save): New function.
17924 (TARGET_OPTION_SAVE): Redefine.
17925 (arm_option_restore): Restore string options.
17926 (arm_option_print): Print string options.
17928 2017-06-16 Martin Sebor <msebor@redhat.com>
17930 PR tree-optimization/80933
17931 PR tree-optimization/80934
17932 * builtins.c (fold_builtin_3): Do not handle bcmp here.
17933 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
17934 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
17935 (gimple_fold_builtin): Call them.
17937 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17939 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
17940 as unlikely; update profile.
17942 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17944 * predict.c (force_edge_cold): Handle declaring edges impossible
17947 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17949 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
17951 (try_unroll_loop_completely): Fix reporting.
17953 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17955 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
17957 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
17960 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
17961 if given a non-constant argument for an intrinsic which requires a
17964 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17966 * profile.c (compare_freqs): New function.
17967 (branch_prob): Sort edge list.
17968 (find_spanning_tree): Assume that the list is priority sorted.
17970 2017-06-16 Richard Biener <rguenther@suse.de>
17972 PR tree-optimization/81090
17973 * passes.def (pass_record_bounds): Remove.
17974 * tree-pass.h (make_pass_record_bounds): Likewise.
17975 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
17976 make_pass_record_bounds): Likewise.
17977 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
17978 not free niter estimates at the beginning but at the end.
17979 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
17981 2017-06-16 Richard Biener <rguenther@suse.de>
17983 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
17984 initializer to workaround ICE in host GCC 4.8.
17986 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17988 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
17990 (clone_inlined_nodes): Update.
17992 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17994 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
17995 prefetch settings, and enable prefetching by default at -O3.
17997 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17999 * config/aarch64/aarch64.c (aarch64_override_options_internal):
18000 Set flag_prefetch_loop_arrays according to tuning data.
18002 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18004 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
18005 New tune structure.
18006 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
18007 [Unrelated to main purpose of the patch] Place the pointer field last
18008 to enable type checking errors when tune structure are wrongly merged.
18009 * config/aarch64/aarch64.c (generic_prefetch_tune,)
18010 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
18011 (thunderx2t99_prefetch_tune): New tune constants.
18012 (tune_params *_tunings): Update all tunings (no functional change).
18013 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
18014 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
18015 from tunings structures.
18017 2017-06-16 Jakub Jelinek <jakub@redhat.com>
18020 * ubsan.c (instrument_null): Add T argument, use it instead
18021 of computing it based on IS_LHS.
18022 (instrument_object_size): Likewise.
18023 (pass_ubsan::execute): Adjust instrument_null and
18024 instrument_object_size callers to pass gimple_get_lhs or
18025 gimple_assign_rhs1 result to it. Use instrument_null instead of
18026 calling get_base_address and instrument_mem_ref. Handle
18027 aggregate call arguments for object-size sanitization.
18029 2017-06-16 Yury Gribov <tetra2005@gmail.com>
18031 PR tree-optimization/81089
18032 * tree-vrp.c (is_masked_range_test): Validate operands of
18035 2017-06-15 Martin Sebor <msebor@redhat.com>
18038 * dumpfile.c (dump_register): Avoid calling memset to initialize
18039 a class with a default ctor.
18040 * gcc.c (struct compiler): Remove const qualification.
18041 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
18042 * hash-table.h: Ditto.
18043 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
18045 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
18046 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
18048 * params.h (struct param_info): Make struct members non-const.
18049 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
18050 with default initialization.
18051 * vec.h (vec_copy_construct, vec_default_construct): New helper
18053 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
18054 with vec_copy_construct.
18055 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
18056 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
18057 * doc/invoke.texi (-Wclass-memaccess): Document.
18059 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18061 * emit-rtl.h (is_leaf): Update comment about local
18062 register allocator.
18064 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18067 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
18068 for a variable to have a section before checking if the section has a
18070 Set section to.persistent if persistent attribute is set.
18071 Warn if .persistent attribute is used on an automatic variable.
18073 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
18075 PR rtl-optimization/80474
18076 * reorg.c (update_block): Do not ignore instructions in a delay slot.
18078 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
18080 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
18083 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
18085 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
18086 (casesi): Emit bounds checking as RTL.
18087 (casesi_internal_mips16_<mode>): Remove bounds checking.
18089 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
18091 * config/xtensa/xtensa.c (xtensa_option_override): Append
18092 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
18093 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
18094 xtensa_doloop_hooks): Define unconditionally.
18095 (xtensa_reorg_loops): Only call reorg_loops in the presence of
18097 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
18098 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
18099 for it in xtensa_option_override.
18100 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
18101 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
18103 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
18105 * doc/cppopts.texi: Document '-' special value to -MF.
18107 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
18109 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
18110 (cortex_a53_fconst): Likewise.
18111 (cortex_a53_fpmul): Likewise.
18112 (cortex_a53_f_load_64): Likewise.
18113 (cortex_a53_f_load_many): Likewise.
18114 (cortex_a53_advsimd_alu): Likewise.
18115 (cortex_a53_advsimd_alu_q): Likewise.
18116 (cortex_a53_advsimd_mul): Likewise.
18117 (cortex_a53_advsimd_mul_q): Likewise.
18118 (fpmac bypass): Add new bypass for fpmac-fpmac case.
18119 Add missing fmul, r2f_cvt and fconst cases.
18121 2017-06-14 Richard Biener <rguenther@suse.de>
18123 PR middle-end/81088
18124 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
18126 (fold_binary_loc): When associating do not treat pre-existing
18127 TREE_OVERFLOW on literal constants as a reason to allow
18128 TREE_OVERFLOW on associated literal constants.
18130 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18132 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
18133 (MASK_FEATURES): New macro.
18134 * config/sparc/sparc.c (sparc_option_override): Remove the special
18135 handling of -mfpu and generalize it to all MASK_FEATURES switches.
18137 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18139 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
18140 a division of 0 if non-call exceptions are enabled.
18142 2017-06-14 Andrew Pinski <apinski@cavium.com>
18143 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18146 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
18147 Improve vector initialization code gen for only variable case.
18149 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18151 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
18153 2017-06-14 Richard Biener <rguenther@suse.de>
18155 PR tree-optimization/81083
18156 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
18159 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18161 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
18162 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
18163 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
18164 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
18165 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
18166 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
18168 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18170 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
18171 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
18173 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18175 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
18177 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18179 * config/rs6000/t-rtems: Don't handle SPE.
18181 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18183 * config/rs6000/t-linux: Don't handle SPE.
18185 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18187 * config/rs6000/eabispe.h: Delete file.
18189 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18191 * config/rs6000/t-spe: Delete file.
18193 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18195 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
18196 (rs6000_legitimate_offset_address_p): Return false for anything in
18197 V2SImode or V2SFmode.
18199 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18201 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
18202 except V2SF and V2SI. Rearrange the vector modes, and add comments.
18203 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
18205 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
18206 (rs6000_legitimate_offset_address_p): Ditto.
18207 (rs6000_emit_move): Ditto.
18208 (rs6000_init_builtins): Remove V4HI_type_node.
18210 2017-06-13 Martin Liska <mliska@suse.cz>
18213 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
18214 (gate_asan): Likewise.
18215 * asan.h (asan_no_sanitize_address_p): Remove the function.
18216 (sanitize_flags_p): New function.
18217 * builtins.def: Fix coding style.
18218 * common.opt: Use renamed enum value.
18219 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
18220 * doc/extend.texi: Document no_sanitize attribute.
18221 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
18222 to SANITIZE_UNDEFINED_NONDEFAULT.
18223 * gcc.c (sanitize_spec_function): Use the renamed enum value.
18224 * gimple-fold.c (optimize_atomic_compare_exchange_p):
18225 Use sanitize_flags_p.
18226 * gimplify.c (gimplify_function_tree): Likewise.
18227 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
18228 * opts.c (parse_no_sanitize_attribute): New function.
18229 (common_handle_option): Use renamed enum value.
18230 * opts.h (parse_no_sanitize_attribute): Declare.
18231 * tree.c (sanitize_flags_p): New function.
18232 * tree.h: Declared here.
18233 * tsan.c: Use sanitize_flags_p.
18234 * ubsan.c (ubsan_expand_null_ifn): Likewise.
18235 (instrument_mem_ref): Likewise.
18236 (instrument_bool_enum_load): Likewise.
18237 (do_ubsan_in_current_function): Remove the function.
18238 (pass_ubsan::execute): Use sanitize_flags_p.
18239 * ubsan.h: Remove do_ubsan_in_current_function
18240 * tree-cfg.c (print_no_sanitize_attr_value): New function.
18241 (dump_function_to_file): Use it here.
18243 2017-06-13 Martin Jambor <mjambor@suse.cz>
18245 PR tree-optimization/80803
18246 PR tree-optimization/81063
18247 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
18248 (propagate_subaccesses_across_link): Enqueue subtree whenever
18249 necessary instead of relying on the caller.
18251 2017-06-13 Martin Jambor <mjambor@suse.cz>
18253 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
18254 that have a first_link.
18255 (sort_and_splice_var_accesses): Do not check first_link before
18257 (subtree_mark_written_and_enqueue): Likewise.
18258 (propagate_all_subaccesses): Likewise and do not stop at first
18259 parent with a first_link.
18261 2017-06-13 Martin Jambor <mjambor@suse.cz>
18263 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
18266 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18268 * match.pd: New pattern.
18270 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18272 * tree-vrp.c (is_masked_range_test): New function.
18273 (register_edge_assert_for): Determine ranges for
18276 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18278 PR tree-optimization/67328
18279 * fold-const.c (maskable_range_p): New function.
18280 (build_range_check): Generate bittests if possible.
18282 2017-06-13 Martin Liska <mliska@suse.cz>
18284 * gimple-pretty-print.c (dump_probability): Add new argument.
18285 (dump_edge_probability): Dump both probability and count.
18286 (dump_gimple_label): Likewise.
18287 (dump_gimple_bb_header): Likewise.
18289 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
18292 * config/avr/avr-devices.c: Fix indentation.
18293 * config/avr/gen-avr-mmcu-specs.c: Dito.
18295 2017-06-13 Richard Biener <rguenther@suse.de>
18297 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
18298 instead get vector type from stmt_info.
18299 (vectorizable_reduction): Adjust. Remove dead code.
18301 2017-06-13 Richard Biener <rguenther@suse.de>
18303 PR middle-end/81065
18304 * fold-const.c (extract_muldiv_1): Remove bogus distribution
18305 case of C * (x * C2 + C3).
18306 (fold_addr_of_array_ref_difference): Properly fold index difference.
18308 2017-06-12 David S. Miller <davem@davemloft.net>
18311 * config/sparc/sparc.md (return expander): Emit frame blockage if
18312 function uses alloca.
18314 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
18316 * combine.c (make_field_assignment): Check len rather than the mode
18317 precision when calling force_to_mode.
18319 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
18321 Support multilibs and devices that see flash in RAM address range.
18324 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
18325 (avr_mcu_t) <flash_pm_offset>: New field.
18326 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
18327 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
18328 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
18329 (AVR_TINY_PM_OFFSET): Remove macro.
18330 * config/avr/avr.opt (-mshort-calls): New option.
18331 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18332 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
18333 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
18334 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
18335 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
18336 instead of avr_arch->have_jmp_call.
18337 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
18338 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
18339 avr_arch->flash_pm_offset to define.
18340 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
18341 new field flash_pm_offset. Add entry for avrxmega3.
18342 (avr_texinfo): Add entry for avrxmega3.
18343 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
18344 attiny212, attiny214,
18345 attiny412, attiny414, attiny416, attiny417,
18346 attiny814, attiny816, attiny817,
18347 attiny1614, attiny1616, attiny1617,
18348 attiny3214, attiny3216, attiny3217.
18349 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
18350 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
18351 (avr_print_operand_address) [AVR_TINY]: Same.
18352 (avr_asm_init_sections) <readonly_data_section>: Only patch
18353 callback if avr_arch->flash_pm_offset = 0.
18354 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
18355 for rodata if avr_arch->flash_pm_offset != 0.
18356 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
18357 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
18358 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
18359 (m_options): Append opt_rcall.
18360 (m_dirnames): Append dir_rcall.
18361 * config/avr/t-multilib: Regenerate.
18363 * configure.ac [target=avr]: Check whether avrxmega3 default
18364 linker description file works as needed.
18365 * configure: Regenerate.
18366 * doc/avr-mmcu.texi: Regenerate.
18367 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
18368 <__AVR_ARCH__>: Document avrxmega3 and 103.
18369 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
18370 <__AVR_SHORT_CALLS__>: Document it.
18371 <__AVR_PM_BASE_ADDRESS__>: Document it.
18372 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
18373 (AVR Variable Attributes) <progmem>: Document this is
18374 not needed for avrxmega3.
18375 (AVR Named Address Spaces) <__flash>: Dito.
18377 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
18379 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
18381 2017-06-12 Doug Rupp <rupp@adacore.com>
18383 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
18384 Append vxworks-stdint.h to the tm_file list.
18385 * config/vxworks-stdint.h: New file.
18387 2017-06-12 Martin Liska <mliska@suse.cz>
18389 PR tree-optimization/81041
18390 * tree-profile.c (gimple_gen_ic_func_profiler):
18391 Create an extra BB in profile-generate
18392 (gimple_gen_time_profiler): Likewise.
18394 2017-06-12 Jakub Jelinek <jakub@redhat.com>
18396 PR tree-optimization/81003
18397 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
18398 (update_range_test): Use it instead of force_gimple_operand_gsi.
18400 2017-06-12 Richard Biener <rguenther@suse.de>
18402 PR tree-optimization/81053
18403 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
18404 with backedge value not defined in loop. Simplify def stmt
18407 2017-06-11 Tom de Vries <tom@codesourcery.com>
18410 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
18412 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
18413 nvptx_cannot_force_const_mem.
18415 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18417 * opts.c (finish_options): Move test for flag_split_stack after
18418 it has been initialized.
18420 2017-06-11 Jason Merrill <jason@redhat.com>
18422 * tree.h (id_equal): New.
18423 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
18424 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
18425 instead of strcmp of IDENTIFIER_POINTER.
18427 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18429 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
18430 (mark_all_inlined_calls_cdtor): Fix formating.
18431 (inline_transform): Rescale profile before inlining.
18433 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18435 * cgraph.h (cgraph_edge::clone): Update prototype.
18436 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
18437 (cgraph_node::create_clone): Update.
18438 (cgraph_node::create_version_clone): Update.
18439 * tree-inline.c (copy_bb): Update.
18440 (expand_call_inline): Update.
18442 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
18444 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
18445 factored out from ...
18446 (rs6000_emit_prologue): ... here.
18448 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
18450 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
18451 factored out from ...
18452 (rs6000_emit_prologue): ... here.
18454 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18456 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
18458 (handle_missing_profiles): Fix computation of tp_first_run.
18459 (counts_to_freqs): Do not touch freqs when count is 0.
18461 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18463 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
18466 2017-06-10 Tom de Vries <tom@codesourcery.com>
18468 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
18469 attributes): Document signal effective target.
18471 2017-06-10 Tom de Vries <tom@codesourcery.com>
18473 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
18474 Document effective target stack_size.
18476 2017-06-09 David Malcolm <dmalcolm@redhat.com>
18478 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
18479 to the edit_context if they can be auto-applied.
18481 2017-06-9 Ian Lance Taylor <iant@golang.org>
18483 * opts.c (finish_options): If -fsplit-stack, disable implicit
18484 -forder-blocks-and-partition.
18485 * doc/invoke.texi (Optimize Options): Document that when using
18486 -fsplit-stack -forder-blocks-and-partition is not implicitly
18489 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18491 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
18492 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
18493 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
18494 * builtins.def (abort, trap, unreachable): Declare cold.
18495 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
18496 * tree-core.h (ECF_COLD): New.
18497 * tree.c (set_call_expr_flags): Handle ECF_COLD.
18498 (build_common_builtin_nodes): Mark unreachable and abort as cold.
18500 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18502 * predict.c (unlikely_executed_stmt_p): Cleanup.
18504 2017-06-09 Richard Biener <rguenther@suse.de>
18506 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
18507 model if the ref is always written to.
18509 2017-06-09 Tamar Christina <tamar.christina@arm.com>
18511 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
18513 2017-06-09 Tamar Christina <tamar.christina@arm.com>
18515 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
18518 2017-06-09 Tom de Vries <tom@codesourcery.com>
18521 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
18522 "target cannot support label values" when encountering LABEL_REF.
18524 2017-06-09 Martin Liska <mliska@suse.cz>
18526 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
18527 (gimple_gen_ic_func_profiler): Emit direct comparison
18528 of __gcov_indirect_call_callee with NULL.
18529 (gimple_gen_time_profiler): Change probability from
18530 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
18532 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18534 * profile.c (edge_gcov_counts): Turn to pointer.
18535 (compute_branch_probabilities, compute_branch_probabilities): Update.
18536 (branch_prob): Do not clear edge_gcov_count.
18537 * profile.h (edge_gcov_counts): Turn to pointer.
18538 (edge_gcov_count): Update.
18540 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18542 * gimple.h (gimple_check_failed): Mark cold.
18544 2017-06-09 Richard Biener <rguenther@suse.de>
18546 PR tree-optimization/66623
18547 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
18548 refactor check_reduction into two parts, properly computing
18549 whether we have to check reduction validity for outer loop
18552 2017-06-09 Richard Biener <rguenther@suse.de>
18554 PR tree-optimization/79483
18555 * graphite-scop-detection.c (order): New global.
18556 (get_order): Compute bb to order mapping that satisfies code
18557 generation constraints.
18558 (cmp_pbbs): New helper.
18559 (build_scops): Start domwalk at entry block, sort generated
18562 2017-06-09 Richard Biener <rguenther@suse.de>
18564 PR middle-end/81007
18565 * ipa-polymorphic-call.c
18566 (ipa_polymorphic_call_context::restrict_to_inner_class):
18567 Skip FIELD_DECLs with error_mark_node type.
18568 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
18571 2017-06-09 Martin Liska <mliska@suse.cz>
18573 * predict.c (struct branch_predictor): New struct.
18574 (test_prediction_value_range): New test.
18575 (predict_c_tests): New function.
18576 * selftest-run-tests.c (selftest::run_tests): Run the function.
18577 * selftest.h: Declare new tests.
18579 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
18582 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
18583 gen_add3_insn did not fail.
18584 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
18585 r0, construct that number in a temporary reg and add that reg to r0.
18586 If asked to put the result in r0 as well, fail.
18588 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
18590 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18591 for early expansion of vec_eqv.
18593 2017-06-08 Jakub Jelinek <jakub@redhat.com>
18595 PR middle-end/81005
18596 * ubsan.c (instrument_null): Avoid pointless code temporary.
18597 (pass_ubsan::execute): Instrument aggregate arguments of calls.
18599 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
18603 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
18606 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
18609 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18611 * predict.c (unlikely_executed_edge_p): Move ahead.
18612 (probably_never_executed_edge_p): Use it.
18614 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
18616 PR middle-end/79988
18617 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
18618 gimple_call_builtin_p call.
18620 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18622 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
18623 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
18624 rtl_check_failed_type2, rtl_check_failed_code1,
18625 rtl_check_failed_code2, rtl_check_failed_code_mode,
18626 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
18627 rtvec_check_failed_bounds, rtl_check_failed_flag,
18628 _fatal_insn_not_found, _fatal_insn): Likewise.
18629 * tree.h (tree_contains_struct_check_failed,
18630 tree_check_failed, tree_not_check_failed,
18631 tree_class_check_failed, tree_range_check_failed,
18632 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
18633 tree_vec_elt_check_failed, phi_node_elt_check_failed,
18634 tree_operand_check_failed, omp_clause_check_failed,
18635 omp_clause_operand_check_failed, omp_clause_range_check_failed):
18638 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18640 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
18641 flag_branch_probabilities.
18642 * ipa-inline.c (edge_badness): Likewise.
18643 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
18644 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
18645 * predict.c (maybe_hot_frequency_p): Likewise.
18646 (probably_never_executed): Likewise.
18647 * sched-ebb.c (schedule_ebbs): Likewise.
18648 * sched-rgn.c (find_single_block_region): Likewise.
18649 * tracer.c (tail_duplicate): Likewise.
18651 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18653 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
18654 longer requires x_flag_profile_use.
18656 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18658 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
18659 instead of flag_reorder_blocks_and_partition.
18660 * dbxout.c (dbxout_function_end): Likewise.
18661 * dwarf2out.c (gen_subprogram_die): Likewise.
18662 * haifa-sched.c (sched_create_recovery_edges): Likewise.
18663 * hw-doloop.c (reorg_loops): Likewise.
18664 * varasm.c (assemble_start_function,
18665 assemble_end_function): Likewise.
18666 (decide_function_section): Do not check for
18667 flag_reorder_blocks_and_partition.
18669 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
18671 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
18673 (chkp_get_hard_register_fake_addr_expr): Ditto.
18674 (chkp_build_addr_expr): Add check for hard reg case.
18675 (chkp_parse_array_and_component_ref): Ditto.
18676 (chkp_find_bounds_1): Ditto.
18677 (chkp_process_stmt): Don't generate bounds store for
18680 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18682 * predict.c (maybe_hot_bb_p): Do not check profile status.
18683 (maybe_hot_edge_p): Likewise.
18684 (probably_never_executed): Check for zero counts even if profile
18686 (unlikely_executed_edge_p): New function.
18687 (unlikely_executed_stmt_p): New function.
18688 (unlikely_executed_bb_p): New function.
18689 (set_even_probabilities): Use unlikely predicates.
18690 (combine_predictions_for_bb): Likewise.
18691 (predict_paths_for_bb): Likewise.
18692 (predict_paths_leading_to_edge): Likewise.
18693 (determine_unlikely_bbs): New function.
18694 (estimate_bb_frequencies): Use it.
18695 (compute_function_frequency): Use zero counts even if profile is
18697 * profile-count.h: Fix typo.
18699 2017-08-08 Julia Koval <julia.koval@intel.com>
18701 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
18702 _mm512_mask_cvtsepi16_storeu_epi8,
18703 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
18704 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
18705 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
18706 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
18707 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
18708 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
18709 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
18710 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
18711 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
18712 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
18713 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
18714 __builtin_ia32_pmovuswb256mem_mask,
18715 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
18716 __builtin_ia32_pmovwb512mem_mask): New builtins.
18718 2017-08-08 Julia Koval <julia.koval@intel.com>
18720 PR target/73350,80862
18721 * config/i386/subst.md (round): Fix round pattern.
18722 * config/i386/i386.c (ix86_erase_embedded_rounding):
18723 Fix erasing rounding for the fixed pattern.
18725 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18727 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
18729 2017-06-08 Martin Liska <mliska@suse.cz>
18731 PR gcov-profile/80911
18732 * gcov.c (block_info::block_info): New constructor.
18734 2017-06-07 Carl Love <cel@us.ibm.com>
18736 * config/rs6000/rs6000-c: The return type of the following
18737 built-in functions was implemented as int not long long. Fix sign
18738 of return value for the unsigned version of vec_mulo and vec_mule.
18739 vector unsigned long long vec_bperm (vector unsigned long long,
18740 vector unsigned char)
18741 vector signed long long vec_mule (vector signed int,
18743 vector unsigned long long vec_mule (vector unsigned int,
18744 vector unsigned int)
18745 vector signed long long vec_mulo (vector signed int,
18747 vector unsigned long long vec_mulo (vector unsigned int,
18748 vector unsigned int)
18749 * doc/extend.texi: Fix the documentation for the built-in
18752 2017-06-07 Carl Love <cel@us.ibm.com>
18755 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
18758 2017-06-07 Carl Love <cel@us.ibm.com>
18760 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
18761 support, Generate doublehv for signed int/float for BE case only.
18763 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
18765 * doc/invoke.texi (mcx16): Rewrite.
18767 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18769 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
18770 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
18771 *mov<mode>_softfloat, and an anonymous splitter): Use
18772 nonimmediate_operand instead of rs6000_nonimmediate_operand.
18774 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18776 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
18778 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
18779 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
18780 (rs6000_debug_reg_global): Adjust.
18781 (rs6000_init_hard_regno_mode_ok): Adjust.
18782 (rs6000_dbx_register_number): Adjust.
18783 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
18784 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
18785 Remove SPE_ACC and SPEFSCR.
18786 (REG_ALLOC_ORDER): Ditto.
18787 (FRAME_POINTER_REGNUM): Change to 111.
18788 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
18789 (REG_CLASS_NAMES): Ditto.
18790 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
18791 (REGISTER_NAMES): Ditto.
18792 (ADDITIONAL_REG_NAMES): Ditto.
18793 (rs6000_reg_names): Ditto.
18794 * config/rs6000/rs6000.md: Renumber some register number
18797 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18799 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
18801 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
18802 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
18804 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
18805 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
18806 Delete the SPE high registers.
18807 (REG_ALLOC_ORDER): Ditto.
18808 (enum reg_class): Remove SPE_HIGH_REGS.
18809 (REG_CLASS_NAMES): Ditto.
18810 (REG_CLASS_CONTENTS): Delete the SPE high registers.
18811 (REGISTER_NAMES): Ditto.
18812 (rs6000_reg_names): Ditto.
18813 * doc/tm.texi.in: Remove SPE as example.
18814 * doc/tm.texi: Regenerate.
18816 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18818 * config/rs6000/8540.md (ppc8540_brinc): Delete.
18819 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
18820 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
18821 * config/rs6000/rs6000.md (type): Remove "brinc".
18823 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18825 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
18826 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
18827 * config/rs6000/linuxspe.h: Delete file.
18828 * config/rs6000/rs6000.md: Don't include spe.md.
18829 * config/rs6000/spe.h: Delete file.
18830 * config/rs6000/spe.md: Delete file.
18831 * config/rs6000/t-rs6000: Remove spe.md.
18833 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18835 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
18836 (reg_or_none500mem_operand): Delete.
18837 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
18838 instead of reg_or_none500mem_operand.
18840 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18842 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
18843 handling of SPE flags.
18844 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
18846 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18848 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
18850 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
18851 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
18852 paired_divv2sf3): Similar.
18853 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
18854 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
18855 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
18857 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
18858 Rename the paired_* instruction patterns.
18859 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
18861 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
18862 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
18863 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
18864 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
18865 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
18866 PAIRED_VECTOR_MODE.
18867 (struct machine_function): Delete field spe_insn_chain_scanned_p.
18868 (spe_func_has_64bit_regs_p): Delete.
18869 (spe_expand_predicate_builtin): Delete.
18870 (spe_expand_evsel_builtin): Delete.
18871 (TARGET_DWARF_REGISTER_SPAN): Do not define.
18872 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
18873 (invalid_e500_subreg): Delete.
18874 (rs6000_legitimize_address): Always force_reg op2 as well, for
18875 paired single memory accesses.
18876 (rs6000_member_type_forces_blk): Delete.
18877 (rs6000_spe_function_arg): Delete.
18878 (rs6000_expand_unop_builtin): Delete SPE handling.
18879 (rs6000_expand_binop_builtin): Ditto.
18880 (spe_expand_stv_builtin): Delete.
18881 (bdesc_2arg_spe): Delete.
18882 (spe_expand_builtin): Delete.
18883 (spe_expand_predicate_builtin): Delete.
18884 (spe_expand_evsel_builtin): Delete.
18885 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
18886 (spe_init_builtins): Delete.
18887 (spe_func_has_64bit_regs_p): Delete.
18888 (savres_routine_name): Delete "info" parameter. Adjust callers.
18889 (rs6000_emit_stack_reset): Ditto.
18890 (rs6000_dwarf_register_span): Delete.
18891 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
18892 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
18893 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
18895 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
18897 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
18898 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
18899 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
18900 mulv2sf3, divv2sf3): Delete expanders.
18902 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18904 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
18906 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18908 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
18909 * config/rs6000/rs6000.c: Ditto.
18911 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18913 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
18914 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
18915 comparison_operator.
18917 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18919 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
18920 * config/rs6000/rs6000.opt: Ditto.
18921 * config/rs6000/t-rtems: Ditto.
18923 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18925 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
18926 TARGET_E500_SINGLE by 0, simplify.
18927 * config/rs6000/rs6000.c: Ditto.
18928 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
18929 (spe_build_register_parallel): Delete.
18930 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
18931 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
18932 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
18933 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
18934 (E500_CONVERT): Delete.
18935 * config/rs6000/spe.md: Remove many patterns and all define_constants.
18937 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18939 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
18940 * config/rs6000/dfp.md: Ditto.
18941 (negdd2, *negdd2_fpr): Merge.
18942 (absdd2, *absdd2_fpr): Merge.
18943 (negtd2, *negtd2_fpr): Merge.
18944 (abstd2, *abstd2_fpr): Merge.
18945 * config/rs6000/e500.h: Delete file.
18946 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
18947 TARGET_FPRS by 1 and simplify.
18948 * config/rs6000/rs6000-c.c: Ditto.
18949 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
18950 TARGET_DF_SPE by 0.
18951 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
18953 * config/rs6000/rs6000.md: Ditto.
18954 (floatdidf2, *floatdidf2_fpr): Merge.
18955 (move_from_CR_gt_bit): Delete.
18956 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
18957 (E500_CR_IOR_COMPARE): Delete.
18958 (All patterns that require !TARGET_FPRS): Delete.
18959 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
18961 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18963 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
18965 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18967 * graphds.c (add_edge): Intitialize edge's attached data.
18968 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
18969 pointer parameter. Call pointed function on each edge during
18970 graph traversing. Skip traversing the edge when the function
18972 (graphds_dfs, graphds_scc): Ditto.
18973 (for_each_edge): New parameter. Pass the new parameter to callback
18975 * graphds.h (skip_edge_callback): New function pointer type.
18976 (graphds_dfs, graphds_scc): New function pointer parameter.
18977 (graphds_edge_callback, for_each_edge): New parameter.
18979 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18981 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
18982 out code checking if runtime alias check is possible to below ...
18983 Call the new function.
18984 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
18985 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
18987 2017-06-07 Marek Polacek <polacek@redhat.com>
18990 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
18991 TYPE_OVERFLOW_WRAPS check.
18993 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18995 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
18996 if versioning is required.
18997 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
18998 peeling with the check for versioning.
19000 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19002 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
19003 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
19004 Set true to new parameter if new ssa variable is defined.
19005 (vect_gen_vector_loop_niters): Refactor. Set range information
19006 for the new vector loop bound variable.
19007 (vect_do_peeling): Ditto.
19009 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19011 * tree-affine.c (ssa.h): Include header file.
19012 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
19013 has wrapping overflow behavior.
19015 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19017 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
19019 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19021 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
19022 (tree_to_aff_combination): ... here.
19024 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19026 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
19027 reg_pressure model function.
19028 (ivopts_global_cost_for_size): Delete.
19029 (determine_set_costs, iv_ca_recount_cost): Call new model function
19030 ivopts_estimate_reg_pressure.
19032 2017-06-07 Tamar Christina <tamar.christina@arm.com>
19034 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
19035 expensive than udiv. Remove floating point cases from mod.
19037 2017-06-07 Tamar Christina <tamar.christina@arm.com>
19039 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
19040 Increase idiv cost.
19042 2017-06-07 Tamar Christina <tamar.christina@arm.com>
19044 * config/aarch64/aarch64.md
19045 (copysignsf3): Fix mask generation.
19047 2017-06-07 Jakub Jelinek <jakub@redhat.com>
19049 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
19051 (class dump_manager): Add register_dumps method.
19052 * dumpfile.c: Include langhooks.h.
19053 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
19054 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
19055 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
19056 (dump_manager::dump_register): Start with 512 entries instead of 32.
19057 (dump_manager::register_dumps): New method.
19058 * toplev.c (general_init): Instead of invoking register_dumps
19059 langhook, invoke register_dumps method on the dump manager.
19060 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
19063 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
19065 * doc/md.texi: Clarify the restrictions on a define_insn condition.
19066 Say that # requires an associated define_split to exist, and that
19067 the define_split must be suitable for use after register allocation.
19069 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19071 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
19072 (compute_outgoing_frequencies): Also initialize zero counts.
19073 (find_many_sub_basic_blocks): Do not produce uninitialized profile
19074 around loops; preserve more of profile when nothing changes.
19076 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
19078 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
19080 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
19081 * config/arm/arm-cpu-cdata.h: Regenerate.
19082 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
19083 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
19084 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
19085 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
19086 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
19088 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
19090 * config/arm/t-rmprofile: Likewise.
19091 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
19093 2017-06-06 David S. Miller <davem@davemloft.net>
19096 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
19097 blockage if function uses alloca.
19099 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19101 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
19102 New "uid" fields to hold pretty-print IDs of group and ref.
19103 Memory references are now identified as <group_id>:<ref_id>
19104 instead of using [random] addresses.
19105 (dump_mem_details): Simplify, no functional change.
19106 (dump_mem_ref): Simplify and make output more concise.
19107 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
19108 (find_or_create_group): Initialize group uid.
19109 (record_ref): Initialize ref uid. Improve debug output.
19110 (prune_group_by_reuse, should_issue_prefetch_p,)
19111 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
19112 (mark_nontemporal_store, determine_loop_nest_reuse):
19113 Improve debug output.
19115 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19117 * dbgcnt.def (prefetch): New debug counter.
19118 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
19119 (schedule_prefetches): Stop issueing prefetches if debug counter
19122 2017-06-06 Tom de Vries <tom@codesourcery.com>
19124 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
19125 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
19127 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19129 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
19130 Use aarch64_reg_or_zero predicate for operand 4.
19131 (aarch64_compare_and_swap<mode> define_insn_and_split):
19132 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
19133 (aarch64_store_exclusive<mode>): Likewise for operand 2.
19135 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
19137 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
19138 (arm_compute_save_core_reg_mask): This.
19139 (thumb1_compute_save_reg_mask): Rename into ...
19140 (thumb1_compute_save_core_reg_mask): This.
19141 (arm_compute_save_reg0_reg12_mask): Adapt comment.
19142 (arm_compute_frame_layout): Likewise.
19144 2017-06-06 Richard Biener <rguenther@suse.de>
19146 PR tree-optimization/80974
19147 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
19148 keep or clear leaders SSA info.
19150 2017-06-06 Tom de Vries <tom@codesourcery.com>
19152 * config/nvptx/nvptx.c (split_mode_p): New function.
19153 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
19155 2017-06-06 Tom de Vries <tom@codesourcery.com>
19157 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
19159 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19162 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
19165 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19167 * shrink-wrap.c (handle_simple_exit): Update profile.
19168 (try_shrink_wrapping): Upate profile.
19170 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19172 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
19173 (tree_guess_outgoing_edge_probabilities): New.
19174 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
19175 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
19177 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19179 * ipa-split.c (split_function): Initialize return bb profile.
19181 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19183 * profile.c (compute_branch_probabilities): Also initialize
19184 EXIT_BLOCK profile.
19186 2017-06-06 Richard Biener <rguenther@suse.de>
19188 PR tree-optimization/80928
19189 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
19190 (vect_analyze_loop_operations): Properly guard analysis for
19192 (vect_transform_loop): Likewise.
19193 (vect_analyze_loop_2): Also reset SLP type on PHIs.
19194 (vect_model_induction_cost): Do not cost for pure SLP.
19195 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
19196 of induction in inner loop vectorization.
19197 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
19198 (vect_get_and_check_slp_defs): Handle vect_induction_def.
19199 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
19201 (vect_analyze_slp_cost_1): Cost induction.
19202 (vect_detect_hybrid_slp_stmts): Handle PHIs.
19203 (vect_get_slp_vect_defs): Likewise.
19204 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
19205 (vect_transform_stmt): Handle SLP reductions.
19206 * tree-vectorizer.h (vectorizable_induction): Adjust.
19208 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19210 * config/rs6000/rs6000.c (make_resolver_func): Update
19211 init_lowered_empty_function call.
19213 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
19215 * doc/invoke.texi: Document the -fprofile-abs-path option.
19216 * common.opt (fprofile-abs-path): New option.
19217 * gcov-io.h (gcov_write_filename): Declare.
19218 * gcov-io.c (gcov_write_filename): New function.
19219 * coverage.c (coverage_begin_function): Use gcov_write_filename.
19220 * profile.c (output_location): Likewise.
19222 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19224 * shring-wrap.c: Revert accidental commit.
19226 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
19228 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
19230 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19232 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
19234 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
19235 profile in callgraph edge.
19236 * profile-count.h (apply_probability): If THIS is 0, then result is 0
19237 (apply_scale): Likewise.
19238 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
19239 Also scale profile when inlining function with zero profile.
19240 (initialize_cfun): Update exit block profile even when it is zero.
19241 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
19242 when profile is read.
19244 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19246 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
19247 (CLONE_*): New constants to define the processors we can generate
19248 code for with the target_clone attribute.
19249 (rs6000_clone_map): New array to identify which clone processors
19250 the current program is running on.
19251 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
19252 target_clone attribute.
19253 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
19254 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
19255 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
19256 (cpu_expand_builtin): Add support for target_clone attribute.
19257 (rs6000_valid_attribute_p): Allow "default" attribute.
19258 (get_decl_name): New debug function to simplify printing the
19259 current function name in debugging statements.
19260 (rs6000_clone_priority): New functions to support the target_clone
19261 attribute, and be able to generate code to switch between ISA 2.05
19262 through ISA 3.0 (power6 through power9).
19263 (rs6000_compare_version_priority): Likewise.
19264 (rs6000_get_function_versions_dispatcher): Likewise.
19265 (make_resolver_func): Likewise.
19266 (add_condition_to_bb): Likewise.
19267 (dispatch_function_versions): Likewise.
19268 (rs6000_generate_version_dispatcher_body): Likewise.
19269 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
19270 (fusion_gpr_load_p): Fix a spacing issue.
19271 * doc/extend.texi (Common Function Attributes): Document that the
19272 PowerPC supports the target_clone attribute.
19274 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
19276 * config/arm/arm.h: explain F symbol found in description of ARM
19277 register allocation in its legend.
19279 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19281 * config/mips/frame-header-opt.c: Include profile-count.h.
19282 * config/riscv/riscv.c: Include profile-count.h
19284 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19286 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
19288 (sm_set_flag_if_changed): Add bbs field.
19289 (execute_sm_if_changed_flag_set): Pass BBS.
19290 (execute_sm): Update.
19292 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19294 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
19297 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19299 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
19300 (peephole2): New peephole2 to emit the above.
19301 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
19303 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19305 * config/aarch64/aarch64.c (define_peephole2 above
19306 *sub_<shift>_<mode>): New peephole.
19308 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19310 * config/i386/i386.c (make_resolver_func): Update.
19311 * Makefile.in: Add profile-count.h and profile-count.o
19312 * auto-profile.c (afdo_indirect_call): Update to new API.
19313 (afdo_set_bb_count): Update.
19314 (afdo_propagate_edge): Update.
19315 (afdo_propagate_circuit): Update.
19316 (afdo_calculate_branch_prob): Update.
19317 (afdo_annotate_cfg): Update.
19318 * basic-block.h: Include profile-count.h
19319 (struct edge_def): Turn count to profile_count.
19320 (struct basic_block_def): Likewie.
19321 (REG_BR_PROB_BASE): Move to profile-count.h
19322 (RDIV): Move to profile-count.h
19323 * bb-reorder.c (max_entry_count): Turn to profile_count.
19324 (find_traces): Update.
19325 (rotate_loop):Update.
19326 (connect_traces):Update.
19327 (sanitize_hot_paths):Update.
19328 * bt-load.c (migrate_btr_defs): Update.
19329 * cfg.c (RDIV): Remove.
19330 (init_flow): Use alloc_block.
19331 (alloc_block): Uninitialize count.
19332 (unchecked_make_edge): Uninitialize count.
19333 (check_bb_profile): Update.
19334 (dump_edge_info): Update.
19335 (dump_bb_info): Update.
19336 (update_bb_profile_for_threading): Update.
19337 (scale_bbs_frequencies_int): Update.
19338 (scale_bbs_frequencies_gcov_type): Update.
19339 (scale_bbs_frequencies_profile_count): New.
19340 * cfg.h (update_bb_profile_for_threading): Update.
19341 (scale_bbs_frequencies_profile_count): Declare.
19342 * cfgbuild.c (compute_outgoing_frequencies): Update.
19343 (find_many_sub_basic_blocks): Update.
19344 * cfgcleanup.c (try_forward_edges): Update.
19345 (try_crossjump_to_edge): Update.
19346 * cfgexpand.c (expand_gimple_tailcall): Update.
19347 (construct_exit_block): Update.
19348 * cfghooks.c (verify_flow_info): Update.
19349 (dump_bb_for_graph): Update.
19350 (split_edge): Update.
19351 (make_forwarder_block): Update.
19352 (duplicate_block): Update.
19353 (account_profile_record): Update.
19354 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
19355 (get_estimated_loop_iterations): Update.
19356 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
19357 (single_likely_exit): Update.
19358 * cfgloopmanip.c (scale_loop_profile): Update.
19360 (set_zero_probability): Update.
19361 (lv_adjust_loop_entry_edge): Update.
19362 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
19363 (purge_dead_edges): Update.
19364 (rtl_account_profile_record): Update.
19365 * cgraph.c (cgraph_node::create): Uninitialize count.
19366 (symbol_table::create_edge): Uninitialize count.
19367 (cgraph_update_edges_for_call_stmt_node): Update.
19368 (cgraph_edge::dump_edge_flags): Update.
19369 (cgraph_node::dump): Update.
19370 (cgraph_edge::maybe_hot_p): Update.
19371 * cgraph.h: Include profile-count.h
19372 (create_clone), create_edge, create_indirect_edge): Update.
19373 (cgraph_node): Turn count to profile_count.
19374 (cgraph_edge0: Likewise.
19375 (make_speculative, clone): Update.
19376 (create_edge): Update.
19377 (init_lowered_empty_function): Update.
19378 * cgraphclones.c (cgraph_edge::clone): Update.
19379 (duplicate_thunk_for_node): Update.
19380 (cgraph_node::create_clone): Update.
19381 * cgraphunit.c (cgraph_node::analyze): Update.
19382 (cgraph_node::expand_thunk): Update.
19383 * final.c (dump_basic_block_info): Update.
19384 * gimple-streamer-in.c (input_bb): Update.
19385 * gimple-streamer-out.c (output_bb): Update.
19386 * graphite.c (print_global_statistics): Update.
19387 (print_graphite_scop_statistics): Update.
19388 * hsa-brig.c: Include basic-block.h.
19389 * hsa-dump.c: Include basic-block.h.
19390 * hsa-gen.c (T sum_slice): Update.
19391 (convert_switch_statements):Update.
19392 * hsa-regalloc.c: Include basic-block.h.
19393 * ipa-chkp.c (chkp_produce_thunks): Update.
19394 * ipa-cp.c (struct caller_statistics): Update.
19395 (init_caller_stats): Update.
19396 (gather_caller_stats): Update.
19397 (ipcp_cloning_candidate_p): Update.
19398 (good_cloning_opportunity_p): Update.
19399 (get_info_about_necessary_edges): Update.
19400 (dump_profile_updates): Update.
19401 (update_profiling_info): Update.
19402 (update_specialized_profile): Update.
19403 (perhaps_add_new_callers): Update.
19404 (decide_about_value): Update.
19405 (ipa_cp_c_finalize): Update.
19406 * ipa-devirt.c (struct odr_type_warn_count): Update.
19407 (struct decl_warn_count): Update.
19408 (struct final_warning_record): Update.
19409 (possible_polymorphic_call_targets): Update.
19410 (ipa_devirt): Update.
19411 * ipa-fnsummary.c (redirect_to_unreachable): Update.
19412 * ipa-icf.c (sem_function::merge): Update.
19413 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
19414 * ipa-inline.c (compute_uninlined_call_time): Update.
19415 (compute_inlined_call_time): Update.
19416 (want_inline_small_function_p): Update.
19417 (want_inline_self_recursive_call_p): Update.
19418 (edge_badness): Update.
19419 (lookup_recursive_calls): Update.
19420 (recursive_inlining): Update.
19421 (inline_small_functions): Update.
19422 (dump_overall_stats): Update.
19423 (dump_inline_stats): Update.
19424 * ipa-profile.c (ipa_profile_generate_summary): Update.
19425 (ipa_propagate_frequency): Update.
19426 (ipa_profile): Update.
19427 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19428 * ipa-utils.c (ipa_merge_profiles): Update.
19429 * loop-doloop.c (doloop_modify): Update.
19430 * loop-unroll.c (report_unroll): Update.
19431 (unroll_loop_runtime_iterations): Update.
19432 * lto-cgraph.c (lto_output_edge): Update.
19433 (lto_output_node): Update.
19434 (input_node): Update.
19435 (input_edge): Update.
19436 (merge_profile_summaries): Update.
19437 * lto-streamer-in.c (input_cfg): Update.
19438 * lto-streamer-out.c (output_cfg): Update.
19439 * mcf.c (create_fixup_graph): Update.
19440 (adjust_cfg_counts): Update.
19441 (sum_edge_counts): Update.
19442 * modulo-sched.c (sms_schedule): Update.
19443 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
19444 * predict.c (maybe_hot_count_p): Update.
19445 (probably_never_executed): Update.
19446 (dump_prediction): Update.
19447 (combine_predictions_for_bb): Update.
19448 (propagate_freq): Update.
19449 (handle_missing_profiles): Update.
19450 (counts_to_freqs): Update.
19451 (rebuild_frequencies): Update.
19452 (force_edge_cold): Update.
19453 * predict.h: Include profile-count.h
19454 (maybe_hot_count_p, counts_to_freqs): UPdate.
19455 * print-rtl-function.c: Do not include cfg.h
19456 * print-rtl.c: Include basic-block.h
19457 * profile-count.c: New file.
19458 * profile-count.h: New file.
19459 * profile.c (is_edge_inconsistent): Update.
19460 (correct_negative_edge_counts): Update.
19461 (is_inconsistent): Update.
19462 (set_bb_counts): Update.
19463 (read_profile_edge_counts): Update.
19464 (compute_frequency_overlap): Update.
19465 (compute_branch_probabilities): Update; Initialize and deinitialize
19467 (branch_prob): Update.
19468 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
19469 (edge_gcov_count): New.
19470 (bb_gcov_count): New.
19471 * shrink-wrap.c (try_shrink_wrapping): Update.
19472 * tracer.c (better_p): Update.
19473 * trans-mem.c (expand_transaction): Update.
19474 (ipa_tm_insert_irr_call): Update.
19475 (ipa_tm_insert_gettmclone_call): Update.
19476 * tree-call-cdce.c: Update.
19477 * tree-cfg.c (gimple_duplicate_sese_region): Update.
19478 (gimple_duplicate_sese_tail): Update.
19479 (gimple_account_profile_record): Update.
19480 (execute_fixup_cfg): Update.
19481 * tree-inline.c (copy_bb): Update.
19482 (copy_edges_for_bb): Update.
19483 (initialize_cfun): Update.
19484 (freqs_to_counts): Update.
19485 (copy_cfg_body): Update.
19486 (expand_call_inline): Update.
19487 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19488 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
19489 (try_unroll_loop_completely): Update.
19490 (try_peel_loop): Update.
19491 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19492 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
19493 * tree-ssa-loop-split.c (connect_loops): Update.
19494 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
19495 * tree-ssa-reassoc.c (branch_fixup): Update.
19496 * tree-ssa-tail-merge.c (replace_block_by): Update.
19497 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
19498 (compute_path_counts): Update.
19499 (update_profile): Update.
19500 (recompute_probabilities): Update.
19501 (update_joiner_offpath_counts): Update.
19502 (estimated_freqs_path): Update.
19503 (freqs_to_counts_path): Update.
19504 (clear_counts_path): Update.
19505 (ssa_fix_duplicate_block_edges): Update.
19506 (duplicate_thread_path): Update.
19507 * tree-switch-conversion.c (case_bit_test_cmp): Update.
19508 (struct switch_conv_info): Update.
19509 * tree-tailcall.c (decrease_profile): Update.
19510 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19511 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19512 * value-prof.c (check_counter): Update.
19513 (gimple_divmod_fixed_value): Update.
19514 (gimple_mod_pow2): Update.
19515 (gimple_mod_subtract): Update.
19516 (gimple_ic_transform): Update.
19517 (gimple_stringop_fixed_value): Update.
19518 * value-prof.h (gimple_ic): Update.
19520 2017-06-02 Carl Love <cel@us.ibm.com>
19522 * config/rs6000/rs6000-c: Add support for built-in functions
19523 vector double vec_doublee (vector signed int);
19524 vector double vec_doublee (vector unsigned int);
19525 vector double vec_doublee (vector float);
19526 vector double vec_doubleh (vector signed int);
19527 vector double vec_doubleh (vector unsigned int);
19528 vector double vec_doubleh (vector float);
19529 vector double vec_doublel (vector signed int);
19530 vector double vec_doublel (vector unsigned int);
19531 vector double vec_doublel (vector float);
19532 vector double vec_doubleo (vector signed int);
19533 vector double vec_doubleo (vector unsigned int);
19534 vector double vec_doubleo (vector float);.
19535 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
19536 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
19538 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
19539 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
19540 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
19542 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
19543 vec_doublel, vec_doubleh.
19544 * doc/extend.texi: Update the built-in documentation file for the
19545 new built-in functions.
19547 2017-06-02 David Malcolm <dmalcolm@redhat.com>
19550 * ipa-inline-analysis.c (free_growth_caches): Set
19551 edge_removal_hook_holder to NULL after removing it.
19553 2017-06-02 Sudakshina Das <sudi.das@arm.com>
19555 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
19556 comparision with zero.
19558 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
19559 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19560 for early expansion of vec_min and vec_max builtins.
19561 (builtin_function_type): Add min/max unsigned variants to those
19562 identified as having unsigned arguments.
19564 2017-06-02 Olivier Hainque <hainque@adacore.com>
19566 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
19568 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19570 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
19571 Use VALL_F16 iterator rather than VALL.
19573 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19575 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
19576 Emit CBNZ inside loop when doing a strong exchange and comparing
19577 against zero. Generate the CC flags after the loop.
19579 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
19581 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
19582 (dl_section_ref): New.
19583 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
19584 On AIX, append an expression to subtract the size of the
19585 section length to dl_section_ref.
19587 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
19589 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19590 for early expansion of vector absolute builtins.
19592 2017-06-02 Richard Biener <rguenther@suse.de>
19594 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
19595 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
19597 2017-06-02 Richard Biener <rguenther@suse.de>
19599 PR tree-optimization/80948
19600 * tree-tailcall.c (find_tail_calls): Track stmts to move in
19601 stmt order as well.
19603 2017-06-02 Richard Biener <rguenther@suse.de>
19605 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
19607 * tree-vect-stmts.c (process_use): Do not mark backedge defs
19608 for inductions as relevant.
19610 2017-06-02 Richard Biener <rguenther@suse.de>
19612 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
19613 (vectorizable_induction): ... this. Remove dead code.
19615 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
19617 * builtins. (expand_builtin_alloca): Remove second parameter and
19618 infer its value from the first parameter instead.
19619 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
19621 2017-06-02 Jakub Jelinek <jakub@redhat.com>
19623 PR rtl-optimization/80903
19624 * loop-doloop.c (add_test): Unshare sequence.
19626 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
19628 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
19630 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
19632 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
19634 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
19635 xlogue_layout::get_instance, logue_layout::xlogue_layout,
19636 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
19637 (xlogue_layout::get_stub_rtx): Make static.
19638 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
19639 (xlogue_layout::compute_stub_managed_regs): Rename to...
19640 (xlogue_layout::count_stub_managed_regs): ...this.
19641 (xlogue_layout::is_stub_managed_reg): New function.
19642 (xlogue_layout::m_stub_names): Rename to...
19643 (xlogue_layout::s_stub_names): ...this, make static.
19644 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
19645 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
19646 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
19647 xlogue_layout::s_stub_names): Instantiate statics.
19648 (stub_managed_regs): Remove.
19649 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
19650 (disable_call_ms2sysv_xlogues): Rename to...
19651 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
19652 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
19654 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
19655 change after reload_completed.
19656 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
19658 (ix86_expand_prologue): Likewise.
19659 (ix86_expand_epilogue): Likewise.
19660 (ix86_expand_split_stack_prologue): Likewise.
19661 (ix86_compute_frame_layout): Remove frame parameter ...
19662 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
19663 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
19665 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
19666 (ix86_frame): Move from here ...
19667 * config/i386/i386.h (ix86_frame): ... to here.
19668 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
19669 complete ix86_frame data structure instead. Remove some_ld_name.
19671 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
19673 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
19674 symbols that hold a DECL_VALUE_EXPR.
19676 2017-06-01 Martin Jambor <mjambor@suse.cz>
19678 PR tree-optimization/80898
19679 * tree-sra.c (process_subtree_disqualification): Removed.
19680 (disqualify_candidate): Do not acll
19681 process_subtree_disqualification.
19682 (subtree_mark_written_and_enqueue): New function.
19683 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
19684 RHS has been disqualified and re-queue LHS if necessary. Apart
19685 from that, ignore disqualified RHS.
19687 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19689 * config/s390/s390.c (s390_emit_epilogue): Disable early return
19690 address fetch for z10 or later.
19692 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19694 * config/arc/arc.md (tst_movb): Add guard when splitting.
19696 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19698 * config/arc/arc.c (arc_can_eliminate): Test against
19699 arc_frame_pointer_needed.
19701 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19703 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
19704 to prevent store reordering.
19705 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
19706 (type): Add block type.
19707 (stack_tie): Define special instruction to be used in
19710 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19712 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
19713 constraint. It is not valid for the pattern.
19714 (noncommutative_binary_comparison): Likewise.
19716 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19718 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
19721 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19723 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
19724 be used by the reg-alloc.
19726 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19728 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
19729 reg-alloc when having mul64 or mul32x16 instructions.
19730 (mulsidi3): Likewise.
19731 (umulsidi3): Likewise.
19732 (mulsi32x16): New pattern.
19733 (mulsi64): Likewise.
19734 (mulsidi64): Likewise.
19735 (umulsidi64): Likewise.
19736 (MUL32x16_REG): Define.
19737 (mul64_600): Use MUL32x16_REG.
19738 (mac64_600): Likewise.
19739 (umul64_600): Likewise.
19740 (umac64_600): Likewise.
19742 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19744 * config/arc/arc.md (mulsi3_700): Make it commutative.
19746 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
19748 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
19750 (*sign_extendsidi2_insn): Likewise for movstosw.
19752 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
19754 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
19755 the type of the input discriminant value. Convert the
19756 discriminant value of signedness vary.
19758 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
19760 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
19761 Add to -Wall section.
19763 2017-06-01 Richard Biener <rguenther@suse.de>
19765 PR middle-end/66313
19766 * fold-const.c (fold_plusminus_mult_expr): If the factored
19767 factor may be zero use a wrapping type for the inner operation.
19768 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
19769 and handle moved defs.
19770 (process_assignment): Properly guard the unary op case. Return a
19771 tri-state indicating that moving the stmt before the call may allow
19772 to continue. Pass through to_move.
19773 (find_tail_calls): Handle moving unrelated defs before
19776 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
19779 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
19780 splitter result in the canonical way.
19782 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
19784 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
19785 also for 32bit target. Update insn attributes.
19786 (zero-extendsidi2 splitter): Allow all registers for operand 1.
19788 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
19790 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
19791 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
19792 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
19793 (_mm_maskz_min_ss): New intrinsics.
19795 2017-05-31 Martin Liska <mliska@suse.cz>
19797 * tree-vect-loop.c (vect_create_epilog_for_reduction):
19798 Change comment style to one we normally use.
19799 (vectorizable_reduction): Likewise.
19800 (vectorizable_induction): Likewise.
19801 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19802 (vectorizable_call): Likewise.
19803 (vectorizable_simd_clone_call): Likewise.
19804 (vectorizable_conversion): Likewise.
19805 (vectorizable_assignment): Likewise.
19806 (vectorizable_shift): Likewise.
19807 (vectorizable_operation): Likewise.
19808 (vectorizable_store): Likewise.
19809 (vectorizable_load): Likewise.
19810 * tree-vectorizer.h: Likewise.
19812 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
19814 * passes.c (emergency_dump_function): New.
19815 * tree-pass.h (emergency_dump_function): Declare.
19816 * plugin.c (plugins_internal_error_function): Remove.
19817 * plugin.h (plugins_internal_error_function): Remove declaration.
19818 * toplev.c (internal_error_function): New static function. Use it...
19819 (general_init): ...here.
19821 2017-05-31 Graham Markall <graham.markall@embecosm.com>
19823 * config/arc/arc.c (arc_print_operand): Handle constant operands.
19824 (arc_rtx_costs): Add costs for new patterns.
19825 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
19826 * config/arc/predicates.md: Add _1_2_3_operand predicate.
19828 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
19830 * tree-ssa-strlen.c (get_next_strinfo): New function.
19831 (get_stridx_plus_constant): Use it.
19832 (zero_length_string): Likewise.
19833 (adjust_related_strinfos): Likewise.
19834 (adjust_last_stmt): Likewise.
19836 2017-05-31 Richard Biener <rguenther@suse.de>
19839 * config/i386/i386.c (ix86_expand_builtin): Remove assert
19840 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
19842 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
19844 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
19845 loop_vinfo argument and use of dependence distance vectors.
19846 Check instead whether the two references differ only in their
19847 initial value and assume that they have the same alignment if the
19848 difference is a multiple of the vector alignment.
19849 (vect_analyze_data_refs_alignment): Update call accordingly.
19851 2017-05-31 Martin Liska <mliska@suse.cz>
19854 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
19856 2017-05-31 Bin Cheng <bin.cheng@arm.com>
19858 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
19859 (create_intersect_range_checks): Move from ...
19860 * tree-data-ref.c (create_intersect_range_checks_index)
19861 (create_intersect_range_checks): ... to here.
19862 (create_runtime_alias_checks): New function factored from ...
19863 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
19864 here. Call above function.
19865 * tree-data-ref.h (create_runtime_alias_checks): New function.
19867 2017-05-31 Bin Cheng <bin.cheng@arm.com>
19869 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
19870 segment length for dr_b and compute it in wide_int.
19872 2017-05-31 Richard Biener <rguenther@suse.de>
19874 PR tree-optimization/80906
19875 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
19876 and pass through iv_map.
19877 (copy_bb_and_scalar_dependences): Adjust.
19878 (translate_pending_phi_nodes): Likewise.
19879 (copy_loop_close_phi_args): Handle code-generating IVs instead
19882 2017-05-30 David Malcolm <dmalcolm@redhat.com>
19884 * diagnostic-color.c (color_dict): Add "type-diff".
19885 (parse_gcc_colors): Update comment.
19886 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
19887 -fdiagnostics-show-template-tree and -fno-elide-type.
19888 (GCC_COLORS): Add type-diff to example.
19890 (-fdiagnostics-show-template-tree): New.
19891 (-fno-elide-type): New.
19892 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
19893 the pp_format_decoder callback. Call any m_format_postprocessor's
19895 (pretty_printer::pretty_printer): Initialize
19896 m_format_postprocessor.
19897 (pretty_printer::~pretty_printer): Delete any
19898 m_format_postprocessor.
19899 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
19900 (class format_postprocessor): New class.
19901 (struct pretty_printer::format_decoder): Document the new parameters.
19902 (struct pretty_printer::m_format_postprocessor): New field.
19903 * tree-diagnostic.c (default_tree_printer): Update for new
19904 bool and const char ** params.
19905 * tree-diagnostic.h (default_tree_printer): Likewise.
19907 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
19909 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
19910 (lwa_operand): Delete rs6000_gen_cell_microcode test.
19911 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
19912 rs6000_gen_cell_microcode code.
19913 (rs6000_final_prescan_insn): Delete.
19914 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
19915 "warn-cell-microcode" entries.
19916 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
19917 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
19918 throughout. Change cc_reg_not_micro_cr0_operand to
19919 cc_reg_not_cr0_operand throughout.
19920 (*extendhi<mode>2_noload): Delete.
19921 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
19922 (mwarn-cell-microcode): Delete.
19923 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
19924 -mgen-cell-microcode and -mwarn-cell-microcode.
19926 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
19929 * config/i386/constraints.md (Yd): New constraint.
19931 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
19932 and (?Yd, r) alternatives. Update insn attributes.
19933 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
19934 and (?*Yd, r) alternatives. Update insn attributes.
19935 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
19937 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
19939 * gimplify.c (gimplify_modify_expr): Don't create a
19940 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
19943 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
19945 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
19947 2017-05-30 Richard Biener <rguenther@suse.de>
19949 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
19950 and reduc_def fields.
19951 (STMT_VINFO_REDUC_TYPE): New define.
19952 (STMT_VINFO_REDUC_DEF): Likewise.
19953 (vect_force_simple_reduction): Adjust prototype.
19954 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
19955 (vect_is_simple_reduction): Remove check_reduction argument.
19956 (vect_force_simple_reduction): Adjust and set
19957 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
19958 (vectorizable_reduction): Do not re-do reduction analysis
19959 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
19960 * tree-parloops.c (gather_scalar_reductions): Adjust.
19962 2017-05-30 Richard Biener <rguenther@suse.de>
19964 PR middle-end/80901
19965 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
19968 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
19970 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
19971 Introduce unknown_misalignment parameter and remove vf.
19972 (vect_peeling_hash_get_lowest_cost):
19973 Pass unknown_misalignment parameter.
19974 (vect_enhance_data_refs_alignment):
19975 Fix unsupportable data ref treatment.
19977 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19979 * tree-vect-data-refs.c (vect_get_data_access_cost):
19980 Workaround for SLP handling.
19981 (vect_enhance_data_refs_alignment):
19982 Compute costs for doing no peeling at all, compare to the best
19983 peeling costs so far and avoid peeling if cheaper.
19985 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19987 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
19988 Return peeling info and set costs to zero for unlimited cost
19990 (vect_enhance_data_refs_alignment): Also inspect all datarefs
19991 with unknown misalignment. Compute and costs for unknown
19992 misalignment, compare them to the costs for known misalignment
19993 and choose the cheapest for peeling.
19995 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19997 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
19998 (vect_get_peeling_costs_all_drs): Create function.
19999 (vect_peeling_hash_get_lowest_cost):
20000 Use vect_get_peeling_costs_all_drs.
20001 (vect_peeling_supportable): Create function.
20002 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
20004 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20006 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
20007 DR_HAS_NEGATIVE_STEP.
20008 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
20009 (vect_enhance_data_refs_alignment): Use.
20010 (vect_duplicate_ssa_name_ptr_info): Use.
20011 * tree-vectorizer.h (dr_misalignment): Use.
20012 (known_alignment_for_access_p): Use.
20014 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20017 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
20019 (has_section_name): New function.
20021 2017-05-30 Martin Liska <mliska@suse.cz>
20024 * auto-profile.c (get_function_decl_from_block): Fix
20027 2017-05-30 Richard Biener <rguenther@suse.de>
20029 PR middle-end/80876
20030 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
20032 2017-05-30 Martin Liska <mliska@suse.cz>
20034 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
20035 * dumpfile.h (struct dump_file_info): Remove ctors.
20037 2017-05-30 Martin Liska <mliska@suse.cz>
20039 * predict.def: Fix GNU coding style.
20041 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
20043 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
20044 Mark 'to' argument with ATTRIBUTE_UNUSED.
20046 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
20048 * config/xtensa/xtensa.c (xtensa_emit_call): Use
20049 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
20050 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
20053 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
20055 * doc/install.texi (Options specification): Restore entry of
20056 --enable-sjlj-exceptions.
20058 2017-05-27 Michael Eager <eager@eagercon.com>
20061 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
20063 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
20065 * config/microblaze/microblaze.h
20066 (FIXED_REGISTERS): Update in macro.
20067 (CALL_USED_REGISTERS): Update in macro.
20069 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
20071 * doc/install.texi: Add links to macOS binary distributions.
20073 2017-05-27 Jakub Jelinek <jakub@redhat.com>
20077 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20079 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
20081 2017-05-26 Martin Liska <mliska@suse.cz>
20083 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
20085 2017-05-26 Martin Liska <mliska@suse.cz>
20087 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
20088 always leading ';; '.
20089 (dump_bb_info): Likewise.
20090 (brief_dump_cfg): Likewise.
20091 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
20092 * dumpfile.c: Remove usage of TDF_VERBOSE.
20093 * dumpfile.h (enum dump_kind): Likewise.
20094 (dump_gimple_bb_header): Do not use TDF_COMMENT.
20095 * print-tree.c (debug_verbose): Remove.
20096 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
20097 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
20098 * tree-diagnostic.c (default_tree_printer): Replace
20099 TDF_DIAGNOSTIC with TDF_SLIM.
20101 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20103 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
20104 in parameter loop, rather than loop_vinfo.
20105 (create_intersect_range_checks): Ditto.
20106 (vect_create_cond_for_alias_checks): Update call to above functions.
20108 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20110 PR tree-optimization/80815
20111 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
20112 for merging runtime alias checks. Handle negative DR_STEPs.
20114 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20116 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
20118 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
20119 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
20120 out code pruning runtime alias checks.
20121 * tree-data-ref.c (prune_runtime_alias_test_list): New function
20122 factored out from above.
20123 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
20125 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
20127 (prune_runtime_alias_test_list): New decalaration.
20129 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20131 * tree-vect-data-refs.c (compare_tree): Rename and move ...
20132 * tree-data-ref.c (data_ref_compare_tree): ... to here.
20133 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
20134 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
20135 (operator==, comp_dr_with_seg_len_pair): Ditto.
20136 (vect_prune_runtime_alias_test_list): Ditto.
20138 2017-05-26 Martin Liska <mliska@suse.cz>
20141 * params.def: Bound partial-inlining-entry-probability param.
20143 2017-05-26 Marek Polacek <polacek@redhat.com>
20146 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
20149 2017-05-26 Richard Biener <rguenther@suse.de>
20151 PR tree-optimization/80842
20152 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
20155 2017-05-26 Richard Biener <rguenther@suse.de>
20157 PR tree-optimization/80844
20158 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
20160 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
20162 * doc/md.texi (Machine Constraints): Update x86 family
20163 machine constraints section to match 'config/i386/constraints.md'.
20165 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
20167 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
20169 2017-05-25 Nathan Sidwell <nathan@acm.org>
20171 * doc/invoke.texi (--enable-languages): Update documentation.
20173 2017-05-25 Martin Liska <mliska@suse.cz>
20175 * dumpfile.c: Add TDF_FOLDING.
20176 * dumpfile.h (enum dump_kind): Likewise.
20177 * genmatch.c (dt_simplify::gen_1): Use it.
20179 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20181 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
20183 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20185 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
20186 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
20188 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20190 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
20191 * match.pd (X == C): Rewrite it here.
20192 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
20193 with_certain_nonzero_bits2): New predicates.
20194 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
20196 2017-05-24 Nathan Sidwell <nathan@acm.org>
20198 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
20201 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
20204 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
20206 * config/powerpcspe: New port. Files are copied from the rs6000
20207 port, with "rs6000" in filenames replaced by "powerpcspe".
20209 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
20211 PR rtl-optimization/80754
20212 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
20214 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
20216 * config/sparc/sparc.md (length): Return the correct value for -mflat
20217 sibcalls to match output_sibcall.
20219 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
20223 * config/rs6000/rs6000.c (struct machine_function): Add new field
20225 (rs6000_get_separate_components): Init that field, use it.
20226 (rs6000_components_for_bb): Use the field.
20228 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
20230 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
20232 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
20234 PR middle-end/80823
20235 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
20237 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20240 * config/s390/s390.c (s390_check_qrst_address): Check incoming
20241 address against address_operand predicate.
20242 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
20244 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
20246 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
20247 parameters passed indirectly.
20249 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
20251 * config/i386/i386.md (*movdi_internal): Remove SSE4
20252 alternative 18 (?r, *v). Update insn attributes.
20253 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
20254 Update insn attributes.
20255 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
20256 Update insn attributes.
20257 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
20258 alternative 1 (r, v). Remove isa attribute.
20259 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
20260 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
20261 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
20263 2017-05-23 Tom de Vries <tom@codesourcery.com>
20265 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
20268 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20270 * cgraphunit.c (symbol_table::process_new_functions): Update.
20271 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
20272 (inline_generate_summary): Rename to ...
20273 (ipa_fn_summary_generate): ... this one.
20274 (inline_read_summary): Rename to ...
20275 (ipa_fn_summary_read): ... this one.
20276 (inline_write_summary): Rename to ...
20277 (ipa_fn_summary_write): ... this one.
20278 (inline_free_summary): Rename to ...
20279 (ipa_free_fn_summary): ... this one.
20280 (pass_data_local_fn_summary, pass_local_fn_summary,
20281 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
20282 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
20283 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
20284 make_pass_ipa_fn_summary): New.
20285 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
20286 inline_write_summary, inline_free_summary): Remove.
20287 (ipa_free_fn_summary) : New.
20288 * ipa-inline.c (ipa_inline): Update.
20289 (pass_ipa_inline): Do not generate summaries.
20290 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
20292 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
20293 and add pass_ipa_fn_summary.
20294 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
20296 (make_pass_inline_parameters): Remove.
20298 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
20300 * omp-low.c (struct omp_context): Remove "default_kind" member.
20303 * omp-offload.c (execute_oacc_device_lower): Remove the
20304 parallelism dimensions function attributes for unparallelized
20305 OpenACC kernels constructs.
20307 2017-05-23 Martin Liska <mliska@suse.cz>
20309 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
20311 (cgraph_edge::make_speculative): Likewise.
20312 (cgraph_edge::resolve_speculation): Likewise.
20313 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
20314 (cgraph_node::dump): Likewise.
20315 * cgraph.h: Likewise.
20316 * cgraphunit.c (analyze_functions): Likewise.
20317 (symbol_table::compile): Likewise.
20318 * ipa-cp.c (print_all_lattices): Likewise.
20319 (determine_versionability): Likewise.
20320 (initialize_node_lattices): Likewise.
20321 (ipcp_verify_propagated_values): Likewise.
20322 (estimate_local_effects): Likewise.
20323 (update_profiling_info): Likewise.
20324 (create_specialized_node): Likewise.
20325 (perhaps_add_new_callers): Likewise.
20326 (decide_about_value): Likewise.
20327 (decide_whether_version_node): Likewise.
20328 (identify_dead_nodes): Likewise.
20329 (ipcp_store_bits_results): Likewise.
20330 * ipa-devirt.c (dump_targets): Likewise.
20331 (ipa_devirt): Likewise.
20332 * ipa-icf.c (sem_item::dump): Likewise.
20333 (sem_function::equals): Likewise.
20334 (sem_variable::equals): Likewise.
20335 (sem_item_optimizer::read_section): Likewise.
20336 (sem_item_optimizer::execute): Likewise.
20337 (congruence_class::dump): Likewise.
20338 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
20339 (dump_inline_summary): Likewise.
20340 (estimate_node_size_and_time): Likewise.
20341 (inline_analyze_function): Likewise.
20342 * ipa-inline-transform.c (inline_call): Likewise.
20343 * ipa-inline.c (report_inline_failed_reason): Likewise.
20344 (want_early_inline_function_p): Likewise.
20345 (edge_badness): Likewise.
20346 (update_edge_key): Likewise.
20347 (inline_small_functions): Likewise.
20348 * ipa-profile.c (ipa_profile): Likewise.
20349 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
20350 (ipa_make_edge_direct_to_target): Likewise.
20351 (remove_described_reference): Likewise.
20352 (ipa_impossible_devirt_target): Likewise.
20353 (propagate_controlled_uses): Likewise.
20354 (ipa_print_node_params): Likewise.
20355 (ipcp_transform_function): Likewise.
20356 * ipa-pure-const.c (pure_const_read_summary): Likewise.
20357 (propagate_pure_const): Likewise.
20358 * ipa-reference.c (generate_summary): Likewise.
20359 (read_write_all_from_decl): Likewise.
20360 (propagate): Likewise.
20361 (ipa_reference_read_optimization_summary): Likewise.
20362 * ipa-utils.c (ipa_merge_profiles): Likewise.
20363 * ipa.c (walk_polymorphic_call_targets): Likewise.
20364 (symbol_table::remove_unreachable_nodes): Likewise.
20365 (ipa_single_use): Likewise.
20366 * passes.c (execute_todo): Likewise.
20367 * predict.c (drop_profile): Likewise.
20368 * symtab.c (symtab_node::get_dump_name): New function.
20369 (symtab_node::dump_name): Likewise.
20370 (symtab_node::dump_asm_name): Likewise.
20371 (symtab_node::dump_references): Likewise.
20372 (symtab_node::dump_referring): Likewise.
20373 (symtab_node::dump_base): Likewise.
20374 (symtab_node::debug_symtab): Likewise.
20375 * tree-sra.c (convert_callers_for_node): Likewise.
20376 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
20377 * value-prof.c (init_node_map): Likewise.
20379 2017-05-23 Martin Liska <mliska@suse.cz>
20381 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
20382 and symtab_node::debug_symtab to symbol_table::debug.
20383 * cgraphunit.c (analyze_functions): Use the renamed function.
20384 (symbol_table::compile): Likewise.
20385 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
20386 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
20387 * passes.c (execute_todo): Likewise.
20388 * symtab.c (symbol_table::dump): New function.
20389 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
20391 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20393 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
20394 that nonconst implies exec.
20396 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20398 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
20399 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
20400 (inline_edge_summary_vec): Turn into ...
20401 (ipa_call_summaries): ... this one.
20402 (redirect_to_unreachable, edge_set_predicate,
20403 evaluate_properties_for_edge, inline_summary_alloc,
20404 reset_ipa_call_summary, reset_inline_summary,
20405 inline_summary_t::duplicate): Update.
20406 (inline_edge_duplication_hook): Turn to ...
20407 (ipa_call_summary_t::duplicate): ... this one.
20408 (inline_edge_removal_hook): Turn to ...
20409 (ipa_call_summary_t::remove): ... this one.
20410 (dump_inline_edge_summary): Turn to ...
20411 (dump_ipa_call_summary): ... this one.
20412 (estimate_function_body_sizes): Update.
20413 (inline_update_callee_summaries): Update.
20414 (remap_edge_change_prob): Update.
20415 (remap_edge_summaries): Update.
20416 (inline_merge_summary): Update.
20417 (do_estimate_edge_time): Update.
20418 (inline_generate_summary): Update.
20419 (inline_read_section): Update.
20420 (inline_read_summary): Update.
20421 (inline_free_summary): Update.
20422 * ipa-inline.c (can_inline_edge_p): Update.
20423 (compute_inlined_call_time): Update.
20424 (want_inline_small_function_p): Update.
20425 (edge_badness): Update.
20426 (early_inliner): Update.
20427 * ipa-inline.h (inline_edge_summary): Turn to ...
20428 (ipa_call_summary): ... this one.
20429 (ipa_call_summary_t): New class.
20430 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
20431 (ipa_call_summaries): New.
20432 (inline_edge_summary): Remove.
20433 (estimate_edge_growth): Update.
20434 * ipa-profile.c (ipa_propagate_frequency_1): Update.
20435 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20436 * ipa-split.c (execute_split_functions): Update.
20437 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
20439 2017-05-23 Tom de Vries <tom@codesourcery.com>
20441 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
20442 attributes): Document rdrand effective target.
20444 2017-05-23 Tom de Vries <tom@codesourcery.com>
20446 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
20447 attributes): Sort alphabetically.
20449 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
20451 * config/avr/genmultilib.awk: Use gsub instead of gensub.
20453 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
20456 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
20457 V2DF/V2DI splat into two separate patterns, one that handles
20458 registers, and the other that only handles memory. Drop support
20459 for splatting from a GPR on ISA 2.07 and then splitting the
20460 splat into direct move and splat.
20461 (vsx_splat_<mode>_reg): Likewise.
20462 (vsx_splat_<mode>_mem): Likewise.
20464 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
20466 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
20468 2017-05-22 Jakub Jelinek <jakub@redhat.com>
20470 PR middle-end/80809
20471 * omp-low.c (finish_taskreg_remap): New function.
20472 (finish_taskreg_scan): If unit size of ctx->record_type
20473 is non-constant, unshare the size expression and replace
20474 decls in it with possible outer var refs.
20476 PR middle-end/80809
20477 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
20478 GOVD_SHARED rather than GOVD_PRIVATE with it.
20479 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
20480 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
20482 PR middle-end/80853
20483 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
20484 as last argument to build_outer_var_ref for pointer bases of array
20485 section reductions.
20487 2017-05-19 Martin Sebor <msebor@redhat.com>
20489 * print-tree.c (print_node): Print DECL_READ_P flag.
20491 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20493 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
20494 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
20495 * cgraph.c: Likewise.
20496 * cgraphunit.c: Likewise.
20497 * gengtype.c: Likewise.
20498 * ipa-cp.c: Likewise.
20499 * ipa-devirt.c: Likewise.
20500 * ipa-icf.c: Likewise.
20501 * ipa-predicate.c: Likewise.
20502 * ipa-profile.c: Likewise.
20503 * ipa-prop.c: Likewise.
20504 * ipa-split.c: Likewise.
20506 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
20507 edge_predicate_pool, dump_inline_hints,
20508 inline_summary::account_size_time, redirect_to_unreachable,
20509 edge_set_predicate, set_hint_predicate,
20510 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
20511 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
20512 inline_summary_t::remove, remap_hint_predicate_after_duplication,
20513 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
20514 ipa_call_summary_t::remove, initialize_growth_caches,
20515 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
20516 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
20517 mark_modified, unmodified_parm_1, unmodified_parm,
20518 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
20519 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
20520 compute_bb_predicates, will_be_nonconstant_expr_predicate,
20521 will_be_nonconstant_predicate, record_modified_bb_info,
20522 get_minimal_bb, record_modified, param_change_prob,
20523 phi_result_unknown_predicate, predicate_for_phi_result,
20524 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
20525 estimate_function_body_sizes, compute_inline_parameters,
20526 compute_inline_parameters_for_curren, pass_data_inline_parameters,
20527 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
20528 inline_update_callee_summaries, remap_edge_change_prob,
20529 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
20530 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
20531 inline_analyze_function, inline_summary_t::insert,
20532 inline_generate_summary, read_ipa_call_summary, inline_read_section,
20533 inline_read_summary, write_ipa_call_summary, inline_write_summary,
20534 inline_free_summary): Move to ipa-fnsummary.h
20535 (predicate_t): Remove.
20536 * ipa-fnsummary.c: New file.
20537 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
20538 (enum inline_hints_vals, inline_hints, agg_position_info,
20539 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
20540 inline_summaries, ipa_call_summary, ipa_call_summary_t,
20541 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
20542 dump_inline_summary, dump_inline_hints, inline_generate_summary,
20543 inline_read_summary, inline_write_summary, inline_free_summary,
20544 inline_analyze_function, initialize_inline_failed,
20545 inline_merge_summary, inline_update_overall_summary,
20546 compute_inline_parameters): Move to ipa-fnsummary.h
20547 * ipa-fnsummary.h: New file.
20548 * ipa-inline-transform.h: Include ipa-inline.h.
20549 * ipa-inline.c: LIkewise.
20551 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20553 * ipa-inline.c (edge_badness): Use inlined_time instead of
20554 inline_summaries->get.
20556 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20558 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
20560 2017-05-22 Nathan Sidwell <nathan@acm.org>
20562 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
20563 (fdump-lang): Document 'raw' option.
20564 * dumpfile.h (TDI_tu): Delete.
20565 * dumpfile.c (dump_files): Remove translation-unit.
20566 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
20568 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
20570 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
20571 command option from $(AWK) call.
20572 * config/avr/genmultilib.awk: Simplify and rewrite so that it
20573 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
20574 [FORMAT]: Remove handling of variable.
20575 * config/avr/t-multilib: Regenerate.
20577 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20579 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
20581 (dump_inline_summary): Do not print self_time.
20582 (estimate_function_body_sizes): Do not set self_time.
20583 (compute_inline_parameters): Likewise.
20584 (inline_read_section, inline_write_summary): Do not stream self_time.
20585 * ipa-inline.h (inline_summary): Drop self_time.
20587 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20589 * ipa-inline-analysis.c (account_size_time): Rename to ...
20590 (inline_summary::account_size_time): ... this one.
20591 (reset_ipa_call_summary): Turn to ...
20592 (ipa_call_summary::reset): ... this one.
20593 (reset_inline_summary): Turn to ...
20594 (inline_summary::reset): ... this one.
20595 (inline_summary_t::remove): Update.
20596 (inline_summary_t::duplicate): Update.
20597 (ipa_call_summary_t::remove): Update.
20598 (dump_inline_summary): Update.
20599 (estimate_function_body_sizes): Update.
20600 (compute_inline_parameters): Update.
20601 (estimate_node_size_and_time): Update.
20602 (inline_merge_summary): Update.
20603 (inline_update_overall_summary): Update.
20604 (inline_read_section): Update.
20605 (inline_write_summary): Update.
20606 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
20607 add account_size_time and reset member functions.
20608 (ipa_call_summary): Add reset function.
20609 * ipa-predicate.h (predicate::operator &): Constify.
20611 2017-05-22 Richard Biener <rguenther@suse.de>
20613 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
20615 2017-05-19 Jason Merrill <jason@redhat.com>
20617 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
20619 2017-05-19 Marek Polacek <polacek@redhat.com>
20622 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
20623 TYPE_OVERFLOW_WRAPS checks.
20625 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
20627 * tree-core.h (enum omp_clause_default_kind): Add
20628 "OMP_CLAUSE_DEFAULT_PRESENT".
20629 * tree-pretty-print.c (dump_omp_clause): Handle it.
20630 * gimplify.c (enum gimplify_omp_var_data): Add
20631 "GOVD_MAP_FORCE_PRESENT".
20632 (gimplify_adjust_omp_clauses_1): Map it to
20633 "GOMP_MAP_FORCE_PRESENT".
20634 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
20636 * gimplify.c (oacc_default_clause): Clarify.
20638 2017-05-19 Nathan Sidwell <nathan@acm.org>
20640 LANG_HOOK_REGISTER_DUMPS
20641 * toplev.c (general_init): Call register dump lang hook.
20642 * doc/invoke.texi: Document -fdump-lang option family.
20643 * dumpfile.c (dump_files): Remove class dump here.
20644 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
20645 * dumpfile.h (tree_dump_index): Remove TDI_class.
20646 * langhooks-def.h (lhd_register_dumps): Declare.
20647 (LANG_HOOKS_REGISTER_DUMPS): Define.
20648 (LANG_HOOKS_INITIALIZER): Add it.
20649 * langhooks.c (lhd_register_dumps): Define.
20650 * langhooks.h (struct lang_hooks): Add register_dumps.
20652 2017-05-19 Nathan Sidwell <nathan@acm.org>
20654 * context.h (context::set_passes): New.
20655 * context.c (context::context): Do not create pass manager.
20656 * toplev.c (general_init): Create pass manager here.
20658 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
20660 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
20661 use this splitter if two add or or instructions would also work for
20662 the constant we want to generate.
20664 2017-05-19 Richard Biener <rguenther@suse.de>
20667 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
20668 predicate evaluation.
20670 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
20672 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
20674 * ipa-inline.c (want_inline_small_function_p): Do not cast to
20677 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
20679 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
20680 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
20681 (inline_edge_summary_vec): Turn into ...
20682 (ipa_call_summaries): ... this one.
20683 (redirect_to_unreachable, edge_set_predicate,
20684 evaluate_properties_for_edge, inline_summary_alloc,
20685 reset_ipa_call_summary, reset_inline_summary,
20686 inline_summary_t::duplicate): Update.
20687 (inline_edge_duplication_hook): Turn to ...
20688 (ipa_call_summary_t::duplicate): ... this one.
20689 (inline_edge_removal_hook): Turn to ...
20690 (ipa_call_summary_t::remove): ... this one.
20691 (dump_inline_edge_summary): Turn to ...
20692 (dump_ipa_call_summary): ... this one.
20693 (estimate_function_body_sizes): Update.
20694 (inline_update_callee_summaries): Update.
20695 (remap_edge_change_prob): Update.
20696 (remap_edge_summaries): Update.
20697 (inline_merge_summary): Update.
20698 (do_estimate_edge_time): Update.
20699 (inline_generate_summary): Update.
20700 (inline_read_section): Update.
20701 (inline_read_summary): Update.
20702 (inline_free_summary): Update.
20703 * ipa-inline.c (can_inline_edge_p): Update.
20704 (compute_inlined_call_time): Update.
20705 (want_inline_small_function_p): Update.
20706 (edge_badness): Update.
20707 (early_inliner): Update.
20708 * ipa-inline.h (inline_edge_summary): Turn to ...
20709 (ipa_call_summary): ... this one.
20710 (ipa_call_summary_t): New class.
20711 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
20712 (ipa_call_summaries): New.
20713 (inline_edge_summary): Remove.
20714 (estimate_edge_growth): Update.
20715 * ipa-profile.c (ipa_propagate_frequency_1): Update.
20716 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20717 * ipa-split.c (execute_split_functions): Update.
20718 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
20720 2017-05-19 Richard Biener <rguenther@suse.de>
20722 PR middle-end/80764
20723 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
20725 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
20727 * config/rs6000/rs6000.c (struct machine_function): Add field
20728 fpr_is_wrapped_separately.
20729 (rs6000_get_separate_components): Use 64 components. Handle the
20730 new FPR components.
20731 (rs6000_components_for_bb): Handle the FPR components.
20732 (rs6000_emit_prologue_components): Handle the FPR components.
20733 (rs6000_emit_epilogue_components): Handle the FPR components.
20734 (rs6000_set_handled_components): Handle the FPR components.
20735 (rs6000_emit_prologue): Don't output prologue code for those FPRs
20736 that are already separately shrink-wrapped.
20737 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
20738 that are already separately shrink-wrapped.
20740 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
20743 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
20746 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
20747 (define_peephole2 for Altivec d-form load): Add peepholes to catch
20748 cases where the register allocator uses a move and an offsettable
20749 memory operation to/from a FPR register on ISA 2.06/2.07.
20750 (define_peephole2 for Altivec d-form store): Likewise.
20752 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
20755 * config/i386/mmx.md (*mov<mode>_internal): Enable
20756 alternatives 11, 12, 13 and 14 also for 32bit targets.
20757 Remove alternatives 15, 16, 17 and 18.
20758 * config/i386/sse.md (vec_concatv2di): Change
20759 alternative (!x, *y) to (x, ?!*Yn).
20761 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
20763 * dumpfile.h (enum dump_kind): Remove stray comma.
20765 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
20767 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
20768 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
20769 predicate::num_conditions
20770 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
20771 (CHANGED): turn into predicate::changed.
20772 (agg_position_info): Move to ipa-predicate.h
20773 (add_condition, predicate::add_clause, predicate::operator &=,
20774 predicate::or_with, predicate::evaluate, predicate::probability,
20775 dump_condition, dump_clause, predicate::dump,
20776 predicate::remap_after_duplication, predicate::remap_after_inlining,
20777 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
20778 (evaluate_conditions_for_known_args): Update.
20779 (set_cond_stmt_execution_predicate): Update.
20780 * ipa-inline.h: Include ipa-predicate.h
20781 (condition, inline_param_summary, conditions, agg_position_info,
20782 predicate): Move to ipa-predicate.h
20783 * ipa-predicate.c: New file.
20784 * ipa-predicate.h: New file.
20786 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
20788 * final.c (leaf_function_p): Check we are not in a sequence.
20790 2017-05-18 Martin Liska <mliska@suse.cz>
20792 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
20793 * dumpfile.c (dump_register): Use new enum dump_kind.
20794 (get_dump_file_name): Likewise.
20795 (dump_enable_all): Likewise.
20796 (dump_switch_p_1): Likewise.
20797 (enable_rtl_dump_file): Remove usage of TDF_RTL.
20798 * dumpfile.h (enum dump_kind): New enum type.
20799 (struct dump_file_info): Create constructor and
20800 format fields and comments.
20801 * passes.c (pass_manager::register_one_dump_file):
20803 * statistics.c (statistics_early_init): Likewise.
20804 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
20805 TDF_TREE with TDF_SLIM.
20806 (gather_memory_references_ref): Likewise.
20808 2017-05-18 Martin Liska <mliska@suse.cz>
20810 * vec.h (struct vnull): Use it.
20812 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
20814 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
20815 (true_predicate, false_predicate, true_predicate_p,
20816 false_predicate_p): Remove.
20817 (single_cond_predicate, not_inlined_predicate): Turn to member function
20819 (add_condition): Update.
20820 (add_clause): Turn to...
20821 (predicate::add_clause): ... this one; update; allow passing NULL
20823 (and_predicates): Turn to ...
20824 (predicate::operator &=): ... this one.
20825 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
20826 (or_predicates): Turn to ...
20827 (predicate::or_with): ... this one.
20828 (evaluate_predicate): Turn to ...
20829 (predicate::evaluate): ... this one.
20830 (predicate_probability): Turn to ...
20831 (predicate::probability): ... this one.
20832 (dump_condition): Update.
20833 (dump_predicate): Turn to ...
20834 (predicate::dump): ... this one.
20835 (account_size_time): Update.
20836 (edge_set_predicate): Update.
20837 (set_hint_predicate): UPdate.
20838 (evaluate_conditions_for_known_args): Update.
20839 (evaluate_properties_for_edge): Update.
20840 (remap_predicate_after_duplication): Turn to...
20841 (predicate::remap_after_duplication): ... this one.
20842 (remap_hint_predicate_after_duplication): Update.
20843 (inline_summary_t::duplicate): UPdate.
20844 (dump_inline_edge_summary): Update.
20845 (dump_inline_summary): Update.
20846 (set_cond_stmt_execution_predicate): Update.
20847 (set_switch_stmt_execution_predicate): Update.
20848 (compute_bb_predicates): Update.
20849 (will_be_nonconstant_expr_predicate): Update.
20850 (will_be_nonconstant_predicate): Update.
20851 (phi_result_unknown_predicate): Update.
20852 (predicate_for_phi_result): Update.
20853 (array_index_predicate): Update.
20854 (estimate_function_body_sizes): Update.
20855 (estimate_node_size_and_time): Update.
20856 (estimate_ipcp_clone_size_and_time): Update.
20857 (remap_predicate): Rename to ...
20858 (predicate::remap_after_inlining): ... this one.
20859 (remap_hint_predicate): Update.
20860 (inline_merge_summary): Update.
20861 (inline_update_overall_summary): Update.
20862 (estimate_size_after_inlining): Update.
20863 (read_predicate): Rename to ...
20864 (predicate::stream_in): ... this one.
20865 (read_inline_edge_summary): Update.
20866 (write_predicate): Rename to ...
20867 (predicate::stream_out): ... this one.
20868 (write_inline_edge_summary): Update.
20869 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
20870 (clause_t): Turn to uint32_t
20871 (predicate): Turn to class; implement constructor and operators
20873 (size_time_entry): Update.
20874 (inline_summary): Update.
20875 (inline_edge_summary): Update.
20877 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
20879 * fold-const.c (fold_binary_loc): Move transformation...
20880 * match.pd (C - X CMP X): ... here.
20882 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
20884 * config/sparc/sparc.c (sparc_option_override): Set function
20885 alignment for -mcpu=niagara7 to 64 to match the I$ line.
20886 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
20888 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
20890 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
20892 2017-05-18 Marek Polacek <polacek@redhat.com>
20895 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
20896 (pass_ubsan::execute): Call gimple_assign_single_p instead of
20897 gimple_assign_load_p.
20899 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
20901 PR middle-end/80692
20902 * real.c (do_compare): Give decimal_do_compare preference over
20903 comparing just the signs.
20905 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
20907 * doc/md.texi (Canonicalization of Instructions): Describe the
20908 canonical form of instructions that inherently set a condition
20911 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
20913 PR middle-end/80775
20914 * tree-cfg.c: Move deletion of unreachable case statements to after
20915 the merging of consecutive case labels.
20917 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
20919 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
20920 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
20921 restoring of callee-saved registers.
20923 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
20925 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
20926 * config/visium/visium.c (single_set_and_flags): Likewise.
20927 * config/visium/visium.md (Substitutions): Likewise.
20929 2017-05-17 Martin Liska <mliska@suse.cz>
20931 * cfg.c: Introduce dump_flags_t type and
20932 use it instead of int type.
20934 * cfghooks.c: Likewise.
20935 * cfghooks.h (struct cfg_hooks): Likewise.
20936 * cfgrtl.c: Likewise.
20937 * cfgrtl.h: Likewise.
20938 * cgraph.c (cgraph_node::get_body): Likewise.
20939 * coretypes.h: Likewise.
20940 * domwalk.c: Likewise.
20941 * domwalk.h: Likewise.
20942 * dumpfile.c (struct dump_option_value_info): Likewise.
20943 (dump_enable_all): Likewise.
20944 (dump_switch_p_1): Likewise.
20945 (opt_info_switch_p): Likewise.
20946 * dumpfile.h (enum tree_dump_index): Likewise.
20947 (struct dump_file_info): Likewise.
20948 * genemit.c: Likewise.
20949 * generic-match-head.c: Likewise.
20950 * gengtype.c (open_base_files): Likewise.
20951 * gimple-pretty-print.c: Likewise.
20952 * gimple-pretty-print.h: Likewise.
20953 * graph.c (print_graph_cfg): Likewise.
20954 * graphite-scop-detection.c (dot_all_sese): Likewise.
20955 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
20956 * loop-unroll.c (report_unroll): Likewise.
20957 * passes.c (pass_manager::register_one_dump_file): Likewise.
20958 * print-tree.c: Likewise.
20959 * statistics.c: Likewise.
20960 * tree-cfg.c: Likewise.
20961 * tree-cfg.h: Likewise.
20962 * tree-dfa.c: Likewise.
20963 * tree-dfa.h: Likewise.
20964 * tree-dump.c (dump_function): Likewise.
20965 * tree-dump.h (struct dump_info): Likewise.
20966 * tree-pretty-print.c: Likewise.
20967 * tree-pretty-print.h: Likewise.
20968 * tree-ssa-live.c: Likewise.
20969 * tree-ssa-live.h: Likewise.
20970 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
20971 * tree-vect-loop.c: Likewise.
20972 * tree-vect-slp.c: Likewise.
20974 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
20975 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20977 PR tree-optimization/80457
20978 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
20979 of all arguments to a statement as scalar_to_vec operations.
20980 (vectorizable_call): Adjust call to vect_model_simple_cost for
20982 (vectorizable_conversion): Likewise.
20983 (vectorizable_assignment): Likewise.
20984 (vectorizable_shift): Likewise.
20985 (vectorizable_operation): Likewise.
20986 (vectorizable_comparison): Likewise.
20987 (vect_is_simple_cond): Record the def types for operands.
20988 (vectorizable_condition): Likewise, call vect_model_simple_cost.
20989 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
20990 for statement argument count.
20992 2017-05-16 Carl Love <cel@us.ibm.com>
20994 * config/rs6000/rs6000-c: Add support for built-in functions
20995 vector unsigned long long vec_bperm (vector unsigned long long,
20996 vector unsigned char)
20997 vector signed long long vec_mule (vector signed int,
20999 vector unsigned long long vec_mule (vector unsigned int,
21000 vector unsigned int)
21001 vector signed long long vec_mulo (vector signed int,
21003 vector unsigned long long vec_mulo (vector unsigned int,
21004 vector unsigned int)
21005 vector signed char vec_sldw (vector signed char,
21006 vector signed char,
21008 vector unsigned char vec_sldw (vector unsigned char,
21009 vector unsigned char,
21011 vector signed short vec_sldw (vector signed short,
21012 vector signed short,
21014 vector unsigned short vec_sldw (vector unsigned short,
21015 vector unsigned short,
21017 vector signed int vec_sldw (vector signed int,
21020 vector unsigned int vec_sldw (vector unsigned int,
21021 vector unsigned int,
21023 vector signed long long vec_sldw (vector signed long long,
21024 vector signed long long,
21026 vector unsigned long long vec_sldw (vector unsigned long long,
21027 vector unsigned long long,
21029 * config/rs6000/rs6000-c: Add support for built-in functions
21030 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
21031 * config/rs6000/altivec.h: Add defintion for vec_sldw.
21032 * doc/extend.texi: Update the built-in documentation for the
21033 new built-in functions.
21035 2017-05-16 Marek Polacek <polacek@redhat.com>
21039 * tree.c (save_expr): Don't fold the expression.
21041 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
21043 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
21044 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
21045 and (?*y,m). Update insn attributes.
21047 2017-05-16 Martin Liska <mliska@suse.cz>
21049 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
21050 flags argument of print_gimple_stmt, print_gimple_expr,
21051 print_generic_stmt and print_generic_expr.
21052 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
21053 * coretypes.h: Likewise.
21054 * except.c (dump_eh_tree): Likewise.
21055 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
21056 * gimple-pretty-print.h: Likewise.
21057 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
21058 (backprop::push_to_worklist): Likewise.
21059 (backprop::pop_from_worklist): Likewise.
21060 (backprop::process_use): Likewise.
21061 (backprop::intersect_uses): Likewise.
21062 (note_replacement): Likewise.
21063 * gimple-ssa-store-merging.c
21064 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
21065 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
21066 (pass_store_merging::execute): Likewise.
21067 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
21068 (ssa_base_cand_dump_callback): Likewise.
21069 (dump_incr_vec): Likewise.
21070 (replace_refs): Likewise.
21071 (replace_mult_candidate): Likewise.
21072 (create_add_on_incoming_edge): Likewise.
21073 (create_phi_basis): Likewise.
21074 (insert_initializers): Likewise.
21075 (all_phi_incrs_profitable): Likewise.
21076 (introduce_cast_before_cand): Likewise.
21077 (replace_one_candidate): Likewise.
21078 * gimplify.c (gimplify_expr): Likewise.
21079 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
21080 (set_rename): Likewise.
21081 (rename_uses): Likewise.
21082 (copy_loop_phi_nodes): Likewise.
21083 (add_close_phis_to_merge_points): Likewise.
21084 (copy_loop_close_phi_args): Likewise.
21085 (copy_cond_phi_args): Likewise.
21086 (graphite_copy_stmts_from_block): Likewise.
21087 (translate_pending_phi_nodes): Likewise.
21088 * graphite-poly.c (print_pdr): Likewise.
21089 (dump_gbb_cases): Likewise.
21090 (dump_gbb_conditions): Likewise.
21091 (print_scop_params): Likewise.
21092 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
21093 (build_cross_bb_scalars_use): Likewise.
21094 (gather_bbs::before_dom_children): Likewise.
21095 * hsa-dump.c (dump_hsa_immed): Likewise.
21096 * ipa-cp.c (print_ipcp_constant_value): Likewise.
21097 (get_replacement_map): Likewise.
21098 * ipa-inline-analysis.c (dump_condition): Likewise.
21099 (estimate_function_body_sizes): Likewise.
21100 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
21101 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
21102 * ipa-prop.c (ipa_dump_param): Likewise.
21103 (ipa_print_node_jump_functions_for_edge): Likewise.
21104 (ipa_modify_call_arguments): Likewise.
21105 (ipa_modify_expr): Likewise.
21106 (ipa_dump_param_adjustments): Likewise.
21107 (ipa_dump_agg_replacement_values): Likewise.
21108 (ipcp_modif_dom_walker::before_dom_children): Likewise.
21109 * ipa-pure-const.c (check_stmt): Likewise.
21110 (pass_nothrow::execute): Likewise.
21111 * ipa-split.c (execute_split_functions): Likewise.
21112 * omp-offload.c (dump_oacc_loop_part): Likewise.
21113 (dump_oacc_loop): Likewise.
21114 * trans-mem.c (tm_log_emit): Likewise.
21115 (tm_memopt_accumulate_memops): Likewise.
21116 (dump_tm_memopt_set): Likewise.
21117 (dump_tm_memopt_transform): Likewise.
21118 * tree-cfg.c (gimple_verify_flow_info): Likewise.
21119 (print_loop): Likewise.
21120 * tree-chkp-opt.c (chkp_print_addr): Likewise.
21121 (chkp_gather_checks_info): Likewise.
21122 (chkp_get_check_result): Likewise.
21123 (chkp_remove_check_if_pass): Likewise.
21124 (chkp_use_outer_bounds_if_possible): Likewise.
21125 (chkp_reduce_bounds_lifetime): Likewise.
21126 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
21127 (chkp_mark_completed_bounds): Likewise.
21128 (chkp_register_incomplete_bounds): Likewise.
21129 (chkp_mark_invalid_bounds): Likewise.
21130 (chkp_maybe_copy_and_register_bounds): Likewise.
21131 (chkp_build_returned_bound): Likewise.
21132 (chkp_get_bound_for_parm): Likewise.
21133 (chkp_build_bndldx): Likewise.
21134 (chkp_get_bounds_by_definition): Likewise.
21135 (chkp_generate_extern_var_bounds): Likewise.
21136 (chkp_get_bounds_for_decl_addr): Likewise.
21137 * tree-chrec.c (chrec_apply): Likewise.
21138 * tree-data-ref.c (dump_data_reference): Likewise.
21139 (dump_subscript): Likewise.
21140 (dump_data_dependence_relation): Likewise.
21141 (analyze_overlapping_iterations): Likewise.
21142 * tree-inline.c (expand_call_inline): Likewise.
21143 (tree_function_versioning): Likewise.
21144 * tree-into-ssa.c (dump_defs_stack): Likewise.
21145 (dump_currdefs): Likewise.
21146 (dump_names_replaced_by): Likewise.
21147 (dump_update_ssa): Likewise.
21148 (update_ssa): Likewise.
21149 * tree-object-size.c (pass_object_sizes::execute): Likewise.
21150 * tree-parloops.c (build_new_reduction): Likewise.
21151 (try_create_reduction_list): Likewise.
21152 (ref_conflicts_with_region): Likewise.
21153 (oacc_entry_exit_ok_1): Likewise.
21154 (oacc_entry_exit_single_gang): Likewise.
21155 * tree-pretty-print.h: Likewise.
21156 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
21157 (get_scalar_evolution): Likewise.
21158 (add_to_evolution): Likewise.
21159 (get_loop_exit_condition): Likewise.
21160 (analyze_evolution_in_loop): Likewise.
21161 (analyze_initial_condition): Likewise.
21162 (analyze_scalar_evolution): Likewise.
21163 (instantiate_scev): Likewise.
21164 (number_of_latch_executions): Likewise.
21165 (gather_chrec_stats): Likewise.
21166 (final_value_replacement_loop): Likewise.
21167 (scev_const_prop): Likewise.
21168 * tree-sra.c (dump_access): Likewise.
21169 (disqualify_candidate): Likewise.
21170 (create_access): Likewise.
21171 (reject): Likewise.
21172 (maybe_add_sra_candidate): Likewise.
21173 (create_access_replacement): Likewise.
21174 (analyze_access_subtree): Likewise.
21175 (analyze_all_variable_accesses): Likewise.
21176 (sra_modify_assign): Likewise.
21177 (initialize_constant_pool_replacements): Likewise.
21178 (find_param_candidates): Likewise.
21179 (decide_one_param_reduction): Likewise.
21180 (replace_removed_params_ssa_names): Likewise.
21181 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
21182 * tree-ssa-copy.c (dump_copy_of): Likewise.
21183 (copy_prop_visit_cond_stmt): Likewise.
21184 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
21185 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
21186 (record_equivalences_from_stmt): Likewise.
21187 * tree-ssa-dse.c (compute_trims): Likewise.
21188 (delete_dead_call): Likewise.
21189 (delete_dead_assignment): Likewise.
21190 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
21191 (forward_propagate_into_cond): Likewise.
21192 (pass_forwprop::execute): Likewise.
21193 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
21194 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
21196 (move_computations_worker): Likewise.
21197 (execute_sm): Likewise.
21198 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
21199 (remove_exits_and_undefined_stmts): Likewise.
21200 (remove_redundant_iv_tests): Likewise.
21201 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
21202 (adjust_iv_update_pos): Likewise.
21203 * tree-ssa-math-opts.c (bswap_replace): Likewise.
21204 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
21205 (value_replacement): Likewise.
21206 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
21207 * tree-ssa-pre.c (print_pre_expr): Likewise.
21208 (get_representative_for): Likewise.
21209 (create_expression_by_pieces): Likewise.
21210 (insert_into_preds_of_block): Likewise.
21211 (eliminate_insert): Likewise.
21212 (eliminate_dom_walker::before_dom_children): Likewise.
21213 (eliminate): Likewise.
21214 (remove_dead_inserted_code): Likewise.
21215 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
21216 * tree-ssa-reassoc.c (get_rank): Likewise.
21217 (eliminate_duplicate_pair): Likewise.
21218 (eliminate_plus_minus_pair): Likewise.
21219 (eliminate_not_pairs): Likewise.
21220 (undistribute_ops_list): Likewise.
21221 (eliminate_redundant_comparison): Likewise.
21222 (update_range_test): Likewise.
21223 (optimize_range_tests_var_bound): Likewise.
21224 (optimize_vec_cond_expr): Likewise.
21225 (rewrite_expr_tree): Likewise.
21226 (rewrite_expr_tree_parallel): Likewise.
21227 (linearize_expr): Likewise.
21228 (break_up_subtract): Likewise.
21229 (linearize_expr_tree): Likewise.
21230 (attempt_builtin_powi): Likewise.
21231 (attempt_builtin_copysign): Likewise.
21232 (transform_stmt_to_copy): Likewise.
21233 (transform_stmt_to_multiply): Likewise.
21234 (dump_ops_vector): Likewise.
21235 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
21236 (print_scc): Likewise.
21237 (set_ssa_val_to): Likewise.
21238 (visit_reference_op_store): Likewise.
21239 (visit_use): Likewise.
21240 (sccvn_dom_walker::before_dom_children): Likewise.
21241 (run_scc_vn): Likewise.
21242 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
21244 (expr_hash_elt::print): Likewise.
21245 (const_and_copies::pop_to_marker): Likewise.
21246 (const_and_copies::record_const_or_copy_raw): Likewise.
21247 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
21248 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
21249 (dump_predicates): Likewise.
21250 (find_uninit_use): Likewise.
21251 (warn_uninitialized_phi): Likewise.
21252 (pass_late_warn_uninitialized::execute): Likewise.
21253 * tree-ssa.c (verify_vssa): Likewise.
21254 (verify_ssa): Likewise.
21255 (maybe_optimize_var): Likewise.
21256 * tree-vrp.c (dump_value_range): Likewise.
21257 (dump_all_value_ranges): Likewise.
21258 (dump_asserts_for): Likewise.
21259 (register_edge_assert_for_2): Likewise.
21260 (vrp_visit_cond_stmt): Likewise.
21261 (vrp_visit_switch_stmt): Likewise.
21262 (vrp_visit_stmt): Likewise.
21263 (vrp_visit_phi_node): Likewise.
21264 (simplify_cond_using_ranges_1): Likewise.
21265 (fold_predicate_in): Likewise.
21266 (evrp_dom_walker::before_dom_children): Likewise.
21267 (evrp_dom_walker::push_value_range): Likewise.
21268 (evrp_dom_walker::pop_value_range): Likewise.
21269 (execute_early_vrp): Likewise.
21271 2017-05-16 Richard Biener <rguenther@suse.de>
21273 * dwarf2out.c (loc_list_from_tree_1): Do not create
21274 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
21276 2017-05-16 Richard Biener <rguenther@suse.de>
21278 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
21280 (note_variable_value_in_expr): If we resolved the decl ref
21281 do not push to the stack.
21283 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
21285 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
21286 operations in fast-math mode.
21287 (vaddq_f16): Likewise.
21288 (vmul_f16): Likewise.
21289 (vmulq_f16): Likewise.
21290 (vsub_f16): Likewise.
21291 (vsubq_f16): Likewise.
21292 * config/arm/neon.md (add<mode>3): New.
21294 (fma:<VH:mode>3): New. Also remove outdated comment.
21297 2017-05-16 Martin Liska <mliska@suse.cz>
21301 * ipa-devirt.c (warn_types_mismatch): Fix typo.
21302 (odr_types_equivalent_p): Likewise.
21304 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
21306 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
21308 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
21311 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
21312 non-interunit SSE move alternatives with '?'.
21313 (zero-extendsidi peephole2): New peephole to skip intermediate
21314 general register in SSE zero-extend sequence.
21316 2017-05-15 Jeff Law <law@redhat.com>
21318 * reorg.c (relax_delay_slots): Create a new variable to hold
21319 the temporary target rather than clobbering TARGET_LABEL.
21321 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
21322 missing argument to extract_bit_field call.
21323 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
21325 2017-05-15 Martin Liska <mliska@suse.cz>
21328 * gcc.c (process_command): Do not allow empty argument of -o option.
21330 2017-05-15 Renlin Li <renlin.li@arm.com>
21332 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
21333 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
21334 * config/aarch64/constraints.md (Usf): Add long call check.
21335 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
21336 (call_value): Likewise.
21337 (sibcall): Likewise.
21338 (sibcall_value): Likewise.
21340 (call_value_insn): New.
21341 (sibcall_insn): Update rtx pattern.
21342 (sibcall_value_insn): Likewise.
21343 (call_internal): Remove.
21344 (call_value_internal): Likewise.
21345 (sibcall_internal): Likewise.
21346 (sibcall_value_internal): Likewise.
21347 (call_reg): Likewise.
21348 (call_symbol): Likewise.
21349 (call_value_reg): Likewise.
21350 (call_value_symbol): Likewise.
21352 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
21355 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
21357 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
21359 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
21360 compatible with CCGOCmode and with CCZmode.
21362 2017-05-14 Martin Sebor <msebor@redhat.com>
21364 PR middle-end/77671
21365 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
21366 (gimple_fold_builtin_snprintf): Same.
21367 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
21368 (gimple_fold_builtin_snprintf): Same.
21369 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
21370 of character types.
21371 (is_call_safe): New function.
21372 (try_substitute_return_value): Call it.
21373 (try_simplify_call): New function.
21374 (pass_sprintf_length::handle_gimple_call): Call it.
21376 2017-05-14 Martin Sebor <msebor@redhat.com>
21378 PR middle-end/80669
21379 * builtins.c (expand_builtin_stpncpy): Simplify.
21381 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
21383 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
21384 * config/i386/i386.h
21385 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
21386 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
21387 (struct machine_function): Add new members call_ms2sysv,
21388 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
21389 (struct machine_frame_state): New fields sp_realigned and
21390 sp_realigned_offset.
21391 * config/i386/i386.c
21392 (enum xlogue_stub): New enum.
21393 (enum xlogue_stub_sets): New enum.
21394 (class xlogue_layout): New class.
21395 (struct ix86_frame): New fields stack_realign_allocate_offset,
21396 stack_realign_offset and outlined_save_offset. Modify comments to
21397 detail stack layout when using out-of-line stubs.
21398 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
21399 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
21400 -mcall-ms2sysv-xlogues.
21401 (stub_managed_regs): New static variable.
21402 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
21403 registers managed by out-of-line stub.
21404 (disable_call_ms2sysv_xlogues): New function.
21405 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
21406 m->call_ms2sysv when appropriate and compute frame layout for
21408 (sp_valid_at, fp_valid_at): New inline functions.
21409 (choose_basereg): New function.
21410 (choose_baseaddr): Add align parameter, use choose_basereg and modify
21412 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
21413 Use align parameter of choose_baseaddr to generated aligned SSE movs
21415 (pro_epilogue_adjust_stack): Modify to track
21416 machine_frame_state::sp_realigned.
21417 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
21418 (ix86_nsaved_sseregs): Likewise.
21419 (ix86_emit_save_regs): Likewise.
21420 (ix86_emit_save_regs_using_mov): Likewise.
21421 (ix86_emit_save_sse_regs_using_mov): Likewise.
21422 (get_scratch_register_on_entry): Likewise.
21423 (gen_frame_set): New function.
21424 (gen_frame_load): Likewise.
21425 (gen_frame_store): Likewise.
21426 (emit_outlined_ms2sysv_save): Likewise.
21427 (emit_outlined_ms2sysv_restore): Likewise.
21428 (ix86_expand_prologue): Modify stack re-alignment code and call
21429 emit_outlined_ms2sysv_save when appropriate.
21430 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
21431 parameter rtx_insn *insn, which allows the function to be used to only
21432 generate the notes.
21433 (ix86_expand_epilogue): Modify validity checks of frame and stack
21434 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
21435 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
21436 * config/i386/predicates.md
21437 (save_multiple): New predicate.
21438 (restore_multiple): Likewise.
21439 * config/i386/sse.md
21440 (save_multiple<mode>): New pattern.
21441 (save_multiple_realign<mode>): Likewise.
21442 (restore_multiple<mode>): Likewise.
21443 (restore_multiple_and_return<mode>): Likewise.
21444 (restore_multiple_leave_return<mode>): Likewise.
21445 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
21447 2017-05-14 Julia Koval <julia.koval@intel.com>
21449 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
21450 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
21451 (__builtin_ia32_xsetbv): New builtins.
21452 * config/i386/i386.c (ix86_expand_special_args_builtin):
21454 (ix86_expand_builtin): Special expand for new intrinsics.
21455 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
21456 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
21457 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
21459 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21461 * cfganal.c (inverted_post_order_compute): Change argument type
21463 * cfganal.h (inverted_post_order_compute): Adjust prototype.
21464 * df-core.c (rest_of_handle_df_initialize): Adjust.
21465 (rest_of_handle_df_finish): Likewise.
21466 (df_analyze_1): Likewise.
21467 (df_analyze): Likewise.
21468 (loop_inverted_post_order_compute): Change argument to be a vec *.
21469 (df_analyze_loop): Adjust.
21470 (df_get_n_blocks): Likewise.
21471 (df_get_postorder): Likewise.
21472 * df.h (struct df_d): Change field to be a vec.
21473 * lcm.c (compute_laterin): Adjust.
21474 (compute_available): Likewise.
21475 * lra-lives.c (lra_create_live_ranges_1): Likewise.
21476 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
21477 * tree-ssa-pre.c (compute_antic): Likewise.
21479 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21481 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
21482 (depth_first_search::depth_first_search): Change structure init
21483 function to this constructor.
21484 (depth_first_search::add_bb): Rename function to this member.
21485 (depth_first_search::execute): Likewise.
21486 (flow_dfs_compute_reverse_finish): Adjust.
21488 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21490 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
21491 (longest_simple_path): Likewise.
21492 * shrink-wrap.c (spread_components): Likewise.
21493 (disqualify_problematic_components): Likewise.
21494 (emit_common_heads_for_components): Likewise.
21495 (emit_common_tails_for_components): Likewise.
21496 (insert_prologue_epilogue_for_components): Likewise.
21498 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21500 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
21503 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21505 * df-core.c (df_set_blocks): Start using auto_bitmap.
21506 (df_compact_blocks): Likewise.
21507 * df-problems.c (df_rd_confluence_n): Likewise.
21508 * df-scan.c (df_insn_rescan_all): Likewise.
21509 (df_process_deferred_rescans): Likewise.
21510 (df_update_entry_block_defs): Likewise.
21511 (df_update_exit_block_uses): Likewise.
21512 (df_entry_block_bitmap_verify): Likewise.
21513 (df_exit_block_bitmap_verify): Likewise.
21514 (df_scan_verify): Likewise.
21515 * lra-constraints.c (lra_constraints): Likewise.
21516 (undo_optional_reloads): Likewise.
21517 (lra_undo_inheritance): Likewise.
21518 * lra-remat.c (calculate_gen_cands): Likewise.
21519 (do_remat): Likewise.
21520 * lra-spills.c (assign_spill_hard_regs): Likewise.
21521 (spill_pseudos): Likewise.
21522 * tree-ssa-pre.c (bitmap_set_and): Likewise.
21523 (bitmap_set_subtract_values): Likewise.
21525 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21527 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
21528 management with auto_bitmap.
21529 (fix_inter_tick): Likewise.
21530 (fix_recovery_deps): Likewise.
21531 * ira.c (add_store_equivs): Likewise.
21532 (find_moveable_pseudos): Likewise.
21533 (split_live_ranges_for_shrink_wrap): Likewise.
21534 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
21535 (rtx_reuse_manager::seen_def_p): Likewise.
21536 (rtx_reuse_manager::set_seen_def): Likewise.
21537 * print-rtl.h (class rtx_reuse_manager): Likewise.
21539 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21541 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
21543 (migrate_btr_def): Likewise.
21544 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
21545 * df-core.c (loop_post_order_compute): Likewise.
21546 (loop_inverted_post_order_compute): Likewise.
21547 * hsa-common.h: Likewise.
21548 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
21549 * init-regs.c (initialize_uninitialized_regs): Likewise.
21550 * ipa-inline.c (resolve_noninline_speculation): Likewise.
21551 (inline_small_functions): Likewise.
21552 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
21553 * ira.c (combine_and_move_insns): Likewise.
21554 (build_insn_chain): Likewise.
21555 * loop-invariant.c (find_invariants): Likewise.
21556 * lower-subreg.c (propagate_pseudo_copies): Likewise.
21557 * predict.c (tree_predict_by_opcode): Likewise.
21558 (predict_paths_leading_to): Likewise.
21559 (predict_paths_leading_to_edge): Likewise.
21560 (estimate_loops_at_level): Likewise.
21561 (estimate_loops): Likewise.
21562 * shrink-wrap.c (try_shrink_wrapping): Likewise.
21563 (spread_components): Likewise.
21564 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
21565 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
21566 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
21567 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
21568 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
21569 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
21570 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
21571 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
21572 (mark_threaded_blocks): Likewise.
21573 (thread_through_all_blocks): Likewise.
21574 * tree-ssa.c (verify_ssa): Likewise.
21575 (execute_update_addresses_taken): Likewise.
21576 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
21578 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21580 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
21582 (post_order_compute): Likewise.
21583 (inverted_post_order_compute): Likewise.
21584 (pre_and_rev_post_order_compute_fn): Likewise.
21586 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21588 * genrecog.c (int_set::int_set): Explicitly construct our
21589 auto_vec base class.
21590 * vec.h (auto_vec::auto_vec): New constructor.
21592 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21594 * bitmap.h (class auto_bitmap): New constructor taking
21595 bitmap_obstack * argument.
21597 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21599 * bitmap.h (class auto_bitmap): Change type of m_bits to
21600 bitmap_head, and adjust ctor / dtor and member operators.
21602 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
21604 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
21605 when returned register mode doesn't match original mode.
21607 2017-05-12 Jeff Law <law@redhat.com>
21608 Jakub Jelinek <jakub@redhat.com>
21610 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
21611 we look for cc setter after the compare-elim changes.
21612 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
21613 within the vector to match what compare-elim now expects.
21614 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
21615 (xorsi3_flags, one_cmplsi2_flags): Likewise.
21617 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
21618 after the compare-elim changes.
21619 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
21620 the vector to match what compare-elim now expects.
21621 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
21622 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
21623 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
21624 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
21625 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
21627 * config/visium/visium.c (single_set_and_flags): Fix where
21628 we look for cc setter after the compare-elim changes.
21629 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
21630 with the vector to match what compare-elim now expects.
21631 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
21632 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
21633 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
21634 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
21635 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
21636 (neg<mode>2_insn_set_overflow): Likewise.
21638 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
21640 PR middle-end/79794
21641 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
21642 maybe_expand_insn call, set ops[0].target. If still set after call,
21643 set alt_rtl. Add extra arg to recursive calls.
21644 (extract_bit_field): Add alt_rtl argument. Pass to
21646 * expmed.h (extract_bit_field): Fix prototype.
21647 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
21648 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
21649 to extract_bit_field_calls.
21650 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
21651 Pass alt_rtl to extract_bit_field calls.
21652 * calls.c (store_unaligned_arguments_into_psuedos)
21653 load_register_parameters): Pass extra NULL to extract_bit_field calls.
21654 * optabs.c (maybe_legitimize_operand): Clear op->target when call
21656 * optabs.h (struct expand_operand): Add target bitfield.
21658 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
21660 * compare-elim.c (try_eliminate_compare): Canonicalize
21661 operation with embedded compare to
21662 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
21663 (set (reg) (operation)].
21665 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
21667 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
21670 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
21671 cost of adding a carry flag for ADC instruction.
21672 [case MINUS]: Ignore the cost of subtracting a carry flag
21673 for SBB instruction.
21675 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
21677 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
21679 * config/rs6000/bmiintrin.h: New file.
21680 * config/rs6000/bmi2intrin.h: New file.
21681 * config/rs6000/x86intrin.h: New file.
21683 2017-05-12 Jeff Law <law@redhat.com>
21685 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
21688 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
21690 PR middle-end/80707
21691 * tree-cfg.c: Remove cfg edges of unreachable case statements.
21693 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21695 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
21696 early expansion of vector divide builtins.
21697 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
21698 builtins identified as having unsigned arguments.
21700 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21702 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
21703 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
21704 expansion of vector logical operations (and, andc, or, xor,
21707 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21709 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
21710 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
21712 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21714 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
21715 early GIMPLE expansion of vector multiplies.
21717 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21719 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
21720 TARGET_HAVE_MOVT conditional.
21721 (movt splitter): Likewise.
21723 2017-05-12 Richard Biener <rguenther@suse.de>
21725 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
21726 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
21727 Fold all stmts not inplace.
21729 2017-05-12 Richard Biener <rguenther@suse.de>
21731 PR tree-optimization/80713
21732 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
21733 inserted_exprs bit for not removed stmts.
21735 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
21737 PR middle-end/69921
21738 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
21739 parallelized" attribute for parallelized OpenACC kernels.
21740 * omp-offload.c (execute_oacc_device_lower): Use it.
21742 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
21743 Set "oacc kernels" attribute.
21744 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
21745 parameter. Adjust all users.
21746 (oacc_fn_attrib_kernels_p): Remove function.
21747 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
21748 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
21749 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
21750 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
21751 assert "oacc kernels" attribute is set.
21753 2017-05-11 Carl Love <cel@us.ibm.com>
21755 * config/rs6000/rs6000-c: Add support for built-in functions
21756 vector unsigned char vec_popcnt (vector signed char)
21757 vector unsigned char vec_popcnt (vector unsigned char)
21758 vector unsigned short vec_popcnt (vector signed short)
21759 vector unsigned short vec_popcnt (vector unsigned short)
21760 vector unsigned int vec_popcnt (vector signed int)
21761 vector unsigned int vec_popcnt (vector unsigned int)
21762 vector unsigned long long vec_popcnt (vector signed long long)
21763 vector unsigned long long vec_popcnt (vector unsigned long long)
21764 vector signed long long vec_slo (vector signed long long,
21765 vector signed char)
21766 vector signed long long vec_slo (vector signed long long,
21767 vector unsigned char)
21768 vector unsigned long long vec_slo (vector unsigned long long,
21769 vector signed char)
21770 vector unsigned long long vec_slo (vector unsigned long long,
21771 vector unsigned char)
21772 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
21773 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
21774 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
21775 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
21776 * doc/extend.texi: Update the built-in documentation file for the
21777 new built-in functions.
21779 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
21781 * attribs.h (sorted_attr_string): Move machine independent
21782 functions for target clone support from the i386 port to common
21783 code. Rename ix86_function_versions to common_function_versions.
21784 Rename make_name to make_unique_name.
21785 (common_function_versions): Likewise.
21786 (make_unique_name): Likewise.
21787 (make_dispatcher_decl): Likewise.
21788 (is_function_default_version): Likewise.
21789 * attribs.c (attr_strcmp): Likewise.
21790 (sorted_attr_string): Likewise.
21791 (common_function_versions): Likewise.
21792 (make_unique_name): Likewise.
21793 (make_dispatcher_decl): Likewise.
21794 (is_function_default_version): Likewise.
21795 * config/i386/i386.c (attr_strcmp): Likewise.
21796 (sorted_attr_string): Likewise.
21797 (ix86_function_versions): Likewise.
21798 (make_name): Likewise.
21799 (make_dispatcher_decl): Likewise.
21800 (is_function_default_version): Likewise.
21801 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
21803 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21806 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
21807 Account for direct move costs for vec_construct of integer
21810 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
21813 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
21814 (UNSPEC_STX_ATOMIC): Ditto.
21815 (loaddi_via_sse): New insn.
21816 (storedi_via_sse): Ditto.
21817 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
21818 Update corresponding peephole2 patterns.
21819 (atomic_storedi_fpu): Ditto.
21821 2017-05-11 Julia Koval <julia.koval@intel.com>
21823 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
21824 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
21826 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
21827 (__builtin_ia32_rsqrt14ss_mask): New builtins.
21828 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
21830 2017-05-11 Nathan Sidwell <nathan@acm.org>
21832 * graphite-poly.c: Include dumpfile.h.
21834 * dumpfle.h (dump_function): Declare here ...
21835 * tree-dump.h (dump_function): ... not here.
21836 * dumpfile.c: #include tree-cfg.h.
21837 (dump_function): Move here from ...
21838 * tree-dump.c (dump_function): ... here.
21839 * gimplify.c: #include splay-tree.h, not tree-dump.h.
21840 * graphite-poly.c: Don't include tree-dump.h.
21841 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
21842 * print-tree.c: Likewise.
21843 * stor-layout.c: Likewise.
21844 * tree-nested.c: Likewise.
21846 * dumpfile.c (dump_start): Use TDF_FLAGS.
21847 (dump_enable_all): Fix TDF_KIND check thinko.
21849 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
21851 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21852 array entries to represent two legal parameterizations of the
21853 overloaded __builtin_cmpb function, as represented by the
21854 P6_OV_BUILTIN_CMPB constant.
21855 (altivec_resolve_overloaded_builtin): Add special case handling
21856 for the __builtin_cmpb function, as represented by the
21857 P6_OV_BUILTIN_CMPB constant.
21858 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
21859 (BU_P6_64BIT_2): New macro.
21860 (BU_P6_OVERLOAD_2): New macro
21861 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
21862 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
21863 (CMPB): Add overload support to represent both 32-bit and 64-bit
21864 compare-bytes function.
21865 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
21866 support for TARGET_CMPB.
21867 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
21868 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21869 documentation of the __builtin_cmpb overloaded built-in function.
21871 2017-05-11 Richard Biener <rguenther@suse.de>
21873 PR tree-optimization/80705
21874 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
21875 bases are not vectorizable.
21877 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21879 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
21880 when counting register pressure.
21882 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21884 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
21885 (iv_ca_more_deps): Renamed to ...
21886 (iv_ca_compare_deps): ... this.
21887 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
21889 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21891 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
21893 (determine_group_iv_costs): ... here.
21894 (find_inv_vars_cb): Record inv var if it's not recorded before.
21896 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21898 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
21899 (get_shiftadd_cost): Ditto.
21901 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21903 * tree-ssa-address.c: Include header file.
21904 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
21906 (add_to_parts): Refactor.
21907 (addr_to_parts): New parameter. Update use of move_hint_to_base.
21908 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
21911 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21913 PR tree-optimization/53090
21914 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
21916 (extract_cond_operands): Detect condition with IV on both sides
21917 and return COMP_IV_EXPR_2.
21918 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
21919 (rewrite_use_compare): Simplify by removing call to function
21920 extract_cond_operands.
21922 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21924 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
21925 (extract_cond_operands): Detect condition comparing against non-
21926 invariant bound and return appropriate enum value.
21927 (find_interesting_uses_cond): Update use of extract_cond_operands.
21928 Handle its return value accordingly.
21929 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
21931 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21933 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
21934 nonlinear iv_use computation in loop invariant sensitive way.
21936 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21938 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
21939 (find_iv_candidates): Call relate_compare_use_with_all_cands.
21941 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21943 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
21944 (dump_cand): Support iv_cand.inv_exprs.
21945 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
21947 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
21950 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21952 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
21954 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
21955 as local function. Include necessary header files.
21956 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
21958 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21960 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
21962 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21964 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
21965 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
21966 RSHIFT_EXPR and BIT_NOT_EXPR.
21968 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21970 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
21971 (adjust_setup_cost): New parameter supporting round up adjustment.
21972 (struct address_cost_data): Delete.
21973 (force_expr_to_var_cost): Don't bound cost with spill_cost.
21974 (split_address_cost, ptr_difference_cost): Delete.
21975 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
21976 (struct ainc_cost_data): New struct.
21977 (get_address_cost_ainc): New function.
21978 (get_address_cost, get_computation_cost): Reimplement.
21979 (determine_group_iv_cost_address): Record inv_expr for all uses of
21981 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
21982 (iv_ca_has_deps): Reimplemented to ...
21983 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
21985 (iv_ca_extend): Call iv_ca_more_deps.
21987 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21989 * tree-ssa-address.c (struct mem_address): Move to header file.
21990 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
21991 * tree-ssa-address.h (struct mem_address): Move from C file.
21992 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
21994 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21996 * tree-affine.h (aff_combination_type): New interface.
21997 (aff_combination_zero_p): Remove static.
21998 (aff_combination_const_p): New interface.
21999 (aff_combination_singleton_var_p): New interfaces.
22001 2017-05-11 Richard Biener <rguenther@suse.de>
22003 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
22004 Skip unreachable blocks and destinations.
22005 (eliminate): Move stmt removal and fixup ...
22006 (fini_eliminate): ... here. Skip inserted exprs.
22007 (pass_pre::execute): Move fini_pre after fini_eliminate.
22008 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
22009 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
22010 PRE to get rid of dead code that has invalid SSA form and
22011 split critical edges again.
22013 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22015 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22017 2017-05-11 Richard Biener <rguenther@suse.de>
22019 * passes.c (execute_function_todo): Verify loops if they are
22020 said to be up-to-date.
22021 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
22022 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
22024 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
22027 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
22028 handle calling assemble_external ourself.
22031 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
22032 modes with zero size. Enhance comment.
22034 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22036 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
22037 built-ins for vec_xl and vec_xst with short and char pointer
22040 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
22042 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
22043 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
22044 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
22045 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
22046 (_mm_maskz_min_round_ss): New intrinsics.
22047 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
22048 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
22049 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
22050 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
22051 (__builtin_ia32_minss_mask_round): New builtins.
22052 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22053 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22054 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
22056 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
22057 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
22059 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
22062 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
22064 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
22065 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
22066 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
22067 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
22068 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
22069 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
22070 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
22071 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22072 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
22073 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
22074 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
22075 (__builtin_ia32_mulss_mask_round): New builtins.
22076 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22077 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22078 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
22080 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
22081 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
22083 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
22086 2017-05-10 Julia Koval <julia.koval@intel.com>
22088 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
22089 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
22090 (_mm256_setr_m128i): New intrinsics.
22092 2017-05-10 Julia Koval <julia.koval@intel.com>
22094 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
22095 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
22096 (_mm_maskz_rcp14_ss): New intrinsics.
22097 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
22098 (__builtin_ia32_rcp14ss_mask): New builtins.
22099 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
22101 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
22103 PR tree-optimization/51513
22104 * tree-cfg.c (gimple_seq_unreachable_p): New function.
22105 (assert_unreachable_fallthru_edge_p): Use it.
22106 (group_case_labels_stmt): Likewise.
22107 * tree-cfg.h: Prototype it.
22108 * stmt.c: Include cfghooks.h and tree-cfg.h.
22109 (emit_case_dispatch_table) <gap_label>: New local variable.
22110 Use it to fill dispatch table gaps.
22111 Test for default_label before updating probabilities.
22112 (expand_case) <default_label>: Remove unneeded initialization.
22113 Test for unreachable default case statement and remove its edge.
22114 Set default_label accordingly.
22115 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
22117 2017-05-10 Carl Love <cel@us.ibm.com>
22119 * config/rs6000/rs6000-c: Add support for built-in functions
22120 vector signed char vec_neg (vector signed char)
22121 vector signed short int vec_neg (vector short int)
22122 vector signed int vec_neg (vector signed int)
22123 vector signed long long vec_neg (vector signed long long)
22124 vector float vec_neg (vector float)
22125 vector double vec_neg (vector double)
22126 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
22128 * config/rs6000/altivec.h: Add define for vec_neg
22129 * doc/extend.texi: Update the built-in documentation for the
22130 new built-in functions.
22132 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22134 PR tree-optimization/77644
22135 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
22137 2017-05-10 Nathan Sidwell <nathan@acm.org>
22139 * dumpfile.h (TDI_lang_all): New.
22140 (TDF_KIND): New. Renumber others
22141 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
22143 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
22145 (get_dump_file_name): Adjust suffix generation.
22146 (dump_enable_all): Use TDF_KIND.
22147 * doc/invoke.texi (-fdump-lang-all): Document.
22149 * dumpfile.h: Tabify.
22151 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
22154 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
22155 Move member access before delete.
22157 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
22159 * tree-inline.c (expand_call_inline): Split block at stmt
22162 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
22165 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
22166 are now unused after splitting mov{sf,sd}_hardfloat.
22167 (f32_lr2): Likewise.
22168 (f32_lm): Likewise.
22169 (f32_lm2): Likewise.
22170 (f32_li): Likewise.
22171 (f32_li2): Likewise.
22172 (f32_lv): Likewise.
22173 (f32_sr): Likewise.
22174 (f32_sr2): Likewise.
22175 (f32_sm): Likewise.
22176 (f32_sm2): Likewise.
22177 (f32_si): Likewise.
22178 (f32_si2): Likewise.
22179 (f32_sv): Likewise.
22180 (f32_dm): Likewise.
22181 (f32_vsx): Likewise.
22182 (f32_av): Likewise.
22183 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
22184 For movsf, order stores so the VSX stores occur before the GPR
22185 store which encourages the register allocator to use a traditional
22186 FPR instead of a GPR. For movsd, order the stores so that the GPR
22187 store comes before the VSX stores to allow the power6 to work.
22188 This is due to the power6 not having a 32-bit integer store
22189 instruction from a FPR.
22190 (movsf_hardfloat): Likewise.
22191 (movsd_hardfloat): Likewise.
22193 2017-05-09 Martin Sebor <msebor@redhat.com>
22195 PR translation/80280
22196 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
22199 PR translation/80280
22200 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
22201 data member added in r247778.
22202 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
22204 2017-05-09 Nathan Sidwell <nathan@acm.org>
22206 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
22208 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
22211 2017-05-09 Marek Polacek <polacek@redhat.com>
22213 * doc/invoke.texi: Fix typo.
22215 2017-05-09 Richard Biener <rguenther@suse.de>
22217 * tree-vrp.c (vrp_val_is_max): Adjust comment.
22218 (vrp_val_is_min): Likewise.
22219 (set_value_range_to_value): Likewise.
22220 (set_value_range_to_nonnegative): Likewise.
22221 (gimple_assign_nonzero_p): Likewise.
22222 (gimple_stmt_nonzero_p): Likewise.
22223 (vrp_int_const_binop): Likewise. Remove unreachable case.
22224 (adjust_range_with_scev): Adjust comments.
22225 (compare_range_with_value): Likewise.
22226 (extract_range_from_phi_node): Likewise.
22227 (test_for_singularity): Likewise.
22229 2017-05-09 Richard Biener <rguenther@suse.de>
22231 * tree-vrp.c (get_single_symbol): Add assert that we don't
22232 get overflowed constants as invariant part.
22233 (compare_values_warnv): Add comment before the TREE_NO_WARNING
22234 checks. Use wi::cmp instead of recursing for integer constants.
22235 (compare_values): Just ignore whether we assumed undefined
22236 overflow instead of failing the compare.
22237 (extract_range_for_var_from_comparison_expr): Add comment before the
22238 TREE_NO_WARNING sets.
22239 (test_for_singularity): Likewise.
22240 (extract_range_from_comparison): Do not disable optimization
22241 when we assumed undefined overflow.
22242 (extract_range_basic): Remove init of unused var.
22244 2017-05-09 Richard Biener <rguenther@suse.de>
22246 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
22247 (extract_range_from_multiplicative_op_1): Adjust.
22248 (extract_range_from_binary_expr_1): Use int_const_binop.
22250 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
22253 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
22254 rs6000_store_data_bypass_p in seven define_bypass directives and
22255 in several comments.
22256 * config/rs6000/rs6000-protos.h: Add prototype for
22257 rs6000_store_data_bypass_p function.
22258 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
22259 function implements slightly different (rs6000-specific) semantics
22260 than store_data_bypass_p, returning false rather than aborting
22261 with assertion error when arguments do not satisfy the
22262 requirements of store data bypass.
22263 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
22264 rs6000_store_data_bypass_p.
22266 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
22268 * config/xtensa/xtensa-protos.h
22269 (xtensa_initial_elimination_offset): New declaration.
22270 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22271 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
22272 macro definition, add case for FRAME_POINTER_REGNUM when
22273 FRAME_GROWS_DOWNWARD.
22274 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
22275 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
22276 xtensa_initial_elimination_offset.
22278 2017-05-08 Nathan Sidwell <nathan@acm.org>
22280 * doc/invoke.texi: Alphabetize -fdump options.
22282 2017-05-08 Martin Sebor <msebor@redhat.com>
22284 PR translation/80280
22285 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
22287 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
22289 * target.def (compute_frame_layout): New optional target hook.
22290 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
22291 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
22292 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
22294 * reload1.c (verify_initial_elim_offsets): Likewise.
22295 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
22296 (use_simple_return_p): Call arm_compute_frame_layout if needed.
22297 (arm_get_frame_offsets): Split up into this ...
22298 (arm_compute_frame_layout): ... and this function.
22300 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
22302 * config/aarch64/constraints.md (Usa): New constraint.
22303 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
22305 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
22307 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
22308 with_multilib_list after it has been checked.
22310 2017-05-08 Richard Biener <rguenther@suse.de>
22312 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
22313 (bitmap_set_subtract_values): Likewise.
22315 2017-05-08 Richard Biener <rguenther@suse.de>
22317 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
22318 (gimple_assign_nonzero): ... this and remove strict_overflow_p
22320 (gimple_stmt_nonzero_warnv_p): Rename to ...
22321 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
22323 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
22324 (extract_range_basic): Adjust, do not disable propagation on
22325 strict overflow sensitive simplification.
22326 (vrp_visit_cond_stmt): Likewise.
22328 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
22330 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
22331 body size unconditionally.
22333 2017-05-07 Jeff Law <law@redhat.com>
22336 2017-05-06 Jeff Law <law@redhat.com>
22337 PR tree-optimization/78496
22338 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
22341 PR tree-optimization/78496
22342 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
22343 (simplify_stmt_using_ranges): Call it.
22344 (vrp_dom_walker::before_dom_children): Extract equivalences
22345 from an ASSERT_EXPR with an equality comparison against a
22348 2017-05-06 Jeff Law <law@redhat.com>
22350 PR tree-optimization/78496
22351 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
22354 PR tree-optimization/78496
22355 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
22356 (simplify_stmt_using_ranges): Call it.
22357 (vrp_dom_walker::before_dom_children): Extract equivalences
22358 from an ASSERT_EXPR with an equality comparison against a
22361 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
22363 * lra-constraints.c (lra_copy_reg_equiv): New function.
22364 (split_reg): Use it to copy equivalence information from the
22365 original register to the spill register.
22367 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
22369 PR rtl-optimization/75964
22370 * simplify-rtx.c (simplify_const_relational_operation): Remove
22371 invalid handling of comparisons of integer ABS.
22373 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
22375 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
22376 initialize to zero.
22377 (init_regs): Remove declaration.
22378 (function_arg_advance_32): Initialize error_p as boolean variable.
22380 2017-05-05 Nathan Sidwell <nathan@acm.org>
22382 * store-motion.c (remove_reachable_equiv_notes): Reformat long
22383 lines. Use for (;;).
22385 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22387 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
22388 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
22389 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
22390 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
22391 VF=2 that require versioning.
22393 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22395 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
22398 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22400 * diagnostic.h (diagnostic_override_option_index): Convert from
22401 macro to inline function.
22403 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22405 * diagnostic.c (last_module_changed_p): New function.
22406 (set_last_module): New function.
22407 (diagnostic_report_current_module): Convert macro usage to
22408 the above functions.
22409 * diagnostic.h (diagnostic_context::last_module): Strengthen
22410 from const line_map * to const line_map_ordinary *.
22411 (diagnostic_last_module_changed): Delete macro.
22412 (diagnostic_set_last_module): Delete macro.
22414 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22416 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
22417 with diagnostic_report_diagnostic.
22418 (diagnostic_n_impl_richloc): Likewise.
22419 * diagnostic.h (report_diagnostic): Delete macro.
22420 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
22421 with diagnostic_report_diagnostic.
22422 * substring-locations.c (format_warning_va): Likewise.
22424 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22426 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
22427 save/restor of format_spec. Move option-printing code to...
22428 (print_option_information): ...this new function, and
22429 reimplement by simply printing to the pretty_printer,
22430 rather than appending to the format string.
22432 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22434 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
22435 handling logic into...
22436 (update_effective_level_from_pragmas): ...this new function.
22438 2017-05-04 Andrew Waterman <andrew@sifive.com>
22440 * config/riscv/riscv.opt (mstrict-align): New option.
22441 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
22442 (SLOW_UNALIGNED_ACCESS): Define.
22443 (riscv_slow_unaligned_access): Declare.
22444 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
22446 (riscv_slow_unaligned_access): New variable.
22447 (rocket_tune_info): Set slow_unaligned_access to true.
22448 (optimize_size_tune_info): Set slow_unaligned_access to false.
22449 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
22450 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
22451 (riscv_option_override): Set riscv_slow_unaligned_access.
22452 * doc/invoke.texi: Add -mstrict-align to RISC-V.
22454 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
22456 * config/riscv/riscv.md: Unify indentation.
22458 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22463 * config/rs6000/rs6000.md (u code attribute): Add FIX and
22465 (extendsi<mode>2): Add support for doing sign extension via
22466 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
22467 don't have ISA 3.0 instructions.
22468 (extendsi<mode>2 splitter): Likewise.
22469 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
22470 generate the normal insns since SImode can now go in vector
22471 registers. Disallow the special UNSPECs needed for previous
22472 machines to hide SImode being used. Add new insns
22473 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
22474 (fix_trunc<mode>si2_stfiwx): Likewise.
22475 (fix_trunc<mode>si2_internal): Likewise.
22476 (fixuns_trunc<mode>si2): Likewise.
22477 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22478 (fctiw<u>z_<mode>_smallint): Likewise.
22479 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
22480 of floating point to 32-bit integer from doing a direct move to
22481 the GPR registers to do a store.
22482 (fctiwz_<mode>): Break long line.
22484 2017-05-05 Bin Cheng <bin.cheng@arm.com>
22486 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
22487 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
22488 (addr_list, addr_offset_valid_p): New.
22489 (split_address_groups): Check offset validity with above function.
22490 (gt-tree-ssa-loop-ivopts.h): Include header file.
22492 2017-05-05 Nathan Sidwell <nathan@acm.org>
22494 * config.gcc (arm*-*-*): Add missing 'fi'.
22496 2017-05-05 Steve Ellcey <sellcey@cavium.com>
22498 * doc/invoke.texi (-fopt-info): Explicitly say order of options
22499 included in -fopt-info does not matter.
22500 * doc/optinfo.texi (-fopt-info): Fix description of default
22501 behavour. Explicitly say order of options included in -fopt-info
22504 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22506 * config.gcc: Allow combinations of aprofile and rmprofile values for
22507 --with-multilib-list.
22508 * config/arm/t-multilib: New file.
22509 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
22510 variables. Remove setting of ISA and floating-point ABI in
22511 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
22512 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
22513 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
22514 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
22516 * config/arm/t-rmprofile: Likewise except for the matches changes.
22517 * doc/install.texi (--with-multilib-list): Document the combination of
22518 aprofile and rmprofile values and warn about pitfalls in doing that.
22520 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
22522 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
22523 (movdi_aarch64): Likewise.
22525 2017-05-05 Jakub Jelinek <jakub@redhat.com>
22527 PR tree-optimization/80632
22528 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
22530 (build_arrays): Initialize it for virtual phis.
22531 (fix_phi_nodes): Use it for virtual phis.
22533 PR tree-optimization/80558
22534 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
22535 [x, y] op z into [x op, y op z] for op & or | if conditions
22538 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
22539 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22542 * config/arm/arm.md (use_literal_pool): Remove.
22543 (64-bit immediate split): No longer takes cost into consideration
22544 if arm_disable_literal_pool is enabled.
22545 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
22546 used when arm_disable_literal_pool is enabled.
22547 (arm_max_const_double_inline_cost): Remove use of
22548 arm_disable_literal_pool.
22549 (push_minipool_fix): Add assert.
22550 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
22551 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
22552 (no_literal_pool_sf_immediate): New.
22554 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22556 PR tree-optimization/80613
22557 * tree-ssa-dce.c (propagate_necessity): Remove cases for
22558 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
22560 2017-05-05 Richard Biener <rguenther@suse.de>
22562 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
22564 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
22566 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
22567 of this flag from insn conditions due to removal from r247495.
22569 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
22571 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
22573 (arm_early_store_addr_dep_ptr): Likewise.
22574 * config/arm/aarch-common-protos.h
22575 (arm_early_load_addr_dep_ptr): Add prototype.
22576 (arm_early_store_addr_dep_ptr): Likewise.
22577 * config/arm/cortex-a53.md: Add new bypasses.
22579 2017-05-05 Jakub Jelinek <jakub@redhat.com>
22581 * tree.c (next_type_uid): Change type to unsigned.
22582 (type_hash_canon): Decrement back next_type_uid if
22583 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
22584 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
22587 2017-05-04 Martin Sebor <msebor@redhat.com>
22589 * builtins.c: Fix a trivial typo in a comment.
22591 PR middle-end/79234
22592 * builtins.c (check_sizes): Adjust to handle reading past the end.
22593 Avoid printing excessive upper bound of ranges. Use %E to print
22594 tree nodes instead of converting them to %wu.
22595 (expand_builtin_memchr): New function.
22596 (compute_dest_size): Rename...
22597 (compute_objsize): ...to this.
22598 (expand_builtin_memcpy): Adjust.
22599 (expand_builtin_mempcpy): Adjust.
22600 (expand_builtin_strcat): Adjust.
22601 (expand_builtin_strcpy): Adjust.
22602 (check_strncat_sizes): Adjust.
22603 (expand_builtin_strncat): Adjust.
22604 (expand_builtin_strncpy): Adjust and simplify.
22605 (expand_builtin_memset): Adjust.
22606 (expand_builtin_bzero): Adjust.
22607 (expand_builtin_memcmp): Adjust.
22608 (expand_builtin): Handle memcmp.
22609 (maybe_emit_chk_warning): Check strncat just once.
22611 2017-05-04 Martin Sebor <msebor@redhat.com>
22613 PR preprocessor/79214
22614 PR middle-end/79222
22615 PR middle-end/79223
22616 * builtins.c (check_sizes): Add inlining context and issue
22617 warnings even when -Wno-system-headers is set.
22618 (check_strncat_sizes): Same.
22619 (expand_builtin_strncat): Same.
22620 (expand_builtin_memmove): New function.
22621 (expand_builtin_stpncpy): Same.
22622 (expand_builtin): Handle memmove and stpncpy.
22624 2017-05-04 Bin Cheng <bin.cheng@arm.com>
22626 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
22627 which is not used any more.
22629 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22631 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
22633 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22635 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
22636 (cortexa53_tunings): Likewise.
22637 (cortexa57_tunings): Likewise.
22638 (cortexa72_tunings): Likewise.
22639 (cortexa73_tunings): Likewise.
22641 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22643 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
22644 Set loop alignment to 8.
22646 2017-05-04 Martin Sebor <msebor@redhat.com>
22648 PR translation/80280
22649 * builtins.c (expand_builtin_object_size): Add missing quoting to
22650 %D and like directives.
22651 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
22652 (hsa_type_for_tree_type): Same.
22653 (verify_function_arguments): Same.
22654 * symtab.c (symbol_table::change_decl_assembler_name): Same.
22655 * varasm.c (get_section): Same.
22658 2017-05-04 Martin Sebor <msebor@redhat.com>
22660 PR translation/80280
22661 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
22663 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22665 * config/aarch64/aarch64.c (generic_addrcost_table):
22666 Change HI/TI mode setting.
22668 2017-05-04 Martin Jambor <mjambor@suse.cz>
22670 PR tree-optimization/80622
22671 * tree-sra.c (comes_initialized_p): New function.
22672 (build_accesses_from_assign): Only set write lazily when
22673 comes_initialized_p is false.
22674 (analyze_access_subtree): Use comes_initialized_p.
22675 (propagate_subaccesses_across_link): Assert !comes_initialized_p
22676 instead of testing for PARM_DECL.
22678 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22680 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
22681 constraint on operand 0 to allow more general addressing modes.
22682 Adjust output template.
22683 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
22685 * config/aarch64/aarch64-protos.h
22686 (aarch64_address_valid_for_prefetch_p): Declare prototype.
22687 * config/aarch64/constraints.md (Dp): New address constraint.
22688 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
22691 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
22693 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
22694 update use of estimate_ipcp_clone_size_and_time.
22695 (estimate_local_effects): Update use of
22696 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
22697 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
22698 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
22699 Return nonspecialized time.
22701 2017-05-04 Richard Biener <rguenther@suse.de>
22703 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
22704 for the last VUSE which def dominates the PHI. Directly call
22706 (get_continuation_for_phi_1): Remove.
22708 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
22710 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
22711 to explain the use of truncating division. Cap the number of
22712 iterations to the maximum given by nb_iterations_upper_bound,
22715 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22717 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
22718 * configure: Regenerate.
22719 * config.in: Regenerate.
22720 * config/i386/driver-mingw32.c: new file.
22721 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
22722 * config.host: Link driver-mingw32.o on MinGW host.
22723 * doc/install.texi: Document new --enable-mingw-wildcard configure
22726 2017-05-04 Marek Polacek <polacek@redhat.com>
22728 PR tree-optimization/80612
22729 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
22731 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22732 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
22734 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
22735 (movt splitter): Likewise.
22736 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
22737 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
22738 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
22739 block for Thumb-1 with MOVT.
22740 (thumb2_legitimate_address_p): Move code block ...
22741 (can_avoid_literal_pool_for_label_p): ... into this new function.
22742 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
22744 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
22745 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
22746 "M-profile targets with the MOVT instruction".
22748 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22750 * config/arm/arm-builtins.c (arm_init_builtins): Rename
22751 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
22752 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
22754 2017-05-04 Martin Liska <mliska@suse.cz>
22756 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
22757 variable cond_code.
22759 2017-05-04 Richard Biener <rguenther@suse.de>
22761 * tree.c (array_at_struct_end_p): Handle arrays at struct
22762 end with flexarrays more conservatively. Refactor and treat
22763 arrays of arrays or aggregates more strict. Fix
22764 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
22765 * tree.c (array_at_struct_end_p): Adjust prototype.
22766 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
22767 * gimple-fold.c (get_range_strlen): Likewise.
22768 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
22770 2017-05-04 Richard Biener <rguenther@suse.de>
22772 PR tree-optimization/31130
22773 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
22775 (supports_overflow_infinity): Likewise.
22776 (is_negative_overflow_infinity): Likewise.
22777 (is_positive_overflow_infinity): Likewise.
22778 (is_overflow_infinity): Likewise.
22779 (stmt_overflow_infinity): Likewise.
22780 (overflow_infinity_range_p): Likewise.
22781 (usable_range_p): Remove as always returning true.
22782 (make_overflow_infinity): Remove.
22783 (negative_overflow_infinity): Likewise.
22784 (positive_overflow_infinity): Likewise.
22785 (avoid_overflow_infinity): Likewise.
22786 (set_value_range): Adjust accordingly.
22787 (set_value_range_to_nonnegative): Likewise, remove now unused
22788 overflow_infinity arg.
22789 (vrp_operand_equal_p): Adjust.
22790 (update_value_range): Likewise.
22791 (range_int_cst_singleton_p): Likewise.
22792 (operand_less_p): Likewise.
22793 (compare_values_warnv): Likewise.
22794 (extract_range_for_var_from_comparison_expr): Likewise.
22795 (vrp_int_const_binop): Likewise.
22796 (zero_nonzero_bits_from_vr): Likewise.
22797 (extract_range_from_multiplicative_op_1): Likewise.
22798 (extract_range_from_binary_expr_1): Likewise.
22799 (extract_range_from_unary_expr): Likewise.
22800 (extract_range_from_comparison): Likewise.
22801 (extract_range_basic): Likewise.
22802 (adjust_range_with_scev): Likewise.
22803 (compare_ranges): Likewise.
22804 (compare_range_with_value): Likewise.
22805 (dump_value_range): Likewise.
22806 (test_for_singularity): Likewise, remove strict_overflow_p parameter
22808 (simplify_cond_using_ranges): Adjust.
22810 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22812 * brig-builtins.def: Added a builtin for class_f64.
22813 * builtin-types.def: Added a builtin type needed by class_f64.
22815 2017-05-03 Jason Merrill <jason@redhat.com>
22817 * timevar.def: Add TV_CONSTEXPR.
22819 2017-05-03 David Malcolm <dmalcolm@redhat.com>
22821 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
22823 2017-05-03 Martin Jambor <mjambor@suse.cz>
22825 * ipa-prop.c (ipa_update_after_lto_read): Removed.
22826 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
22827 * ipa-cp.c (ipcp_propagate_stage): Do not call
22828 ipa_update_after_lto_read.
22829 * ipa-inline.c (ipa_inline): Likewise.
22831 2017-05-03 Martin Jambor <mjambor@suse.cz>
22833 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
22834 tag. Added a default constructor and a destructor.
22835 (ipa_edge_args_sum_t): New class;
22836 (ipa_edge_args_sum): Declare.
22837 (ipa_edge_args_vector): Remove declaration.
22838 (IPA_EDGE_REF): Use ipa_edge_args_sum.
22839 (ipa_free_edge_args_substructures): Remove declaration.
22840 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
22841 (ipa_edge_args_info_available_for_edge_p): Likewise.
22842 * ipa-prop.c (ipa_edge_args_vector): Removed.
22843 (edge_removal_hook_holder): Likewise.
22844 (edge_duplication_hook_holder): Likewise.
22845 (ipa_edge_args_sum): New variable.
22846 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
22847 ipa_edge_args_vector.
22848 (ipa_free_edge_args_substructures): Likewise.
22849 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
22850 ipa_edge_args_vector.
22851 (ipa_edge_removal_hook): Turned into method
22852 ipa_edge_args_sum_t::remove.
22853 (ipa_edge_duplication_hook): Turned into method
22854 ipa_edge_args_sum_t::duplicate.
22855 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
22856 registering edge hooks.
22857 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
22858 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
22859 ipa_edge_args_sum instead of ipa_edge_args_vector.
22860 * ipa-profile.c (ipa_profile): Likewise.
22862 2017-05-03 Martin Jambor <mjambor@suse.cz>
22864 * symbol-summary.h (function_summary): New method exists.
22865 (function_summary::symtab_removal): Deallocate through release.
22866 (call_summary): New class.
22867 (gt_ggc_mx): New overload.
22868 (gt_pch_nx): Likewise.
22869 (gt_pch_nx): Likewise.
22871 2017-05-03 Jeff Law <law@redhat.com>
22873 PR tree-optimization/78496
22874 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
22875 from simplify_cond_using_ranges. Split off code to walk
22876 backwards through casts into ...
22877 (simplify_cond_using_ranges_2): New function.
22878 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
22879 (execute_vrp): After identifying jump threads, call
22880 simplify_cond_using_ranges_2.
22882 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
22885 * ipa-inline.h (inline_summary): Add ctor.
22886 (create_ggc): Do not use ggc_cleared_alloc.
22888 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
22889 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22891 * gcc.c (handle_braces): Support escaping in switch matching
22893 * doc/invoke.texi (Spec Files): Document it.
22894 Remove superfluous @code markup in items.
22896 2017-05-03 David Malcolm <dmalcolm@redhat.com>
22898 * diagnostic-show-locus.c (struct column_range): New struct.
22899 (get_affected_columns): New function.
22900 (get_printed_columns): New function.
22901 (struct correction): New struct.
22902 (correction::ensure_capacity): New function.
22903 (correction::ensure_terminated): New function.
22904 (struct line_corrections): New struct.
22905 (line_corrections::~line_corrections): New dtor.
22906 (line_corrections::add_hint): New function.
22907 (layout::print_trailing_fixits): Reimplement in terms of the new
22909 (selftest::test_overlapped_fixit_printing): New function.
22910 (selftest::diagnostic_show_locus_c_tests): Call it.
22912 2017-05-03 Nathan Sidwell <nathan@acm.org>
22914 Canonicalize canonical type hashing
22915 * tree.h (type_hash_canon_hash): Declare.
22916 * tree.c (type_hash_list, attribute_hash_list): Move into
22917 type_hash_canon_hash.
22918 (build_type_attribute_qual_variant): Break out hash code calc into
22919 type_hash_canon_hash.
22920 (type_hash_canon_hash): New. Generic type hash computation.
22921 (build_range_type_1, build_array_type_1, build_function_type,
22922 build_method_type_directly, build_offset_type, build_complex_type,
22923 make_vector_type): Call it.
22925 2017-05-03 Richard Biener <rguenther@suse.de>
22927 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
22928 When all DRs have unknown misaligned do not always peel
22929 when there is a store but apply the same costing model as if
22930 there were only loads.
22932 2017-05-03 Richard Biener <rguenther@suse.de>
22935 PR tree-optimization/80492
22936 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
22937 compare_base_decls returning dont-know properly.
22939 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
22941 * config/arm/iterators.md (CCSI): New mode iterator.
22942 (arch): New mode attribute.
22943 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
22944 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
22945 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
22946 code iterator for success result mode.
22947 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
22948 the corresponding new insn generators.
22950 2017-05-03 Bin Cheng <bin.cheng@arm.com>
22953 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22954 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22956 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
22958 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
22959 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
22960 (DDR_A): Wrap DDR argument in brackets.
22961 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
22962 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
22963 (DDR_REVERSED_P): Likewise.
22965 2017-05-03 Jakub Jelinek <jakub@redhat.com>
22967 PR tree-optimization/79472
22968 * tree-switch-conversion.c (struct switch_conv_info): Add
22969 contiguous_range and default_case_nonstandard fields.
22970 (collect_switch_conv_info): Compute contiguous_range and
22971 default_case_nonstandard fields, don't clear final_bb if
22972 contiguous_range and only the default case doesn't have the required
22974 (check_all_empty_except_final): Set default_case_nonstandard instead
22975 of failing if contiguous_range and the default case doesn't have empty
22977 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
22978 and only the default case doesn't have the required constants. Skip
22980 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
22981 if default_case_nonstandard.
22982 (build_constructors): Build constant 1 just once. Assert that default
22983 values aren't inserted in between cases if contiguous_range. Skip
22985 (build_arrays): Skip virtual phis.
22986 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
22987 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
22988 Handle virtual phis.
22989 (gen_inbound_check): Handle default_case_nonstandard case.
22990 (process_switch): Adjust check_final_bb caller. Call
22991 gather_default_values with the first non-default case instead of
22992 default case if default_case_nonstandard.
22994 2017-05-02 Nathan Sidwell <nathan@acm.org>
22996 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
22997 check. Fix formatting.
22999 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
23001 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
23002 errors when comparing specialized and unspecialized times.
23004 2017-05-02 David Malcolm <dmalcolm@redhat.com>
23006 * diagnostic-show-locus.c
23007 (layout::should_print_annotation_line_p): Make private.
23008 (layout::print_annotation_line): Make private.
23009 (layout::annotation_line_showed_range_p): Make private.
23010 (layout::show_ruler): Make private.
23011 (layout::print_source_line): Make private. Pass in line and
23012 line_width, rather than calling location_get_source_line. Drop
23014 (layout::print_leading_fixits): New method.
23015 (layout::print_any_fixits): Rename to...
23016 (layout::print_trailing_fixits): ...this, and make private.
23017 Don't print newline fixits.
23018 (diagnostic_show_locus): Move logic for printing one row into...
23019 (layout::print_line): ...this new function. Move the
23020 location_get_source_line call and error-handling from
23021 print_source_line to here. Call print_leading_fixits, and rename
23022 print_any_fixits to print_trailing_fixits.
23023 (selftest::test_fixit_insert_containing_newline): Update now that
23024 newlines are partially supported.
23025 (selftest::test_fixit_insert_containing_newline_2): New test.
23026 (selftest::test_fixit_replace_containing_newline): Update comments.
23027 (selftest::diagnostic_show_locus_c_tests): Call the new test.
23028 * edit-context.c (class added_line): New class.
23029 (class edited_line): Describe newline handling in comment.
23030 (edited_line::actually_edited_p): New method.
23031 (edited_line::print_content): Delete redundant decl.
23032 (edited_line::m_predecessors): New field.
23033 (edited_file::print_content): Call edited_line::print_content.
23034 (edited_file::print_diff): Update to support newlines.
23035 (edited_file::print_diff_hunk): Likewise.
23036 (edited_file::print_run_of_changed_lines): New function.
23037 (edited_file::print_diff_line): Convert to...
23038 (print_diff_line): ...this.
23039 (edited_file::get_effective_line_count): New function.
23040 (edited_line::edited_line): Initialize new field m_predecessors.
23041 (edited_line::~edited_line): Clean up m_predecessors.
23042 (edited_line::apply_fixit): Handle newlines.
23043 (edited_line::get_effective_line_count): New function.
23044 (edited_line::print_content): New function.
23045 (edited_line::print_diff_lines): New function.
23046 (selftest::test_applying_fixits_insert_containing_newline): New
23048 (selftest::test_applying_fixits_replace_containing_newline): New
23050 (selftest::insert_line): New function.
23051 (selftest::test_applying_fixits_multiple_lines): Add example of
23053 (selftest::edit_context_c_tests): Call the new tests.
23055 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23057 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
23058 parameter cand. Update dump information.
23059 (get_computation_cost): Update uses.
23061 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23063 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
23064 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
23065 (get_computation_at, rewrite_use_address): Update use of
23066 get_computation_aff.
23068 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23070 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
23071 (get_computation): Delete.
23072 (get_computation_cost): Implement like get_computation_cost_at.
23073 Use get_computation_at.
23074 (get_computation_cost_at): Delete.
23075 (rewrite_use_nonlinear_expr): Use get_computation_at.
23076 (rewrite_use_compare, remove_unused_ivs): Ditto.
23078 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23080 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
23082 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23084 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
23085 (ivopts_global_cost_for_size): Rename parameter and update uses.
23086 (iv_ca_recount_cost): Update uses.
23087 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
23088 candidates seperately in n_invs and n_cands.
23089 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
23091 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23093 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
23094 (find_inv_vars_cb): New.
23095 (find_depends): Renamed to ...
23096 (find_inv_vars): ... this.
23097 (add_candidate_1, force_var_cost): Call find_inv_vars.
23098 (split_address_cost, determine_group_iv_cost_cond): Ditto.
23100 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23102 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
23103 inv_vars. Add inv_exprs.
23104 (struct iv_cand): Rename depends_on to inv_vars.
23105 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
23106 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
23107 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
23108 (dump_cand): Dump inv_vars.
23109 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
23110 (record_invariant, find_depends, add_candidate_1): Ditto.
23111 (set_group_iv_cost, force_var_cost): Ditto.
23112 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
23113 (get_computation_cost_at, get_computation_cost): Ditto.
23114 (determine_group_iv_cost_generic): Ditto.
23115 (determine_group_iv_cost_address): Ditto.
23116 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
23117 (determine_group_iv_costs): Ditto.
23118 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
23119 (iv_ca_set_remove_invariants): Renamed to ...
23120 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
23121 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
23122 (iv_ca_set_add_invariants): Renamed to ...
23123 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
23124 (iv_ca_set_cp): Use iv_ca_set_add_invs.
23125 (iv_ca_has_deps): Support inv_vars and inv_exprs.
23126 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
23127 (create_new_ivs): Remove useless dump.
23129 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23131 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
23133 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
23134 (iv_ca_set_no_cp, create_new_iv): Ditto.
23136 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23138 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23140 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
23142 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
23143 function tree_check2.
23145 2017-05-02 Martin Liska <mliska@suse.cz>
23147 * doc/gcov.texi: Add missing preposition.
23148 * gcov.c (function_info::function_info): Properly fill up
23149 all member variables.
23151 2017-05-02 Tamar Christina <tamar.christina@arm.com>
23153 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
23155 2017-05-02 Tamar Christina <tamar.christina@arm.com>
23157 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
23159 2017-05-02 Martin Liska <mliska@suse.cz>
23162 * lto-streamer-in.c (lto_read_tree_1): Remove
23163 LTO_STREAMER_DEBUG.
23164 * lto-streamer.c (struct tree_hash_entry): Likewise.
23165 (struct tree_entry_hasher): Likewise.
23166 (tree_entry_hasher::hash): Likewise.
23167 (tree_entry_hasher::equal): Likewise.
23168 (lto_streamer_init): Likewise.
23169 (lto_orig_address_map): Likewise.
23170 (lto_orig_address_get): Likewise.
23171 (lto_orig_address_remove): Likewise.
23172 * lto-streamer.h: Likewise.
23173 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
23174 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
23176 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
23178 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
23179 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
23180 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
23181 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
23182 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
23183 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
23184 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
23185 (mm_maskz_sub_ss): New intrinsics.
23186 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23187 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23188 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
23189 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
23190 (__builtin_ia32_subss_mask_round): New builtins.
23191 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23192 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23193 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
23195 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
23196 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23198 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23201 2017-05-02 Martin Jambor <mjambor@suse.cz>
23203 PR tree-optimization/78687
23204 * tree-sra.c (access): New field parent.
23205 (process_subtree_disqualification): New function.
23206 (disqualify_candidate): Call it.
23207 (build_accesses_from_assign): Reset write flag if creating an
23209 (build_access_subtree): Fill in parent field and also prpagate
23210 down grp_write flag.
23211 (create_artificial_child_access): New parameter set_grp_write, set
23212 grp_write to its value.
23213 (propagate_subaccesses_across_link): Also propagate grp_write flag
23215 (propagate_all_subaccesses): Push the closest parent back to work
23216 queue if add_access_to_work_queue returned true.
23218 2017-05-02 Richard Biener <rguenther@suse.de>
23220 * common.opt (fstrict-overflow): Alias negative to fwrapv.
23221 * doc/invoke.texi (fstrict-overflow): Remove all traces of
23222 -fstrict-overflow documentation.
23223 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
23224 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
23225 flag_strict_overflow.
23226 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
23227 * lto-opts.c (lto_write_options): Do not stream it.
23228 * lto-wrapper.c (merge_and_complain): Do not handle it.
23229 * opts.c (default_options_table): Do not set -fstrict-overflow.
23230 (finish_options): Likewise do not clear it when sanitizing.
23231 * simplify-rtx.c (simplify_const_relational_operation): Do not
23232 test flag_strict_overflow.
23234 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
23236 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
23237 using enabled attribute.
23238 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
23239 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
23240 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
23241 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
23242 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
23243 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
23244 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
23245 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
23246 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
23247 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
23249 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
23251 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
23253 2017-05-02 Richard Biener <rguenther@suse.de>
23255 PR tree-optimization/80591
23257 2017-04-10 Richard Biener <rguenther@suse.de>
23259 * tree-ssa-structalias.c (find_func_aliases): Properly handle
23262 2017-05-02 Richard Biener <rguenther@suse.de>
23264 PR tree-optimization/80549
23265 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
23266 (cleanup_tree_cfg_noloop): Create forwarders to known loop
23267 headers if they do not have a preheader.
23269 2017-05-02 Martin Liska <mliska@suse.cz>
23272 * common.opt: Fix typo.
23273 * doc/invoke.texi: Likewise.
23275 2017-05-01 Jan Beulich <jbeulich@suse.com>
23277 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
23278 swapping, add (x,x,m,x,n) alternative.
23280 2017-05-01 Nathan Sidwell <nathan@acm.org>
23282 * calls.c (combine_pending_stack_adjustment_and_call): Remove
23283 unnecessary unadjusted_alignment check.
23285 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
23288 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
23290 * gimplify.c (gimplify_cilk_detach): New function.
23291 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
23292 * tree-core.h: Document EXPR_CILK_SPAWN.
23293 * tree.h (EXPR_CILK_SPAWN): Define.
23295 2017-05-01 David Malcolm <dmalcolm@redhat.com>
23297 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
23298 to use new fixit_hint representation, using the "replace" logic.
23299 (get_line_span_for_fixit_hint): Likewise.
23300 (layout::print_any_fixits): Likewise.
23301 (selftest::test_one_liner_many_fixits): Rename to...
23302 (selftest::test_one_liner_many_fixits_1): ...this, and update
23303 comment and expected output to reflect that the multiple fix-it
23304 hints are now consolidated into one insertion.
23305 (selftest::test_one_liner_many_fixits_2): New test.
23306 (selftest::test_diagnostic_show_locus_one_liner): Update for
23308 (selftest::test_fixit_consolidation): Update for fix-it API
23310 * diagnostic.c (print_parseable_fixits): Likewise.
23311 * edit-context.c (edited_line::m_line_events): Convert from
23312 auto_vec <line_event *> to auto_vec <line_event>.
23313 (class line_event): Convert from abstract base class to a concrete
23314 class, taking over the role of replace_event.
23315 (class insert_event): Delete.
23316 (class replace_event): Rename to class line_event. Convert to
23318 (edit_context::add_fixits): Reimplement.
23319 (edit_context::apply_insert): Delete.
23320 (edit_context::apply_replace): Rename to...
23321 (edit_context::apply_fixit): ...this. Convert to half-open range.
23322 (edited_file::apply_insert): Delete.
23323 (edited_file::apply_replace): Rename to...
23324 (edited_file::apply_fixit): ...this.
23325 (edited_line::~edited_line): Drop deletion of events.
23326 (edited_line::apply_insert): Delete.
23327 (edited_line::apply_replace): Rename to...
23328 (edited_line::apply_fixit): ...this. Convert to half-open range.
23329 Update for change to type of m_line_events.
23330 * edit-context.h (edit_context::apply_insert): Delete.
23331 (edit_context::apply_replace): Rename to...
23332 (edit_context::apply_fixit): ...this.
23334 2017-05-01 Martin Sebor <msebor@redhat.com>
23336 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
23339 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
23342 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
23343 __get_cpuid_max returns 0.
23344 (__get_cpuid_count): Ditto.
23346 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
23348 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
23349 replacement expression is another instance of one of its arguments.
23351 2017-05-01 Jakub Jelinek <jakub@redhat.com>
23354 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
23355 check for stack push/pop autoinc.
23356 * config/i386/i386.c (ix86_agi_dependent): Return false
23357 if the only reason why modified_in_p returned true is that
23358 addr is SP based and set_insn is a push or pop.
23360 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
23362 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
23365 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
23368 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
23369 (account_size_time): Use two predicates - exec_pred and
23371 (evaluate_conditions_for_known_args): Compute both clause and
23373 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
23374 (inline_summary_t::duplicate): Update.
23375 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
23377 (compute_inline_parameters): Likewise.
23378 (estimate_edge_size_and_time): Update caluclation of time.
23379 (estimate_node_size_and_time): Compute both time and nonspecialized
23381 (estimate_ipcp_clone_size_and_time): Update.
23382 (inline_merge_summary): Update.
23383 (do_estimate_edge_time): Update.
23384 (do_estimate_edge_size): Update.
23385 (do_estimate_edge_hints): Update.
23386 (inline_read_section, inline_write_summary): Stream both new predicates.
23387 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
23389 (compute_inlined_call_time): Cleanup.
23390 (big_speedup_p): Update.
23391 (edge_badness): Update.
23392 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
23393 (size_time_entry): Replace predicate by exec_predicate and
23394 nonconst_predicate.
23395 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
23396 (estimate_edge_time): Return also nonspec_time.
23397 (reset_edge_growth_cache): Update.
23399 2017-04-29 Jakub Jelinek <jakub@redhat.com>
23401 PR rtl-optimization/80491
23402 * ifcvt.c (noce_process_if_block): When looking for x setter
23403 with missing else_bb, don't check only the insn right before
23404 cond_earliest, but look for the last insn that x is modified in
23405 within the same bb.
23407 PR rtl-optimization/80491
23408 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
23410 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
23412 PR tree-optimization/80487
23413 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
23415 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23417 PR tree-optimization/79697
23418 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
23419 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
23420 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
23422 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
23423 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
23425 2017-04-28 Martin Sebor <msebor@redhat.com>
23427 PR tree-optimization/80523
23428 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
23429 (init_target_to_host_charmap, target_to_host, target_strtol10): New
23431 (maybe_warn, format_directive, parse_directive): Use new functions.
23432 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
23434 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
23436 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
23438 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
23440 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
23441 target_header_dir): Set correctly.
23442 * configure: Regenerated.
23443 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23444 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23445 instead of SYSTEM_HEADER_DIR.
23447 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
23449 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
23450 (estimate_local_effects): Likewise.
23451 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
23452 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
23453 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
23454 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
23455 do_estimate_edge_time, estimate_edge_time): Likewise.
23456 * ipa-inline-analysis.c (estimate_node_size_and_time,
23457 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
23458 (estimate_time_after_inlining): Remove.
23460 2017-04-28 Martin Liska <mliska@suse.cz>
23462 * doc/gcov.texi: Enhance documentation of gcov.
23464 2017-04-28 Martin Liska <mliska@suse.cz>
23466 * doc/gcov.texi: Sort options in alphabetic order.
23467 * doc/gcov-dump.texi: Likewise.
23468 * doc/gcov-tool.texi: Likewise.
23469 * gcov.c (print_usage): Likewise.
23470 * gcov-dump.c (print_usage): Likewise.
23471 * gcov-tool.c (print_merge_usage_message): Likewise.
23472 (print_rewrite_usage_message): Likewise.
23473 (print_overlap_usage_message): Likewise.
23475 2017-04-28 Martin Liska <mliska@suse.cz>
23477 PR gcov-profile/53915
23478 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
23480 2017-04-28 Martin Liska <mliska@suse.cz>
23482 PR gcov-profile/79891
23483 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
23484 is marked by compiler as living on a line.
23485 (get_cycles_count): Remove usage of the union.
23486 (output_intermediate_file): Likewise.
23487 (find_source): Fix GNU coding style.
23488 (accumulate_line_counts): Remove old non-all block mode.
23489 (output_lines): Remove usage of the union.
23490 * profile.c (output_location): Include all BBs, even if
23491 belonging to a same line (and file) as a previous BB.
23493 2017-04-28 Martin Liska <mliska@suse.cz>
23495 * gcov.c (process_args): Handle new argument 'w'.
23496 (read_graph_file): Assign ID to BBs.
23497 (output_branch_count): Display BB # if verbose flag is set.
23498 (output_lines): Likewise for arcs.
23499 (print_usage): Add '--verbose' option help.
23500 * doc/gcov.texi: Document --verbose (-w) option.
23502 2017-04-28 Martin Liska <mliska@suse.cz>
23504 * gcov.c (struct block_location_info): New struct.
23505 (process_file): Fill up the new structure.
23506 (read_graph_file): Replace usage of encoding by the newly added
23508 (add_line_counts): Likewise.
23509 (accumulate_line_counts): Remove usage of the union.
23510 (function_info::function_info): New function.
23511 (function_info::~function_info): Likewise.
23512 (process_file): Call delete instead of release_function.
23513 (release_function): Release the function.
23514 (release_structures): Call delete instead of release_function.
23515 (solve_flow_graph): Replace usage of num_blocks.
23516 (find_exception_blocks): Likewise.
23517 (output_lines): Fix GNU coding style.
23519 2017-04-28 Martin Liska <mliska@suse.cz>
23522 * coverage.c (coverage_remove_note_file): New function.
23523 * coverage.h: Declare the function.
23524 * toplev.c (finalize): Clean if an error has been seen.
23526 2017-04-28 Martin Liska <mliska@suse.cz>
23528 PR gcov-profile/80031
23529 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
23530 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
23531 * gcov.c (read_graph_file): Read just number of blocks.
23532 * profile.c (branch_prob): Do not stream 0 flags per a basic
23535 2017-04-28 Martin Liska <mliska@suse.cz>
23537 * gcov-dump.c (tag_*): Add new argument to declarations.
23538 (dump_gcov_file): Likewise.
23539 (tag_blocks): Add and use new argument depth.
23540 (tag_arcs): Likewise.
23541 (tag_lines): Likewise.
23542 (tag_counters): Likewise.
23543 (tag_summary): Likewise.
23544 (dump_working_sets): Use depth to do a proper indentation.
23546 2017-04-28 Jakub Jelinek <jakub@redhat.com>
23549 * cgraph.h (symtab_node::debug_symtab): No longer inline.
23550 * symtab.c (symtab_node::debug_symtab): Move definition here.
23552 2017-04-28 Richard Biener <rguenther@suse.de>
23554 * lto-streamer.h (LTO_major_version): Bump to 7.
23556 2017-04-28 Richard Biener <rguenther@suse.de>
23558 * tree-vrp.c (assert_info): New struct.
23559 (add_assert_info): New helper.
23560 (register_edge_assert_for_2): Refactor to add asserts to a vector
23562 (register_edge_assert_for_1): Likewise.
23563 (register_edge_assert_for): Likewise.
23564 (finish_register_edge_assert_for): New helper actually registering
23565 asserts where live on edge.
23566 (find_conditional_asserts): Adjust.
23567 (find_switch_asserts): Likewise.
23568 (evrp_dom_walker::try_find_new_range): Generalize.
23569 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
23571 2017-04-27 Marek Polacek <polacek@redhat.com>
23574 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
23575 arg10 and arg11 to itype.
23577 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
23579 * doc/extend.texi (Object Size Checking): Improve grammar.
23581 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
23584 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
23585 that the logic for permitting reciprocal estimates matches that
23588 2017-04-27 Jakub Jelinek <jakub@redhat.com>
23591 * tree.c (type_cache_hasher::equal): Only compare
23592 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
23593 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
23594 non-aggregate element types.
23595 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
23596 about the flag on ARRAY_TYPEs in the comment, formatting fix.
23598 2017-04-27 Richard Biener <rguenther@suse.de>
23600 PR middle-end/80533
23601 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
23602 stripping ARRAY_REFs from MEM_EXPR make sure we're not
23603 keeping a reference to a trailing array.
23605 2017-04-27 Richard Biener <rguenther@suse.de>
23607 PR middle-end/80539
23608 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
23609 being in loop-closed SSA form conservatively.
23610 (chrec_fold_multiply_poly_poly): Likewise.
23612 2017-04-27 Tamar Christina <tamar.christina@arm.com>
23614 PR middle-end/79665
23615 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
23616 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
23618 2017-04-27 Jakub Jelinek <jakub@redhat.com>
23621 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
23622 (aarch64_function_arg_alignment): Return unsigned int again, but still
23623 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
23624 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
23625 Don't emit -Wpsabi note.
23626 (aarch64_function_arg_boundary): Likewise.
23627 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
23630 2017-04-26 Nathan Sidwell <nathan@acm.org>
23632 * tree.h (crc32_unsigned_n): Declare.
23633 (crc32_unsigned, crc32_unsigned): Make inline.
23634 * tree.c (crc32_unsigned_bits): Replace with ...
23635 (crc32_unsigned_n): ... this.
23636 (crc32_unsigned, crc32_byte): Remove.
23637 (crc32_string): Remove unnecessary braces.
23639 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
23641 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
23642 * ipa-inline-analysis.c (MAX_TIME): Remove.
23643 (account_size_time): Use sreal for time.
23644 (dump_inline_summary): Update.
23645 (estimate_function_body_sizes): Update.
23646 (estimate_edge_size_and_time): Update.
23647 (estimate_calls_size_and_time): Update.
23648 (estimate_node_size_and_time): Update.
23649 (inline_merge_summary): Update.
23650 (inline_update_overall_summary): Update.
23651 (estimate_time_after_inlining): Update.
23652 (inline_read_section): Update.
23653 (inline_write_summary): Update.
23654 * ipa-inline.c (compute_uninlined_call_time): Update.
23655 (compute_inlined_call_time): Update.
23656 (recursive_inlining): Update.
23657 (inline_small_functions): Update.
23658 (dump_overall_stats): Update.
23659 * ipa-inline.h: Include sreal.h.
23660 (size_time_entry): Turn time to sreal.
23661 (inline_summary): Turn self_time nad time to sreal.
23663 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
23665 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
23667 (sreal::stream_out, sreal::stream_in): New.
23668 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
23670 2017-04-25 Jakub Jelinek <jakub@redhat.com>
23672 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
23675 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
23678 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
23680 (dimode_scalar_chain::compute_convert_gain): Ditto.
23681 (dimode_scalar_chain::make_vector_copies): Ditto.
23682 (dimode_scalar_chain::convert_reg): Ditto.
23683 (dimode_scalar_chain::convert_insn): Ditto.
23684 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
23685 (VI248_AVX512BW_1): New mode iterator.
23686 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
23687 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
23690 2017-04-25 Martin Sebor <msebor@redhat.com>
23692 PR tree-optimization/80497
23693 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
23694 constants are representable in HOST_WIDE_INT.
23695 (parse_directive): Ditto.
23697 2017-04-25 Martin Sebor <msebor@redhat.com>
23700 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
23701 (new_zero_array): Adjust signature.
23702 (dom_info::dom_init): Used unsigned rather that size_t.
23703 (dom_info::dom_info): Same.
23705 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23706 Jakub Jelinek <jakub@redhat.com>
23709 * config/arm/arm.c: Include gimple.h.
23710 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
23711 returns negative, increment ncrn only if it returned positive.
23712 (arm_needs_doubleword_align): Return int instead of bool,
23713 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
23714 members, but if there is any such non-FIELD_DECL
23715 > PARM_BOUNDARY aligned decl, return -1 instead of false.
23716 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
23717 returns negative, increment nregs only if it returned positive.
23718 (arm_setup_incoming_varargs): Likewise.
23719 (arm_function_arg_boundary): Emit -Wpsabi note if
23720 arm_needs_doubleword_align returns negative, return
23721 DOUBLEWORD_ALIGNMENT only if it returned positive.
23723 2017-04-25 Marek Polacek <polacek@redhat.com>
23726 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
23727 first argument to type.
23729 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
23732 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
23733 type checks to test for compatibility instead of equality.
23735 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23736 Jakub Jelinek <jakub@redhat.com>
23739 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
23741 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
23742 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
23743 the alignment computation, but return their maximum in warn_alignment.
23744 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
23745 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
23747 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
23748 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
23751 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23753 * config/arc/simdext.md (dmpyh): Fix typo.
23755 2017-04-25 Richard Biener <rguenther@suse.de>
23757 PR tree-optimization/80492
23758 * alias.c (compare_base_decls): Handle registers with asm
23759 specification conservatively.
23760 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
23761 compare_base_decls returning dont-know properly.
23763 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23765 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
23766 (legitimate_offset_address_p): New function.
23767 (arc_legitimate_address_p): Use above function.
23769 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23771 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
23773 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23775 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
23776 ACCH registers whenever they are available.
23778 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23780 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
23781 double regs fix when not used.
23783 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23785 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
23787 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
23788 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
23790 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23792 * config/arc/arc.c (arc_output_addsi): Check for h-register class
23793 when emitting short ADD instructions.
23795 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23797 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
23799 (cmpsi_cc_c_insn): Likewise.
23800 (cbranchsi4_scratch): Compute proper instruction length using
23801 compact_hreg_operand.
23802 * config/arc/predicates.md (compact_hreg_operand): New predicate.
23804 2017-04-25 Richard Biener <rguenther@suse.de>
23806 PR middle-end/80509
23807 * passes.c (pass_manager::pass_manager): Initialize
23808 m_name_to_pass_map.
23810 2017-04-25 Richard Biener <rguenther@suse.de>
23812 PR tree-optimization/79201
23813 * tree-ssa-sink.c (statement_sink_location): Handle calls.
23815 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23818 * config/s390/vector.md: Split MEM->GPR vector moves for
23819 non-s_operand addresses.
23821 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23824 * config/s390/predicates.md (reload_const_wide_int_operand): New
23826 * config/s390/s390.md ("movti"): Remove d/P alternative.
23827 ("movti_bigconst"): New pattern definition.
23829 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23832 * s390-protos.h (s390_expand_cs_hqi): Removed.
23833 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
23834 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
23835 modes as well as CCZ1mode and CCZmode.
23836 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
23837 signature of s390_emit_compare_and_swap.
23838 (s390_expand_cs_hqi): Likewise, make static.
23839 (s390_expand_cs_tdsi): Generate an explicit compare before trying
23840 compare-and-swap, in some cases.
23841 (s390_expand_cs): Wrapper function.
23842 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
23844 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
23845 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
23846 patterns for small and large integers. Forbid symref memory operands.
23847 Move expander to s390.c. Require cc register.
23848 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
23849 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
23850 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
23851 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
23852 symref memory operands. Remove CC mode and call s390_match_ccmode
23854 ("atomic_exchange<mode>"): Allow and implement all integer modes.
23856 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23858 * config/s390/s390.md (define_peephole2): New peephole to help
23859 combining the load-and-test pattern with volatile memory.
23861 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23863 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
23864 with CCZmode for TARGET_Z196.
23866 2017-04-25 Jakub Jelinek <jakub@redhat.com>
23868 PR rtl-optimization/80501
23869 * combine.c (make_compound_operation_int): Set subreg_code to SET
23870 even for AND with mask of the sign bit of mode.
23872 PR rtl-optimization/80500
23873 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
23874 sum's initial value.
23876 2017-04-25 Julian Brown <julian@codesourcery.com>
23877 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23879 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
23881 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
23883 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
23885 2017-04-25 Julian Brown <julian@codesourcery.com>
23886 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23888 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
23889 (thunderx2t99_sha): New Reservation.
23891 2017-04-25 Julian Brown <julian@codesourcery.com>
23892 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23894 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
23895 type for 1-element load.
23897 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
23899 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
23901 2017-04-24 Martin Jambor <mjambor@suse.cz>
23903 PR tree-optimization/80293
23904 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
23905 char arrays not totally scalarizable if it is false.
23906 (analyze_all_variable_accesses): Pass correct value in the new
23907 parameter. Add a statistics counter.
23909 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
23911 PR middle-end/79931
23912 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
23914 2017-04-24 Richard Biener <rguenther@suse.de>
23916 PR tree-optimization/80494
23917 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
23918 out for complex types.
23920 2017-04-24 Richard Biener <rguenther@suse.de>
23922 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
23923 * tree-ssa-sccvn.c (print_scc): Print SCC size.
23924 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
23925 (DFS): Adjust and never fail.
23926 (sccvn_dom_walker::fail): Remove.
23927 (sccvn_dom_walker::before_dom_children): Adjust.
23928 (run_scc_vn): Likewise and never fail.
23929 * tree-ssa-pre.c (pass_pre::execute): Adjust.
23930 (pass_fre::execute): Likewise.
23932 2017-04-24 Richard Biener <rguenther@suse.de>
23934 PR tree-optimization/79725
23935 * tree-ssa-sink.c (statement_sink_location): Return whether
23936 failure reason was zero uses. Move that check later.
23937 (sink_code_in_bb): Deal with zero uses by removing the stmt
23940 2017-04-24 Richard Biener <rguenther@suse.de>
23943 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
23944 pointer-based references.
23946 2017-04-24 Richard Biener <rguenther@suse.de>
23949 * pass_manager.h (pass_manager::operator new): Remove.
23950 (pass_manager::operator delete): Likewise.
23951 * passes.c (pass_manager::operator new): Remove.
23952 (pass_manager::operator delete): Likewise.
23953 (pass_manager::pass_manager): Zero individual pass members.
23955 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
23958 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
23959 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
23960 Check "XEXP (src, 1)" operand here.
23961 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
23962 Check "XEXP (src, 1)" operand here.
23963 (dimode_scalar_chain::make_vector_copies): Detect count register
23964 of a shift instruction. Zero extend count register from QImode
23965 to DImode to satisfy vector shift pattern count operand predicate.
23966 Substitute vector shift count operand with a DImode copy.
23967 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
23970 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
23972 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
23973 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
23974 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
23975 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
23976 (UNSPEC_NOREX_MEM): Remove definition.
23978 2017-04-21 Richard Biener <rguenther@suse.de>
23980 PR tree-optimization/79547
23981 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
23982 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
23983 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
23984 without any constraints.
23986 2017-04-21 Richard Biener <rguenther@suse.de>
23988 PR tree-optimization/78847
23989 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
23991 2017-04-21 Richard Biener <rguenther@suse.de>
23993 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
23994 (build_distinct_type_copy): Likewise.
23995 (build_variant_type_copy): Likewise.
23996 * tree.c (build_qualified_type): Pass down mem-stat info.
23997 (build_distinct_type_copy): Likewise.
23998 (build_variant_type_copy): Likewise.
24000 2017-04-21 Richard Biener <rguenther@suse.de>
24002 PR tree-optimization/80237
24003 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
24005 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
24006 for a simplified result.
24008 2016-04-21 Richard Biener <rguenther@suse.de>
24010 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
24011 sth as strict as a simple_iv but a chrec without symbols and an
24012 operand defined in the loop we are peeling (and not some subloop).
24013 (propagate_constants_for_unrolling): Propagate all constants.
24015 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
24018 * config/i386/i386.c (print_reg): Remove assert for disalowed
24019 regno values, call output_operand_lossage instead.
24021 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
24024 * config/i386/constraints.md (Yc): New register constraint.
24025 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
24026 Use Yc constraint for alternative 2 of operand 0. Remove
24027 preferred_for_speed attribute.
24029 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
24031 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
24032 lastprivate clauses in SIMT case.
24034 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
24036 * doc/invoke.texi (-Wextra-semi): Document new warning option.
24038 2017-04-20 Richard Biener <rguenther@suse.de>
24040 PR tree-optimization/57796
24041 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
24042 as N scalar stores.
24043 (vect_model_load_cost): Cost gathers as N scalar loads.
24045 2017-04-20 Richard Biener <rguenther@suse.de>
24047 * ggc-page.c (ggc_allocated_p): Rename to ...
24048 (safe_lookup_page_table_entry): ... this and return the lookup
24050 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
24052 2017-04-20 Richard Biener <rguenther@suse.de>
24054 PR tree-optimization/80453
24055 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
24056 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
24057 from the conditions.
24058 (vn_phi_eq): Pass them down.
24059 (vn_phi_lookup): Record them.
24060 (vn_phi_insert): Likewise.
24062 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
24064 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
24065 uninitialized variable warning to avoid buffer overrun.
24067 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
24070 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
24071 is suppressed for '{ 0 }' in C.
24073 2017-04-20 Jakub Jelinek <jakub@redhat.com>
24075 * BASE-VER: Set to 8.0.0.
24077 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
24079 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
24080 priority .init_array and .fini_array section with SECTION_NOTYPE
24083 2017-04-20 Jakub Jelinek <jakub@redhat.com>
24085 PR middle-end/80423
24086 * tree.h (build_array_type): Add typeless_storage default argument.
24087 * tree.c (type_cache_hasher::equal): Also compare
24088 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
24089 (build_array_type): Add typeless_storage argument, set
24090 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
24092 (build_nonshared_array_type): Adjust build_array_type_1 caller.
24093 (build_array_type): Likewise. Add typeless_storage argument.
24095 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
24096 Jakub Jelinek <jakub@redhat.com>
24098 PR tree-optimization/80426
24099 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
24100 operation on symbolic operands, also compute the overflow for the
24101 invariant part when the operation degenerates into a negation.
24103 2017-04-19 Jakub Jelinek <jakub@redhat.com>
24106 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
24107 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
24110 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
24112 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
24115 * config/avr/avr.c (tree.h): Include it.
24116 (cgraph.h): Include it.
24117 (avr_encode_section_info): Don't warn for uninitialized progmem
24118 variable if it's just an alias.
24120 2017-04-19 Richard Biener <rguenther@suse.de>
24123 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
24124 when needed by AutoPGO.
24126 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
24129 * doc/lto.texi: Remove an extra 'that'.
24131 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
24133 PR rtl-optimization/80429
24134 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
24135 are only used in debug insns.
24137 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
24138 Vladimir Makarov <vmakarov@redhat.com>
24140 * config/sparc/predicates.md (input_operand): Add comment. Return
24141 true for any memory operand when LRA is in progress.
24142 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
24144 2017-04-18 Jeff Law <law@redhat.com>
24147 * mips.md ({return,simple_return}_internal): Do not overwrite
24150 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24152 PR tree-optimization/80443
24153 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
24154 instead of adding 1, subtract -1 and similarly instead of subtracting
24157 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
24159 PR rtl-optimization/80357
24160 * haifa-sched.c (tmp_bitmap): New variable.
24161 (model_recompute): Handle duplicate use records.
24162 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
24163 (free_global_sched_pressure_data): Free it.
24165 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
24168 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
24169 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24170 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24171 instead of SYSTEM_HEADER_DIR.
24173 2017-04-18 Jeff Law <law@redhat.com>
24175 PR middle-end/80422
24176 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
24177 predecessors after walking up the insn chain.
24179 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24182 * dwarf2out.c (modified_type_die): Try harder not to emit internal
24183 sizetype type into debug info.
24185 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24188 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
24189 unneeded test for TARGET_UPPER_REGS_SF.
24190 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
24192 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24195 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
24196 instead of gsi_after_labels.
24198 2017-04-18 Jeff Law <law@redhat.com>
24200 * regcprop.c (maybe_mode_change): Avoid creating copies of the
24204 2017-04-13 Jeff Law <law@redhat.com>
24205 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
24206 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
24208 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
24211 * config/avr/avr.c (intl.h): Include it.
24212 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
24214 2017-04-18 Martin Liska <mliska@suse.cz>
24216 PR gcov-profile/78783
24217 * gcov-tool.c (gcov_output_files): Validate that destination
24218 file is either removed by the tool or by a user.
24220 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
24221 Guy Benyei <guybe@mellanox.com>
24223 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
24224 block, and do not negate it, the stored id is already negative.
24226 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
24228 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
24230 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
24233 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
24234 masks of options that should be turned off if the VSX vector
24235 options are turned off.
24236 (OTHER_P8_VECTOR_MASKS): Likewise.
24237 (OTHER_VSX_VECTOR_MASKS): Likewise.
24238 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
24239 rs6000_disable_incompatible_switches to validate no type switches
24241 (rs6000_incompatible_switch): New function to disallow turning on
24242 other vector options if -mno-vsx, -mno-power8-vector, or
24243 -mno-power9-vector are specified.
24245 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24247 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
24249 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24251 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
24252 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
24253 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
24254 (ARG_POINTER_CFA_OFFSET): Likewise.
24256 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24258 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
24259 conditions to take advantage of various optimizations.
24261 2017-04-13 Jeff Law <law@redhat.com>
24263 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
24264 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
24265 (zero_extendsidi2_dext): Likewise.
24267 2017-04-13 Jakub Jelinek <jakub@redhat.com>
24270 * fold-const.c (fold_ternary_loc): Revert
24271 use op0 instead of fold_convert_loc (loc, type, arg0) part of
24274 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
24276 PR rtl-optimization/80343
24277 * lra-remat.c (update_scratch_ops): Assign original hard reg to
24278 new scratch pseudo.
24280 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
24283 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
24284 to ubsan_encode_value.
24286 2017-04-13 Jeff Law <law@redhat.com>
24288 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
24289 appearing in DEBUG_INSNs.
24291 2017-04-13 Martin Liska <mliska@suse.cz>
24293 PR gcov-profile/80413
24294 * gcov-io.c (gcov_write_string): Copy to buffer just when
24295 allocated size is greater than zero.
24297 2017-04-13 Jakub Jelinek <jakub@redhat.com>
24300 * dwarf2out.c (decls_for_scope): Ignore declarations of
24301 current_function_decl in BLOCK_NONLOCALIZED_VARS.
24303 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
24306 * ipa-visibility.c (non_local_p): Fix typos.
24307 (localize_node): When localizing symbol in same comdat group,
24308 dissolve the group only when we know external symbols are going
24310 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
24312 2017-04-12 Jakub Jelinek <jakub@redhat.com>
24314 PR tree-optimization/79390
24315 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
24316 order does not result in usable sequence, retry with reversed operand
24322 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
24323 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
24324 op0 instead of fold_convert_loc (loc, type, arg0).
24326 2017-04-12 Jeff Law <law@redhat.com>
24328 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
24329 has a delay slot in the generated code.
24331 * config/cris/cris.md (cris_preferred_reload_class): Return
24332 GENNONACR_REGS rather than GENERAL_REGS.
24334 2017-04-12 Jakub Jelinek <jakub@redhat.com>
24337 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
24338 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
24339 signedness of the result type.
24341 2017-04-12 Richard Biener <rguenther@suse.de>
24342 Jeff Law <law@redhat.com>
24344 PR tree-optimization/80359
24345 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
24346 trim stores to TARGET_MEM_REFs.
24348 2017-04-12 Richard Biener <rguenther@suse.de>
24350 PR tree-optimization/79390
24351 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
24352 threading case even more.
24354 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
24357 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
24358 for quad_address_p for TImode, instead of just not indexed_address.
24360 2017-04-12 Richard Biener <rguenther@suse.de>
24361 Bernd Edlinger <bernd.edlinger@hotmail.de>
24363 PR middle-end/79671
24364 * alias.c (component_uses_parent_alias_set_from): Handle
24365 TYPE_TYPELESS_STORAGE.
24366 (get_alias_set): Likewise.
24367 * tree-core.h (tree_type_common): Add typeless_storage flag.
24368 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
24369 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
24370 for types containing members with TYPE_TYPELESS_STORAGE.
24371 (place_field): Likewise.
24372 (layout_type): Likewise for ARRAY_TYPE.
24373 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
24374 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
24375 TYPE_TYPELESS_STORAGE.
24376 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
24378 2017-04-12 Jakub Jelinek <jakub@redhat.com>
24381 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
24382 first argument to type.
24384 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24388 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
24389 CONST0_RTX (mode) rather than const0_rtx where appropriate.
24390 (rs6000_expand_binop_builtin): Likewise.
24391 (rs6000_expand_ternop_builtin): Likewise; also add missing
24392 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
24393 vshasigma built-ins.
24394 * doc/extend.texi: Document that vec_xxpermdi's third argument
24395 must be a constant.
24397 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
24399 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
24400 Use shift_const cost parameter when calculating gain of STV shifts.
24402 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
24404 PR rtl-optimization/70478
24405 * lra-constraints.c (process_alt_operands): Check memory for
24406 disfavoring memory insn operand.
24408 2017-04-11 Jakub Jelinek <jakub@redhat.com>
24410 PR middle-end/80100
24411 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
24412 left shift in unsigned HOST_WIDE_INT type.
24414 PR rtl-optimization/80385
24415 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
24416 (not (neg X)) into (plus X -1) for complex or non-integral modes.
24419 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
24420 if they have any depend clauses.
24422 2017-04-11 Martin Liska <mliska@suse.cz>
24425 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
24426 * ipa-split.c (split_function): Create a local comdat symbol
24427 if caller is in a comdat group.
24429 2017-04-11 Martin Liska <mliska@suse.cz>
24432 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
24435 2017-04-11 Martin Sebor <msebor@redhat.com>
24437 PR middle-end/80364
24438 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
24439 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
24441 (directive::set_width, directive::set_precision, format_character):
24443 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
24446 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
24449 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
24450 conflict, set target->arch_name instead of target->cpu_name.
24452 2017-04-11 Richard Biener <rguenther@suse.de>
24454 PR tree-optimization/80374
24455 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
24456 build_zero_cst, remove fold_convertible_p check again.
24458 2017-04-11 Martin Liska <mliska@suse.cz>
24461 * ubsan.c (instrument_object_size): Do not instrument register
24464 2017-04-11 Jakub Jelinek <jakub@redhat.com>
24467 * config/i386/i386-builtin-types.def
24468 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
24469 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
24470 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
24471 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
24472 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
24473 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
24474 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
24475 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
24476 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
24477 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
24478 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
24479 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
24480 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
24481 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
24482 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
24483 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
24484 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
24485 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
24486 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
24487 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
24488 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
24489 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
24490 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
24491 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
24492 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
24493 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
24494 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
24495 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
24496 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
24497 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
24498 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
24499 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
24500 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
24501 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
24502 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
24503 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
24504 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
24505 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
24506 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
24507 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
24508 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
24509 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
24510 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
24511 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
24512 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
24513 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
24515 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
24516 flag to second_arg_count, handle 4 argument function type _COUNT
24517 aliases, handle second_arg_count on second argument rather than last.
24519 2017-04-10 Jeff Law <law@redhat.com>
24521 PR tree-optimization/80374
24522 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
24523 record anything if we can not convert integer_zero_node to the
24526 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24529 * config/rs6000/rs6000.c (rs6000_option_override_internal):
24530 Enhance special handling given to the TARGET_P9_MINMAX option in
24531 relation to certain other options.
24533 2017-04-10 Bin Cheng <bin.cheng@arm.com>
24535 PR tree-optimization/80153
24536 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
24537 remove POINTER_PLUS_EXPR's base part directly, rather than through
24540 2017-04-10 Richard Biener <rguenther@suse.de>
24541 Bin Cheng <bin.cheng@arm.com>
24543 PR tree-optimization/80153
24544 * tree-affine.c (aff_combination_to_tree): Get base pointer from
24545 the first element of pointer type aff_tree. Build result expr in
24547 (add_elt_to_tree): Convert to type unconditionally. Remove other
24548 fold_convert calls.
24549 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
24550 (rewrite_use_nonlinear_expr): Check invariant using iv information.
24552 2017-04-10 Richard Biener <rguenther@suse.de>
24554 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24557 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
24559 PR rtl-optimization/70478
24560 * lra-constraints.c (curr_small_class_check): New.
24561 (update_and_check_small_class_inputs): New.
24562 (process_alt_operands): Update curr_small_class_check. Disfavor
24563 alternative insn memory operands. Check available regs for small
24566 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
24569 * config/mips/mips.opt (-mvirt): Update description.
24570 * doc/invoke.texi (-mvirt): Likewise.
24572 2017-04-10 Richard Biener <rguenther@suse.de>
24574 PR middle-end/80362
24575 * fold-const.c (fold_binary_loc): Look at unstripped ops when
24576 looking for NEGATE_EXPR in -A / -B to A / B folding.
24578 2017-04-10 Martin Liska <mliska@suse.cz>
24580 PR gcov-profile/80224
24581 * gcov.c (print_usage): Fix usage string.
24582 (get_gcov_intermediate_filename): Remove.
24583 (output_gcov_file): Use both for normal and intermediate format.
24584 (generate_results): Do not initialize special file for
24585 intermediate format.
24587 2017-04-10 Richard Biener <rguenther@suse.de>
24589 PR tree-optimization/80304
24590 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
24593 2017-04-10 Nathan Sidwell <nathan@acm.org>
24596 * config/rs6000/rs6000.c (rs6000_vector_type): New.
24597 (rs6000_init_builtins): Use it.
24599 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24601 * config/arm/arm.md (<mrc>): Add mode to SET source.
24602 (<mrrc>): Likewise.
24604 2017-04-10 Richard Biener <rguenther@suse.de>
24606 PR middle-end/80344
24607 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
24609 2017-04-10 Jakub Jelinek <jakub@redhat.com>
24612 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
24613 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
24614 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
24615 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
24616 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
24617 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
24618 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
24619 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
24620 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
24621 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
24622 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
24623 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
24624 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
24625 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
24626 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
24627 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
24628 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
24629 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
24630 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
24631 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
24632 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
24633 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
24634 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
24636 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
24638 PR rtl-optimization/70478
24639 * lra-constraints.c: Reverse the last patch.
24641 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
24643 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
24644 Add comment for WCHAR_T.
24646 2017-04-08 Martin Liska <mliska@suse.cz>
24649 2017-04-07 Martin Liska <mliska@suse.cz>
24652 * ipa-split.c (split_function): Add function part to a same comdat
24655 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24658 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
24660 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
24662 * rs6000/rs6000.c (vec_load_pendulum): Rename...
24663 (vec_pairing): ...to this.
24664 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
24665 (rs6000_sched_init): Adjust for name change.
24666 (struct rs6000_sched_context): Likewise.
24667 (rs6000_init_sched_context): Likewise.
24668 (rs6000_set_sched_context): Likewise.
24670 2017-04-07 Jakub Jelinek <jakub@redhat.com>
24676 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
24678 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
24679 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
24680 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
24682 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
24684 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
24686 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
24688 PR rtl-optimization/70703
24689 * ira-color.c (update_conflict_hard_regno_costs): Use
24690 int64_t instead of HOST_WIDE_INT.
24692 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
24694 PR rtl-optimization/70478
24695 * lra-constraints.c (process_alt_operands): Disfavor alternative
24696 insn memory operands.
24698 2017-04-07 Jeff Law <law@redhat.com>
24700 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
24701 CALL and NOTE_INSN_CALL_ARG_LOCATION.
24703 2017-04-07 Martin Liska <mliska@suse.cz>
24706 * config/aarch64/aarch64.c (aarch64_process_target_attr):
24707 Show error message instead of an ICE.
24709 2017-04-07 Martin Liska <mliska@suse.cz>
24712 * ipa-split.c (split_function): Add function part to a same comdat
24715 2017-04-07 Richard Biener <rguenther@suse.de>
24717 PR middle-end/80341
24718 * tree.c (get_unwidened): Also handle ! for_type case for
24720 * convert.c (do_narrow): Split out from ...
24721 (convert_to_integer_1): ... here. Do not pass final truncation
24722 type to get_unwidened for TRUNC_DIV_EXPR.
24724 2017-04-07 Richard Biener <rguenther@suse.de>
24726 * tree-affine.c (wide_int_ext_for_comb): Take type rather
24728 (aff_combination_const): Adjust.
24729 (aff_combination_scale): Likewise.
24730 (aff_combination_add_elt): Likewise.
24731 (aff_combination_add_cst): Likewise.
24732 (aff_combination_convert): Likewise.
24733 (add_elt_to_tree): Likewise. Remove unused argument.
24734 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
24736 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
24738 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
24740 * config/arm/arm.c (arm_default_short_enums): Use
24741 ARM_DEFAULT_SHORT_ENUMS.
24742 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
24744 2017-04-06 Jakub Jelinek <jakub@redhat.com>
24747 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
24748 members with redundant out-of-class redeclaration.
24750 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
24753 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
24754 * config/i386/i386.md (*zero_extendsidi2):
24755 Add (?*x,*x) and (?*v,*v) alternatives.
24757 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
24760 * config/i386/i386.c (ix86_expand_builtin)
24761 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
24762 mode from insn data. Convert operands to insn operand mode.
24763 Copy operands that don't satisfy insn predicate to a register.
24765 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
24767 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
24770 2017-04-06 Richard Biener <rguenther@suse.de>
24772 PR tree-optimization/80334
24773 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
24774 preserve alignment of accesses.
24776 2017-04-06 Richard Biener <rguenther@suse.de>
24778 PR tree-optimization/80262
24779 * tree-sra.c (build_ref_for_offset): Preserve address-space
24781 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
24782 Drop useless address-space information on MEM_REF offsets.
24784 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
24786 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
24788 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
24790 PR rtl-optimization/70703
24791 * ira-color.c (update_conflict_hard_regno_costs): Use
24792 HOST_WIDE_INT instead of long.
24794 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
24797 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
24798 not defined for x86_64 target. Add -mmmx target option when __SSE2__
24800 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
24801 for x86_64 target. Handle -m3dnowa option.
24803 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
24805 PR rtl-optimization/70703
24806 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
24807 (update_conflict_hard_regno_costs): Use long instead of unsigned
24808 arithmetic for cost calculation.
24810 2017-04-05 Jakub Jelinek <jakub@redhat.com>
24811 Bernd Edlinger <bernd.edlinger@hotmail.de>
24814 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
24817 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
24820 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
24821 ptr_mode with Pmode throughout.
24822 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
24823 into probe_stack_range and use DImode.
24825 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
24828 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
24829 call_eh_return is true.
24831 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24833 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
24834 Initialize last_match_fntype_index.
24836 2017-04-05 Jakub Jelinek <jakub@redhat.com>
24839 * tree-nvr.c: Include internal-fn.h.
24840 (pass_return_slot::execute): Ignore internal calls without
24843 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24844 Richard Biener <rguenther@suse.de>
24847 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
24848 captures used multiple times, except for the last use.
24849 * generic-match-head.c: Include gimplify.h.
24851 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24853 PR tree-optimization/79390
24854 * target.h (struct noce_if_info): Declare.
24855 * targhooks.h (default_noce_conversion_profitable_p): Declare.
24856 * target.def (noce_conversion_profitable_p): New target hook.
24857 * ifcvt.h (struct noce_if_info): New type, moved from ...
24858 * ifcvt.c (struct noce_if_info): ... here.
24859 (noce_conversion_profitable_p): Renamed to ...
24860 (default_noce_conversion_profitable_p): ... this. No longer
24862 (noce_try_store_flag_constants, noce_try_addcc,
24863 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
24864 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
24865 instead of noce_conversion_profitable_p.
24866 * config/i386/i386.c: Include ifcvt.h.
24867 (ix86_option_override_internal): Don't override
24868 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
24869 (ix86_noce_conversion_profitable_p): New function.
24870 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
24871 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
24872 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
24873 * doc/tm.texi: Regenerated.
24875 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24877 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
24880 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24883 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
24884 instructions for small multiply cores.
24886 2017-04-04 Jeff Law <law@redhat.com>
24888 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
24890 (mips_expand_vec_perm_const): Initialize elements in orig_perm
24891 that are not set by the loop over the elements.
24893 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24896 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
24897 int mode, convert_modes it to mode as unsigned, otherwise use
24898 lowpart_subreg to mode rather than SImode.
24899 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
24900 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
24901 Use DImode instead of SImode for the shift count operand.
24902 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
24905 2017-04-04 Richard Biener <rguenther@suse.de>
24907 PR middle-end/80281
24908 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
24909 arithmetic done for the negate or the plus. Simplify.
24910 (A - (-B) -> A + B): Likewise.
24911 * fold-const.c (split_tree): Make sure to not negate pointers.
24913 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
24915 PR rtl-optimization/60818
24916 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
24917 a compare of comparisons with the thing compared if this results
24918 in a different machine mode.
24920 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
24922 * alias.c (base_alias_check): Fix typo in comment.
24923 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
24924 * cgraphunit.c (symbol_table::compile): Likewise.
24925 * collect2.c (maybe_run_lto_and_relink): Likewise.
24926 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
24927 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
24928 * config/avr/avr.c (avr_map_op_t): Likewise.
24929 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
24930 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
24931 * config/epiphany/epiphany.md (movcc): Likewise.
24932 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
24933 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
24935 * config/mips/mips.c (mips_save_restore_reg): Likewise.
24936 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
24937 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
24938 * config/sh/sh.c (sh_rtx_costs): Likewise.
24939 * fold-const.c (fold_truth_andor): Likewise.
24940 * genautomata.c (collapse_flag): Likewise.
24941 * gengtype.h (struct type::u::s): Likewise.
24942 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
24943 * input.c (FORMAT_AMOUNT): Likewise.
24944 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
24945 (known_aggs_to_agg_replacement_list): Likewise.
24946 * ipa-inline-analysis.c: Likewise.
24947 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
24948 * ipa-polymorphic-call.c
24949 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
24950 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
24951 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
24953 * modulo-sched.c (apply_reg_moves): Likewise.
24954 * omp-expand.c (build_omp_regions_1): Likewise.
24955 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
24956 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
24957 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
24958 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
24959 * value-prof.c: Likewise.
24960 * var-tracking.c (val_reset): Likewise.
24962 2017-04-03 Richard Biener <rguenther@suse.de>
24964 PR tree-optimization/80275
24965 * fold-const.c (split_address_to_core_and_offset): Handle
24968 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
24970 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
24971 descriptors is at least equal to that of functions.
24973 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
24975 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
24977 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
24980 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
24981 (mov<IMOD4:mode>): New expander.
24982 (*mov<IMOD4:mode>_internal): New insn and split pattern.
24984 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
24986 PR rtl-optimization/79405
24987 * fwprop.c (propagations_left): New variable.
24988 (forward_propagate_into): Decrement it.
24989 (fwprop_init): Initialize it.
24990 (fw_prop): If the variable has reached zero, stop propagating.
24991 (fwprop_addr): Ditto.
24993 2017-03-31 Jakub Jelinek <jakub@redhat.com>
24996 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
24997 a FUNCTION_DECL, pass it as decl instead of origin to
25000 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
25002 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
25005 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
25008 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
25009 TARGET_VSX_SMALL_INTEGER.
25011 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25013 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25014 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
25016 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
25018 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
25019 extraction from odd-numbered MSA register.
25021 2017-03-31 Jakub Jelinek <jakub@redhat.com>
25023 PR middle-end/80173
25024 * expmed.c (store_bit_field_1): Don't attempt to create
25025 a word subreg out of hard registers wider than word if they
25026 have HARD_REGNO_NREGS of 1 for their mode.
25028 PR middle-end/80163
25029 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
25030 conversions to integer types wider than word and pointer.
25033 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
25034 (rtx_equal_for_cselib_p): Pass 0 to it.
25035 * cselib.c (cselib_hasher::equal): Likewise.
25036 (rtx_equal_for_cselib_1): Add depth argument. If depth
25037 is 128, don't look up VALUE locs and punt. Increment
25038 depth in recursive calls when walking VALUE locs.
25040 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
25042 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
25043 (make_gcov_file_name): Use the canonical path name for generating
25045 (read_line): Fix handling of files with ascii null bytes.
25047 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
25049 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
25050 to initialise a vector register instead
25051 of using a const_int.
25053 2017-03-30 Jakub Jelinek <jakub@redhat.com>
25055 PR translation/80189
25056 * gimplify.c (omp_default_clause): Use %qs instead of %s in
25057 diagnostic messages.
25059 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
25062 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
25063 (dfp_diex_<mode>): Update mode of operand 1.
25064 * doc/extend.texi (dxex, dxexq): Document change to return type.
25065 (diex, diexq): Document change to argument type.
25067 2017-03-30 Martin Jambor <mjambor@suse.cz>
25070 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
25071 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
25072 it reflects the signature changes performed at the callee side.
25073 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
25074 to cgraph_build_function_type_skip_args.
25075 (build_function_decl_skip_args): Adjust call to the above function.
25077 2017-03-30 Jakub Jelinek <jakub@redhat.com>
25080 * config/i386/sse.md
25081 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
25082 register as dest whenever it is a MEM not rtx_equal_p to the
25083 corresponding dup operand, and when forcing into reg move the
25084 reg into the memory afterwards.
25085 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
25086 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
25087 for the force_reg mode.
25088 (avx512vl_vextractf128<mode>): Use register as dest either
25089 always when a MEM, or when it is a MEM not rtx_equal_p to the
25090 corresponding dup operand, or even not when it is a CONST_VECTOR
25091 depending on the mode and lo vs. hi.
25092 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
25094 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
25095 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
25096 Likewise. Require that operands[2] is even.
25097 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
25098 Remove extraneous parens. Require that operands[2] is a multiple
25100 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
25101 operands[0] is a MEM if <mask_applied>, the predicates/constraints
25102 disallow memory then.
25104 2017-03-30 Richard Biener <rguenther@suse.de>
25106 PR tree-optimization/77498
25107 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
25108 to non-constants over backedges.
25110 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
25112 PR rtl-optimization/80233
25113 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
25114 as last_combined_insn. Do not test for BARRIER_P separately.
25116 2017-03-29 Andreas Schwab <schwab@suse.de>
25119 * calls.c (prepare_call_address): Convert funexp to Pmode before
25120 copying to temp reg.
25122 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25124 PR tree-optimization/80158
25125 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
25126 Handle possible future case of more than one alternate
25128 (replace_rhs_if_not_dup): Likewise.
25129 (replace_one_candidate): Likewise.
25131 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
25133 PR rtl-optimization/80193
25134 * ira.c (ira): Do not check allocation for LRA.
25136 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
25138 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
25139 (nvptx_output_simt_exit): Declare.
25140 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
25141 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
25142 (init_softstack_frame): Move initialization of crtl->is_leaf to...
25143 (nvptx_declare_function_name): ...here. Emit declaration of local
25144 memory space buffer for omp_simt_enter insn.
25145 (nvptx_output_unisimt_switch): New.
25146 (nvptx_output_softstack_switch): New.
25147 (nvptx_output_simt_enter): New.
25148 (nvptx_output_simt_exit): New.
25149 * config/nvptx/nvptx.h (struct machine_function): New fields
25150 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
25151 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
25152 (UNSPECV_SIMT_EXIT): Ditto.
25153 (omp_simt_enter_insn): New insn.
25154 (omp_simt_enter): New expansion.
25155 (omp_simt_exit): New insn.
25156 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
25158 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
25159 (expand_GOMP_SIMT_ENTER_ALLOC): New.
25160 (expand_GOMP_SIMT_EXIT): New.
25161 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
25162 (GOMP_SIMT_ENTER_ALLOC): Ditto.
25163 (GOMP_SIMT_EXIT): Ditto.
25164 * target-insns.def (omp_simt_enter): New insn.
25165 (omp_simt_exit): Ditto.
25166 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
25168 (lower_rec_simd_input_clauses): Implement SIMT privatization.
25169 (lower_rec_input_clauses): Likewise.
25170 (lower_lastprivate_clauses): Handle SIMT privatization.
25172 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
25173 (ompdevlow_adjust_simt_enter): New.
25174 (find_simtpriv_var_op): New.
25175 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
25176 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
25178 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
25179 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
25180 (copy_decl_for_dup_finish): Ditto.
25182 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
25184 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
25187 * config/i386/i386.c (ix86_option_override_internal): Always
25188 allow -mpreferred-stack-boundary=3 for 64-bit targets.
25190 2017-03-28 Bin Cheng <bin.cheng@arm.com>
25192 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
25194 2017-03-28 Bin Cheng <bin.cheng@arm.com>
25196 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
25197 mark new edge's irreducible flag accordign to it.
25198 (vect_do_peeling): Check loop preheader edge's irreducible flag
25199 and pass it to function slpeel_add_loop_guard.
25201 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
25203 PR tree-optimization/80218
25204 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
25205 Update block frequencies and counts.
25207 2017-03-28 Richard Biener <rguenther@suse.de>
25209 PR tree-optimization/78644
25210 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
25211 of a simplification result we may not use it at all.
25213 2017-03-28 Richard Biener <rguenther@suse.de>
25216 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
25217 without arguments, generate default definition of a SSA name.
25219 2017-03-28 Richard Biener <rguenther@suse.de>
25221 PR middle-end/80222
25222 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
25223 TYPE_REF_CAN_ALIAS_ALL references.
25224 * fold-const.c (fold_indirect_ref_1): Likewise.
25226 2017-03-28 Martin Liska <mliska@suse.cz>
25229 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
25230 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
25232 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
25233 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
25235 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
25236 (EXTRA_SPECS): Define.
25237 (SUBTARGET_EXTRA_SPECS): Likewise.
25238 (SUBTARGET_CPP_SPEC): Likewise.
25239 * config/arc/elf.h (EXTRA_SPECS): Renamed to
25240 SUBTARGET_EXTRA_SPECS.
25241 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
25243 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
25245 * config/arc/simdext.md (vst64_insn): Update pattern.
25246 (vld32wh_insn): Likewise.
25247 (vld32wl_insn): Likewise.
25248 (vld64_insn): Likewise.
25249 (vld32_insn): Likewise.
25251 2017-03-28 Marek Polacek <polacek@redhat.com>
25254 * fold-const.c (fold_comparison): Use protected_set_expr_location
25255 instead of SET_EXPR_LOCATION.
25257 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
25259 * tree.c (add_expr): Avoid name lookup warning.
25261 2017-03-27 Jeff Law <law@redhat.com>
25263 PR tree-optimization/80216
25264 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
25265 function name. Limit recursion depth.
25266 (record_temporary_equivalences): Corresponding changes.
25268 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
25270 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
25273 2017-03-27 Jakub Jelinek <jakub@redhat.com>
25276 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
25278 * cfgcleanup.c (reg_note_cfa_p): New array.
25279 (insns_have_identical_cfa_notes): New function.
25280 (old_insns_match_p): Don't cross-jump in between /f
25281 and non-/f instructions. If both i1 and i2 are frame related,
25282 verify all CFA notes, their order and content.
25284 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
25287 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
25288 HImode and SImode with zero extend to DImode to one insn.
25289 (bswap<mode>2_extenddi): Likewise.
25290 (bswapsi2_extenddi): Likewise.
25291 (bswaphi2_extendsi): Likewise.
25292 (bswaphi2): Combine bswap HImode and SImode into one insn.
25293 Separate memory insns from swapping register.
25294 (bswapsi2): Likewise.
25295 (bswap<mode>2): Likewise.
25296 (bswaphi2_internal): Delete, no longer used.
25297 (bswapsi2_internal): Likewise.
25298 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
25299 store, and gpr<-gpr swap insns.
25300 (bswap<mode>2_store): Likewise.
25301 (bswaphi2_reg): Register only splitter, combine with the splitter.
25302 (bswaphi2 splitter): Likewise.
25303 (bswapsi2_reg): Likewise.
25304 (bswapsi2 splitter): Likewise.
25305 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
25306 the insns into load, store, and register/register insns.
25307 (bswapdi2_ldbrx): Likewise.
25308 (bswapdi2_load): Likewise.
25309 (bswapdi2_store): Likewise.
25310 (bswapdi2_reg): Likewise.
25312 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
25314 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
25315 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
25317 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
25320 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
25322 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25323 special handling for target option conflicts between dform
25324 options (-mpower9-dform, -mpower9-dform-vector,
25325 -mpower9-dform-scalar) and -mno-direct-move.
25327 2017-03-27 Richard Biener <rguenther@suse.de>
25329 PR tree-optimization/80181
25330 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
25332 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25334 * config/arc/predicates.md (move_double_src_operand): Replace the
25335 call to move_double_src_operand with a call to address_operand.
25337 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25339 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
25340 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
25341 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
25343 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25345 * config/arc/predicates.md (long_immediate_loadstore_operand):
25346 Consider scaled addresses cases.
25348 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25350 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
25351 restored when in interrupt.
25352 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
25353 doesn't have delay slot.
25355 2017-03-27 Richard Biener <rguenther@suse.de>
25358 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
25359 inlined thunk clones.
25361 2017-03-27 Jakub Jelinek <jakub@redhat.com>
25364 * asan.c (instrument_derefs): Copy over last operand from
25365 original COMPONENT_REF to the new COMPONENT_REF with
25366 DECL_BIT_FIELD_REPRESENTATIVE.
25367 * ubsan.c (instrument_object_size): Likewise.
25369 2017-03-27 Richard Biener <rguenther@suse.de>
25371 PR tree-optimization/80170
25372 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
25373 sure DR/SCEV didnt fold in constants we do not see when looking
25374 at the reference base alignment.
25376 2017-03-27 Richard Biener <rguenther@suse.de>
25378 PR middle-end/80171
25379 * gimple-fold.c (fold_ctor_reference): Properly guard against
25380 NULL return value from canonicalize_constructor_val.
25382 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
25385 * config/i386/i386.c (ix86_expand_builtin)
25386 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
25387 flags reg setting and flags reg using instructions.
25388 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
25389 clobbering instructions to zero extend op2.
25391 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
25393 * doc/install.texi (Configuration) <--with-aix-soname>:
25394 Update link to AIX ld.
25396 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
25398 PR rtl-optimization/80160
25399 PR rtl-optimization/80159
25400 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
25401 reg_alternate_class into account.
25403 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
25406 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
25407 to consider in curr_insn_transform.
25409 2017-03-24 Jakub Jelinek <jakub@redhat.com>
25411 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
25412 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
25413 and emit_mode_inner.
25415 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25417 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
25418 argument to the overloaded builtin variants. Use the new flag to
25419 deprecate certain builtin variants.
25420 * config/s390/s390-builtin-types.def: Add new builtin types.
25421 * config/s390/s390-builtins.h: Support new flags field for
25422 overloaded builtins.
25423 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
25424 (s390_macro_to_expand): Enable vector float data type.
25425 (s390_cpu_cpp_builtins_internal): Indicate support of the new
25426 builtins by incrementing the __VEC__ version number.
25427 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
25429 (s390_resolve_overloaded_builtin): Emit error messages depending
25430 on the builtin flags.
25431 * config/s390/s390.c (s390_expand_builtin): Support additional
25432 flags argument. Change error message to match the messages
25433 emitted in s390-c.c.
25434 * config/s390/s390.md: New UNSPEC_* constants.
25435 (op_type): Add new instruction types.
25436 * config/s390/vecintrin.h: Add new builtins and test data class
25438 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
25439 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
25440 (VEC_INEXACT, VEC_NOINEXACT): New constants.
25441 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
25442 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
25443 ("vec_mergel<mode>"): V_HW -> VEC_HW.
25445 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
25446 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
25447 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
25448 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
25450 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
25451 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
25452 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
25453 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
25455 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
25456 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
25457 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
25458 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
25459 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
25460 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
25461 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
25463 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
25464 ("vec_scatter_element<V_HW_4:mode>_DI")
25465 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
25466 ("vec_fpint<mode>", "vflls")
25467 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
25468 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
25469 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
25470 ("*vec_cmphe<mode>_cc"): ... these.
25472 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
25473 mode constant instead of magic value.
25475 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25477 * config/s390/s390.c (s390_expand_vec_compare): Support other
25478 vector floating point modes than just V2DF.
25479 (s390_expand_vcond): Likewise.
25480 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
25481 (s390_cannot_change_mode_class): Prevent mode changes between TF
25482 and V1TF in vector registers.
25483 * config/s390/s390.md (DF, SF): New mode attributes.
25484 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
25485 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
25486 SFmode support for VRs.
25487 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
25489 (VFT, VF_HW): New mode iterators.
25490 (vw, sdx): New mode attributes.
25491 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
25492 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
25493 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
25494 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
25495 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
25496 also the new vector floating point modes. Renaming to ...
25498 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
25499 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
25500 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
25501 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
25502 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
25503 ("vec_unordered<mode>"): ... these.
25505 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
25506 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
25507 ("*vec_extendv2df"): New insn definitions.
25509 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25511 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
25512 ("mulditi3_2", "*muldi3_sign"): New patterns.
25513 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
25514 rename the pattern definition.
25516 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25518 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
25520 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
25522 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25524 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
25525 instruction if possible.
25526 * config/s390/vector.md (vec_halfnumelts): New mode
25528 ("*vec_vllezlf<mode>"): New pattern.
25530 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25532 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
25533 ("popcountv4si2", "popcountv2di2"): Rename to ...
25534 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
25535 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
25537 ("popcount<mode>2_vxe"): New pattern.
25539 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25541 * common/config/s390/s390-common.c (processor_flags_table): Add
25543 * config.gcc: Add arch12.
25544 * config/s390/driver-native.c (s390_host_detect_local_cpu):
25545 Default to arch12 for unknown CPU model numbers.
25546 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
25547 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
25548 PROCESSOR_max sanity check.
25549 * config/s390/s390-opts.h (enum processor_type): Add
25551 * config/s390/s390.c (processor_table): Add arch12.
25552 (s390_expand_builtin): Add check for B_VXE flag.
25553 (s390_issue_rate): Add PROCESSOR_ARCH12.
25554 (s390_get_sched_attrmask): Likewise.
25555 (s390_get_unit_mask): Likewise.
25556 (s390_sched_score): Enable z13 scheduling for arch12.
25557 (s390_sched_reorder): Likewise.
25558 (s390_sched_variable_issue): Likewise.
25559 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
25561 (s390_tune_attr): Use z13 scheduling also for arch12.
25562 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
25563 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
25564 (TARGET_VXE_P): New macros.
25565 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
25566 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
25567 * config/s390/s390.opt: Add arch12 as processor_type.
25569 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25571 * config/s390/s390.md
25572 ("fixuns_truncdddi2", "fixuns_trunctddi2")
25573 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
25574 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
25576 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
25577 Rename expanders to ...
25579 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
25580 ("fixuns_truncdddi2_emu"): ... these.
25582 ("fixuns_trunc<mode>si2_emu"): New expander.
25584 ("*fixuns_truncdfdi2_z13"): Rename to ...
25585 ("*fixuns_truncdfdi2_vx"): ... this.
25587 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25589 * config/s390/2964.md: Remove the single element vector compare
25590 instructions which are no longer used.
25591 * config/s390/s390.c (s390_select_ccmode): Remove handling of
25593 (s390_canonicalize_comparison): Remove handling of DFmode
25595 (s390_expand_vec_compare_scalar): Remove function.
25596 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
25597 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
25599 ("*cmp<mode>_ccs"): Add wfcdb instruction.
25601 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25603 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
25605 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
25606 will anyway by matched by mov<mode>_64dfp.
25608 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25610 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
25611 vlef/vstef. Add missing operand to vleif.
25613 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25615 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
25616 pair for all vector types with 64 bit elements.
25617 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
25618 * config/s390/vector.md (V_HW_64): ... here.
25619 (V_128_NOSINGLE): New mode iterator.
25620 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
25621 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
25622 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
25623 ("*vec_load_pairv2di"): Change to ...
25624 ("*vec_load_pair<mode>"): ... this one.
25626 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25628 * config/s390/constraints.md: Add comments.
25629 (jKK): Reject element sizes > 8 bytes.
25630 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
25632 * config/s390/s390.md: Add the s_operand checks formerly in
25633 s390_split_ok_p to various splitters where they are still
25635 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
25636 for 128 bit vectors. Plus two splitters.
25638 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25640 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
25643 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25646 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
25647 error if the boundary argument is not constant.
25649 2017-03-24 Jakub Jelinek <jakub@redhat.com>
25651 PR rtl-optimization/80112
25652 * loop-doloop.c (doloop_condition_get): Don't check condition
25653 if cmp isn't SET with IF_THEN_ELSE src.
25655 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25657 PR tree-optimization/80158
25658 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
25659 replacing a candidate statement, also replace it for the
25660 candidate's alternate interpretation.
25661 (replace_rhs_if_not_dup): Likewise.
25662 (replace_one_candidate): Likewise.
25664 2017-03-24 Richard Biener <rguenther@suse.de>
25666 PR tree-optimization/80167
25667 * graphite-isl-ast-to-gimple.c
25668 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
25670 (translate_isl_ast_to_gimple::get_rename): Likewise.
25672 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
25674 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
25675 handling of certain combinations of target options, including the
25676 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
25677 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
25679 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25682 * config/arm/arm.md (*load_multiple): Add reload_completed to
25683 matching condition.
25685 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25686 Richard Biener <rguenth@suse.de>
25688 PR tree-optimization/79908
25689 PR tree-optimization/80136
25690 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
25691 been cast away, gimplify_and_add suffices.
25693 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
25695 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
25697 2017-03-23 Richard Biener <rguenther@suse.de>
25699 PR tree-optimization/80032
25700 * gimplify.c (gimple_push_cleanup): Forced unconditional
25701 cleanups still have to go to the conditional_cleanups
25704 2017-03-22 Jakub Jelinek <jakub@redhat.com>
25706 PR tree-optimization/80072
25707 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
25709 (next_operand_entry_id): Change type to unsigned int.
25710 (sort_by_operand_rank): Make sure to return the right return value
25711 even if unsigned fields are bigger than INT_MAX.
25712 (struct oecount): Change cnt and id type to unsigned int.
25713 (oecount_hasher::equal): Formatting fix.
25714 (oecount_cmp): Make sure to return the right return value
25715 even if unsigned fields are bigger than INT_MAX.
25716 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
25719 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
25720 TREE_READONLY on result if writing it more than once.
25723 * doc/invoke.texi (-fsanitize=thread): Document that with
25724 -fnon-call-exceptions atomics are not able to throw
25728 * tsan.c: Include tree-eh.h.
25729 (instrument_builtin_call): Call maybe_clean_eh_stmt or
25730 maybe_clean_or_replace_eh_stmt where needed.
25731 (instrument_memory_accesses): Add cfg_changed argument.
25732 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
25733 if it returned true.
25734 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
25736 PR rtl-optimization/63191
25737 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
25738 wrapper function, moved the whole old content into ...
25739 (ix86_delegitimize_address_1): ... this. New inline function.
25740 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
25741 true as last argument instead of ix86_delegitimize_address.
25743 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
25745 * config/aarch64/aarch64.c (generic_branch_cost): Copy
25746 cortexa57_branch_cost.
25748 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
25750 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
25752 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25755 * doc/md.texi (Constraints): Document wA constraint.
25756 * config/rs6000/constraints.md (wA): New.
25757 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
25758 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
25759 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
25760 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
25762 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
25765 * gimplify.c (is_oacc_declared): New function.
25766 (oacc_default_clause): Use it to set default flags for acc declared
25767 variables inside parallel regions.
25768 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
25769 declared variables.
25770 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
25771 declare attribute to any decl as necessary.
25773 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
25776 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
25777 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
25778 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
25779 (arm_arch_lpae): This.
25780 * config/arm/arm.c (arm_arch7ve): Rename into ...
25781 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
25782 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
25785 2017-03-22 Martin Liska <mliska@suse.cz>
25788 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
25789 error message instead of an ICE.
25791 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25793 * doc/extend.texi (6.11 Additional Floating Types): Revise.
25795 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
25797 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
25799 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25802 2017-03-21 Martin Sebor <msebor@redhat.com>
25804 * doc/extend.texi: Use "cannot" instead of "can't."
25805 * doc/hostconfig.texi: Same.
25806 * doc/install.texi: Same.
25807 * doc/invoke.texi: Same.
25808 * doc/loop.texi: Same.
25809 * doc/md.texi: Same.
25810 * doc/objc.texi: Same.
25811 * doc/rtl.texi: Same.
25812 * doc/tm.texi: Same.
25813 * doc/tm.texi.in: Same.
25814 * doc/trouble.texi: Same.
25816 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
25819 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
25820 (collect_checksum_attributes): Set it.
25821 (die_checksum_ordered): Use it.
25823 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25825 PR tree-optimization/79908
25826 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
25827 change: For a VA_ARG whose LHS has been cast away, use
25828 force_gimple_operand to construct the side effects.
25830 2017-03-21 David Malcolm <dmalcolm@redhat.com>
25832 PR translation/80001
25833 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
25834 more amenable to translation.
25835 (oacc_loop_auto_partitions): Likewise.
25837 2017-03-21 Marek Polacek <polacek@redhat.com>
25838 Martin Sebor <msebor@redhat.com>
25840 PR tree-optimization/80109
25841 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
25842 on INTEGRAL_TYPE_P.
25844 2017-03-21 Jakub Jelinek <jakub@redhat.com>
25845 Segher Boessenkool <segher@kernel.crashing.org>
25848 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
25849 check reg_used_between_p between insn and one of succ or succ2
25850 depending on if succ is artificial insn not inserted into insn
25853 2017-03-21 Martin Liska <mliska@suse.cz>
25855 PR gcov-profile/80081
25856 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
25857 * doc/gcc.texi: Include gcov-dump stuff.
25858 * doc/gcov-dump.texi: New file.
25860 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
25862 PR rtl-optimization/79150
25863 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
25864 conditional jump, if the jump is the last insn of the loop.
25866 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25867 Richard Biener <rguenth@suse.de>
25869 PR tree-optimization/79908
25870 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
25871 been cast away, use force_gimple_operand to construct the side
25874 2017-03-21 Martin Liska <mliska@suse.cz>
25876 PR libfortran/79956
25877 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
25880 2017-03-21 Brad Spengler <spender@grsecurity.net>
25883 * plugin.c (htab_hash_plugin): New function.
25884 (add_new_plugin): Use it and adjust.
25885 (parse_plugin_arg_opt): Adjust.
25886 (init_one_plugin): Likewise.
25888 2017-03-21 Richard Biener <rguenther@suse.de>
25890 PR tree-optimization/80032
25891 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
25892 if set force the cleanup to happen unconditionally.
25893 (gimplify_target_expr): Push inserted clobbers with force_uncond
25894 to avoid them being removed by control-dependent DCE.
25896 2017-03-21 Richard Biener <rguenther@suse.de>
25898 PR tree-optimization/80122
25899 * tree-inline.c (copy_bb): Do not expans va-arg packs or
25900 va_arg_pack_len when the inlined call stmt requires pack
25902 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
25904 2017-03-21 Jakub Jelinek <jakub@redhat.com>
25907 * tsan.c (instrument_builtin_call): If the memory model argument
25908 is not a constant, assume it is valid.
25911 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
25914 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
25916 PR rtl-optimization/79910
25917 * combine.c (can_combine_p): Do not allow combining an I0 or I1
25918 if its dest is used by an insn before I2 (other than the combined
25919 insns themselves, which are properly handled already).
25921 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
25924 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
25926 * combine.c (record_used_regs): New static function.
25927 (try_combine): Handle situations where there is an additional
25928 instruction between I2 and I3 which needs to have a LOG_LINK
25932 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
25934 * combine.c (try_combine): Delete redundant i1 test. Call
25935 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
25937 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25940 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
25941 alternatives 13/14.
25943 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25945 PR tree-optimization/80054
25946 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
25947 the optimization if a PHI or any of its arguments is not dominated
25948 by the candidate's basis. Use gphi* rather than gimple* as
25950 (replace_profitable_candidates): Clean up a gimple* variable that
25951 should be a gphi* variable.
25953 2017-03-20 Martin Sebor <msebor@redhat.com>
25956 * doc/extend.texi (attribute constructor): Document present limitation.
25958 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
25961 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
25962 __POWER9_VECTOR__ #ifdef control, change template definition to
25963 use Power9-specific built-in function.
25964 (vec_any_eq): Likewise.
25965 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
25966 to control outcomes from this test.
25967 (vector_ae_<mode>p): For VEC_F modes, likewise.
25969 2017-03-20 Ian Lance Taylor <iant@google.com>
25971 * config/i386/i386.c (ix86_function_regparm): Save an extra
25972 register for -fsplit-stack with DECL_STATIC_CHAIN.
25974 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
25977 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
25978 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
25980 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
25982 * config/riscv/riscv.c (riscv_print_operand): Use "fence
25984 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
25987 2017-03-20 Marek Polacek <polacek@redhat.com>
25990 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
25992 2017-03-20 Richard Biener <rguenther@suse.de>
25994 PR tree-optimization/80113
25995 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
25996 allocate extra SSA name for PHI def.
25997 (add_close_phis_to_outer_loops): Likewise.
25998 (add_close_phis_to_merge_points): Likewise.
25999 (copy_loop_close_phi_args): Likewise.
26000 (copy_cond_phi_nodes): Likewise.
26002 2017-03-20 Martin Liska <mliska@suse.cz>
26004 PR middle-end/79753
26005 * tree-chkp.c (chkp_build_returned_bound): Do not build
26006 returned bounds for a LHS that's not a BOUNDED_P type.
26008 2017-03-20 Martin Liska <mliska@suse.cz>
26012 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
26013 COMPLEX_CST and VECTOR_CST.
26015 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26018 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
26019 target operand. A new splitter adds the clobber statement in case
26020 the target operand is dead anyway.
26022 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
26024 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
26025 to age-old versions of binutils and glibc.
26027 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
26029 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
26031 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
26033 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
26035 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
26037 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
26038 requirement for binutils 2.13.
26040 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
26042 * combine.c (try_combine): Delete redundant i1 test. Call
26043 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
26045 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
26047 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
26048 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
26050 <riscv64-*-elf>: Re-arrange section
26051 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
26052 <riscv32-*-linux>: Likewise.
26053 <riscv64-*-elf>: Likewise
26054 <riscv64-*-linux>: Likewise.
26056 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
26059 * aarch64.opt(verbose-cost-dump): Fix typo.
26061 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
26064 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
26065 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
26067 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
26069 * reload.c (find_reloads): When reloading a nonoffsettable address,
26070 use RELOAD_OTHER for it and its address reloads.
26072 PR rtl-optimization/79910
26073 * combine.c (record_used_regs): New static function.
26074 (try_combine): Handle situations where there is an additional
26075 instruction between I2 and I3 which needs to have a LOG_LINK
26078 2017-03-17 Jeff Law <law@redhat.com>
26080 PR tree-optimization/71437
26081 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
26082 conditional in the hash table first.
26083 (vrp_dom_walker::before_dom_children): Extract condition from
26084 ASSERT_EXPR. Record condition, its inverion and any implied
26085 conditions as well.
26087 2017-03-17 Marek Polacek <polacek@redhat.com>
26088 Markus Trippelsdorf <markus@trippelsdorf.de>
26090 PR tree-optimization/80079
26091 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
26094 2017-03-17 Richard Biener <rguenther@suse.de>
26096 PR middle-end/80075
26097 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
26098 Properly verify the LHS before the RHS possibly claims to be
26100 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
26103 2017-03-17 Martin Jambor <mjambor@suse.cz>
26105 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
26106 (List of -O2 options): Likewise.
26107 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
26110 2017-03-17 Tom de Vries <tom@codesourcery.com>
26112 * gcov-dump.c (print_usage): Print bug_report_url.
26114 2017-03-17 Richard Biener <rguenther@suse.de>
26116 PR middle-end/80050
26117 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
26118 (parser::peek): Likewise.
26120 2017-03-17 Richard Biener <rguenther@suse.de>
26122 PR tree-optimization/80048
26123 * sese.c (free_sese_info): Properly release rename_map and
26124 copied_bb_map elements.
26126 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
26128 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
26129 Add linked-list forward and backlinks. Insert on
26130 construction, remove on destruction.
26131 (class pass_store_merging): Add m_stores_head field.
26132 (pass_store_merging::terminate_and_process_all_chains):
26133 Iterate over m_stores_head list.
26134 (pass_store_merging::terminate_all_aliasing_chains):
26136 (pass_store_merging::execute): Check for debug stmts first.
26137 Push new chains onto the m_stores_head stack.
26139 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
26142 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
26143 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
26144 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
26146 2017-03-16 Jeff Law <law@redhat.com>
26148 PR tree-optimization/71437
26149 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
26150 member function. Implementation moved into after_dom_children
26151 member function and into the threader's thread_outgoing_edges
26153 (dom_opt_dom_walker::after_dom_children): Simplify by moving
26154 some code into new thread_outgoing_edges.
26155 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
26156 definition. Simplify marker handling (do it here). Assume we always
26157 have the available expression and the const/copies tables.
26158 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
26160 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
26161 * tree-vrp.c (equiv_stack): No longer file scoped.
26162 (vrp_dom_walker): New class.
26163 (vrp_dom_walker::before_dom_children): New member function.
26164 (vrp_dom_walker::after_dom_children): Likewise.
26165 (identify_jump_threads): Setup domwalker. Use it rather than
26166 walking edges in a random order by hand. Simplify setup/finalization.
26167 (finalize_jump_threads): Remove.
26168 (vrp_finalize): Do not call identify_jump_threads here.
26169 (execute_vrp): Do it here instead and call thread_through_all_blocks
26172 PR tree-optimization/71437
26173 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
26175 (simplify_stmt_for_jump_threading): Add basic_block argument. All
26177 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
26178 (dom_opt_dom_walker::thread_across_edge): Remove
26179 handle_dominating_asserts argument. All callers changed.
26180 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
26181 changes. Remove calls to lhs_of_dominating_assert. Other
26182 uses of handle_dominating_asserts turn into unconditional code
26183 (simplify_control_stmt_condition_1): Likewise.
26184 (simplify_control_stmt_condition): Likewise.
26185 (thread_through_normal_block, thread_across_edge): Likewise.
26186 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
26187 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
26188 object if it is not an SSA_NAME.
26189 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
26190 before calling into the VRP specific simplifiers.
26191 (identify_jump_threads): Remove handle_dominating_asserts
26194 2017-03-16 Jakub Jelinek <jakub@redhat.com>
26197 * tree-diagnostic.c (default_tree_printer): No longer static.
26198 * tree-diagnostic.h (default_tree_printer): New prototype.
26200 2017-03-16 Tamar Christina <tamar.christina@arm.com>
26202 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
26203 Change ins into fmov.
26205 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26207 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
26208 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
26209 Use h_con constraint for operand 1.
26210 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
26211 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
26213 2017-03-15 Jeff Law <law@redhat.com>
26215 PR tree-optimization/71437
26216 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
26217 (record_temporary_equivalences): Use it.
26219 PR tree-optimization/71437
26220 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
26221 tree-ssa-scopedtables.
26222 (lookup_avail_expr, build_and_record_new_cond): Likewise.
26223 (record_conditions, record_cond, vuse_eq): Likewise.
26224 (record_edge_info): Adjust to API tweak of record_conditions.
26225 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
26226 (record_temporary_equivalences, optimize_stmt): Likewise.
26227 (eliminate_redundant_computations): Likewise.
26228 (record_equivalences_from_stmt): Likewise.
26229 * tree-ssa-scopedtables.c: Include options.h and params.h.
26230 (vuse_eq): New function, moved from tree-ssa-dom.c
26231 (build_and_record_new_cond): Likewise.
26232 (record_conditions): Likewise. Accept vector of conditions rather
26233 than edge_equivalence structure for first argument.
26234 for the first argument.
26235 (avail_exprs_stack::lookup_avail_expr): New member function, moved
26236 from tree-ssa-dom.c.
26237 (avail_exprs_stack::record_cond): Likewise.
26238 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
26239 from tree-ssa-dom.c.
26240 (avail_exprs_stack): Add new member functions lookup_avail_expr
26242 (record_conditions): Declare.
26244 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
26247 * lra-constraints.c (process_alt_operands): Increase reject for
26248 reloading an input/output operand.
26250 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
26253 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
26254 insns to convert from signed/unsigned char/short to IEEE 128-bit
26256 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
26258 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
26261 * config/i386/i386.c (ix86_vector_duplicate_value): Create
26262 subreg of inner mode for values already in registers.
26264 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
26266 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
26267 iteration reg is used after the loop.
26269 2017-03-14 Martin Sebor <msebor@redhat.com>
26271 PR tree-optimization/79800
26272 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
26273 precision in negative-positive range.
26274 (format_floating): Call non-const overload with adjusted precision.
26276 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26279 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
26282 2017-03-14 Martin Sebor <msebor@redhat.com>
26284 PR middle-end/80020
26285 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
26286 * builtins.def (aligned_alloc): Use it.
26289 * Makefile.in (GTFILES): Add calls.c.
26290 * calls.c: Include "gt-calls.h".
26292 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
26294 PR rtl-optimization/79728
26295 * regs.h (struct target_regs): New field
26296 x_contains_allocatable_regs_of_mode.
26297 (contains_allocatable_regs_of_mode): New macro.
26298 * reginfo.c (init_reg_sets_1): Initialize it, and change
26299 contains_reg_of_mode so it includes global regs as well.
26300 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
26301 rather than contains_regs_of_mode.
26303 2017-03-14 Martin Liska <mliska@suse.cz>
26305 * doc/invoke.texi: Document options that can't be combined with
26306 -fcheck-pointer-bounds.
26308 2017-03-14 Martin Liska <mliska@suse.cz>
26310 PR middle-end/79831
26311 * doc/invoke.texi (-Wchkp): Document the option.
26313 2017-03-14 Martin Liska <mliska@suse.cz>
26315 * Makefile.in: Install gcov-dump.
26317 2017-03-14 Martin Liska <mliska@suse.cz>
26319 * multiple_target.c (expand_target_clones): Bail out for
26320 an invalid attribute.
26322 2017-03-14 Richard Biener <rguenther@suse.de>
26324 * alias.c (struct alias_set_entry): Pack properly.
26325 * cfgloop.h (struct loop): Likewise.
26326 * cse.c (struct set): Likewise.
26327 * ipa-utils.c (struct searchc_env): Likewise.
26328 * loop-invariant.c (struct invariant): Likewise.
26329 * lra-remat.c (struct cand): Likewise.
26330 * recog.c (struct change_t): Likewise.
26331 * rtl.h (struct address_info): Likewise.
26332 * symbol-summary.h (function_summary): Likewise.
26333 * tree-loop-distribution.c (struct partition): Likewise.
26334 * tree-object-size.c (struct object_size_info): Likewise.
26335 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
26336 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
26337 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
26338 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
26339 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
26340 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
26341 (struct _stmt_vec_info): Likewise.
26343 2017-03-14 Martin Liska <mliska@suse.cz>
26346 * multiple_target.c (create_dispatcher_calls): Check that
26347 a target can create a function dispatcher.
26349 2017-03-14 Martin Liska <mliska@suse.cz>
26352 * multiple_target.c (expand_target_clones): Drop local.local
26353 flag for default implementation.
26355 2017-03-14 Richard Biener <rguenther@suse.de>
26357 PR tree-optimization/80030
26358 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
26360 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
26362 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
26363 gcc_fallthrough() instead of __attribute__((fallthrough));
26365 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
26367 * doc/gcc.texi: Remove "up" link to (DIR).
26368 * doc/gccint.texi: Ditto.
26370 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
26372 * doc/install.texi (Specific) <avr>: Remove reference to
26375 2017-03-13 Jeff Law <law@redhat.com>
26377 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
26378 attribute rather than comments.
26380 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
26381 match_scratch operand is highest.
26383 2017-03-13 Martin Liska <mliska@suse.cz>
26385 PR middle-end/78339
26386 * ipa-pure-const.c (warn_function_noreturn): If the declarations
26387 is a CHKP clone, use original declaration.
26389 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26391 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
26392 (arc_conditional_register_usage): Use a different allocation order
26393 when optimizing for size.
26394 * common/config/arc/arc-common.c (arc_option_optimization_table):
26395 Section anchors default on when optimizing for size.
26397 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26399 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
26401 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26403 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
26404 * config/arc/arc.md (cpu_facility): Add cd variant.
26405 (*movqi_insn): Add code density variant.
26406 (*movhi_insn): Likewise.
26407 (*movqi_insn): Likewise.
26408 (*addsi3_mixed): Likewise.
26409 (subsi3_insn): Likewise.
26411 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26413 * config/arc/arc.md (movsi_cond_exec): Update constraint.
26415 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26417 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
26418 expressions with MINUS and UNARY ops.
26420 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26423 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
26425 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
26426 between vec_select and vector argument.
26427 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
26428 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
26429 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
26430 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
26431 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
26432 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
26434 2017-03-13 Richard Biener <rguenther@suse.de>
26437 * params.def (vect-max-peeling-for-alignment): Fix typo.
26439 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26441 * doc/install.texi (Specific) <mips-*-*>: Remove description of
26442 issue that only occurred with binutils below 2.18.
26444 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26446 * doc/install.texi (Specific) <cris-axis-elf>: No longer
26447 refer to binutils 2.11/2.12 minimum.
26449 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26451 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
26452 ftp.kernel.org and simplify binutils requirement.
26454 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
26456 * doc/invoke.texi (Warning Options): Fix spelling of link-time
26458 (Optimize Options): Ditto. Also remove redundancy.
26460 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26462 PR translation/79848
26463 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
26465 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
26466 to G_ to avoid double translation.
26468 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26470 PR translation/79923
26471 * auto-profile.c (get_combined_location): Convert leading
26472 character of diagnostics to lower case and remove trailing period.
26473 (read_profile): Likewise for various diagnostics.
26474 * config/arm/arm.c (arm_option_override): Remove trailing period
26475 from various diagnostics.
26476 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
26477 (msp430_expand_delay_cycles): Likewise.
26479 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26482 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
26483 full command-line argument, rather than just "str".
26484 (aarch64_validate_march): Likewise.
26485 (aarch64_validate_mtune): Likewise.
26487 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
26489 PR rtl-optimization/78911
26490 * lra-assigns.c (must_not_spill_p): New function.
26491 (spill_for): Use it.
26493 2017-03-10 Jakub Jelinek <jakub@redhat.com>
26495 PR tree-optimization/79981
26496 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
26497 ATOMIC_COMPARE_EXCHANGE ifn result.
26498 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
26499 IFN_ATOMIC_COMPARE_EXCHANGE.
26501 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26504 * opts.c (parse_sanitizer_options): Add missing question mark to
26505 "did you mean" message.
26507 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26509 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
26511 (VMULEUH_UNS): Likewise.
26512 (VMULOUB_UNS): Likewise.
26513 (VMULOUH_UNS): Likewise.
26514 * config/rs6000/rs6000.c (builtin_function_type): Remove
26515 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
26517 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26520 * read-rtl-function.c (function_reader::read_rtx_operand): Update
26521 x with result of extra_parsing_for_operand_code_0.
26522 (function_reader::extra_parsing_for_operand_code_0): Convert
26523 return type from void to rtx, returning x. When reading
26524 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
26525 larger size containing struct block_symbol.
26527 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
26529 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
26530 -mfloat128-hardware without -m64.
26532 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
26535 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
26536 entries to the case statement that marks unsigned arguments to
26537 overloaded functions.
26539 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26541 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
26542 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
26544 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
26547 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
26548 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
26550 2017-03-10 Martin Liska <mliska@suse.cz>
26554 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
26555 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
26558 2017-03-10 Olivier Hainque <hainque@adacore.com>
26560 * tree-switch-conversion (array_value_type): Start by resetting
26561 candidate type to it's main variant.
26563 2017-03-10 Jakub Jelinek <jakub@redhat.com>
26565 PR rtl-optimization/79909
26566 * combine.c (try_combine): Use simplify_replace_rtx on individual
26567 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
26568 of the whole CALL_INSN_FUNCTION_USAGE.
26570 PR tree-optimization/79972
26571 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
26572 get_range_info on SSA_NAMEs. Formatting fixes.
26574 2017-03-10 Richard Biener <rguenther@suse.de>
26575 Jakub Jelinek <jakub@redhat.com>
26577 PR tree-optimization/77975
26578 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
26579 edge to be constant.
26580 (get_val_for): For constant x return it. Formatting fix.
26581 (loop_niter_by_eval): Avoid pointless looping if the next iteration
26582 would use the same bases as the current one.
26584 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26586 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
26587 instead of vec_select for V1TImode.
26588 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
26590 (VSX_LE_128): Add V1TI to this mode iterator.
26591 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
26592 (*vsx_le_perm_store_<mode>): Likewise.
26593 (pre-reload splitter for VSX stores): Likewise.
26594 (post-reload splitter for VSX stores): Likewise.
26595 (*vsx_xxpermdi2_le_<mode>): Likewise.
26596 (*vsx_lxvd2x2_le_<mode>): Likewise.
26597 (*vsx_stxvd2x2_le_<mode>): Likewise.
26599 2017-03-09 Michael Eager <eager@eagercon.com>
26601 Correct failures with --enable-checking=yes,rtl.
26603 * config/microblaze/microblaze.c (microblaze_expand_shift):
26604 Replace GET_CODE test with CONST_INT_P and INTVAL test with
26605 test for const0_rtx.
26606 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
26607 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
26609 2017-03-09 Richard Biener <rguenther@suse.de>
26611 PR tree-optimization/79977
26612 * graphite-scop-detection.c (scop_detection::merge_sese):
26613 Handle the case of extra exits to blocks dominating the entry.
26615 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
26617 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26620 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
26622 PR rtl-optimization/79949
26623 * lra-constraints.c (process_alt_operands): Check memory when
26624 trying to predict a cycle. Print about the overall increase.
26626 2017-03-09 Richard Biener <rguenther@suse.de>
26628 PR middle-end/79971
26629 * gimple-expr.c (useless_type_conversion_p): Preserve
26630 TYPE_SATURATING for fixed-point types.
26632 2017-03-09 Richard Biener <rguenther@suse.de>
26635 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
26636 alignment of BLKmode params.
26638 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26641 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
26642 (VALL_NO_V2Q): Likewise.
26644 * config/aarch64/aarch64-simd.md
26645 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
26647 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
26648 VALL_NO_V2Q mode iterator.
26649 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
26651 2017-03-09 Martin Liska <mliska@suse.cz>
26653 PR tree-optimization/79631
26654 * tree-chkp-opt.c (chkp_is_constant_addr): Call
26655 tree_int_cst_sign_bit just for INTEGER constants.
26657 2017-03-09 Martin Liska <mliska@suse.cz>
26661 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
26664 2017-03-09 Marek Polacek <polacek@redhat.com>
26667 * tree.c (inchash::add_expr): Handle TREE_VEC.
26669 2017-03-09 Martin Liska <mliska@suse.cz>
26672 (chkp_narrow_size_and_offset): New function.
26673 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
26674 (void chkp_parse_bit_field_ref): New function.
26675 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
26676 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
26678 2017-03-09 Martin Liska <mliska@suse.cz>
26681 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
26682 (chkp_find_bounds_1): Remove gcc_unreachable.
26684 2017-03-09 Jakub Jelinek <jakub@redhat.com>
26687 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
26688 BUILT_IN_SYNC*, determine the access type from the size suffix and
26689 always build a MEM_REF with that type. Handle forgotten
26690 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
26693 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
26694 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
26695 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
26696 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
26697 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
26698 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
26699 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
26700 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
26701 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
26702 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
26703 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
26704 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
26705 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
26706 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
26707 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
26708 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
26709 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
26710 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
26711 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
26712 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
26713 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
26714 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
26715 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
26716 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
26717 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
26718 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
26719 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
26720 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
26721 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
26722 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
26723 definitions outside of __OPTIMIZE__ guarded section.
26726 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
26727 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
26728 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
26729 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
26732 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26734 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
26735 ("vfenez<mode>"): Add missing constraints.
26737 2017-03-08 Martin Sebor <msebor@redhat.com>
26740 * config/nds32/nds32.c (nds32_option_override):
26741 Fix misspelled diagnostic.
26743 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26746 * gimplify.c (gimplify_omp_for): Replace index var in outer
26747 taskloop statement with an artificial variable and add
26748 OMP_CLAUSE_PRIVATE clause for it.
26750 2017-03-08 Richard Biener <rguenther@suse.de>
26752 PR tree-optimization/79955
26753 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
26754 for accesses that are completely outside of the variable.
26756 2017-03-08 Andrew Haley <aph@redhat.com>
26758 PR tree-optimization/79943
26759 * tree-ssa-loop-split.c (compute_new_first_bound): When
26760 calculating the new upper bound, (END-BEG) should be added, not
26763 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26765 * config/avr/avr.md (setmemhi): Make sure match_dup
26766 operand number comes before match_scratch.
26768 2017-03-08 Richard Biener <rguenther@suse.de>
26770 PR tree-optimization/79920
26771 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
26772 with ncopies == 1 to ...
26773 (vect_transform_slp_perm_load): ... here. Properly compute
26774 all element loads by iterating VF times over the group. Do
26775 not handle ncopies (computed in a broken way) in
26776 vect_create_mask_and_perm.
26778 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26781 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
26782 is a uniform vector, use uniform_vector_p return value instead of
26783 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
26785 2017-03-07 Marek Polacek <polacek@redhat.com>
26787 PR middle-end/79809
26788 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
26789 (alloca_call_type): Likewise.
26791 2017-03-07 Martin Liska <mliska@suse.cz>
26793 * gcov.c (process_args): Put comment to correct location.
26795 2017-03-07 Martin Liska <mliska@suse.cz>
26797 PR middle-end/68270
26798 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
26799 Use array_at_struct_end_p instead of DECL_CHAIN (field).
26800 (chkp_narrow_bounds_for_field): Likewise.
26801 (chkp_parse_array_and_component_ref): Pass one more argument to
26804 2017-03-07 Richard Biener <rguenther@suse.de>
26806 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
26809 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
26811 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
26812 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
26814 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26817 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
26818 to end of description.
26819 (PARAM_MAX_STORES_TO_MERGE): Likewise.
26821 2017-03-07 Jakub Jelinek <jakub@redhat.com>
26823 PR rtl-optimization/79901
26824 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
26826 (*avx512f_<code><mode>3<mask_name>): ... this.
26827 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
26828 iterator instead of VI8_AVX2_AVX512BW.
26830 PR rtl-optimization/79901
26831 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
26832 min/max expander, expand it using expand_vec_cond_expr.
26835 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
26838 2017-03-06 Jakub Jelinek <jakub@redhat.com>
26841 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
26842 to void * for PCH reasons.
26843 * dwarf2out.c (output_loc_operands, output_die): Cast
26844 v.val_vec.array to unsigned char *.
26846 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
26849 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
26852 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
26854 PR rtl-optimization/79571
26855 * lra-constraints.c (process_alt_operands): Calculate static
26856 reject and subtract it from overall when only addresses will be
26859 2017-03-06 Julia Koval <julia.koval@intel.com>
26862 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
26863 incoming stack boundary to 128 for 64-bit targets.
26865 2017-03-06 Richard Biener <rguenther@suse.de>
26867 PR tree-optimization/79894
26868 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
26869 to NULL after folding it.
26871 2017-03-06 Richard Biener <rguenther@suse.de>
26873 PR tree-optimization/79824
26874 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
26875 check disabling peeling for gaps.
26877 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
26879 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
26880 attributes): Document gettimeofday.
26882 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
26884 * config/s390/s390.c (s390_option_override_internal): Set
26885 PARAM_MIN_VECT_LOOP_BOUND
26887 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
26889 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
26890 * config/s390/s390.md: Likewise.
26892 2017-03-06 Jakub Jelinek <jakub@redhat.com>
26895 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
26896 (<avx2_avx512>_perm<mode>): Rename to ...
26897 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
26899 (<avx512>_perm<mode>_mask): Rename to ...
26900 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
26902 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
26903 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
26904 instead of VI8F_256_512.
26905 (avx512f_perm<mode>): New define_expand.
26906 (avx512f_perm<mode>_mask): Likewise.
26907 (avx512f_perm<mode>_1<mask_name>): New define_insn.
26908 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
26910 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26912 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
26913 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
26915 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
26917 2017-03-06 Martin Liska <mliska@suse.cz>
26920 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
26921 when having a SSA NAME w/o VAR_DECL assigned to it.
26923 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26925 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
26926 msa_dpsub_<su>_d): Fix MODE for vec_select.
26928 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26930 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
26932 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
26934 2017-03-06 Richard Biener <rguenther@suse.de>
26936 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
26937 * plugin.c (register_plugin_info): Likewise.
26938 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
26940 2017-03-05 Jakub Jelinek <jakub@redhat.com>
26942 * config/i386/sse.md (sse_storehps, sse_storelps,
26943 avx_<castmode><avxsizesuffix>_<castmode>,
26944 avx512f_<castmode><avxsizesuffix>_<castmode>,
26945 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
26946 in condition that at least one operand is not a MEM.
26948 2017-03-03 Jakub Jelinek <jakub@redhat.com>
26950 PR middle-end/79805
26951 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
26952 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
26954 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
26955 gimple_call_nothrow_p flag based on whether original builtin can throw.
26956 If it can, emit following stmts on the fallthrough edge.
26957 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
26958 don't create new bb if inserting just debug stmts on the edge, try to
26959 insert them on the fallthru bb or just reset debug stmts.
26961 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
26964 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
26965 restore recog_data (including the operand rtxes inside it) around
26966 the call to get_insn_template.
26968 2017-03-03 Martin Sebor <msebor@redhat.com>
26970 PR tree-optimization/79699
26971 * context.c (context::~context): Free MPFR caches to avoid
26972 a memory leak on program exit.
26974 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26976 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
26977 Use wide_int::ulow () instead of .elt (0).
26979 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
26981 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
26982 (*pushxf): Limit oF constraint to 32bit targets and add oC
26983 constraint for 64bit targets.
26984 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
26985 (*pushdf): Change rmF constraint to rmC.
26987 2017-03-03 Martin Liska <mliska@suse.cz>
26989 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
26990 Remove unused variable.
26992 2017-03-03 Jakub Jelinek <jakub@redhat.com>
26995 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
26996 is a memory operand, increase num_memory.
26997 (ix86_expand_args_builtin): Likewise.
26999 2017-03-03 Jan Hubicka <jh@suse.cz>
27002 * ipa-devirt.c (maybe_record_node): Properly handle
27003 __cxa_pure_virtual visibility.
27005 2017-03-03 Martin Liska <mliska@suse.cz>
27007 PR tree-optimization/79803
27008 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
27010 (pass_loop_prefetch::execute): Disabled optimization if an
27011 assumption about L1 cache size is not met.
27013 2017-03-03 Martin Liska <mliska@suse.cz>
27015 PR rtl-optimization/79574
27016 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
27017 (hash_scan_set): Likewise.
27018 (dump_hash_table): Likewise.
27019 (hoist_code): Likewise.
27021 2017-03-03 Richard Biener <rguenther@suse.de>
27023 * fixed-value.c (fixed_from_string): Restore use of elt (1)
27024 in place of uhigh ().
27025 (fixed_convert_from_real): Likewise.
27027 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
27030 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
27032 2017-03-03 Richard Biener <rguenther@suse.de>
27034 PR middle-end/79818
27035 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
27036 TYPE_OVERFLOW_UNDEFINED check.
27038 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27040 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
27042 (vector_ae_<mode>_p): Likewise.
27043 (vector_nez_<mode>_p): Likewise.
27044 (vector_ne_v2di_p): Likewise.
27045 (vector_ae_v2di_p): Likewise.
27046 (vector_ne_<mode>_p): Likewise.
27047 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
27049 (vsx_tsqrt<mode>2_fe): Likewise.
27051 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
27054 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
27056 2017-03-02 Jakub Jelinek <jakub@redhat.com>
27058 PR rtl-optimization/79780
27059 * cprop.c (one_cprop_pass): When second and further conditional trap
27060 in a single basic block is turned into an unconditional trap, turn it
27061 into a deleted note to avoid RTL verification failures.
27063 2017-03-02 Richard Biener <rguenther@suse.de>
27065 * fold-const.c (const_binop): Use ulow () instead of elt (0).
27067 2017-03-02 Richard Biener <rguenther@suse.de>
27069 PR tree-optimization/79345
27071 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
27072 param and abort the walk, returning -1 if it is hit.
27073 (walk_aliased_vdefs): Take a limit param and pass it on.
27074 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
27075 defaulting to 0 and return a signed int.
27076 * tree-ssa-uninit.c (struct check_defs_data): New struct.
27077 (check_defs): New helper.
27078 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
27079 about uninitialized memory.
27080 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
27081 bogus uninitialized warning.
27082 (fixed_convert_from_real): Likewise.
27084 2017-03-02 Bin Cheng <bin.cheng@arm.com>
27086 PR tree-optimization/66768
27087 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
27088 iv_use if base object can't be determined.
27090 2017-03-02 Jakub Jelinek <jakub@redhat.com>
27092 PR tree-optimization/79345
27093 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
27094 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
27095 (get_pattern_stats): Initialize it.
27096 * genemit.c (gen_expand): Verify match_scratch numbers come after
27097 match_operand/match_dup numbers.
27098 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
27099 match_scratch numbers.
27100 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
27102 * config/s390/s390.md (trunctdsd2): Likewise.
27104 2017-03-02 Richard Biener <rguenther@suse.de>
27106 * wide-int.h (wide_int_storage::operator=): Implement in terms
27109 2017-03-02 Richard Biener <rguenther@suse.de>
27111 PR tree-optimization/79777
27112 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
27113 the to insert expression to sth existing.
27115 2017-03-01 Martin Sebor <msebor@redhat.com>
27117 PR middle-end/79692
27118 * gimple-ssa-sprintf.c
27119 (directive::known_width_and_precision): New function.
27120 (format_integer): Use it.
27121 (get_mpfr_format_length): Consider the full range of precision
27122 when computing %g output with the # flag. Set the likely byte
27123 count to 3 rather than 1 when precision is indeterminate.
27124 (format_floating): Correct the lower bound of precision.
27126 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27128 * doc/invoke.texi: Document default code model for 64-bit Linux.
27130 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27133 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
27134 udiv rather than div since input pattern is unsigned.
27136 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
27138 * config/i386/i386.c (print_reg): Warn for values of
27139 unsupported size in integer register.
27141 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
27144 * config/rs6000/predicates.md (current_file_function_operand): Do
27145 not allow self calls to be local if the function is replaceable.
27147 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
27150 * config/rs6000/altivec.h (vec_ctz and others): Change the
27151 preprocessor macro that controls conditional compilation from
27152 _ARCH_PWR9 to __POWER9_VECTOR__.
27153 (vec_all_ne): Change parameterization of __altivec_scalar_pred
27154 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
27155 control (instead of _ARCH_PWR9 control) so that template
27156 definition uses power9-specific function.
27157 (vec_any_eq): Likewise.
27158 (vec_all_ne): Change macro definition to use a power9-specific
27159 expansion under #ifdef __POWER9_VECTOR__ control (instead of
27160 _ARCH_PWR9 control).
27161 (vec_any_eq) Likewise.
27162 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
27163 expansion for CMPNEF to remove support for xvcmpnesp instruction.
27164 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
27165 support for xvcmpnedp instruction.
27166 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
27167 macro expansion so that Power9 implementation of vec_all_ne does
27168 not use the AltiVec predicate framework.
27169 (VCMPNEH_P): Likewise.
27170 (VCMPNEW_P): Likewise.
27171 (VCMPNED_P): Likewise.
27172 (VCMPNEFP_P): Likewise.
27173 (VCMPNEDP_P): Likewise.
27174 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
27175 implementation of vec_any_eq to not use AltiVec predicate
27177 (VCMPAEH_P): Likewise.
27178 (VCMPAEW_P): Likewise.
27179 (VCMPAED_P): Likewise.
27180 (VCMPAEFP_P): Likewise.
27181 (VCMPAEDP_P): Likewise.
27182 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
27183 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
27184 not use the AltiVec predicate framework.
27185 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
27186 of vec_any_eq to not use AltiVec predicate framework.
27187 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27188 support for predefined __POWER9_VECTOR__ macro to indicate that
27189 Power9 instruction selection is enabled.
27190 (altivec_overloaded_builtins): Remove extraneous
27191 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
27192 function argument types RS6000_BTI_bool_V16QI and
27193 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
27194 entry for overloaded function argument types RS6000_BTI_bool_V4SI
27195 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
27196 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
27197 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
27198 Power9 for implementations of vec_cmpne. Change the signature for
27199 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
27200 (representing vec_all_ne) to remove the previously described first
27201 argument of type RS6000_BTI_INTSI, as this was an artifact of
27202 reliance on the AltiVec predicate framework, which is no longer
27203 used in the implementation of these functions. Add
27204 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
27205 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
27206 since, unlike the AltiVec predicate framework implementation, we
27207 do not share function descriptors between vec_alle and vec_anyeq.
27208 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
27209 set of modes that receive special treatment even when
27210 TARGET_P9_VECTOR is true. The special treatment emits code that
27211 does not depend on Power9 instructions.
27212 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
27213 define_expand to not rely on AltiVec predicate framework.
27214 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
27216 (vector_ne_v2di_p): Change this define_expand to not rely on
27217 AltiVec predicate framework.
27218 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
27220 (vector_ne_<mode>_p): Change this define_expand to not rely on
27221 AltiVec predicate framework.
27222 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
27224 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
27225 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
27226 define_insn pattern.
27227 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
27228 define_insn pattern because the xvcmpne<VSs>. instruction is not
27230 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
27231 instruction is not supported.
27233 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27235 * config/nvptx/nvptx.c: Include intl.h.
27237 2017-03-01 Martin Jambor <mjambor@suse.cz>
27240 * ipa-prop.h (ipa_bits): Removed field known.
27241 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
27242 to pointers. Adjusted their comments to warn about their sharing.
27243 (ipcp_transformation_summary): Change bits to a vector of pointers.
27244 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
27245 (ipa_get_ipa_bits_for_value): Declare.
27246 * tree-vrp.h (value_range): Mark as GTY((for_user)).
27247 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
27248 (ipa_bits_hash_table): Likewise.
27249 (ipa_vr_ggc_hash_traits): Likewise.
27250 (ipa_vr_hash_table): Likewise.
27251 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
27252 being pointers and vr_known being removed.
27253 (ipa_set_jf_unknown): Likewise.
27254 (ipa_get_ipa_bits_for_value): New function.
27255 (ipa_set_jfunc_bits): Likewise.
27256 (ipa_get_value_range): New overloaded functions.
27257 (ipa_set_jfunc_vr): Likewise.
27258 (ipa_compute_jump_functions_for_edge): Use the above functions to
27259 construct bits and vr parts of jump functions.
27260 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
27261 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
27263 (ipcp_grow_transformations_if_necessary): Also allocate
27264 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
27266 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
27267 them. Fix too long lines.
27268 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
27269 vr_known being removed.
27270 (ipa_read_jump_function): Use new setter functions to construct bits
27271 and vr parts of jump functions or set them to NULL.
27272 (write_ipcp_transformation_info): Adjust for bits being pointers.
27273 (read_ipcp_transformation_info): Likewise.
27274 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
27276 Include gt-ipa-prop.h.
27277 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
27279 (ipcp_store_bits_results): Likewise.
27280 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
27281 Do not write to existing jump functions but use a temporary instead.
27283 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27286 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
27287 attempt to use its first operand as BIT_FIELD_REF base.
27289 2017-03-01 Richard Biener <rguenther@suse.de>
27291 PR middle-end/79721
27292 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
27293 interpolating formula in wrapping arithmetic.
27294 (chrec_apply): Convert chrec_evaluate return value to wanted type.
27296 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27298 PR tree-optimization/79734
27299 * tree-vect-generic.c (expand_vector_condition): Optimize
27300 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
27301 Handle VEC_COND_EXPR where comparison has different inner width from
27302 type's inner width.
27304 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
27306 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
27307 markup, and similar issues. Remove @opindex entries for things
27308 that aren't options. Add missing -mmpy-option entries.
27310 2017-02-28 Jakub Jelinek <jakub@redhat.com>
27312 PR tree-optimization/79737
27313 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
27314 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
27315 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
27316 instead of byte_size. Formatting fix.
27317 (shift_bytes_in_array_right): Formatting fix.
27319 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
27322 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
27323 condition on optimize for the leaf function test.
27325 2017-02-28 Martin Liska <mliska@suse.cz>
27328 * read-rtl-function.c (function_reader::handle_unknown_directive):
27329 Bail out when one uses -flto.
27331 2017-02-28 Martin Liska <mliska@suse.cz>
27333 * common.opt: Replace space with tabular for options of <number>
27335 * config/i386/i386.opt: Show <number> value for
27336 -mlarge-data-threshold.
27337 * opts.c (print_filtered_help): Do not display number in hexadecimal
27340 2017-02-28 Martin Liska <mliska@suse.cz>
27342 * common.opt: Fix --help=option -Q for options which are of
27345 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
27347 * config/i386/i386.c (print_reg): Error out for values
27348 of 8-bit size in invalid integer register.
27350 2017-02-28 Martin Sebor <msebor@redhat.com>
27352 PR tree-optimization/79691
27353 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
27355 2017-02-28 Jakub Jelinek <jakub@redhat.com>
27358 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
27359 gcc_unreachable with output_operand_lossage.
27361 2017-02-28 Richard Biener <rguenther@suse.de>
27363 PR tree-optimization/79740
27364 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
27366 (visit_nary_op): Insert the nary into the hashtable if we
27367 pattern-matched sth.
27368 * tree-ssa-pre.c (eliminate_insert): Robustify.
27370 2017-02-28 Richard Biener <rguenther@suse.de>
27372 PR middle-end/79731
27373 * fold-const.c (decode_field_reference): Reject out-of-bound
27376 2017-02-28 Jakub Jelinek <jakub@redhat.com>
27378 * config/i386/i386.c: Include intl.h.
27379 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
27380 instead of just cond ? "..." : "...".
27381 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
27382 * coverage.c (read_counts_file): Likewise.
27383 * omp-offload.c: Include intl.h.
27384 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
27385 of just cond ? "..." : "...".
27386 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
27387 of just cond ? "..." : "...".
27389 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
27392 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
27394 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
27395 'tune for' CPU name.
27396 * config/arm/arm-cpu-data.h: Regenerated.
27398 2017-02-28 Richard Biener <rguenther@suse.de>
27400 PR tree-optimization/79732
27401 * tree-inline.c (expand_call_inline): Do not shadow var.
27403 2017-02-28 Richard Biener <rguenther@suse.de>
27405 PR tree-optimization/79723
27406 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
27407 address-space properly.
27409 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
27411 * doc/optinfo.texi (Optimization groups): Fix option used for
27413 * doc/invoke.texi (-fopt-info): Document "omp".
27414 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
27415 (OPTGROUP_ALL): Add OPTGROUP_OMP.
27416 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
27417 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
27418 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
27420 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
27422 * dumpfile.c (optgroup_options): Instead of "openmp", associate
27423 OPTGROUP_OMP with "omp".
27425 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
27428 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
27429 for arithmetic shift of unsigned V2DI.
27431 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
27433 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
27434 arc/linux.h headers.
27435 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
27436 (LINK_SPEC): Likewise.
27437 (ARC_TLS_EXTRA_START_SPEC): Likewise.
27438 (EXTRA_SPECS): Likewise.
27439 (STARTFILE_SPEC): Likewise.
27440 (ENDFILE_SPEC): Likewise.
27441 (LIB_SPEC): Likewise.
27442 (TARGET_SDATA_DEFAULT): Likewise.
27443 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
27444 (MULTILIB_DEFAULTS): Likewise.
27445 (DWARF2_UNWIND_INFO): Likewise.
27446 * config/arc/big.h: New file.
27447 * config/arc/elf.h: Likewise.
27448 * config/arc/linux.h: Likewise.
27449 * config/arc/t-uClibc: Remove.
27451 2017-02-27 Bin Cheng <bin.cheng@arm.com>
27453 PR tree-optimization/77536
27454 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
27455 (tree_transform_and_unroll_loop): Use above function to compute the
27456 estimated niter of unrolled loop and use it when scaling profile.
27457 Also use count info rather than frequency if it's non-zero.
27458 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
27459 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
27460 (vect_transform_loop): Call above function.
27462 2017-02-27 Richard Biener <rguenther@suse.de>
27464 PR tree-optimization/45397
27465 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
27466 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
27467 (visit_nary_op): Add pattern matching for CSEing sign-changed
27468 or truncated operations with wider ones.
27470 2017-02-27 Richard Biener <rguenther@suse.de>
27472 PR tree-optimization/79690
27473 * tree-vect-stmts.c (vectorizable_store): Use vector type
27474 built from the DR with address-space.
27476 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
27478 * doc/invoke.texi (Optimize Options): Refine the description
27479 of asan-use-after-return.
27481 2017-02-25 Alan Modra <amodra@gmail.com>
27483 PR rtl-optimization/79584
27484 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
27485 base, not ad->base_term, the reg within base. Remove assertion
27486 that ad->base == ad->base_term. Replace gen_int_mode using
27487 bogus mode with const0_rtx.
27489 2017-02-25 Jakub Jelinek <jakub@redhat.com>
27491 PR middle-end/79396
27492 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
27493 FMA_EXPR like tcc_binary or tcc_unary.
27495 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
27498 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
27500 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
27501 (output_loc_operands): Handle DW_OP_call_ref and
27502 DW_OP_GNU_variable_value.
27503 (struct variable_value_struct): New type.
27504 (struct variable_value_hasher): Likewise.
27505 (variable_value_hash): New variable.
27506 (string_types): Remove.
27507 (copy_loc_descr): New function.
27508 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
27509 (prepend_loc_descr_to_each): New function.
27510 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
27511 instead of add_loc_descr_to_each if the first argument is single
27512 location list and the second has multiple.
27513 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
27514 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
27515 when looking for variable value which doesn't have other location info.
27516 (loc_list_from_tree): Formatting fix.
27517 (gen_array_type_die): Simplify DW_AT_string_length handling.
27518 (adjust_string_types): Remove.
27519 (gen_subprogram_die): Don't call adjust_string_types nor test/set
27520 string_types. Call resolve_variable_values.
27521 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
27522 (resolve_addr_in_expr): Likewise. Add A argument.
27523 (copy_deref_exprloc): Remove deref argument. Adjust for the
27524 original expression being DW_OP_GNU_variable_value with optionally
27525 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
27526 optionally after it.
27527 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
27528 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
27529 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
27530 (variable_value_hasher::hash, variable_value_hasher::equal): New
27532 (resolve_variable_value_in_expr, resolve_variable_value,
27533 resolve_variable_values, note_variable_value_in_expr,
27534 note_variable_value): New functions.
27535 (dwarf2out_early_finish): Call note_variable_value on all toplevel
27538 2017-02-24 Jakub Jelinek <jakub@redhat.com>
27541 * opts.h (handle_generated_option): Add GENERATED_P argument.
27542 * opts-common.c (handle_option): Adjust function comment.
27543 (handle_generated_option): Add GENERATED_P argument, pass it to
27545 (control_warning_option): Pass false to handle_generated_option
27547 * opts.c (maybe_default_option): Pass true to handle_generated_option
27549 * optc-gen.awk: Likewise.
27551 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27553 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
27554 a REG, look at the REG it is a SUBREG of.
27555 (splitter for cmpeqsi_t): Ditto.
27557 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27559 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
27560 the special USEs with the pattern of the insn, not the insn itself.
27562 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
27565 * doc/invoke.texi: Document -mload-store-pairs.
27567 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27568 Sandra Loosemore <sandra@codesourcery.com>
27570 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
27571 argument isn't a CONST_INT.
27572 (nios2_alternate_compare_const): Assert op is a CONST_INT.
27573 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
27574 (nios2_validate_compare): Bypass alternate compare logic if *op2
27575 is not a CONST_INT.
27576 (ldstwm_operation_p): Return false if first_base is not a REG or
27577 if first_offset is not a CONST_INT.
27579 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27581 * config/cris/cris.md: Use correct operand in a define_peephole2.
27583 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27585 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
27587 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27589 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
27590 this_insn if it is an INSN or JUMP_INSN.
27591 (force_offsettable): Look at base, not at addr.
27592 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
27593 on things that aren't necessarily CONST_INTs.
27595 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
27597 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
27598 -mfpmath=sse is the default also for x86-32 targets with SSE2
27599 instruction set when @option{-ffast-math} is enabled
27601 2017-02-24 Jeff Law <law@redhat.com>
27603 PR rtl-optimizatoin/79286
27604 * ira.c (update_equiv_regs): Drop may_trap_p exception to
27607 2017-02-24 Richard Biener <rguenther@suse.de>
27609 PR tree-optimization/79389
27610 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
27613 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
27615 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
27616 function comment to reflect reality.
27617 (loop_exits_before_overflow): Fix typo in function description.
27619 2017-02-24 Richard Biener <rguenther@suse.de>
27621 PR tree-optimization/79389
27622 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
27623 properly that a threading opportunity exists. Detect conditional
27624 copy/constant propagation opportunities.
27626 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
27628 * config/visium/visium.md (type): Add trap.
27629 (b): New mode attribute.
27630 (*btst): Rename into...
27631 (*btst<mode>): ...this and adjust.
27632 (*cbranchsi4_btst_insn): Rename into...
27633 (*cbranch<mode>4_btst_insn): ...this and adjust.
27634 (trap): New define_insn.
27636 2017-02-23 Jakub Jelinek <jakub@redhat.com>
27638 PR tree-optimization/79389
27639 * ifcvt.c (struct noce_if_info): Add rev_cond field.
27640 (noce_reversed_cond_code): New function.
27641 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
27642 reversed_comparison_code. Formatting fix.
27643 (noce_try_store_flag): Test rev_cond != NULL in addition to
27644 reversed_comparison_code.
27645 (noce_try_store_flag_constants): Likewise.
27646 (noce_try_store_flag_mask): Likewise.
27647 (noce_try_addcc): Use rev_cond if non-NULL instead of
27648 reversed_comparison_code.
27649 (noce_try_cmove_arith): Likewise. Formatting fixes.
27650 (noce_try_minmax, noce_try_abs): Clear rev_cond.
27651 (noce_find_if_block): Initialize rev_cond.
27652 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
27653 instead of false as last argument never attempt to reverse it
27656 2017-02-23 Bin Cheng <bin.cheng@arm.com>
27658 PR tree-optimization/79663
27659 * tree-predcom.c (combine_chains): Process refs in reverse order
27660 only for ZERO length chains, and add explaining comment.
27662 2017-02-23 Jeff Law <law@redhat.com>
27664 PR tree-optimization/79578
27665 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
27666 in call to operand_equal_p.
27668 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
27671 * config/i386/cpuid.h: Fix another undefined behavior.
27673 2017-02-23 Richard Biener <rguenther@suse.de>
27675 PR tree-optimization/79683
27676 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
27677 vector types for data-refs.
27679 2017-02-23 Martin Liska <mliska@suse.cz>
27681 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
27683 2017-02-23 Jakub Jelinek <jakub@redhat.com>
27685 PR middle-end/79665
27686 * internal-fn.c (get_range_pos_neg): Moved to ...
27687 * tree.c (get_range_pos_neg): ... here. No longer static.
27688 * tree.h (get_range_pos_neg): New prototype.
27689 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
27690 are known to be in between 0 and signed maximum inclusive, try to
27691 expand both unsigned and signed divmod and use the cheaper one from
27694 2017-02-22 Jeff Law <law@redhat.com>
27696 PR tree-optimization/79578
27697 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
27698 to compare base operands.
27700 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
27703 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
27704 gpc_reg_operand instead of fpr_reg_operand.
27706 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
27708 * config/mips/mips.c (mips_return_in_memory): Force FP
27709 vector types to be returned in memory for o32 ABI.
27711 2017-02-22 Jakub Jelinek <jakub@redhat.com>
27713 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
27714 instead of DW_TAG_member for static data member declarations and don't
27715 set no_linkage_name for static inline data members.
27716 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
27719 2017-02-22 Martin Liska <mliska@suse.cz>
27721 * doc/invoke.texi: Replace inequality signs with square brackets
27724 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27726 PR tree-optimization/68644
27727 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
27729 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
27732 * lra-constraints.c (simplify_operand_subreg): Handle
27733 WORD_REGISTER_OPERATIONS targets.
27735 2017-02-22 Jakub Jelinek <jakub@redhat.com>
27738 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
27739 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
27740 elimination by swapping fld*.
27742 2017-02-22 Richard Biener <rguenther@suse.de>
27744 PR tree-optimization/79673
27745 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
27746 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
27747 irrelevant address-space qualifiers and avoiding a
27748 ADDR_SPACE_CONVERT_EXPR from fold_convert.
27750 2017-02-22 Richard Biener <rguenther@suse.de>
27752 PR tree-optimization/79666
27753 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
27754 to not symbolically negate if that may introduce undefined
27757 2017-02-22 Martin Liska <mliska@suse.cz>
27760 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
27761 * data-streamer-out.c (streamer_write_gcov_count_stream):
27763 * value-prof.c (stream_out_histogram_value): Make assert more
27764 precise based on type of counter.
27766 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
27769 * config/i386/i386.md (standard_x87sse_constant_load splitter):
27770 Use nonimmediate_operand instead of memory_operand for operand 1.
27771 (float-extend standard_x87sse_constant_load splitter): Ditto.
27773 2017-02-21 Jeff Law <law@redhat.com>
27775 PR tree-optimization/79621
27776 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
27777 blocks with edges to themselves.
27779 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27782 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
27783 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
27784 Use gimple_call_builtin_p.
27787 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
27788 on temporarily removed DEBUG_INSNs.
27790 PR tree-optimization/79649
27791 * tree-loop-distribution.c (classify_partition): Give up on
27792 non-generic address space loads/stores.
27794 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
27796 * doc/loop.texi (Loop manipulation): Remove nonexistent
27797 tree_ssa_loop_version from the documentation.
27798 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
27800 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27803 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
27804 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
27805 * config/rs6000/rs6000.c: Include except.h.
27806 (rs6000_expand_split_stack_prologue): Call
27807 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
27809 2017-02-21 Martin Jambor <mjambor@suse.cz>
27812 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
27813 have been analyzed.
27815 2017-02-21 Martin Jambor <mjambor@suse.cz>
27817 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
27818 for backward compatibility only.
27819 * doc/invoke.texi (Option Summary): Remove all references to
27820 -fipa-cp-alignment.
27822 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
27826 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27828 * lra-constraints.c (curr_insn_transform): Handle
27829 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
27831 2017-02-21 Martin Liska <mliska@suse.cz>
27833 * config/i386/i386.opt: Replace -masm-dialect with -masm.
27835 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
27837 PR translation/79638
27838 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
27840 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
27843 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
27844 (arm_function_ok_for_sibcall): Return false for an indirect call by
27845 descriptor if all the argument registers are used.
27846 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
27847 alignment of the function.
27849 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27851 PR tree-optimization/61441
27852 * simplify-rtx.c (simplify_const_unary_operation): For
27853 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
27854 the sNaN unmodified.
27856 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27858 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27859 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27860 instead of SYSTEM_HEADER_DIR.
27862 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
27863 Martin Liška <mliska@suse.cz>
27865 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
27866 Fix typos and grammar, use active voice, and clarify.
27868 2017-02-20 Marek Polacek <polacek@redhat.com>
27870 PR middle-end/79537
27871 * gimplify.c (gimplify_expr): Handle unused *&&L;.
27874 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
27876 2017-02-20 Jakub Jelinek <jakub@redhat.com>
27879 * config/i386/i386.c (ix86_expand_builtin): Handle
27880 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
27881 ix86_builtins_isa[fcode].isa as a requirement of those
27882 flags and any other flag in the bitmask.
27883 (ix86_init_mmx_sse_builtins): Use 0 instead of
27884 ~OPTION_MASK_ISA_64BIT as mask.
27885 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
27886 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
27887 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
27888 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
27890 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27893 * lra-constraints.c (split_reg): Check requested split mode
27894 is supported by the register.
27896 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27898 * lra-constraints.c (simplify_operand_subreg): Remove early
27901 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27904 * lra-constraints.c (curr_insn_transform): Tighten condition
27905 for converting SUBREG reloads from OP_OUT to OP_INOUT.
27907 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27910 * lra-constraints.c (curr_insn_transform): Handle
27911 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
27913 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
27916 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
27918 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
27920 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
27923 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
27926 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27928 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
27931 2017-02-18 Jakub Jelinek <jakub@redhat.com>
27933 * final.c (last_columnnum, override_columnnum): New variables.
27934 (final_start_function): Set last_columnnum, pass it to begin_prologue
27935 hook and pass 0 to dwarf2out_begin_prologue.
27936 (final_scan_insn): Update override_columnnum. Pass last_columnnum
27937 to source_line debug hook.
27938 (notice_source_line): Compute last_columnnum and for debug_column_info
27939 return true on column changes.
27940 * debug.h (struct gcc_debug_hooks): Add column argument to
27941 source_line and begin_prologue hooks.
27942 (debug_nothing_int_charstar_int_bool): Remove prototype.
27943 (debug_nothing_int_int_charstar,
27944 debug_nothing_int_int_charstar_int_bool): New prototypes.
27945 (dwarf2out_begin_prologue): Add column argument.
27946 * debug.c (do_nothing_debug_hooks): Adjust source_line and
27947 begin_prologue hooks.
27948 (debug_nothing_int_charstar_int_bool): Remove.
27949 (debug_nothing_int_int_charstar,
27950 debug_nothing_int_int_charstar_int_bool): New functions.
27951 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
27952 through to dwarf2out_source_line.
27953 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
27954 (dwarf2out_source_line): Add column argument, emit it if requested.
27955 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
27957 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
27958 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
27959 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
27960 through to dwarf2out_begin_prologue.
27961 (vmsdbgout_source_line): Add column argument, pass it through to
27962 dwarf2out_source_line.
27963 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
27964 dbxout_source_line caller.
27965 (dbxout_source_line): Add column argument.
27967 * common.opt (gno-column-info, gcolumn-info): New options.
27968 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
27969 (check_die): Also test for multiple DW_AT_decl_column attributes.
27970 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
27971 DW_AT_decl_column if requested.
27972 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
27974 (gen_variable_die): Likewise.
27975 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
27976 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
27979 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
27980 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
27981 (ix86_handle_option): Handle OPT_m3dnowa.
27982 * doc/invoke.texi (-m3dnowa): Document.
27983 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
27984 -m3dnowa instead of -m3dnow -march=athlon.
27987 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
27988 instead of gcc_assert for K, r and R code checks. Formatting fixes.
27990 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27993 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
27994 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
27995 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
27996 generator for vsx_xxpermdi_<mode>_be.
27997 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
27998 force big-endian semantics.
27999 (vsx_xxpermdi_<mode>_be): New define_expand with same
28000 implementation as previous version of vsx_xxpermdi_<mode>.
28002 2017-02-17 Jakub Jelinek <jakub@redhat.com>
28004 PR tree-optimization/79327
28005 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
28006 variable, its initialization and use.
28008 2017-02-17 Julia Koval <julia.koval@intel.com>
28010 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
28011 (OPTION_MASK_ISA_PKU_UNSET): New.
28012 (ix86_handle_option): Handle -mrdpid.
28013 * config/i386/cpuid.h (bit_RDPID): New.
28014 * config/i386/driver-i386.c (host_detect_local_cpu):
28015 Detect RDPID feature.
28016 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
28017 * config/i386/i386-c.c (ix86_target_macros_internal):
28019 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
28020 (ix86_valid_target_attribute_inner_p): Add "rdpid".
28021 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
28022 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
28023 * config/i386/i386.md (define_insn "rdpid"): New.
28024 * config/i386/i386.opt Add -mrdpid.
28025 * config/i386/immintrin.h (_rdpid_u32): New.
28027 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
28029 PR rtl-optimization/79541
28030 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
28031 instead of transforming it into USE.
28033 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
28035 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
28036 If HONOR_SNANS (SFmode) force the input to a register.
28037 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
28038 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
28039 an frsp or similar insn.
28041 2017-02-17 Martin Liska <mliska@suse.cz>
28043 PR rtl-optimization/79577
28044 * params.def (selsched-max-sched-times): Increase minimum to 1.
28046 2017-02-17 Martin Liska <mliska@suse.cz>
28048 PR rtl-optimization/79574
28049 * gcse.c (want_to_gcse_p): Prevent integer overflow.
28051 2017-02-17 Martin Liska <mliska@suse.cz>
28053 PR tree-optimization/79529
28054 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
28055 ssa_defined_default_def_p to handle cases which are implicitly
28057 * tree-ssa.c (ssa_defined_default_def_p): New function.
28058 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
28059 which are implicitly defined.
28060 * tree-ssa.h (ssa_defined_default_def_p): Declare.
28062 2017-02-17 Richard Biener <rguenther@suse.de>
28064 PR middle-end/79576
28065 * params.def (max-ssa-name-query-depth): Limit to 10.
28067 2017-02-17 Richard Biener <rguenther@suse.de>
28069 PR tree-optimization/79552
28070 * tree-ssa-structalias.c (visit_loadstore): Properly verify
28073 2017-02-17 Richard Biener <rguenther@suse.de>
28076 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
28078 2017-02-17 Marek Polacek <polacek@redhat.com>
28080 PR middle-end/79536
28081 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
28082 (fold_negate_expr): New wrapper.
28084 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
28086 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
28087 Correct terminology and de-emphasize pre-standard behavior.
28089 2017-02-16 Alan Modra <amodra@gmail.com>
28091 PR rtl-optimization/79286
28092 * ira.c (def_dominates_uses): New function.
28093 (update_equiv_regs): Don't create an equivalence for insns that
28094 may trap where the register def does not dominate the use.
28096 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
28098 PR rtl-optimization/78127
28099 * lra.c (lra): Call lra_eliminate before finish the loop after
28102 2017-02-16 Richard Biener <rguenther@suse.de>
28104 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
28106 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
28107 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
28108 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
28109 (isl_val_int_from_wi): New function.
28110 (extract_affine_gmp): Rename to ...
28111 (extract_affine_wi): ... this, take a widest_int.
28112 (extract_affine_int): Just wrap extract_affine_wi.
28113 (add_param_constraints): Use isl_val_int_from_wi.
28114 (add_loop_constraints): Likewise, and extract_affine_wi.
28116 2017-02-15 Jeff Law <law@redhat.com>
28118 PR middle-end/79521
28119 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
28120 ira_init_register_move_cost_if_necessary.
28122 2017-02-15 Martin Sebor <msebor@redhat.com>
28124 PR middle-end/32003
28125 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
28126 removed in a prior commit.
28128 2017-02-15 Bin Cheng <bin.cheng@arm.com>
28130 PR tree-optimization/79347
28131 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
28132 counters during peeling.
28134 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
28136 * Makefile.in (site.exp): Remove "set ISLVER".
28138 2017-02-15 Jakub Jelinek <jakub@redhat.com>
28141 * real.c (real_from_integer): Call real_convert even for decimal.
28143 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
28146 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
28148 2017-02-14 Andrew Pinski <apinski@cavium.com>
28150 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
28151 cores and change the partno/implementer to be correct.
28152 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
28153 the 'B" as the implementer.
28154 * config/aarch64/aarch64-tune.md: Regenerate.
28156 2017-02-14 Carl Love <cel@us.ibm.com>
28158 * config/rs6000/rs6000.c: Add case statement entry to make the
28159 xvcvuxdsp built-in argument unsigned.
28160 * config/rs6000/vsx.md: Fix the source and return operand types so they
28161 match the instruction definitions from the ISA document. Fix typo
28162 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
28165 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
28168 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
28169 member early_clobber_alts.
28170 * lra-lives.c (reg_early_clobber_p): New.
28171 (process_bb_lives): Use it.
28172 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
28173 (debug_operand_data): Initialize early_clobber_alts.
28174 (setup_operand_alternative): Set up early_clobber_alts.
28175 (collect_non_operand_hard_regs): Ditto. Pass early clobber
28176 alternatives to new_insn_reg.
28177 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
28179 (lra_update_insn_regno_info): Pass the new arg.
28181 2017-02-14 Jakub Jelinek <jakub@redhat.com>
28183 PR middle-end/79505
28184 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
28185 (new_oacc_loop_raw): Don't clear already cleared fields.
28188 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
28189 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
28190 _mm512_prefetch_i64gather_ps): New inline functions and macros.
28192 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
28195 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
28197 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
28200 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
28201 the extra instruction to the right place to store 128-bit constant
28204 2017-02-14 Martin Sebor <msebor@redhat.com>
28206 PR middle-end/79448
28207 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
28208 warning for strings of unknown length.
28210 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
28212 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
28214 2017-02-14 Jeff Law <law@redhat.com>
28217 * ira-costs.c (scan_one_insn): Initialize register move costs
28218 for pseudos seen in USE/CLOBBER insns.
28220 PR tree-optimization/79095
28221 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
28222 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
28223 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
28224 if the operands are known to be not equal, then the resulting range
28226 (intersect_ranges): If the new range is ~[0,0] and the old range is
28227 wide, then prefer ~[0,0].
28228 * tree-vrp.c (overflow_comparison_p_1): New function.
28229 (overflow_comparison_p): New function.
28230 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
28231 if NAME is used in an overflow test.
28232 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
28233 overflow check that can be expressed as an equality test, then adjust
28234 ops to be that equality test.
28236 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28238 * config/s390/s390-builtin-types.def: Remove flags argument.
28239 * config/s390/s390.c (s390_init_builtins): Likewise.
28241 2017-02-14 Martin Liska <mliska@suse.cz>
28243 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
28244 vector. Fix trailing white spaces.
28246 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
28248 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
28251 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28253 PR rtl-optimization/68664
28254 * config/arm/arm.c (arm_sched_can_speculate_insn):
28255 New function. Declare prototype.
28256 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
28258 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28260 PR rtl-optimization/68664
28261 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
28263 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
28265 2017-02-14 Amit Pawar <amit.pawar@amd.com>
28267 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
28268 max skip bytes for function, loop and jump.
28270 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28272 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
28273 ABS_EXPR for gimple dump.
28275 2017-02-14 Jakub Jelinek <jakub@redhat.com>
28278 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
28280 PR tree-optimization/79408
28281 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
28282 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
28283 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
28284 also if rhs1 is INTEGER_CST.
28286 2017-02-14 Richard Biener <rguenther@suse.de>
28288 PR middle-end/79432
28289 * tree-into-ssa.c (insert_phi_nodes): When the function can
28290 have abnormal edges rewrite SSA names with broken use-def
28291 dominance out of SSA and register them for PHI insertion.
28293 2017-02-13 Martin Sebor <msebor@redhat.com>
28295 PR middle-end/79496
28296 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
28297 clearing info.nowrite flag when snprintf size argument is a range.
28299 2017-02-13 Jakub Jelinek <jakub@redhat.com>
28301 * cprop.c (cprop_jump): Add missing space in string literal.
28302 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
28303 (get_constraint_for_component_ref): Likewise.
28304 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
28305 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
28306 * lra-constraints.c (process_alt_operands): Likewise.
28307 * ipa-inline.c (inline_small_functions): Likewise.
28308 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
28309 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
28310 * trans-mem.c (diagnose_tm_1_op): Likewise.
28311 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
28312 (grid_parallel_clauses_gridifiable): Likewise.
28314 * config/nvptx/mkoffload.c (process): Add space in between
28317 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
28318 "MOD4_SSE_REGS" and "ALL_REGS".
28320 * spellcheck.c (test_data): Add , in between "foo" and "food".
28322 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28325 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
28326 boundary crossing check and subsequent code generation agree.
28328 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28330 * config/aarch64/aarch64.c (has_memory_op): Delete.
28331 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
28334 2017-02-13 Jakub Jelinek <jakub@redhat.com>
28336 PR rtl-optimization/79388
28337 PR rtl-optimization/79450
28338 * combine.c (distribute_notes): When removing TEM_INSN for which
28339 corresponding dest has last value recorded, invalidate that last
28342 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28344 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
28345 of explicit '@'. Add missing assembly comment marker on branch costs
28348 2017-02-13 Nathan Sidwell <nathan@acm.org>
28350 * gengtype-lex.l (<in_struct>): Add '/'.
28352 2017-02-13 Martin Liska <mliska@suse.cz>
28355 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
28357 2017-02-13 Richard Biener <rguenther@suse.de>
28359 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
28361 * configure: Re-generate.
28362 * config.in: Likewise.
28363 * graphite-dependences.c: Simplify as if
28364 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
28365 * graphite-isl-ast-to-gimple.c: Likewise.
28366 * graphite-optimize-isl.c: Likewise.
28367 * graphite-poly.c: Likewise.
28368 * graphite-sese-to-poly.c: Likewise.
28369 * graphite.h: Likewise.
28370 * toplev.c: Include isl/version.h and use isl_version () for
28371 printing the ISL version.
28372 * doc/install.texi: Update ISL requirement.
28374 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
28376 * doc/standards.texi (Standards): Update reference to
28379 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
28381 * doc/extend.texi (Named Address Spaces): sourceware.org now
28383 * doc/install.texi (Binaries): Ditto.
28386 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
28388 * doc/cpp.texi: Replace "stringify"/"stringification" with C
28389 standard terminology "stringize"/"stringizing" throughout.
28390 * doc/cppinternals.texi: Likewise.
28392 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
28394 * doc/extend.texi: Fix some spelling mistakes and typos.
28395 * doc/invoke.texi: Likewise.
28397 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
28400 * params.def (inline-min-speedup) Change from 10 to 8.
28402 2017-02-11 Jakub Jelinek <jakub@redhat.com>
28404 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
28407 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
28410 * ipa-inline-analysis.c (get_minimal_bb): New function.
28411 (record_modified): Use it.
28412 (remap_edge_change_prob): Handle also ancestor functions.
28414 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
28416 * doc/contrib.texi (Contributors): Remove broken link into
28417 the Mauve CVS repository.
28419 2017-02-11 Jakub Jelinek <jakub@redhat.com>
28421 PR middle-end/79454
28422 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
28423 result computation whenever lhs doesn't have vector mode, not
28424 just when it has BLKmode.
28426 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
28428 * doc/makefile.texi (profiledbootstrap): Refer to the
28429 installation instructions only in textual form.
28431 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28434 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
28436 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
28438 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
28439 (Specific): Update mingw-w64 reference.
28441 (Specific): Remove broken link to Renesas RX processor.
28443 2017-02-10 Richard Biener <rguenther@suse.de>
28445 * toplev.c (process_options): Do not mention obsolete graphite
28446 options when printing sorry message about missing graphite support.
28447 Mention -floop-nest-optimize.
28449 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
28451 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
28452 (vtst_p16): Likewise.
28453 (vtstq_p8): Likewise.
28454 (vtstq_p16): Likewise.
28456 (vtstq_p64): Likewise.
28457 * config/arm/arm_neon.h (vgetq_lane_p64): New.
28458 (vset_lane_p64): New.
28459 (vsetq_lane_p64): New.
28461 2017-02-10 Jakub Jelinek <jakub@redhat.com>
28463 PR tree-optimization/79411
28464 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
28465 stmt operands are SSA_NAMEs used in abnormal phis.
28466 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
28469 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
28472 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
28475 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
28477 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
28479 2017-02-09 Jakub Jelinek <jakub@redhat.com>
28481 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
28485 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
28486 not arbitrary TREE_CONSTANT.
28489 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
28490 "omp declare target link" attribute unless is_global_var.
28491 * omp-offload.c (find_link_var_op): Likewise.
28493 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
28494 Chung-Lin Tang <cltang@codesourcery.com>
28496 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
28498 (gimplify_adjust_omp_clauses): Don't delete TILE.
28499 (gimplify_omp_for): Deal with TILE.
28500 * internal-fn.c (expand_GOACC_TILE): New function.
28501 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
28503 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
28504 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
28506 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
28507 avoid DIV for outermost collapse var.
28508 (expand_oacc_for): Insert tile element loop as needed. Adjust.
28509 Remove out of date comments, fix whitespace.
28510 * omp-general.c (omp_extract_for_data): Deal with tiling.
28511 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
28512 adjust OLF_DIM_BASE value.
28513 (struct omp_for_data): Add tiling field.
28514 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
28515 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
28516 for auto loops. Remove default auto determining, moved to
28517 oacc_loop_fixed_partitions.
28518 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
28519 stmts, add e_mask field.
28520 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
28521 (oacc_thread_numbers): Use oacc_dim_call.
28522 (oacc_xform_tile): New.
28523 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
28524 (finish_oacc_loop): Adjust for ifns vector.
28525 (oacc_loop_discover_walk): Append loop abstraction sites to list,
28526 add case for GOACC_TILE fns.
28527 (oacc_loop_xform_loop): Delete.
28528 (oacc_loop_process): Iterate over call list directly, and add
28529 handling for GOACC_TILE fns.
28530 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
28532 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
28533 vector partitioning to outer loops. Assign 2 partitions to loops
28534 when available. Add TILE handling.
28535 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
28536 (execite_oacc_device_lower): Process GOACC_TILE fns,
28537 ignore unknown specs.
28538 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
28539 * tree.c (omp_clause_num_ops): Adjust TILE ops.
28540 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
28542 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
28544 * configure.ac (ACX_BUGURL): Update.
28545 * configure: Regenerate.
28547 2017-02-09 Richard Biener <rguenther@suse.de>
28549 PR tree-optimization/69823
28550 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
28551 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
28553 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
28555 * config/arc/arc-c.def: Add __NPS400__ definition.
28556 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
28557 (TARGET_NPS400): Define.
28559 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
28561 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
28563 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
28564 pointer, arch_info.
28565 (arc_cpu_types): Fill the arch_info field with a pointer into the
28566 arc_arch_types table.
28567 (arc_selected_cpu): Declare.
28568 * config/arc/arc.c (arc_selected_cpu): Make global.
28569 (arc_selected_arch): Delete.
28570 (arc_base_cpu): Delete.
28571 (arc_override_options): Remove references to deleted variables,
28572 update access to arch information.
28573 (ARC_OPT): Update access to arch information.
28574 (ARC_OPTX): Likewise.
28575 * config/arc/arc.h (arc_base_cpu): Remove declaration.
28576 (TARGET_ARC600): Update access to arch information.
28577 (TARGET_ARC601): Likewise.
28578 (TARGET_ARC700): Likewise.
28579 (TARGET_EM): Likewise.
28580 (TARGET_HS): Likewise.
28581 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
28584 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
28587 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
28588 condition/operands for integer GE/LE/GEU/LEU operations.
28590 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
28592 PR translation/79397
28593 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
28596 2017-02-08 Martin Jambor <mjambor@suse.cz>
28599 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
28600 whether allocation happened.
28601 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
28602 nothing was allocated.
28604 2017-02-08 Jakub Jelinek <jakub@redhat.com>
28606 PR tree-optimization/79408
28607 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
28608 constant, but SSA_NAME with a known integer range, use the minimum
28609 of that range instead of op1 to determine if modulo can be replaced
28610 with its first operand.
28612 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28614 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
28616 2017-02-08 Richard Biener <rguenther@suse.de>
28618 PR tree-optimization/71824
28619 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
28620 Check all loops contained in the merged region.
28622 2017-02-07 Andrew Pinski <apinski@cavium.com>
28624 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
28626 2017-02-07 Andrew Pinski <apinski@cavium.com>
28628 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
28629 (thunderxt88): Likewise.
28630 (thunderxt81): Disable LSE and change v8.1 to v8.
28631 (thunderxt83): Likewise.
28633 2017-02-07 Jakub Jelinek <jakub@redhat.com>
28634 Richard Biener <rguenther@suse.de>
28636 PR middle-end/79399
28637 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
28638 type from int to size_t.
28639 * ira-costs.c (struct_costs_size): Change type from int to size_t.
28641 2017-02-07 Jakub Jelinek <jakub@redhat.com>
28643 PR rtl-optimization/79386
28644 * cprop.c (bypass_conditional_jumps): Initialize
28645 bypass_last_basic_block already before splitting bbs after
28646 unconditional traps...
28647 (bypass_conditional_jumps): ... rather than here.
28650 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
28651 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
28652 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
28653 fix -masm=intel patterns.
28655 2017-02-07 Richard Biener <rguenther@suse.de>
28657 PR tree-optimization/79256
28658 PR middle-end/79278
28659 * builtins.c (get_object_alignment_2): Use min_align_of_type
28660 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
28661 and ADJUST_FIELD_ALIGN.
28663 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
28665 * doc/tm.texi: Regenerate.
28666 * stor-layout.c (layout_decl): Adjust.
28667 (update_alignment_for_field): Likewise.
28668 (place_field): Likewise.
28669 (min_align_of_type): Likewise.
28670 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
28671 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
28672 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
28673 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
28674 * config/frv/frv.c (frv_adjust_field_align): Likewise.
28675 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
28676 * config/i386/i386.c (x86_field_alignment): Likewise.
28677 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
28678 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
28679 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
28680 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
28681 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
28682 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
28686 2017-01-30 Richard Biener <rguenther@suse.de>
28688 PR tree-optimization/79256
28689 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
28690 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
28693 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
28695 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
28696 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
28697 builtins to SImode and emit a zero-extend, if necessary.
28699 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28701 * docs/invoke.texi (RISC-V Options): Alphabetize.
28703 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28705 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
28708 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28710 * config/riscv/riscv.c: New file.
28711 * common/config/riscv/riscv-common.c: Likewise.
28712 * config.gcc: Likewise.
28713 * config/riscv/constraints.md: Likewise.
28714 * config/riscv/elf.h: Likewise.
28715 * config/riscv/generic.md: Likewise.
28716 * config/riscv/linux.h: Likewise.
28717 * config/riscv/multilib-generator: Likewise.
28718 * config/riscv/peephole.md: Likewise.
28719 * config/riscv/pic.md: Likewise.
28720 * config/riscv/predicates.md: Likewise.
28721 * config/riscv/riscv-builtins.c: Likewise.
28722 * config/riscv/riscv-c.c: Likewise.
28723 * config/riscv/riscv-ftypes.def: Likewise.
28724 * config/riscv/riscv-modes.def: Likewise.
28725 * config/riscv/riscv-opts.h: Likewise.
28726 * config/riscv/riscv-protos.h: Likewise.
28727 * config/riscv/riscv.h: Likewise.
28728 * config/riscv/riscv.md: Likewise.
28729 * config/riscv/riscv.opt: Likewise.
28730 * config/riscv/sync.md: Likewise.
28731 * config/riscv/t-elf-multilib: Likewise.
28732 * config/riscv/t-linux: Likewise.
28733 * config/riscv/t-linux-multilib: Likewise.
28734 * config/riscv/t-riscv: Likewise.
28735 * configure.ac: Likewise.
28736 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
28737 Waterman as RISC-V maintainers.
28738 * doc/install.texi: Add RISC-V entries.
28739 * doc/invoke.texi: Add RISC-V options section.
28740 * doc/md.texi: Add RISC-V constraints section.
28741 * configure: Regenerated.
28743 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
28746 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
28747 false values to be constant vectors with all 0 or all 1 bits set.
28748 (vcondu<mode><mode>): Likewise.
28749 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
28751 (fpmask_comparison_operator): Update comment.
28752 (vecint_comparison_operator): New predicate.
28753 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
28754 vector conditionals when the true and false values are constant
28755 vectors with all 0 bits or all 1 bits set.
28757 2017-02-06 Martin Sebor <msebor@redhat.com>
28759 PR tree-optimization/79376
28760 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
28762 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
28764 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
28765 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
28766 to simplify split condition.
28768 2017-02-06 Jakub Jelinek <jakub@redhat.com>
28770 * omp-expand.c (oxpand_omp_atomic_fetch_op,
28771 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
28774 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
28776 PR rtl-optimization/68664
28777 * target.def (can_speculate_insn): New hook.
28778 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
28779 * doc/tm.texi: Regenerate.
28780 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
28781 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
28782 (rs6000_sched_can_speculate_insn): New function.
28784 2017-02-06 Jakub Jelinek <jakub@redhat.com>
28786 PR tree-optimization/79284
28787 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
28788 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
28789 vectorizable_mask_load_store, vectorizable_operation,
28790 vect_is_simple_cond, get_same_sized_vectype): Use it instead
28791 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
28792 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
28793 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
28794 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
28795 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28796 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
28797 is_gimple_assign (stmt). Replace another such test with
28798 is_gimple_assign (stmt).
28800 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
28803 * config/avr/avr.c (rtl-iter.h): Include it.
28804 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
28805 (avr_legitimate_combined_insn): ...and implementation.
28807 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
28809 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
28810 * config/s390/s390.c (s390_const_operand_ok)
28811 (s390_canonicalize_comparison, s390_extract_part)
28812 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
28813 (s390_contiguous_bitmask_p, s390_rtx_costs)
28814 (legitimize_pic_address): Likewise.
28815 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
28816 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
28817 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
28818 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
28819 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
28821 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
28823 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
28824 REGNO($0) == REGNO($1).
28826 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28828 * config/s390/linux.h(SIZE_TYPE): Add comment.
28830 2017-02-06 Julian Brown <julian@codesourcery.com>
28831 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28832 Virendra Pathak <virendra.pathak@broadcom.com>
28834 * config/aarch64/aarch64-cores.def: Change the scheduler
28836 * config/aarch64/aarch64.md: Include thunderx2t99.md.
28837 * config/aarch64/thunderx2t99.md: New file.
28839 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28841 * doc/standards.texi (Go Language): Update link to language
28844 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
28846 * tree-eh.c (lower_resx): Sanitize profile.
28847 (cleanup_empty_eh_move_lp): Likewise.
28849 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
28852 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
28854 * cfgloopmanip.h (loop_version): Update prototype.
28855 * modulo-sched.c (sms_schedule): Update call of loop_version.
28856 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
28857 * tree-parloops.c (gen_parallel_loop): Likewise.
28858 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
28859 * tree-ssa-loop-split.c (split_loop): Likewise.
28860 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
28861 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
28863 2017-02-05 Martin Liska <mliska@suse.cz>
28866 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
28868 (print_operand_address): Initialize a struct to zero.
28870 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28872 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
28873 garbage collector only in textual form.
28875 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28877 * doc/extend.texi (x86 specific memory model extensions for
28878 transactional memory): Simplify a phrase.
28880 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
28883 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
28884 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
28885 (atomic_storedi_1): Likewise.
28887 2017-02-04 Jakub Jelinek <jakub@redhat.com>
28889 PR tree-optimization/79338
28890 * tree-parloops.c (gather_scalar_reductions): Don't call
28891 vect_analyze_loop_form for loop->inner before destroying loop's
28894 2017-02-03 Martin Sebor <msebor@redhat.com>
28896 PR tree-optimization/79327
28897 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
28898 when precision has resulted in leading zeros.
28899 (format_integer): Adjust the likely counter to assume an unknown
28900 argument that may be zero is non-zero.
28902 2017-02-03 Jason Merrill <jason@redhat.com>
28905 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
28906 avoid copying non-taken branch.
28908 2017-02-03 Jakub Jelinek <jakub@redhat.com>
28910 PR tree-optimization/79340
28911 * tree-vect-loop.c (vectorizable_reduction): Release
28912 vec_defs elements after safe_splicing them into other vectors.
28915 PR tree-optimization/79327
28916 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
28917 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
28919 (format_integer): Use wide_int_to_tree instead of build_int_cst
28920 + to_?hwi. If argmin is NULL, just set argmin and argmax to
28921 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
28922 of shortest and longest sequence.
28924 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
28926 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
28927 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
28929 2017-02-03 Walter Lee <walt@tilera.com>
28932 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
28933 after initial stackframe link reg save.
28934 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
28936 2017-02-03 Jakub Jelinek <jakub@redhat.com>
28939 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
28940 wu for stxssp alternative.
28942 2017-02-03 Martin Sebor <msebor@redhat.com>
28944 PR tree-optimization/79352
28945 * gimple-fold.c (get_range_strlen): Add argument.
28946 (get_range_strlen): Change return type to bool.
28947 (get_maxval_strlen): Pass in a dummy argument.
28948 * gimple-fold.h (get_range_strlen): Change return type to bool.
28949 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
28950 * tree.h (array_at_struct_end_p): Add argument.
28951 * tree.c (array_at_struct_end_p): Handle it.
28953 2017-02-03 Martin Liska <mliska@suse.cz>
28956 * multiple_target.c (create_dispatcher_calls): Redirect edge
28957 from a caller of a dispatcher.
28958 (expand_target_clones): Make the clones local.
28959 (ipa_target_clone): Do both target clones and resolvers.
28960 (ipa_dispatcher_calls): Remove the pass.
28961 (pass_dispatcher_calls::gate): Likewise.
28962 (make_pass_dispatcher_calls): Likewise.
28963 * passes.def (pass_target_clone): Put as very first IPA early
28966 2017-02-03 Martin Liska <mliska@suse.cz>
28968 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
28969 in case of a function with ifunc attribute.
28971 2017-02-03 Martin Liska <mliska@suse.cz>
28973 * cgraph.c (cgraph_node::dump): Dump function version info.
28974 * symtab.c (symtab_node::dump_base): Add missing new line.
28976 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
28978 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
28979 (ifcombine_ifandif): Use it.
28981 2017-02-03 Martin Liska <mliska@suse.cz>
28983 * doc/invoke.texi: Document default value for
28984 use-after-scope-direct-emission-threshold.
28986 2017-02-03 Martin Liska <mliska@suse.cz>
28988 PR tree-optimization/79339
28989 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
28990 (format_floating): Likewise.
28992 2017-02-03 Martin Liska <mliska@suse.cz>
28995 * ipa-prop.c (ipa_node_params_t::insert): Remove current
28997 (ipa_node_params_t::remove): Likewise.
28998 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
28999 initialization from removed ipa_node_params_t::insert.
29000 (ipa_node_params::~ipa_node_params): Move from removed
29001 ipa_node_params_t::release.
29002 * symbol-summary.h (symbol_summary::m_released): New member.
29003 Do not release a summary twice. Do not allow to call finalizer
29004 for types of a summary that live in GGC memory.
29006 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29008 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
29011 2017-02-02 Martin Sebor <msebor@redhat.com>
29013 PR middle-end/79275
29014 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
29015 (format_string): Tighten up the range of output for non-constant
29016 strings and correct the expected range for wide non-constant strings.
29018 2017-02-02 Martin Sebor <msebor@redhat.com>
29020 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
29022 PR middle-end/32003
29023 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
29025 (-fdump-tree-@var): Add to index and document how to come up
29026 with pass-specific option and dump file names.
29027 (-fdump-passes): Clarify where to look for output.
29029 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
29031 PR middle-end/77445
29032 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
29033 statistics of the analyzed path; allow threading for speed when
29034 any of BBs along the path are optimized for speed.
29036 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
29038 PR middle-end/78468
29039 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
29040 settings of the virtual registers.
29043 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
29045 * explow.c (get_dynamic_stack_size): Take known alignment of stack
29046 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
29049 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29051 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
29052 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
29054 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29056 * config/s390/s390.md: Add missing comments with the expanded
29058 * config/s390/vector.md: Likewise.
29059 * config/s390/vx-builtins.md: Likewise.
29061 2017-02-02 Jakub Jelinek <jakub@redhat.com>
29064 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
29065 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
29066 conditions on a single line.
29068 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29070 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
29071 __S390_VX__ to __VX__.
29073 2017-02-01 Andrew Pinski <apinski@cavium.com>
29075 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
29076 stmt_info to record_stmt_cost.
29077 (vect_get_known_peeling_cost): Pass stmt_info if known to
29079 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
29080 cpu_vector_cost field into
29081 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
29082 field into vec_int_stmt_cost and vec_fp_stmt_cost.
29083 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
29084 splitting of scalar_stmt_cost and vec_stmt_cost.
29085 (thunderx_vector_cost): Likewise.
29086 (cortexa57_vector_cost): LIkewise.
29087 (exynosm1_vector_cost): Likewise.
29088 (xgene1_vector_cost): Likewise.
29089 (thunderx2t99_vector_cost): Improve after the splitting of the two
29091 (aarch64_builtin_vectorization_cost): Update for the splitting of
29092 scalar_stmt_cost and vec_stmt_cost.
29094 2017-02-01 Torvald Riegel <triegel@redhat.com>
29095 Richard Henderson <rth@redhat.com>
29097 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
29098 conditional on existance of a fast atomic load.
29099 * optabs-query.c (can_atomic_load_p): New function.
29100 * optabs-query.h (can_atomic_load_p): Declare it.
29101 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
29102 no fast atomic load is available for the particular size of access.
29103 (expand_atomic_compare_and_swap): Likewise.
29104 (expand_atomic_load): Likewise.
29105 (expand_atomic_store): Likewise.
29106 (expand_atomic_fetch_op): Likewise.
29107 * testsuite/lib/target-supports.exp
29108 (check_effective_target_sync_int_128): Remove x86 because it provides
29109 no fast atomic load.
29110 (check_effective_target_sync_int_128_runtime): Likewise.
29112 2017-02-01 Richard Biener <rguenther@suse.de>
29114 * graphite.c: Include tree-vectorizer.h for find_loop_location.
29115 (graphite_transform_loops): Provide opt-info for optimized nests.
29116 * tree-parloop.c (parallelize_loops): Provide opt-info for
29117 parallelized loops.
29119 2017-02-01 Richard Biener <rguenther@suse.de>
29121 PR middle-end/79315
29122 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
29123 was not set before.
29125 2017-02-01 Richard Biener <rguenther@suse.de>
29127 PR tree-optimization/71824
29128 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
29129 Verify the loops are valid in the merged SESE region.
29130 (scop_detection::can_represent_loop_1): Check analyzing the
29131 evolution of the number of iterations in the region succeeds.
29133 2017-01-31 Ian Lance Taylor <iant@golang.org>
29135 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
29136 REG_ARGS_SIZE note to 32-bit push insns and call insn.
29138 2017-01-31 David Malcolm <dmalcolm@redhat.com>
29140 PR preprocessor/79210
29141 * input.c (get_substring_ranges_for_loc): Replace line_width
29142 assertion with error-handling.
29144 2017-01-31 Richard Biener <rguenther@suse.de>
29146 PR tree-optimization/77318
29147 * graphite-sese-to-poly.c (extract_affine): Fix assert.
29148 (create_pw_aff_from_tree): Take loop parameter.
29149 (add_condition_to_pbb): Pass loop of the condition to
29150 create_pw_aff_from_tree.
29152 2017-01-31 Jakub Jelinek <jakub@redhat.com>
29154 * config/s390/s390.c (s390_asan_shadow_offset): New function.
29155 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
29157 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
29161 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
29163 (convert_int_to_float128): Likewise.
29164 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
29165 (convert_int_to_float128): Likewise.
29166 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
29167 (UNSPEC_IEEE128_CONVERT): Likewise.
29168 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
29169 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
29170 Use local variables for IBM extended format.
29171 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
29172 (fix_trunc<mode>si2_fprs): Likewise.
29173 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
29174 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
29175 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
29176 to know that we can now have integers of all sizes in vector
29178 (fix<uns>_<mode>di2_hw): Likewise.
29179 (float<uns>_<mode>si2_hw): Likewise.
29180 (fix_<mode>si2_hw): Likewise.
29181 (fixuns_<mode>si2_hw): Likewise.
29182 (float<uns>_<mode>di2_hw): Likewise.
29183 (float_<mode>di2_hw): Likewise.
29184 (float_<mode>si2_hw): Likewise.
29185 (floatuns_<mode>di2_hw): Likewise.
29186 (floatuns_<mode>si2_hw): Likewise.
29187 (xscvqp<su>wz_<mode>): Delete, no longer used.
29188 (xscvqp<su>dz_<mode>): Likewise.
29189 (xscv<su>dqp_<mode>): Likewise.
29190 (ieee128_mfvsrd_64bit): Likewise.
29191 (ieee128_mfvsrd_32bit): Likewise.
29192 (ieee128_mfvsrwz): Likewise.
29193 (ieee128_mtvsrw): Likewise.
29194 (ieee128_mtvsrd_64bit): Likewise.
29195 (ieee128_mtvsrd_32bit): Likewise.
29197 2017-01-31 Martin Liska <mliska@suse.cz>
29200 * ipa-prop.c (ipa_free_all_node_params): Call release method
29201 instead of ~sumbol_summary to not to trigger double times
29204 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
29206 PR tree-optimization/71691
29207 * bitmap.h (class auto_bitmap): New.
29208 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
29209 is_maybe_undefined instead of ssa_undefined_value_p.
29211 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29213 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
29214 __S390_ARCH_LEVEL__ to __ARCH__.
29216 2017-01-31 Jakub Jelinek <jakub@redhat.com>
29218 PR tree-optimization/79267
29219 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
29220 if should_remove_lhs_p is true.
29222 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
29225 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
29226 (add_alignment_attribute): New.
29227 (base_type_die): Add alignment attribute.
29228 (subrange_type_die): Likewise.
29229 (modified_type_die): Likewise.
29230 (gen_array_type_die): Likewise.
29231 (gen_descr_array_type_die: Likewise.
29232 (gen_enumeration_type_die): Likewise.
29233 (gen_subprogram_die): Likewise.
29234 (gen_variable_die): Likewise.
29235 (gen_field_die): Likewise.
29236 (gen_ptr_to_mbr_type_die): Likewise.
29237 (gen_struct_or_union_type_die): Likewise.
29238 (gen_subroutine_type_die): Likewise.
29239 (gen_typedef_die): Likewise.
29240 (base_type_cmp): Compare alignment attribute.
29242 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29245 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
29246 (setb_unsigned) New pattern for setb with CCUNS.
29247 * config/rs6000/rs6000.c (expand_block_compare): Use a different
29248 subfc./subfe sequence to avoid overflow problems. Generate a
29249 shorter sequence with cmpld/setb for power9.
29250 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
29251 for generating subfc. instruction.
29252 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
29253 now uses this instruction.
29255 2017-01-30 Ian Lance Taylor <iant@google.com>
29258 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
29259 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
29261 2017-01-30 Martin Sebor <msebor@redhat.com>
29263 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
29264 Move constant to the right of a relational operator.
29265 (get_mpfr_format_length, format_character, format_string): Ditto.
29266 (should_warn_p, maybe_warn): Same.
29268 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
29270 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
29273 * asan.c (get_translation_unit_decl): Remove function.
29274 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
29276 2017-01-30 Martin Liska <mliska@suse.cz>
29278 PR gcov-profile/79259
29279 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
29280 -fprofile-generate.
29282 2017-01-30 Martin Liska <mliska@suse.cz>
29285 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
29286 Initialize variables with NULL value.
29288 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
29291 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
29293 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
29295 2017-01-30 Richard Biener <rguenther@suse.de>
29297 PR tree-optimization/79276
29298 * tree-vrp.c (process_assert_insertions): Properly adjust common
29299 when removing a duplicate.
29301 * gcc.dg/torture/pr79276.c: New testcase.
29303 2017-01-30 Richard Biener <rguenther@suse.de>
29305 PR tree-optimization/79256
29306 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
29307 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
29309 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
29311 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
29314 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
29315 ("*r<noxa>sbg_<mode>_sll_bitmask")
29316 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
29317 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
29318 Use contiguous_bitmask_nowrap_operand.
29320 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29323 * config/rs6000/altivec.h (vec_xl): Revise #define.
29324 (vec_xst): Likewise.
29326 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
29328 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
29330 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
29332 PR rtl-optimization/79194
29333 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
29334 traps before call to bypass_conditional_jumps.
29336 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
29338 PR tree-optimization/71374
29339 * lra-constraints.c (check_conflict_input_operands): New.
29340 (match_reload): Use it.
29342 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
29345 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
29346 account to calculate conflict_set.
29348 2017-01-27 Bin Cheng <bin.cheng@arm.com>
29350 PR rtl-optimization/78559
29351 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
29352 other_insn in combine.
29354 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29356 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
29357 uint16_type_node for BT_UINT16.
29359 2017-01-27 David Malcolm <dmalcolm@redhat.com>
29361 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
29362 "RTL Tests" to menu.
29363 (GIMPLE Tests): New node.
29364 (RTL Tests): New node.
29366 2017-01-27 Richard Biener <rguenther@suse.de>
29368 PR tree-optimization/79245
29369 * tree-loop-distribution.c (distribute_loop): Apply cost
29370 modeling also to detected patterns.
29372 2017-01-27 Richard Biener <rguenther@suse.de>
29374 PR tree-optimization/71433
29375 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
29376 (compare_assert_loc): New function.
29377 (process_assert_insertions): Sort and optimize assert locations
29378 to remove duplicates and push down identical assertions on
29379 edges to their destination block.
29381 2017-01-27 Richard Biener <rguenther@suse.de>
29383 PR tree-optimization/79244
29384 * tree-vrp.c (remove_range_assertions): Forcefully propagate
29385 out SSA names even if abnormal.
29387 2017-01-27 Jakub Jelinek <jakub@redhat.com>
29389 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
29390 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
29391 instead of MPFR_RNDN.
29393 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
29396 * arm.c (arm_option_override): Don't call build_target_option_node
29397 until after doing all option overrides.
29398 (arm_valid_target_attribute_tree): Likewise.
29400 2017-01-27 Martin Liska <mliska@suse.cz>
29402 * doc/invoke.texi (-fprofile-arcs): Document profiling support
29403 for {cd}tors and C++ {cd}tors.
29405 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
29407 * config/s390/s390.md ("*setmem_long_and")
29408 ("*setmem_long_and_31z"): Use zero_extend instead of and.
29410 2017-01-26 Martin Sebor <msebor@redhat.com>
29412 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
29415 2017-01-26 Martin Sebor <msebor@redhat.com>
29417 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
29418 HAVE_DFmode before using XFmode or DFmode.
29419 (parse_directive): Avoid using the z length modifier to avoid
29420 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
29422 PR middle-end/78703
29423 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
29424 to accept adjustment as an array.
29425 (get_int_range): New function.
29426 (struct directive): Make width and prec arrays.
29427 (directive::set_width, directive::set_precision): Call get_int_range.
29428 (format_integer, format_floating): Handle width and precision ranges.
29429 (format_string, parse_directive): Same.
29431 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29434 * dwarf2out.c (generate_skeleton_bottom_up): For children with
29435 comdat_type_p set, just clone them, but keep the children in the
29439 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
29440 which have direct callers with -fvar-tracking-assignments enabled
29442 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
29443 inside of type units.
29445 2017-01-26 Martin Sebor <msebor@redhat.com>
29447 PR middle-end/78703
29448 * gimple-ssa-sprintf.c (struct result_range): Add likely and
29450 (struct format_result): Replace number_chars, number_chars_min,
29451 and number_chars_max with a single member of struct result_range.
29453 (format_result::operator+=): Adjust.
29454 (struct fmtresult): Remove bounded. Handle likely and unlikely
29456 (fmtresult::adjust_for_width_or_precision): New function.
29457 (fmtresult:type_max_digits): New function.
29458 (bytes_remaining): Handle likely and unlikely counters.
29459 (min_bytes_remaining): Remove.
29460 (format_percent): Simplify.
29461 (format_integer, format_floating): Set likely and unlikely counters.
29462 (get_string_length, format_character, format_string): Same.
29463 (format_plain, should_warn_p): New function.
29464 (maybe_warn): Call should_warn_p. Update diagnostic messages
29465 and handle those for all directives, including plain strings.
29466 (format_directive): Handle likely and unlikely counters.
29467 Remove unnecessary quoting from diagnostics. Add an informational
29469 (add_bytes): Remove.
29470 (pass_sprintf_length::compute_format_length): Simplify.
29471 (try_substitute_return_value): Handle likely and unlikely counters.
29473 2017-01-26 Carl Love <cel@us.ibm.com>
29475 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
29476 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
29478 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
29481 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
29482 endianess for subregs into account.
29483 * lra-constraints.c (lra_constraints): Do risky transformations
29484 always on the first iteration.
29485 * lra-lives.c (check_pseudos_live_through_calls): Add arg
29486 last_call_used_reg_set.
29487 (process_bb_lives): Define and use last_call_used_reg_set.
29488 * lra.c (lra): Always continue after lra_constraints on the first
29491 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
29493 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
29495 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
29497 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29499 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
29500 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
29501 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
29502 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
29503 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
29504 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
29505 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
29506 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
29507 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
29509 2017-01-26 Marek Polacek <polacek@redhat.com>
29512 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
29513 for the third operand.
29515 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29517 PR middle-end/79236
29518 * omp-low.c (struct omp_context): Add simt_stmt field.
29519 (scan_omp_for): Return omp_context *.
29520 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
29521 context to the _simt_ SIMD stmt.
29522 (lower_omp_for): For combined SIMD with sibling _simt_
29523 SIMD, make sure to use the same decls in _looptemp_
29524 clauses as in the sibling.
29526 2017-01-26 David Sherwood <david.sherwood@arm.com>
29528 PR middle-end/79212
29529 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
29532 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29535 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
29536 emit fld b; fld a; if possible.
29538 * brig-builtins.def: Update copyright years.
29539 * config/arm/arm_acle_builtins.def: Update copyright years.
29541 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
29544 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
29545 constraint instead of o for the stxsd instruction.
29547 2017-01-25 Carl Love <cel@us.ibm.com>
29549 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
29550 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
29552 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
29554 * doc/invoke.texi (C++ Dialect Options): Fix typo.
29556 2017-01-25 Richard Biener <rguenther@suse.de>
29558 PR tree-optimization/69264
29559 * target.def (vector_alignment_reachable): Improve documentation.
29560 * doc/tm.texi: Regenerate.
29561 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
29563 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
29564 earlier changes with respect to TYPE_USER_ALIGN.
29565 (vector_alignment_reachable_p): Likewise. Improve dumping.
29567 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29570 * config/arm/arm.md (xordi3): Force constant operand into a register
29573 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29575 * doc/invoke.texi (-fstore-merging): Correct default optimization
29576 levels at which it is enabled.
29577 (-O): Move -fstore-merging from list to...
29580 2017-01-25 Richard Biener <rguenther@suse.de>
29583 * omp-expand.c: Include debug.h.
29584 (expand_omp_taskreg): Make sure to generate early debug before
29585 outlining anything from a function.
29586 (expand_omp_target): Likewise.
29587 (grid_expand_target_grid_body): Likewise.
29589 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
29592 * asan.c (get_translation_unit_decl): New function.
29593 (asan_add_global): Extract modules file name from globals
29594 TRANSLATION_UNIT_DECL name.
29596 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
29599 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
29600 for long calls with APCS frame and VFP.
29602 2017-01-24 David Malcolm <dmalcolm@redhat.com>
29604 * cfg.c (original_copy_tables_initialized_p): New function.
29605 * cfg.h (original_copy_tables_initialized_p): New decl.
29606 * cfgrtl.c (relink_block_chain): Guard the call to
29607 free_original_copy_tables with a call to
29608 original_copy_tables_initialized_p.
29609 * cgraph.h (symtab_node::native_rtl_p): New decl.
29610 * cgraphunit.c (symtab_node::native_rtl_p): New function.
29611 (symtab_node::needed_p): Don't assert for early assembly output
29612 for __RTL functions.
29613 (cgraph_node::finalize_function): Set "force_output" for __RTL
29615 (cgraph_node::analyze): Bail out early for __RTL functions.
29616 (analyze_functions): Update assertion to support __RTL functions.
29617 (cgraph_node::expand): Bail out early for __RTL functions.
29618 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
29620 * function.h (struct function): Update comment for field
29622 * gimple-expr.c: Include "tree-pass.h".
29623 (gimple_has_body_p): Return false for __RTL functions.
29624 * Makefile.in (OBJS): Add run-rtl-passes.o.
29625 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
29627 (gcc::pass_manager::get_clean_slate): New accessor.
29628 * passes.c: Include "insn-addr.h".
29629 (should_skip_pass_p): Add logging. Update logic for running
29630 "expand" to be compatible with both __GIMPLE and __RTL. Guard
29631 property-provider override so it is only done for gimple passes.
29633 (skip_pass): New function.
29634 (execute_one_pass): Call skip_pass when skipping passes.
29635 * read-md.c (md_reader::read_char): Support filtering
29636 the input to a subset of line numbers.
29637 (md_reader::md_reader): Initialize fields
29638 m_first_line and m_last_line.
29639 (md_reader::read_file_fragment): New function.
29640 * read-md.h (md_reader::read_file_fragment): New decl.
29641 (md_reader::m_first_line): New field.
29642 (md_reader::m_last_line): New field.
29643 * read-rtl-function.c (function_reader::create_function): Only
29644 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
29645 curr_properties. Set DECL_INITIAL to a dummy block.
29646 (read_rtl_function_body_from_file_range): New function.
29647 * read-rtl-function.h (read_rtl_function_body_from_file_range):
29649 * run-rtl-passes.c: New file.
29650 * run-rtl-passes.h: New file.
29652 2017-01-24 Jeff Law <law@redhat.com>
29654 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
29657 2017-01-24 Bin Cheng <bin.cheng@arm.com>
29659 PR tree-optimization/79159
29660 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
29661 (record_nonwrapping_iv): Improve boundary using above function if no
29662 value range information.
29664 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
29665 Martin Jambor <mjambor@suse.cz>
29667 * brig-builtins.def: New file.
29668 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
29669 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
29670 (DEF_HSAIL_SAT_BUILTIN): Likewise.
29671 (DEF_HSAIL_INTR_BUILTIN): Likewise.
29672 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
29673 * builtin-types.def (BT_INT8): New.
29674 (BT_INT16): Likewise.
29675 (BT_UINT8): Likewise.
29676 (BT_UINT16): Likewise.
29677 (BT_FN_ULONG): Likewise.
29678 (BT_FN_UINT_INT): Likewise.
29679 (BT_FN_UINT_ULONG): Likewise.
29680 (BT_FN_UINT_LONG): Likewise.
29681 (BT_FN_UINT_PTR): Likewise.
29682 (BT_FN_ULONG_PTR): Likewise.
29683 (BT_FN_INT8_FLOAT): Likewise.
29684 (BT_FN_INT16_FLOAT): Likewise.
29685 (BT_FN_UINT32_FLOAT): Likewise.
29686 (BT_FN_UINT16_FLOAT): Likewise.
29687 (BT_FN_UINT8_FLOAT): Likewise.
29688 (BT_FN_UINT64_FLOAT): Likewise.
29689 (BT_FN_UINT16_UINT32): Likewise.
29690 (BT_FN_UINT32_UINT16): Likewise.
29691 (BT_FN_UINT16_UINT16_UINT16): Likewise.
29692 (BT_FN_INT_PTR_INT): Likewise.
29693 (BT_FN_UINT_PTR_UINT): Likewise.
29694 (BT_FN_LONG_PTR_LONG): Likewise.
29695 (BT_FN_ULONG_PTR_ULONG): Likewise.
29696 (BT_FN_VOID_UINT64_UINT64): Likewise.
29697 (BT_FN_UINT8_UINT8_UINT8): Likewise.
29698 (BT_FN_INT8_INT8_INT8): Likewise.
29699 (BT_FN_INT16_INT16_INT16): Likewise.
29700 (BT_FN_INT_INT_INT): Likewise.
29701 (BT_FN_UINT_FLOAT_UINT): Likewise.
29702 (BT_FN_FLOAT_UINT_UINT): Likewise.
29703 (BT_FN_ULONG_UINT_UINT): Likewise.
29704 (BT_FN_ULONG_UINT_PTR): Likewise.
29705 (BT_FN_ULONG_ULONG_ULONG): Likewise.
29706 (BT_FN_UINT_UINT_UINT): Likewise.
29707 (BT_FN_VOID_UINT_PTR): Likewise.
29708 (BT_FN_UINT_UINT_PTR: Likewise.
29709 (BT_FN_UINT32_UINT64_PTR): Likewise.
29710 (BT_FN_INT_INT_UINT_UINT): Likewise.
29711 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
29712 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
29713 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
29714 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
29715 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
29716 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
29717 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
29718 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
29719 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
29720 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
29721 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
29722 * doc/frontends.texi: List BRIG FE.
29723 * doc/install.texi (Testing): Add BRIG tesring requirements.
29724 * doc/invoke.texi (Overall Options): Mention BRIG.
29725 * doc/standards.texi (Standards): Doucment BRIG HSA version.
29727 2017-01-24 Richard Biener <rguenther@suse.de>
29729 PR translation/79208
29730 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
29732 2017-01-24 Martin Jambor <mjambor@suse.cz>
29735 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
29736 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
29737 and known_contexts.
29739 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
29741 PR middle-end/79123
29742 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
29743 casts from signed to unsigned really don't have a range.
29745 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
29747 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
29748 GMP_RNDx for compatiblity.
29750 2017-01-24 Martin Liska <mliska@suse.cz>
29753 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
29754 that would prevent us to call alloca with -1 as argument.
29756 2017-01-24 Jakub Jelinek <jakub@redhat.com>
29758 * dwarf2out.c (output_compilation_unit_header, output_file_names):
29759 Avoid -Wformat-security warning.
29761 2017-01-23 Andrew Pinski <apinski@cavium.com>
29763 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
29766 2017-01-23 Martin Sebor <msebor@redhat.com>
29768 PR middle-end/78703
29769 * gimple-ssa-sprintf.c (warn_level): New global.
29770 (format_integer): Use it here and throughout the rest of the file.
29771 Use the same switch to compute sign as base.
29772 (maybe_warn): New function.
29773 (format_directive): Factor out warnings into maybe_warn.
29774 Add debugging output. Use warn_level.
29775 (add_bytes): Use warn_level.
29776 (pass_sprintf_length::compute_format_length): Add debugging output.
29777 (try_substitute_return_value): Same.
29778 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
29780 PR middle-end/78703
29781 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
29782 (struct fmtresult, format_integer, format_floating): Adjust.
29783 (fmtresult::fmtresult): Set max correctly in two argument ctor.
29784 (get_string_length, format_string,format_directive): Same.
29785 (pass_sprintf_length::compute_format_length): Same.
29786 (try_substitute_return_value): Simplify slightly.
29788 PR middle-end/78703
29789 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
29790 (fmtresult::operator+=): Outlined.
29791 (struct fmtresult): Add ctors.
29792 (struct conversion_spec): Rename...
29793 (struct directive): ...to this. Add and remove data members.
29794 (directive::set_width, directive::set_precision): New functions.
29795 (format_percent): Use fmtresult ctor.
29796 (get_width_and_precision): Remove.
29797 (format_integer): Make naming changes. Avoid computing width and
29799 (format_floating): Same. Adjust indentation.
29800 (format_character, format_none): New functions.
29801 (format_string): Moved character handling to format_character.
29802 (format_directive): Remove arguments, change return type.
29803 (parse_directive): New function.
29804 (pass_sprintf_length::compute_format_length): Move directive
29805 parsing to parse_directive.
29807 2017-01-23 Jakub Jelinek <jakub@redhat.com>
29809 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
29810 (assign_assembler_name_if_needed): ... this.
29811 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
29812 (assign_assembler_name_if_needed): ... this.
29813 (free_lang_data_in_cgraph): Adjust callers.
29814 * cgraphunit.c (cgraph_node::analyze): Likewise.
29815 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
29818 2017-01-23 Richard Biener <rguenther@suse.de>
29820 PR tree-optimization/79088
29821 PR tree-optimization/79188
29822 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
29823 resetting loop bounds after last path deletion. Reset loop
29824 bounds of the target loop, make code match the comments.
29825 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
29826 Make sure loops need no fixups.
29828 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29830 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
29831 exponent support with double type for first argument.
29832 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
29833 type returned by __builtin_vec_extract_sig,
29834 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
29835 functions from "vector int" to "vector unsigned int" or from
29836 "vector long long int" to "vector unsigned long long int".
29837 Changed type returned by __builtin_vec_extract_exp,
29838 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
29839 functions from "vector int" to "vector unsigned int" or from
29840 "vector long long int" to "vector unsigned long long int".
29841 Changed return type of __builtin_vec_test_data_class,
29842 __builtin_vec_test_data_class_sp, and
29843 __builtin_vec_test_data_class_dp from "vector int" to
29844 "vector bool int" or from "vector long long int" to "vector bool
29845 long long int" and changed second argument type from "unsigned
29846 int" to "int". Added new overloaded function forms "vector float
29847 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
29848 "vector float __builtin_vec_insert_exp_sp (vector float, vector
29849 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
29850 double, vector unsigned long long int)" and "vector double
29851 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
29852 long int)". Changed return type of
29853 __builtin_scalar_test_data_class and
29854 __builtin_scalar_test_data_class_sp and
29855 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
29856 int" and changed second argument from "unsigned int" to "int".
29857 Changed type returned by __builtin_scalar_test_neg,
29858 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
29859 from "int" to "bool int". Added new overloaded function form
29860 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
29861 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
29862 exponent double-precision with floating point first argument.
29863 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
29864 documentation of scalar_test_data_class, scalar_test_neg,
29865 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
29866 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
29867 vec_test_data_class built-in functions to reflect refinements in
29868 their type signatures.
29870 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
29872 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
29874 (aarch64_elf_asm_destructor): Likewise.
29876 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
29878 PR rtl-optimization/78634
29879 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
29880 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
29881 * ifcvt.c (noce_try_cmove): Add missing cost check.
29883 PR rtl-optimization/71724
29884 * combine.c (if_then_else_cond): Look for situations where it is
29885 beneficial to undo the work of one of the recursive calls.
29887 2017-01-23 Bin Cheng <bin.cheng@arm.com>
29889 PR tree-optimization/70754
29890 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
29891 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
29892 combined stmt before it if not NULL.
29893 (combine_chains): Process refs reversely and compute dominance point
29896 2017-01-23 Martin Liska <mliska@suse.cz>
29898 PR tree-optimization/79196
29899 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
29900 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
29902 (strlen_optimize_stmt): Call the renamed function.
29904 2017-01-23 Michael Matz <matz@suse.de>
29906 PR tree-optimization/78384
29907 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
29909 2017-01-23 Richard Biener <rguenther@suse.de>
29911 PR tree-optimization/79186
29912 * tree-vrp.c (register_new_assert_for): Make sure we've seen
29913 both incoming edges before moving an assert.
29915 2017-01-23 Martin Jambor <mjambor@suse.cz>
29917 * ipa-prop.c (load_from_param_1): Removed.
29918 (load_from_unmodified_param): Bits from load_from_param_1 put back
29920 (load_from_param): Removed.
29921 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
29922 with stmt. Reverted back to use of load_from_unmodified_param.
29924 2017-01-23 Martin Jambor <mjambor@suse.cz>
29927 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
29928 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
29929 field a pointer to garbage collected vector, mark lattices and
29930 ipcp_orig_node with GTY((skip)).
29931 (ipa_get_param_count): Adjust to descriptors being a pointer.
29932 (ipa_get_param): Likewise.
29933 (ipa_get_type): Likewise.
29934 (ipa_get_param_move_cost): Likewise.
29935 (ipa_set_param_used): Likewise.
29936 (ipa_get_controlled_uses): Likewise.
29937 (ipa_set_controlled_uses): Likewise.
29938 (ipa_is_param_used): Likewise.
29939 (ipa_node_params_t): Move into garbage collector. New methods insert
29941 (ipa_node_params_sum): Annotate wth GTY(()).
29942 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
29944 (ipa_load_from_parm_agg): Adjust declaration.
29945 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
29946 * ipa-profile.c (ipa_profile): Likewise.
29947 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
29948 (ipa_populate_param_decls): Make descriptors parameter garbage
29950 (ipa_dump_param): Adjust to descriptors being a pointer.
29951 (ipa_alloc_node_params): Likewise.
29952 (ipa_initialize_node_params): Likewise.
29953 (load_from_param_1): Make descriptors parameter garbage collected.
29954 (load_from_unmodified_param): Likewise.
29955 (load_from_param): Likewise.
29956 (ipa_load_from_parm_agg): Likewise.
29957 (ipa_node_params::~ipa_node_params): Removed.
29958 (ipa_free_all_node_params): Remove call to delete operator.
29959 (ipa_node_params_t::insert): New.
29960 (ipa_node_params_t::remove): Likewise.
29961 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
29962 copy known_csts and known_contexts vectors.
29963 (ipa_read_node_info): Adjust to descriptors being a pointer.
29964 (ipcp_modif_dom_walker): Make m_descriptors field garbage
29966 (ipcp_transform_function): Make descriptors variable garbage
29969 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
29971 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
29972 * config/i386/avx512dqintrin.h: Ditto.
29973 * config/i386/avx512fintrin.h: Ditto.
29974 * config/i386/i386.c: Handle new builtins.
29975 * config/i386/i386-builtin.def: Add new builtins.
29976 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
29977 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
29979 2017-01-23 Jakub Jelinek <jakub@redhat.com>
29980 Martin Liska <mliska@suse.cz>
29982 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
29983 * asan.c (asan_expand_poison_ifn): Support stores and use
29984 appropriate ASAN report function.
29985 * internal-fn.c (expand_ASAN_POISON_USE): New function.
29986 * internal-fn.def (ASAN_POISON_USE): Declare.
29987 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
29988 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
29989 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
29990 ASAN_POISON calls w/o LHS.
29991 * tree-ssa.c (execute_update_addresses_taken): Create clobber
29992 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
29993 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
29994 * gimplify.c (asan_poison_variables): Add attribute
29995 use_after_scope_memory to variables that really needs to live
29997 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
29998 having the attribute.
30000 2017-01-23 Martin Liska <mliska@suse.cz>
30002 * asan.c (create_asan_shadow_var): New function.
30003 (asan_expand_poison_ifn): Likewise.
30004 * asan.h (asan_expand_poison_ifn): New declaration.
30005 * internal-fn.c (expand_ASAN_POISON): Likewise.
30006 * internal-fn.def (ASAN_POISON): New builtin.
30007 * sanopt.c (pass_sanopt::execute): Expand
30008 asan_expand_poison_ifn.
30009 * tree-inline.c (copy_decl_for_dup_finish): Make function
30011 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
30012 * tree-ssa.c (is_asan_mark_p): New function.
30013 (execute_update_addresses_taken): Rewrite local variables
30014 (identified just by use-after-scope as addressable) into SSA.
30016 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
30018 * doc/install.texi (Specific): opensource.apple.com uses https
30019 now. Remove trailing slash.
30021 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
30023 * README.Portability: Remove note on an Irix compatibility issue.
30025 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
30027 * gcov.c (INCLUDE_ALGORITHM): Define.
30028 (INCLUDE_VECTOR): Define.
30029 No longer include <vector> and <algorithm> directly.
30031 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
30033 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
30035 * doc/invoke.texi (Code Gen Options): Ditto.
30037 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
30040 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
30042 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
30044 rtl-optimization/79125
30045 * cprop.c (local_cprop_pass): Handle cases where we make an
30046 unconditional trap.
30048 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
30052 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
30053 read from, for big endian.
30055 2017-01-20 Jiong Wang <jiong.wang@arm.com>
30057 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
30058 register pauth builtins for LP64 only.
30060 2017-01-20 Marek Polacek <polacek@redhat.com>
30063 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
30066 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
30068 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
30070 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
30071 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
30072 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
30074 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30077 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
30078 in big-endian mode when they are not a single duplicated value.
30080 2017-01-20 Richard Biener <rguenther@suse.de>
30082 * BASE-VER: Bump to 7.0.1.
30084 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
30086 * omp-low.c (omplow_simd_context): New struct. Use it...
30087 (lower_rec_simd_input_clauses): ...here and...
30088 (lower_rec_input_clauses): ...here to hold common data. Adjust all
30089 references to idx, lane, max_vf, is_simt.
30091 2017-01-20 Graham Markall <graham.markall@embecosm.com>
30093 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
30096 2017-01-20 Martin Jambor <mjambor@suse.cz>
30098 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
30099 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
30101 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
30102 (GTFILES): Rename hsa.c to hsa-common.c.
30103 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
30104 * hsa-dump.c: Likewise.
30105 * hsa-gen.c: Likewise.
30106 * hsa-regalloc.c: Likewise.
30107 * ipa-hsa.c: Likewise.
30108 * omp-expand.c: Likewise.
30109 * omp-low.c: Likewise.
30110 * toplev.c: Likewise.
30112 2017-01-20 Marek Polacek <polacek@redhat.com>
30115 * doc/invoke.texi: Document -Wduplicated-branches.
30116 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
30117 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
30118 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
30119 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
30120 return 0 only when not OEP_LEXICOGRAPHIC.
30121 (fold_build_cleanup_point_expr): Use the expression
30122 location when building CLEANUP_POINT_EXPR.
30123 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
30124 * tree.c (add_expr): Handle error_mark_node.
30126 2017-01-20 Martin Liska <mliska@suse.cz>
30129 * tree-profile.c (init_ic_make_global_vars): Do not call
30131 (gimple_init_gcov_profiler): Likewise.
30133 2017-01-20 Martin Liska <mliska@suse.cz>
30136 * cgraph.h (maybe_create_reference): Remove argument and
30138 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
30140 * ipa-cp.c (create_specialized_node): Likewise.
30141 * symtab.c (symtab_node::maybe_create_reference): Handle
30142 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
30144 2017-01-20 Martin Liska <mliska@suse.cz>
30146 * read-rtl-function.c (function_reader::create_function): Use
30147 build_decl instread of build_decl_stat.
30149 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
30151 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
30152 * config/i386/avx512dqintrin.h: Ditto.
30153 * config/i386/avx512fintrin.h: Ditto.
30154 * config/i386/i386-builtin-types.def: Add new types.
30155 * config/i386/i386.c: Handle new types.
30156 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
30157 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
30158 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
30159 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
30160 (__builtin_ia32_kshiftridi): New.
30161 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
30163 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
30167 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
30168 define to rs6000_init_stack_protect_guard.
30169 (rs6000_init_stack_protect_guard): New function.
30171 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
30172 Yunqiang Su <yunqiang.su@imgtec.com>
30174 * config.gcc (supported_defaults): Add madd4.
30175 (with_madd4): Add validation.
30176 (all_defaults): Add madd4.
30177 * config/mips/mips.opt (mmadd4): New option.
30178 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
30180 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
30182 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
30183 (ISA_HAS_FUSED_MADD4): Likewise.
30184 * doc/invoke.texi (-mmadd4): Document the new option.
30185 * doc/install.texi (--with-madd4): Document the new option.
30187 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30189 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
30190 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
30191 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
30192 (aarch64_init_pauth_hint_builtins): New.
30193 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
30194 (aarch64_expand_builtin): Expand new builtins.
30196 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30198 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
30199 * combine-stack-adj.c (no_unhandled_cfa): Handle
30200 REG_CFA_TOGGLE_RA_MANGLE.
30201 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
30202 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
30203 info for return address signing.
30204 (aarch64_expand_epilogue): Likewise.
30206 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30208 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
30209 * config/aarch64/aarch64-protos.h
30210 (aarch64_return_address_signing_enabled): New declaration.
30211 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
30213 (aarch64_expand_prologue): Sign return address before it's pushed onto
30215 (aarch64_expand_epilogue): Authenticate return address fetched from
30217 (aarch64_override_options): Sanity check for ILP32 and ISA level.
30218 (aarch64_attributes): New function attributes for "sign-return-address".
30219 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
30220 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
30221 ("*do_return"): Generate combined instructions according to key index.
30222 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
30223 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
30225 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
30226 * config/aarch64/aarch64.opt (msign-return-address=): New.
30227 * doc/extend.texi (AArch64 Function Attributes): Documents
30228 "sign-return-address=".
30229 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
30231 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
30233 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
30234 overall option summary.
30236 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30238 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
30239 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
30240 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
30241 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
30243 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
30245 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
30246 -mpower9-minmax by default for -mcpu=power9.
30247 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
30248 128-bit floating point.
30250 2017-01-20 Alan Modra <amodra@gmail.com>
30252 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
30253 optimizing for size.
30255 2017-01-20 Alan Modra <amodra@gmail.com>
30258 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
30259 for strcmp and strncmp from corresponding builtin decl.
30261 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
30263 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
30264 instead of i386/rtems-64.h.
30265 * config/i386/rtems-64.h: Remove.
30267 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
30271 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
30273 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
30275 2017-01-19 Tamar Christina <tamar.christina@arm.com>
30277 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
30278 Change int to HOST_WIDE_INT.
30279 * config/aarch64/aarch64-protos.h
30280 (aarch64_simd_gen_const_vector_dup): Likewise.
30281 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
30283 2017-01-19 David Malcolm <dmalcolm@redhat.com>
30285 * langhooks-def.h (lhd_type_for_size): New decl.
30286 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
30287 * langhooks.c (lhd_type_for_size): New function, taken from
30290 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
30292 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
30293 define_bypass for CR latency.
30294 (power9-cracked-alu): Update bypass latency and remove power9-branch.
30295 (power9-alu2): Add define_bypass for CR latency.
30297 (power9-mul): Update insn latency.
30298 (power9-mul-compare): Update insn latency, bypass latency and remove
30301 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30303 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
30305 * config/aarch64/aarch64.md
30306 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
30307 aarch64_nopcrelative_literal_loads.
30308 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
30310 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
30312 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
30313 TARGET_LOONGSON_3A.
30314 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
30316 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
30319 * config.gcc (supported_defaults): Add lxc1-sxc1.
30320 (with_lxc1_sxc1): Add validation.
30321 (all_defaults): Add lxc1-sxc1.
30322 * config/mips/mips.opt (mlxc1-sxc1): New option.
30323 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
30325 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
30326 __mips_no_lxc1_sxc1.
30327 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
30328 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
30329 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
30331 2017-01-19 Richard Biener <rguenther@suse.de>
30333 PR tree-optimization/72488
30334 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
30335 sure to restore SSA info.
30336 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
30338 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
30340 PR rtl-optimization/79121
30341 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
30342 of the inner type when shifting an extended value.
30344 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
30347 * symtab.c (symtab_node::equal_address_to): Fix comparing of
30348 interposable aliases.
30350 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
30353 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
30354 Use the evmergelohi instruction.
30355 (mov_si<mode>_e500_subreg4_2_le): Likewise.
30356 (mov_sitf_e500_subreg8_2_be): Likewise.
30357 (mov_sitf_e500_subreg12_2_le): Likewise.
30358 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
30359 (mov_si<mode>_e500_subreg4_2_be): Likewise.
30360 (mov_sitf_e500_subreg8_2_le): Likewise.
30361 (mov_sitf_e500_subreg12_2_be): Likewise.
30363 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30365 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
30366 attribute from vecsimple to vecperm.
30367 (altivec_vbpermq2): Likewise.
30369 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30372 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
30374 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30375 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
30376 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
30377 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
30378 case where N arg is SIZE_MAX.
30379 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
30380 (cmpstrsi): Add pattern.
30382 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30384 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30385 __builtin_vec_revb builtins.
30386 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
30387 built-in functions to support generation of the ISA 3.0 XXBR<x>
30388 vector byte reverse instructions.
30389 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
30390 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
30391 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
30392 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
30393 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
30394 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
30395 (P9V_BUILTIN_VEC_REVB): Likewise.
30396 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
30397 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
30398 (p9_xxbrq_v16qi): Likewise.
30399 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
30400 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
30401 (p9_xxbrh_v8hi): Likewise.
30402 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
30403 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
30404 vec_revb built-in functions.
30406 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
30408 PR rtl-optimization/78952
30409 * config/i386/i386.md (any_extract): New code iterator.
30410 (*insvqi_2): Use any_extract for source operand.
30411 (*insvqi_3): Use any_shiftrt for source operand.
30413 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
30415 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
30417 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
30419 2017-01-18 Matthias Klose <doko@ubuntu.com>
30421 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
30423 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30425 * config/rs6000/altivec.h (vec_bperm): Change #define.
30426 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
30427 (altivec_vbpermq2): New define_insn.
30428 (altivec_vbpermd): Likewise.
30429 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
30430 function interface.
30431 (VBPERMD): Likewise.
30432 (VBPERM): New polymorphic function interface.
30433 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
30434 Add entries for P9V_BUILTIN_VEC_VBPERM.
30435 * doc/extend.texi: Add interfaces for vec_bperm.
30437 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30439 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
30440 first letter of error messages.
30441 (s390_resolve_overloaded_builtin): Likewise.
30442 * config/s390/s390.c (s390_expand_builtin): Likewise.
30443 (s390_invalid_arg_for_unprototyped_fn): Likewise.
30444 (s390_valid_target_attribute_inner_p): Likewise.
30445 * config/s390/s390.md ("tabort"): Likewise.
30447 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
30449 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
30450 (ISA_AVOID_DIV_HILO): New macro.
30451 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
30452 (ISA_HAS_DDIV): Likewise.
30454 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30456 * doc/invoke.texi (fabi-version): Correct number of occurrences.
30458 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30460 * doc/invoke.texi (fabi-version): Spelling fix.
30462 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30465 * doc/invoke.texi (fabi-version): Mention mangling fix for
30468 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30471 * doc/invoke.texi (fabi-version): Document discriminator mangling.
30473 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
30476 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
30477 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
30479 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
30481 (stack_protect_test): Ditto.
30482 * config/rs6000/rs6000.opt (mstack-protector-guard=,
30483 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
30485 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
30486 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
30487 -mstack-protector-guard-offset=.
30488 (RS/6000 and PowerPC Options): Ditto.
30490 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
30492 * config/i386/i386.h (MASK_CLASS_P): New define.
30493 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
30494 there are no registers from different register sets also when
30495 mask registers are used. Update function comment.
30496 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
30497 to (*k/*r) and (*k/*km) alternatives.
30499 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
30501 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
30502 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
30503 (EH_RETURN_HANDLER_RTX): New define.
30504 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
30505 Force frame pointer in EH return functions.
30506 (aarch64_expand_epilogue): Add barrier for eh_return.
30507 (aarch64_final_eh_return_addr): Remove.
30508 (aarch64_eh_return_handler_rtx): New function.
30509 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
30511 (aarch64_eh_return_handler_rtx): New prototype.
30513 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30515 * config/rs6000/altivec.h (vec_rlmi): New #define.
30516 (vec_vrlnm): Likewise.
30517 (vec_rlnm): Likewise.
30518 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
30519 (UNSPEC_VRLNM): Likewise.
30520 (VIlong): New mode iterator.
30521 (altivec_vrl<VI_char>mi): New define_insn.
30522 (altivec_vrl<VI_char>nm): Likewise.
30523 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
30525 (VRLDNM): Likewise.
30526 (RLNM): New polymorphic function entry.
30527 (VRLWMI): New monomorphic function entry.
30528 (VRLDMI): Likewise.
30529 (RLMI): New polymorphic function entry.
30530 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
30531 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
30532 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
30535 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30538 * dwarf2out.c (field_byte_offset): Restore the
30539 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
30540 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
30541 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
30544 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
30547 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
30549 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30552 * dwarf2out.c (add_data_member_location_attribute): For constant
30553 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
30554 instead of DW_AT_data_member_location, DW_AT_bit_offset and
30555 DW_AT_byte_size attributes.
30557 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
30559 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
30560 after forcing to constant memory when the code model is medium.
30562 2017-01-17 Julia Koval <julia.koval@intel.com>
30565 * config/i386/avx512fintrin.h
30566 (_mm512_i32gather_ps): Change __addr type to void const*.
30567 (_mm512_mask_i32gather_ps): Ditto.
30568 (_mm512_i32gather_pd): Ditto.
30569 (_mm512_mask_i32gather_pd): Ditto.
30570 (_mm512_i64gather_ps): Ditto.
30571 (_mm512_mask_i64gather_ps): Ditto.
30572 (_mm512_i64gather_pd): Ditto.
30573 (_mm512_mask_i64gather_pd): Ditto.
30574 (_mm512_i32gather_epi32): Ditto.
30575 (_mm512_mask_i32gather_epi32): Ditto.
30576 (_mm512_i32gather_epi64): Ditto.
30577 (_mm512_mask_i32gather_epi64): Ditto.
30578 (_mm512_i64gather_epi32): Ditto.
30579 (_mm512_mask_i64gather_epi32): Ditto.
30580 (_mm512_i64gather_epi64): Ditto.
30581 (_mm512_mask_i64gather_epi64): Ditto.
30582 (_mm512_i32scatter_ps): Change __addr type to void*.
30583 (_mm512_mask_i32scatter_ps): Ditto.
30584 (_mm512_i32scatter_pd): Ditto.
30585 (_mm512_mask_i32scatter_pd): Ditto.
30586 (_mm512_i64scatter_ps): Ditto.
30587 (_mm512_mask_i64scatter_ps): Ditto.
30588 (_mm512_i64scatter_pd): Ditto.
30589 (_mm512_mask_i64scatter_pd): Ditto.
30590 (_mm512_i32scatter_epi32): Ditto.
30591 (_mm512_mask_i32scatter_epi32): Ditto.
30592 (_mm512_i32scatter_epi64): Ditto.
30593 (_mm512_mask_i32scatter_epi64): Ditto.
30594 (_mm512_i64scatter_epi32): Ditto.
30595 (_mm512_mask_i64scatter_epi32): Ditto.
30596 (_mm512_i64scatter_epi64): Ditto.
30597 (_mm512_mask_i64scatter_epi64): Ditto.
30598 * config/i386/avx512pfintrin.h
30599 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
30600 (_mm512_mask_prefetch_i32gather_ps): Ditto.
30601 (_mm512_mask_prefetch_i64gather_pd): Ditto.
30602 (_mm512_mask_prefetch_i64gather_ps): Ditto.
30603 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
30604 (_mm512_prefetch_i32scatter_ps): Ditto.
30605 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
30606 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
30607 (_mm512_prefetch_i64scatter_pd): Ditto.
30608 (_mm512_prefetch_i64scatter_ps): Ditto.
30609 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
30610 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
30611 * config/i386/avx512vlintrin.h
30612 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
30613 (_mm_mmask_i32gather_ps): Ditto.
30614 (_mm256_mmask_i32gather_pd): Ditto.
30615 (_mm_mmask_i32gather_pd): Ditto.
30616 (_mm256_mmask_i64gather_ps): Ditto.
30617 (_mm_mmask_i64gather_ps): Ditto.
30618 (_mm256_mmask_i64gather_pd): Ditto.
30619 (_mm_mmask_i64gather_pd): Ditto.
30620 (_mm256_mmask_i32gather_epi32): Ditto.
30621 (_mm_mmask_i32gather_epi32): Ditto.
30622 (_mm256_mmask_i32gather_epi64): Ditto.
30623 (_mm_mmask_i32gather_epi64): Ditto.
30624 (_mm256_mmask_i64gather_epi32): Ditto.
30625 (_mm_mmask_i64gather_epi32): Ditto.
30626 (_mm256_mmask_i64gather_epi64): Ditto.
30627 (_mm_mmask_i64gather_epi64): Ditto.
30628 (_mm256_i32scatter_ps): Change __addr type to void*.
30629 (_mm256_mask_i32scatter_ps): Ditto.
30630 (_mm_i32scatter_ps): Ditto.
30631 (_mm_mask_i32scatter_ps): Ditto.
30632 (_mm256_i32scatter_pd): Ditto.
30633 (_mm256_mask_i32scatter_pd): Ditto.
30634 (_mm_i32scatter_pd): Ditto.
30635 (_mm_mask_i32scatter_pd): Ditto.
30636 (_mm256_i64scatter_ps): Ditto.
30637 (_mm256_mask_i64scatter_ps): Ditto.
30638 (_mm_i64scatter_ps): Ditto.
30639 (_mm_mask_i64scatter_ps): Ditto.
30640 (_mm256_i64scatter_pd): Ditto.
30641 (_mm256_mask_i64scatter_pd): Ditto.
30642 (_mm_i64scatter_pd): Ditto.
30643 (_mm_mask_i64scatter_pd): Ditto.
30644 (_mm256_i32scatter_epi32): Ditto.
30645 (_mm256_mask_i32scatter_epi32): Ditto.
30646 (_mm_i32scatter_epi32): Ditto.
30647 (_mm_mask_i32scatter_epi32): Ditto.
30648 (_mm256_i32scatter_epi64): Ditto.
30649 (_mm256_mask_i32scatter_epi64): Ditto.
30650 (_mm_i32scatter_epi64): Ditto.
30651 (_mm_mask_i32scatter_epi64): Ditto.
30652 (_mm256_i64scatter_epi32): Ditto.
30653 (_mm256_mask_i64scatter_epi32): Ditto.
30654 (_mm_i64scatter_epi32): Ditto.
30655 (_mm_mask_i64scatter_epi32): Ditto.
30656 (_mm256_i64scatter_epi64): Ditto.
30657 (_mm256_mask_i64scatter_epi64): Ditto.
30658 (_mm_i64scatter_epi64): Ditto.
30659 (_mm_mask_i64scatter_epi64): Ditto.
30660 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
30661 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
30662 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
30663 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
30664 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
30665 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
30666 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
30667 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
30668 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
30669 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
30670 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
30671 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
30672 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
30673 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
30674 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
30675 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
30676 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
30677 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
30678 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
30679 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
30680 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
30681 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
30682 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
30683 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
30684 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
30685 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
30686 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
30687 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
30688 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
30689 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
30690 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
30691 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
30692 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
30693 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
30694 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
30695 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
30696 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
30697 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
30698 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
30699 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
30700 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
30701 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
30702 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
30703 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
30704 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
30705 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
30706 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
30707 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
30708 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
30709 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
30710 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
30711 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
30712 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
30713 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
30714 definitions accordingly.
30716 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
30717 Kuan-Lin Chen <kuanlinchentw@gmail.com>
30720 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
30723 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
30726 * ira-conflicts.c (ira_build_conflicts): Update total conflict
30727 hard regs for inner regno.
30729 2017-01-17 Martin Liska <mliska@suse.cz>
30732 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
30733 assumption and add comment.
30735 2017-01-17 Nathan Sidwell <nathan@acm.org>
30737 * ipa-visibility.c (localize_node): New function, broken out of ...
30738 (function_and_variable_visibility): ... here. Call it.
30740 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
30742 PR middle-end/77445
30743 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
30744 correctly set frequency of oudgoing edge.
30745 (duplicate_thread_path): Fix profile updating.
30747 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30750 * configure.ac: Add GCC_BASE_VER.
30751 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
30752 version from BASE-VER file.
30753 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
30754 (gcc.o): Depend on $(BASEVER).
30755 * common.opt (dumpfullversion): New option.
30756 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
30757 * doc/invoke.texi: Document -dumpfullversion.
30758 * doc/install.texi: Document --with-gcc-major-version-only.
30759 * configure: Regenerated.
30761 2017-01-17 Richard Biener <rguenther@suse.de>
30763 PR tree-optimization/71433
30764 * tree-vrp.c (register_new_assert_for): Merge same asserts
30765 on all incoming edges.
30766 (process_assert_insertions_for): Handle insertions at the
30769 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
30771 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
30772 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
30774 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
30777 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
30780 2017-01-17 Alan Modra <amodra@gmail.com>
30783 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
30784 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
30785 symbolic stack limit when pic.
30787 2017-01-16 Martin Sebor <msebor@redhat.com>
30789 PR tree-optimization/78608
30790 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
30792 2017-01-16 Jeff Law <law@redhat.com>
30795 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
30796 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
30797 for several include directories that may be relative to sysroot.
30798 * config/i386/x-mingw32 (gplus_includedir): Define.
30799 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
30800 (native_system_includedir): Likewise.
30801 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
30802 override if TARGET_SYSTEM_ROOT is defined.
30803 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
30805 PR tree-optimization/79090
30806 PR tree-optimization/33562
30807 PR tree-optimization/61912
30808 PR tree-optimization/77485
30809 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
30810 and computed trims into the dump file.
30812 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
30814 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
30816 2017-01-16 Jakub Jelinek <jakub@redhat.com>
30819 * gimplify.c (gimplify_init_constructor): If want_value and
30820 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
30824 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
30825 sequence. Formatting fixes.
30826 (doloop_optimize): Formatting fixes.
30829 * gcc.c (debug_level_greater_than_spec_func): New function.
30830 (static_spec_functions): Add debug-level-gt spec function.
30831 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
30833 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
30834 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
30835 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
30836 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
30837 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
30838 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
30840 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
30842 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
30843 QImode fixups to general and mask registers only.
30845 2017-01-16 Carl Love <cel@us.ibm.com>
30847 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
30848 for built-in functions
30849 vector signed char vec_nabs (vector signed char)
30850 vector signed short vec_nabs (vector signed short)
30851 vector signed int vec_nabs (vector signed int)
30852 vector signed long long vec_nabs (vector signed long long)
30853 vector float vec_nabs (vector float)
30854 vector double vec_nabs (vector double)
30855 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
30857 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
30858 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
30859 * doc/extend.texi: Update the documentation file for the new built-in
30862 2017-01-16 Martin Sebor <msebor@redhat.com>
30864 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
30867 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30869 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
30870 UNSPEC_VSX__XXSPLTD to require special splat handling.
30872 2017-01-16 David Malcolm <dmalcolm@redhat.com>
30875 * system.h: Poison strndup.
30877 2017-01-16 Alan Modra <amodra@gmail.com>
30880 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
30883 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
30885 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
30887 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
30889 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
30890 call recog here. Assert that INSN_CODE (insn) is non-negative.
30892 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30895 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
30897 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
30898 in the currently scheduled RTL fragment.
30900 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30902 PR rtl-optimization/78751
30903 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
30906 2017-01-14 Jeff Law <law@redhat.com>
30908 PR tree-optimization/79090
30909 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
30910 variable length stores.
30911 (compute_trims): Delete dead assignment to *trim_tail.
30912 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
30915 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
30917 PR rtl-optimization/78626
30918 PR rtl-optimization/78727
30919 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
30920 of a block, and split such blocks after everything else is finished.
30922 2017-01-14 Alan Modra <amodra@gmail.com>
30925 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
30926 target legitimate_combined_insn.
30927 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
30928 (rs6000_legitimate_combined_insn): New function.
30929 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
30931 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
30932 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
30933 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
30935 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
30937 * doc/frontends.texi (G++ and GCC): Remove references to Java.
30939 2017-01-13 Jeff Law <law@redhat.com>
30941 PR tree-optimization/33562
30942 PR tree-optimization/61912
30943 PR tree-optimization/77485
30944 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
30946 (delete_dead_assignment): Likewise.
30947 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
30948 statement to delete_dead_call and delete_dead_assignment.
30950 2017-01-13 David Malcolm <dmalcolm@redhat.com>
30953 * substring-locations.c (format_warning_va): Strengthen case 1 so
30954 that both endpoints of the substring must be within the format
30955 range for just the substring to be printed.
30957 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
30959 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
30960 * config/i386/i386.c (ix86_target_string): Add missing options
30961 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
30962 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
30963 flags_other and ix86_target_other to flags2_other. Display unknown
30965 (ix86_valid_target_attribute_inner_p): Add missing options and
30966 reorder options by implied ISAs, as in ix86_target_string.
30968 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
30970 * hash-table.h (hash_table::too_empty_p): New function.
30971 (hash_table::expand): Use it.
30972 (hash_table::traverse): Likewise.
30973 (hash_table::empty_slot): Use sizeof (value_type) instead of
30974 sizeof (PTR) to convert bytes to elements. Shrink the table
30975 if the current size is excessive for the current number of
30978 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
30980 * ira-costs.c (record_reg_classes): Break from the inner loop
30981 early once alt_fail is known to be true. Update outer loop
30982 handling accordingly.
30984 2017-01-13 Jeff Law <law@redhat.com>
30986 * tree-ssa-dse.c (decrement_count): New function.
30987 (increment_start_addr, maybe_trim_memstar_call): Likewise.
30988 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
30989 when we know the partially dead statement is a mem* function.
30991 PR tree-optimization/61912
30992 PR tree-optimization/77485
30993 * tree-ssa-dse.c: Include expr.h.
30994 (maybe_trim_constructor_store): New function.
30995 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
30997 PR tree-optimization/33562
30998 PR tree-optimization/61912
30999 PR tree-optimization/77485
31000 * doc/invoke.texi: Document new dse-max-object-size param.
31001 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
31002 * tree-ssa-dse.c: Include params.h.
31003 (dse_store_status): New enum.
31004 (initialize_ao_ref_for_dse): New, partially extracted from
31006 (valid_ao_ref_for_dse, normalize_ref): New.
31007 (setup_live_bytes_from_ref, compute_trims): Likewise.
31008 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
31009 (maybe_trim_partially_dead_store): Likewise.
31010 (maybe_trim_complex_store): Likewise.
31011 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
31012 Track what bytes live from the original store. Return tri-state
31013 for dead, partially dead or live.
31014 (dse_dom_walker): Add constructor, destructor and new private members.
31015 (delete_dead_call, delete_dead_assignment): New extracted from
31017 (dse_optimize_stmt): Make a member of dse_dom_walker.
31018 Use initialize_ao_ref_for_dse.
31020 PR tree-optimization/33562
31021 PR tree-optimization/61912
31022 PR tree-optimization/77485
31023 * sbitmap.h (bitmap_count_bits): Prototype.
31024 (bitmap_clear_range, bitmap_set_range): Likewise.
31025 * sbitmap.c (bitmap_clear_range): New function.
31026 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
31028 2017-01-13 Martin Liska <mliska@suse.cz>
31031 * function.c (set_cfun): Add new argument force.
31032 * function.h (set_cfun): Likewise.
31033 * ipa-inline-transform.c (inline_call): Use the function when
31034 strict alising from is dropped for function we inline to.
31036 2017-01-13 Richard Biener <rguenther@suse.de>
31038 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
31039 for dumping GIMPLE INTEGER_CSTs.
31041 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31043 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
31044 to 201112L since C++17.
31046 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
31049 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
31050 if -fsanitize=kernel-address is present.
31052 2017-01-13 Richard Biener <rguenther@suse.de>
31054 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
31055 as _Literal ( type ) number in case usual suffixes do not
31056 preserve all information.
31058 2017-01-13 Richard Biener <rguenther@suse.de>
31060 PR tree-optimization/77283
31061 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
31062 and ssa-iterators.h.
31063 (is_feasible_trace): Implement a cost model based on joiner
31066 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
31069 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
31070 char or short to __float128/_Float128 directly.
31072 2017-01-12 Martin Sebor <msebor@redhat.com>
31074 to -Wformat-overflow.
31075 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
31076 (min_bytes_remaining): Same.
31077 (get_string_length): Same.
31078 (format_string): Same.
31079 (format_directive): Same.
31081 (pass_sprintf_length::handle_gimple_call): Same.
31083 2017-01-12 Jakub Jelinek <jakub@redhat.com>
31085 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
31086 info.nowrite calls with no lhs that can't throw. Return bool
31087 whether gsi_remove has been called or not.
31088 (pass_sprintf_length::handle_gimple_call): Return bool whether
31089 try_substitute_return_value called gsi_remove. Formatting fix.
31090 (pass_sprintf_length::execute): Don't use gsi_remove if
31091 handle_gimple_call returned true.
31094 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
31095 be removed due to side-effects, don't remove following barrier nor
31096 turn the successor edge into fallthru edge.
31098 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31101 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
31102 element-reversing loads and stores as not swappable.
31104 2017-01-12 Nathan Sidwell <nathan@acm.org>
31105 Nicolai Stange <nicstange@gmail.com>
31107 * combine.c (try_combine): Don't ignore result of overlap checking
31108 loop. Combine overlap & asm check into single loop.
31110 2017-01-12 Richard Biener <rguenther@suse.de>
31112 * tree-pretty-print.c (dump_generic_node): Provide -gimple
31113 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
31115 2017-01-12 Richard Biener <rguenther@suse.de>
31117 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
31118 and TS_TARGET_OPTION directly derive from TS_BASE.
31119 * tree-core.h (tree_optimization_option): Derive from tree_base.
31120 (tree_target_option): Likewise.
31122 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
31124 * config/i386/i386.c (memory_address_length): Increase len
31125 only when rip_relative_addr_p returns false.
31127 2017-01-11 Julia Koval <julia.koval@intel.com>
31129 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
31130 (OPTION_MASK_ISA_SGX_SET): New.
31131 (ix86_handle_option): Handle OPT_msgx.
31132 * config.gcc: Added sgxintrin.h.
31133 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
31134 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
31135 * config/i386/i386.c (ix86_target_string): Add -msgx.
31137 (ix86_option_override_internal): Handle new options.
31138 (ix86_valid_target_attribute_inner_p): Add sgx.
31139 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
31140 * config/i386/i386.opt: Add msgx.
31141 * config/i386/sgxintrin.h: New file.
31142 * config/i386/x86intrin.h: Add sgxintrin.h.
31144 2017-01-11 Jakub Jelinek <jakub@redhat.com>
31147 * fold-const.c (maybe_nonzero_address): Return 1 for function
31149 (tree_single_nonzero_warnv_p): Don't handle function local objects
31153 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
31156 2017-01-11 David Malcolm <dmalcolm@redhat.com>
31159 * opts.c: Include "spellcheck.h"
31160 (struct string_fragment): New struct.
31161 (struct edit_distance_traits<const string_fragment &>): New
31163 (get_closest_sanitizer_option): New function.
31164 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
31166 2017-01-11 Jakub Jelinek <jakub@redhat.com>
31168 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
31170 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
31171 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
31172 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
31173 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
31174 for initial die_offset if dwarf_split_debug_info.
31175 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
31176 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
31178 (output_skeleton_debug_sections): Formatting fix. Use
31179 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
31180 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
31182 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
31184 * config/arm/cortex-a53.md: Add bypasses for
31185 cortex_a53_r2f_cvt.
31186 (cortex_a53_r2f): Only use for transfers.
31187 (cortex_a53_f2r): Likewise.
31188 (cortex_a53_r2f_cvt): Add reservation for conversions.
31189 (cortex_a53_f2r_cvt): Likewise.
31191 2017-01-11 Tamar Christina <tamar.christina@arm.com>
31193 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
31194 to all inlined functions, change static to extern.
31196 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
31199 * config/arm/arm.c (legitimize_pic_address): Handle reference to
31201 (arm_assemble_integer): Likewise.
31203 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
31205 * config.gcc: Use new awk script to check CPU, FPU and architecture
31206 parameters for --with-... options.
31207 * config/arm/parsecpu.awk: New file
31208 * config/arm/arm-cpus.in: New file.
31209 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
31211 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
31213 * config/arm/t-arm: Update dependency rules.
31214 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
31215 of processing .def files.
31216 * config/arm/genopt.sh: Deleted.
31217 * config/arm/gentune.sh: Deleted.
31218 * config/arm/arm-cores.def: Deleted.
31219 * config/arm/arm-arches.def: Deleted.
31220 * config/arm/arm-fpus.def: Deleted.
31221 * config/arm/arm-tune.md: Regenerated.
31222 * config/arm/arm-tables.opt: Regenerated.
31223 * config/arm/arm-cpu.h: New generated file.
31224 * config/arm/arm-cpu-data.h: New generated file.
31225 * config/arm/arm-cpu-cdata.h: New generated file.
31227 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
31230 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
31232 (input_varpool_node): Unpack dynamically_initialized bit.
31234 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
31236 PR rtl-optimization/79032
31237 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
31238 the alignment of the adjusted memory reference against that of MODE,
31239 instead of the alignment of the original memory reference.
31241 2017-01-11 Martin Jambor <mjambor@suse.cz>
31243 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
31245 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
31246 decorated functions.
31248 2017-01-11 Richard Biener <rguenther@suse.de>
31250 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
31251 set range/nonnull info for PHI results. Do not set it on
31252 stmts marked for removal.
31254 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
31256 * expr.c (store_field): In the bitfield case, fetch the return value
31257 from the registers before applying a single big-endian adjustment.
31258 Always do a final load for a BLKmode value not larger than a word.
31260 2017-01-10 David Malcolm <dmalcolm@redhat.com>
31263 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
31264 that we correctly handle column numbers greater than
31265 LINE_MAP_MAX_COLUMN_NUMBER.
31267 2017-01-10 Martin Sebor <msebor@redhat.com>
31269 PR middle-end/78245
31270 * gimple-ssa-sprintf.c (get_destination_size): Call
31271 {init,fini}object_sizes.
31272 * tree-object-size.c (addr_object_size): Adjust.
31273 (pass_through_call): Adjust.
31274 (pass_object_sizes::execute): Adjust.
31275 * tree-object-size.h (fini_object_sizes): Declare.
31277 2017-01-10 Martin Sebor <msebor@redhat.com>
31279 PR tree-optimization/78775
31280 * builtins.c (get_size_range): Move...
31281 * calls.c: ...to here.
31282 (alloc_max_size): Accept zero argument.
31283 (operand_signed_p): Remove.
31284 (maybe_warn_alloc_args_overflow): Call get_size_range.
31285 * calls.h (get_size_range): Declare.
31287 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
31289 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
31290 from TI's devices.csv file as of September 2016.
31291 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
31293 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
31295 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
31296 * doc/invoke.texi: Likewise.
31297 * doc/md.texi: Likewise.
31298 * doc/objc.texi: Likewise.
31300 2017-01-10 Joshua Conner <joshconner@google.com>
31302 * config/arm/fuchsia-elf.h: New file.
31303 * config/fuchsia.h: New file.
31304 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
31305 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
31307 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
31309 2016-01-10 Richard Biener <rguenther@suse.de>
31311 PR tree-optimization/79034
31312 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
31313 Propagate out degenerate PHIs in the joiner.
31315 2017-01-10 Martin Liska <mliska@suse.cz>
31317 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
31318 (sort_congruence_classes_by_decl_uid): Likewise.
31319 (sort_congruence_class_groups_by_decl_uid): Likewise.
31320 (sem_item_optimizer::merge_classes): Sort class, groups in these
31321 classes and members in the groups by DECL_UID of declarations.
31322 This would make merge operations stable.
31324 2017-01-10 Martin Liska <mliska@suse.cz>
31326 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
31327 usage of m_classes_vec.
31328 (sem_item_optimizer::~sem_item_optimizer): Likewise.
31329 (sem_item_optimizer::get_group_by_hash): Likewise.
31330 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
31331 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
31332 (sem_item_optimizer::verify_classes): Likewise.
31333 (sem_item_optimizer::process_cong_reduction): Likewise.
31334 (sem_item_optimizer::dump_cong_classes): Likewise.
31335 (sem_item_optimizer::merge_classes): Likewise.
31336 * ipa-icf.h (congruence_class_hash): Rename from
31337 congruence_class_group_hash. Remove declaration of m_classes_vec.
31339 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
31341 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
31342 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
31343 * config.gcc: Add avx512vpopcntdqintrin.h.
31344 * config/i386/avx512vpopcntdqintrin.h: New.
31345 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
31346 * config/i386/i386-builtin-types.def: Add new types.
31347 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
31348 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
31349 __builtin_ia32_vpopcountq_v8di_mask): New.
31350 * config/i386/i386-c.c (ix86_target_macros_internal): Define
31351 __AVX512VPOPCNTDQ__.
31352 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
31353 (PTA_AVX512VPOPCNTDQ): Define.
31354 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
31355 TARGET_AVX512VPOPCNTDQ_P): Define.
31356 * config/i386/i386.opt: Add mavx512vpopcntdq.
31357 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
31358 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
31360 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31362 PR middle-end/77484
31363 * predict.def (PRED_CALL): Set to 67.
31365 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
31367 * expr.c (store_field): In the bitfield case, if the value comes from
31368 a function call and is of an aggregate type returned in registers, do
31369 not modify the field mode; extract the value in all cases if the mode
31370 is BLKmode and the size is not larger than a word.
31372 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
31375 * config/i386/cpuid.h: Fix undefined behavior.
31377 2017-01-04 Jeff Law <law@redhat.com>
31379 PR tree-optimization/79007
31380 PR tree-optimization/67955
31381 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
31382 conservative for pt.null when flag_non_call_exceptions is on.
31384 2017-01-09 Jakub Jelinek <jakub@redhat.com>
31386 PR translation/79019
31387 PR translation/79020
31388 * params.def (PARAM_INLINE_MIN_SPEEDUP,
31389 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
31390 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
31392 * config/avr/avr.opt (maccumulate-args): Likewise.
31393 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
31394 * common.opt (freport-bug): Likewise.
31395 * cif-code.def (CIF_FINAL_ERROR): Likewise.
31396 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
31397 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
31398 translatable string.
31399 * config/i386/i386.c (function_value_32): Likewise.
31400 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
31401 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
31403 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
31404 * common/config/msp430/msp430-common.c (msp430_handle_option):
31406 * symtab.c (symtab_node::verify_base): Likewise.
31407 * opts.c (set_debug_level): Likewise.
31408 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
31409 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
31410 missing whitespace to translatable strings.
31411 * config/avr/avr.md (bswapsi2): Fix typo in comment.
31412 * config/sh/superh.h: Likewise.
31413 * config/i386/xopintrin.h: Likewise.
31414 * config/i386/znver1.md: Likewise.
31415 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
31416 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
31417 * double-int.h (struct double_int): Likewise.
31418 * double-int.c (div_and_round_double): Likewise.
31419 * wide-int.cc: Likewise.
31420 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
31421 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
31422 * cfgcleanup.c (crossjumps_occured): Renamed to ...
31423 (crossjumps_occurred): ... this.
31424 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
31427 PR tree-optimization/78899
31428 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
31429 returning bool return struct loop *, NULL for failure and the new
31431 (versionable_outer_loop_p): Don't version outer loop if it has
31432 dont_vectorized bit set.
31433 (tree_if_conversion): When versioning outer loop, ensure
31434 tree_if_conversion is performed also on the inner loop of the
31435 non-vectorizable outer loop copy.
31436 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
31437 LOOP_VECTORIZED in inner loop of the scalar outer loop and
31438 prevent vectorization of it.
31439 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
31440 the outer loop vectorization of the non-scalar version is attempted
31441 before vectorization of the inner loop in scalar version. If
31442 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
31443 vectorization of its inner loop.
31444 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
31445 has 2 inner loops, rename also on edges from bb whose single pred
31446 is outer_loop->header. Fix typo in function comment.
31448 2017-01-09 Martin Sebor <msebor@redhat.com>
31451 * asan.c (asan_emit_stack_protection): Increase local buffer size
31452 to avoid snprintf truncation warning.
31454 2017-01-09 Andrew Pinski <apinski@cavium.com>
31456 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
31457 to reference thunderx2t99 for the tuning structure
31458 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
31460 (thunderx2t99_extra_costs): This.
31461 * config/aarch64/aarch64-tune.md: Regenerate.
31462 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
31463 (vulcan_addrcost_table): This.
31464 (vulcan_regmove_cost): Rename to ...
31465 (thunderx2t99_regmove_cost): This.
31466 (vulcan_vector_cost): Rename to ...
31467 (thunderx2t99_vector_cost): this.
31468 (vulcan_branch_cost): Rename to ...
31469 (thunderx2t99_branch_cost): This.
31470 (vulcan_tunings): Rename to ...
31471 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
31472 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
31474 2017-01-09 Martin Jambor <mjambor@suse.cz>
31478 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
31479 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
31480 (propagate_vr_accross_jump_function): Use the above function for all
31481 value range computations for pass-through jump functions and type
31482 converasion from explicit value range values.
31483 (ipcp_propagate_stage): Do not attempt to deduce types of formal
31484 parameters from TYPE_ARG_TYPES.
31485 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
31486 (ipa_write_node_info): Stream type of the actual argument.
31487 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
31489 2017-01-09 Martin Liska <mliska@suse.cz>
31492 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
31493 (lookup_compiler): Do not show error message with have_E.
31495 2017-01-09 Jakub Jelinek <jakub@redhat.com>
31497 PR tree-optimization/78938
31498 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
31499 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
31500 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
31501 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
31504 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31506 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
31509 2017-01-09 Richard Biener <rguenther@suse.de>
31511 PR tree-optimization/78997
31512 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
31513 name condition properly.
31515 2017-01-09 Richard Biener <rguenther@suse.de>
31518 * dwarf2out.c (is_cxx): New overload with context.
31519 (is_naming_typedef_decl): Use it.
31521 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31523 * invoke.texi (Option Summary): Correct spacing in option lists
31524 and add line breaks to fix over-long lines.
31526 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31528 PR middle-end/17660
31530 * extend.texi (Common Variable Attributes): Add xref to GCC
31531 Internals manual to explain mode attribute keywords.
31533 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31536 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
31537 and Preprocessor Options.
31538 (Options for Linking): Document -pthread here....
31539 (RS/6000 and PowerPC Options): ...not here.
31540 (Solaris 2 Options): ...or here.
31541 * doc/cppopts.texi: Document -pthread.
31543 2017-01-08 Martin Sebor <msebor@redhat.com>
31545 PR middle-end/77708
31546 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
31547 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
31548 New member functions.
31549 (format_directive): Used them.
31551 (pass_sprintf_length::handle_gimple_call): Same.
31552 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
31553 to avoid truncation for any argument.
31554 (extract_affine_mul): Same.
31555 * tree.c (get_file_function_name): Same.
31557 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31559 PR middle-end/77484
31560 * predict.def (PRED_INDIR_CALL): Set to 86.
31562 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31564 PR preprocessor/54124
31565 * doc/cppopts.texi: Reformat -d subtable to list the full name
31566 of the options. Add cross-reference to the docs for the general
31567 compiler -d options.
31568 * doc/invoke.texi (Developer Options): Add cross-reference to the
31569 preprocessor-specific -d option documentation.
31571 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31573 PR preprocessor/13498
31574 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
31575 redudant material, and reflect new command-line options.
31576 (System Headers): Likewise.
31578 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31580 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
31581 -isystem, and -idirafter. Copy-edit.
31582 * doc/cppopts.texi: Copy-edit. Remove contradiction about
31583 default for -ftrack-macro-expansion. Delete obsolete and
31584 badly-formatted implementation details about -fdebug-cpp output.
31585 * doc/cppwarnopts.texi: Copy-edit.
31587 2017-01-07 David Malcolm <dmalcolm@redhat.com>
31590 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
31591 that the transition from a max line width >= 1<<10 to narrower
31592 lines works correctly.
31594 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
31596 * doc/options.texi (PerFunction): New.
31597 * opt-functions.awk (switch_flags): Map both Optimization and
31598 PerFunction to CL_OPTIMIZATION.
31599 * opth-gen.awk: Test for PerFunction flag along with
31601 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
31602 it only when the latter is present. Skip those that don't in
31603 the hash function generator.
31604 * common.opt (fvar-tracking): Mark as PerFunction instead of
31606 (fvar-tracking-assignments): Likewise.
31607 (fvar-tracking-assignments-toggle): Likewise.
31608 (fvar-tracking-uninit): Likewise.
31610 2017-01-07 Jakub Jelinek <jakub@redhat.com>
31612 PR translation/79018
31613 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
31616 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
31619 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
31620 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
31621 TARGET_LONG_JUMP_TABLE_OFFSETS.
31622 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
31623 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
31624 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
31625 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
31626 * config/m68k/m68k.md (tablejump expander): Likewise.
31627 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
31628 TARGET_LONG_JUMP_TABLE_OFFSETS.
31629 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
31630 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
31632 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
31633 David Holsgrove <david.holsgrove@xilinx.com>
31635 * common/config/microblaze/microblaze-common.c
31636 (TARGET_EXCEPT_UNWIND_INFO): Remove.
31637 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
31639 * config/microblaze/microblaze.c (microblaze_must_save_register)
31640 (microblaze_expand_epilogue, microblaze_return_addr): Handle
31642 (microblaze_eh_return): New function.
31643 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
31644 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
31645 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
31646 * config/microblaze/microblaze.md (eh_return): New pattern.
31648 2017-01-06 Jakub Jelinek <jakub@redhat.com>
31650 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
31651 GCC_DIAGNOSTIC_STRINGIFY): Define.
31653 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
31655 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31657 * config/arm/arm.md (<mcrr>): New.
31659 * config/arm/arm.c (arm_arch5te): New.
31660 (arm_option_override): Set arm_arch5te.
31661 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
31663 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
31664 (arm_mcrr_qualifiers): ... this. New.
31665 (MRRC_QUALIFIERS): Define to...
31666 (arm_mrrc_qualifiers): ... this. New.
31667 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
31669 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
31670 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
31671 (MRRCI, mrrc, MRRC): New.
31672 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
31673 VUNSPEC_MRRC2): New.
31675 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31677 * config/arm/arm.md (<mcr>): New.
31679 * config/arm/arm.c (arm_coproc_builtin_available): Add
31680 support for mcr, mrc, mcr2 and mrc2.
31681 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
31682 (arm_mcr_qualifiers): ... this. New.
31683 (MRC_QUALIFIERS): Define to ...
31684 (arm_mrc_qualifiers): ... this. New.
31685 (MCR_QUALIFIERS): Define to ...
31686 (arm_mcr_qualifiers): ... this. New.
31687 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
31689 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
31690 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
31691 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
31692 VUNSPEC_MRC2): New.
31694 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31696 * config/arm/arm.md (*ldc): New.
31700 * config/arm/arm.c (arm_coproc_builtin_available): Add
31701 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
31702 (arm_coproc_ldc_stc_legitimate_address): New.
31703 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
31704 'qualifier_const_pointer'.
31705 (LDC_QUALIFIERS): Define to...
31706 (arm_ldc_qualifiers): ... this. New.
31707 (STC_QUALIFIERS): Define to...
31708 (arm_stc_qualifiers): ... this. New.
31709 * config/arm/arm-protos.h
31710 (arm_coproc_ldc_stc_legitimate_address): New.
31711 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
31712 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
31713 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
31714 stc2, stcl, stc2l): New.
31715 * config/arm/constraints.md (Uz): New.
31716 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
31717 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
31718 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
31719 VUNSPEC_STC2L): New.
31721 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31723 * config/arm/arm.md (<cdp>): New.
31724 * config/arm/arm.c (neon_const_bounds): Rename this ...
31725 (arm_const_bounds): ... this.
31726 (arm_coproc_builtin_available): New.
31727 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
31728 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
31729 (CDP_QUALIFIERS): Define to...
31730 (arm_cdp_qualifiers): ... this. New.
31732 (arm_expand_builtin_args): Add case for 6 arguments.
31733 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
31734 (arm_const_bounds): ... this.
31735 (arm_coproc_builtin_available): New.
31736 * config/arm/arm_acle.h (__arm_cdp): New.
31738 * config/arm/arm_acle_builtins.def (cdp): New.
31740 * config/arm/iterators.md (CDPI,CDP,cdp): New.
31741 * config/arm/neon.md: Rename all 'neon_const_bounds' to
31742 'arm_const_bounds'.
31743 * config/arm/types.md (coproc): New.
31744 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
31745 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
31746 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
31747 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
31749 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31751 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
31752 (UBINOP_QUALIFIERS): New.
31754 (acle_builtin_data): New. Change comment.
31755 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
31756 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
31757 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
31758 arm_acle_builtins.def.
31759 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
31760 (arm_init_acle_builtins): New.
31761 (CRC32_BUILTIN): Remove.
31762 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
31763 crc32cb, crc32ch and crc32cw.
31764 (arm_init_crc32_builtins): Remove.
31765 (arm_init_builtins): Use arm_init_acle_builtins rather
31766 than arm_init_crc32_builtins.
31767 (arm_expand_acle_builtin): New.
31768 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
31769 * config/arm/arm_acle_builtins.def: New.
31771 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31773 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
31774 (arm_builtin_datum): ... this.
31775 (arm_init_neon_builtin): Rename to ...
31776 (arm_init_builtin): ... this. Add a new parameters PREFIX
31777 and USE_SIG_IN_NAME.
31778 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
31779 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
31780 'arm_builtin_datum'.
31781 (arm_init_vfp_builtins): Likewise.
31782 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
31783 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
31784 (arm_expand_neon_args): Rename to ...
31785 (arm_expand_builtin_args): ... this. Rename builtin_arg
31786 enum values and differentiate between ARG_BUILTIN_MEMORY
31787 and ARG_BUILTIN_NEON_MEMORY.
31788 (arm_expand_neon_builtin_1): Rename to ...
31789 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
31790 values, arm_expand_builtin_args and add bool parameter NEON.
31791 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
31792 (arm_expand_vfp_builtin): Likewise.
31793 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
31795 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31797 PR middle-end/77484
31798 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
31799 * predict.c (tree_estimate_probability_bb): Reverse direction of
31800 polymorphic call predictor.
31802 2017-01-06 David Malcolm <dmalcolm@redhat.com>
31804 * passes.c (execute_one_pass): Split out pass-skipping logic into...
31805 (determine_pass_name_match): ...this new function and...
31806 (should_skip_pass_p): ...this new function.
31808 2017-01-06 Nathan Sidwell <nathan@acm.org>
31810 * ipa-visibility.c (function_and_variable_visibility): Reformat
31811 comments and long lines. Remove extrneous if.
31812 * symtab.c (symtab_node::make_decl_local): Fix code format.
31813 (symtab_node::set_section_for_node): Fix comment typo.
31815 2017-01-06 Martin Liska <mliska@suse.cz>
31818 * lra-constraints.c: Rename invariant to lra_invariant.
31819 * predict.c (set_even_probabilities): Initialize e to NULL.
31821 2017-01-05 Martin Sebor <msebor@redhat.com>
31823 PR tree-optimization/78910
31824 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
31825 (format_integer): Correct off-by-one error in the handling
31826 of precision with negative numbers in signed conversions..
31828 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
31830 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
31832 2017-01-05 Jakub Jelinek <jakub@redhat.com>
31834 PR tree-optimization/71016
31835 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
31836 factor_out_conditional_conversion. Formatting fix.
31837 (factor_out_conditional_conversion): Add cond_stmt argument.
31838 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
31839 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
31842 2017-01-05 David Malcolm <dmalcolm@redhat.com>
31844 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
31845 read-rtl-function.o, and selftest-rtl.o.
31846 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
31847 (selftest::aarch64_test_loading_full_dump): New function.
31848 (selftest::aarch64_run_selftests): New function.
31849 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
31850 selftest::aarch64_run_selftests.
31851 * config/i386/i386.c
31852 (selftest::ix86_test_loading_dump_fragment_1): New function.
31853 (selftest::ix86_test_loading_call_insn): New function.
31854 (selftest::ix86_test_loading_full_dump): New function.
31855 (selftest::ix86_test_loading_unspec): New function.
31856 (selftest::ix86_run_selftests): Call the new functions.
31857 * emit-rtl.c (maybe_set_max_label_num): New function.
31858 * emit-rtl.h (maybe_set_max_label_num): New decl.
31859 * function.c (instantiate_decls): Guard call to
31860 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
31861 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
31863 * gensupport.c (gen_reader::gen_reader): Pass "false"
31864 for new "compact" param of rtx_reader.
31865 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
31866 rather than an empty string for NULL strings.
31867 * read-md.c: Potentially include config.h rather than bconfig.h.
31868 Wrap include of errors.h with #ifdef GENERATOR_FILE.
31869 (have_error): New global, copied from errors.c.
31870 (md_reader::read_name): Rename to...
31871 (md_reader::read_name_1): ...this, adding "out_loc" param,
31872 and converting "missing name or number" to returning false, rather
31874 (md_reader::read_name): Reimplement in terms of read_name_1.
31875 (md_reader::read_name_or_nil): New function.
31876 (md_reader::read_string): Handle "(nil)" by returning NULL.
31877 (md_reader::md_reader): Add new param "compact".
31878 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
31879 (md_reader::read_file): New method.
31880 * read-md.h (md_reader::md_reader): Add new param "compact".
31881 (md_reader::read_file): New method.
31882 (md_reader::is_compact): New accessor.
31883 (md_reader::read_name): Convert return type from void to file_location.
31884 (md_reader::read_name_or_nil): New decl.
31885 (md_reader::read_name_1): New decl.
31886 (md_reader::m_compact): New field.
31887 (noop_reader::noop_reader): Pass "false" for new "compact" param
31889 (rtx_reader::rtx_reader): Add new "compact" param.
31890 (rtx_reader::read_rtx_operand): Make virtual and convert return
31891 type from void to rtx.
31892 (rtx_reader::read_until): New decl.
31893 (rtx_reader::handle_any_trailing_information): New virtual function.
31894 (rtx_reader::postprocess): New virtual function.
31895 (rtx_reader::finalize_string): New virtual function.
31896 (rtx_reader::m_in_call_function_usage): New field.
31897 (rtx_reader::m_reuse_rtx_by_id): New field.
31898 * read-rtl-function.c: New file.
31899 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
31900 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
31901 (selftest::verify_three_block_rtl_cfg): New decl.
31902 * read-rtl-function.h: New file.
31903 * read-rtl.c: Potentially include config.h rather than bconfig.h.
31904 For host, include function.h, memmodel.h, and emit-rtl.h.
31905 (one_time_initialization): New function.
31906 (struct compact_insn_name): New struct.
31907 (compact_insn_names): New array.
31908 (find_code): Handle insn codes in compact dumps.
31909 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
31910 (bind_subst_iter_and_attr): Likewise.
31911 (add_condition_to_string): Likewise.
31912 (add_condition_to_rtx): Likewise.
31913 (apply_attribute_uses): Likewise.
31914 (add_current_iterators): Likewise.
31915 (apply_iterators): Likewise.
31916 (initialize_iterators): Guard usage of apply_subst_iterator with
31917 #ifdef GENERATOR_FILE.
31918 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
31919 (md_reader::read_mapping): Likewise.
31920 (add_define_attr_for_define_subst): Likewise.
31921 (add_define_subst_attr): Likewise.
31922 (read_subst_mapping): Likewise.
31923 (check_code_iterator): Likewise.
31924 (rtx_reader::read_rtx): Likewise. Move one-time initialization
31926 (one_time_initialization): New function.
31927 (rtx_reader::read_until): New method.
31928 (read_flags): New function.
31929 (parse_reg_note_name): New function.
31930 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
31931 Handle reuse_rtx ids.
31932 Wrap iterator lookup within #ifdef GENERATOR_FILE.
31933 Add parsing support for RTL dumps, mirroring the special-cases in
31934 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
31935 values, and calling handle_any_trailing_information.
31936 (rtx_reader::read_rtx_operand): Convert return type from void
31937 to rtx, returning return_rtx. Handle case 'e'. Call
31938 finalize_string on XSTR and XTMPL fields.
31939 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
31940 "(nil)" values were omitted. Call the postprocess vfunc on the
31942 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
31943 class ctor. Initialize m_in_call_function_usage. Call
31944 one_time_initialization.
31945 * rtl-tests.c (selftest::test_uncond_jump): Call
31946 set_new_first_and_last_insn.
31947 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
31948 * selftest-rtl.c: New file.
31949 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
31950 (selftest::get_insn_by_uid): New decl.
31951 * selftest-run-tests.c (selftest::run_tests): Call
31952 read_rtl_function_c_tests.
31953 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
31954 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
31957 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
31959 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
31960 operands in a special way. Assert that pos+len <= mode precision.
31962 2017-01-05 Jakub Jelinek <jakub@redhat.com>
31964 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
31965 3 argument Alias with unlimited for the negative form.
31966 (fno-vect-cost-model): Removed.
31968 2017-01-05 Martin Liska <mliska@suse.cz>
31970 * hsa-gen.c (gen_hsa_divmod): New function.
31971 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
31973 2017-01-05 Martin Liska <mliska@suse.cz>
31976 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
31979 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31981 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
31982 small constant length operands.
31984 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31986 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
31987 between loop iterations.
31989 2017-01-05 Martin Liska <mliska@suse.cz>
31992 * gimplify.c (gimplify_decl_expr): Compare to
31993 asan_poisoned_variables instread of checking flags.
31994 (gimplify_target_expr): Likewise.
31995 (gimplify_expr): Likewise.
31996 (gimplify_function_tree): Conditionally initialize
31997 asan_poisoned_variables.
31999 2017-01-04 Jeff Law <law@redhat.com>
32001 PR tree-optimizatin/78812
32002 * rtl.h (contains_mem_rtx_p): Prototype.
32003 * ifcvt.c (containts_mem_rtx_p): Move from here to...
32004 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
32005 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
32006 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
32007 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
32009 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32011 * input.c (assert_char_at_range): Default-initialize actual_range.
32013 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32015 * df-scan.c (df_ref_create_structure): Make regno unsigned,
32016 to match the caller.
32018 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32020 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
32021 insns after final jump in test to emit dummy move.
32023 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32025 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
32026 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
32028 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32030 * multiple_target.c (create_dispatcher_calls): Init e_next.
32031 * tree-ssa-loop-split.c (split_loop): Init border.
32032 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
32035 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
32040 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
32041 (altivec_register_operand): Do not return true if the operand
32042 contains a SUBREG mixing SImode and SFmode.
32043 (vsx_register_operand): Likewise.
32044 (vsx_reg_sfsubreg_ok): New predicate.
32045 (vfloat_operand): Do not return true if the operand contains a
32046 SUBREG mixing SImode and SFmode.
32047 (vint_operand): Likewise.
32048 (vlogical_operand): Likewise.
32049 (gpc_reg_operand): Likewise.
32050 (int_reg_operand): Likewise.
32051 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
32052 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
32053 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
32055 (rs6000_emit_move_si_sf_subreg): New helper function.
32056 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
32057 fixup SUBREGs involving SImode and SFmode.
32058 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
32059 numbers for the new peephole2 optimization.
32060 (peephole2 for SFmode unions): New peephole2 to optimize cases in
32061 the GLIBC math library that do AND/IOR/XOR operations on single
32062 precision floating point.
32063 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
32064 target macros to say whether we need to avoid SUBREGs mixing
32066 (TARGET_ALLOW_SF_SUBREG): Likewise.
32067 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
32068 (UNSPEC_SI_FROM_SF): Likewise.
32069 (iorxor): Change spacing.
32070 (and_ior_xor): New iterator for AND, IOR, and XOR.
32071 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
32072 (movdi_from_sf_zero_ext): Likewise.
32073 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
32074 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
32075 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
32076 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
32077 (fms<mode>4): Likewise.
32078 (fnma<mode>4): Likewise.
32079 (fnms<mode>4): Likewise.
32080 (nfma<mode>4): Likewise.
32081 (nfms<mode>4): Likewise.
32083 2017-01-04 Marek Polacek <polacek@redhat.com>
32086 * doc/invoke.texi: Document -Wpointer-compare.
32088 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32090 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
32093 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
32094 descriptions for -gdwarf-5 and emit them as uleb128 instead of
32097 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
32100 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
32101 documentation of the powerpc_popcntb_ok attribute.
32102 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32103 code to issue warning messages if a requested CPU configuration is
32104 not supported by the binary (assembler and loader) toolchain.
32105 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
32106 made to define a built-in function that has been disabled.
32107 (paired_init_builtins): Add assertion to prevent ICE if attempt is
32108 made to define a built-in function that has been disabled.
32109 (altivec_init_builtins): Add comment explaining why definition
32110 of the DST built-in functions is not preceded by an assertion
32111 check. Add assertions to prevent ICE if attempts are made to
32112 define an altivec predicate or an abs* built-in function that has
32114 (htm_init_builtins): Add comment explaining why definition of the
32115 htm built-in functions is not preceded by an assertion check.
32117 2017-01-04 Jeff Law <law@redhat.com>
32119 PR tree-optimizatin/67955
32120 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
32121 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
32122 the points-to solution does not include pt_null. Use DECL_PT_UID
32125 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
32127 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
32128 Use gen_int_mode instead of gen_lopwart for const_int operands.
32130 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32132 PR tree-optimization/71563
32133 * match.pd: Simplify X << Y into X if Y is known to be 0 or
32134 out of range value - has low bits known to be zero.
32136 2017-01-04 Alan Modra <amodra@gmail.com>
32138 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
32139 * configure: Regenerate.
32140 * config.in: Regenerate.
32142 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32145 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
32146 a substring of the message, but strcmp with the whole message. Ifdef
32147 ENABLE_NLS, translate the message first using dgettext.
32149 2017-01-03 Jeff Law <law@redhat.com>
32151 PR tree-optimizatin/78856
32152 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
32153 (mark_threaded_blocks): Remove code to truncate thread paths that
32154 cross multiple loop headers. Instead invalidate the cached loop
32155 iteration information and handle case of a thread path walking
32156 into an irreducible region.
32158 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
32161 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
32162 assertions. Add support for doing the signbit if the IEEE 128-bit
32163 floating point value is in a GPR.
32164 * config/rs6000/rs6000.md (Fsignbit): Delete.
32165 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
32166 Update the length attribute if the value is in a GPR.
32167 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
32168 the sign or zero extension instruction, since the value is always 0/1.
32169 (signbit<mode>2_dm2): Delete using <Fsignbit>.
32172 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
32173 extracting SImode to a GPR register so that we can generate a
32174 store, limit the vector to be in a traditional Altivec register
32175 for the vextuwrx instruction.
32177 2017-01-03 Ian Lance Taylor <iant@google.com>
32179 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
32181 2017-01-03 Martin Sebor <msebor@redhat.com>
32183 PR tree-optimization/78696
32184 * gimple-ssa-sprintf.c (format_floating): Correct handling of
32185 precision. Use MPFR for %f for greater fidelity. Correct handling
32187 (pass_sprintf_length::compute_format_length): Set width and precision
32188 specified by asrerisk to void_node for vararg functions.
32189 (try_substitute_return_value): Adjust dump output.
32191 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
32193 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
32195 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
32197 * doc/invoke.texi (SPARC options): Document -mlra as the default.
32198 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
32199 -mlra/-mno-lra was passed to the compiler.
32201 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
32203 PR rtl-optimization/65618
32204 * emit-rtl.c (try_split): Move initialization of "before" and
32205 "after" to just before the call to emit_insn_after_setloc.
32207 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
32209 * doc/md.texi (Standard Names): Remove reference to Java frontend.
32211 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
32213 * dwarf2out.c (gen_enumeration_type_die): When
32214 -gno-strict-dwarf, add a DW_AT_encoding attribute.
32216 2017-01-03 Jakub Jelinek <jakub@redhat.com>
32218 PR tree-optimization/78965
32219 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
32220 Change first argument from const call_info & to call_info &. For %n
32221 set info.nowrite to false.
32223 PR middle-end/78901
32224 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
32225 possibly throwing calls.
32227 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
32228 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
32229 and fns handling, rather than in a separate case SSA_NAME.
32231 2017-01-02 Jeff Law <law@redhat.com>
32233 * config/darwin-driver.c (darwin_driver_init): Const-correctness
32234 fixes for first_period and second_period variables.
32236 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
32239 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
32240 (*insvqi_1): New insn pattern.
32241 (*insvqi_1_mem_rex64): Ditto.
32242 (*insvqi_2): Ditto.
32243 (*insvqi_3): Rename from *insvqi.
32245 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
32247 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
32249 * doc/cfg.texi (Edges): Remove reference to Java.
32250 (Maintaining the CFG): Ditto.
32252 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32254 PR middle-end/77674
32255 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
32256 transparent aliases.
32258 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32260 PR middle-end/77484
32261 * predict.def (PRED_CALL): Update hitrate.
32262 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
32263 * predict.c (tree_estimate_probability_bb): Split CALL predictor
32264 into direct/indirect/polymorphic variants.
32266 2017-01-01 Jakub Jelinek <jakub@redhat.com>
32268 Update copyright years.
32270 * gcc.c (process_command): Update copyright notice dates.
32271 * gcov-dump.c (print_version): Ditto.
32272 * gcov.c (print_version): Ditto.
32273 * gcov-tool.c (print_version): Ditto.
32274 * gengtype.c (create_file): Ditto.
32275 * doc/cpp.texi: Bump @copying's copyright year.
32276 * doc/cppinternals.texi: Ditto.
32277 * doc/gcc.texi: Ditto.
32278 * doc/gccint.texi: Ditto.
32279 * doc/gcov.texi: Ditto.
32280 * doc/install.texi: Ditto.
32281 * doc/invoke.texi: Ditto.
32283 Copyright (C) 2017 Free Software Foundation, Inc.
32285 Copying and distribution of this file, with or without modification,
32286 are permitted in any medium without royalty provided the copyright
32287 notice and this notice are preserved.