1 2017-11-29 Nathan Sidwell <nathan@acm.org>
4 * tree.c (build_complex_type): Fix canonicalization. Only fill in
7 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
9 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
10 cases for printing LDP/STP memory addresses.
11 (aarch64_print_address_internal): Renamed from
12 aarch64_print_operand_address, added parameter, add Pmode check.
13 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
14 (aarch64_print_operand_address): Indirect to
15 aarch64_print_address_internal.
16 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
17 'y' operand output specifier.
19 2017-11-29 Jakub Jelinek <jakub@redhat.com>
22 * tree.c (build_simple_mem_ref_loc): Handle
23 get_addr_base_and_unit_offset returning a MEM_REF.
26 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
29 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
32 2017-11-28 David Malcolm <dmalcolm@redhat.com>
34 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
36 (layout::move_to_column): Likewise.
38 2017-11-28 Jakub Jelinek <jakub@redhat.com>
41 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
42 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
44 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
45 Martin Jambor <mjambor@suse.cz>
48 * tree.h (expr_type_first_operand_type_p): Declare
49 * tree.c (expr_type_first_operand_type_p): New function.
50 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
51 (ipa_value_from_jfunc): Adjust declaration.
52 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
53 Use it as result type for arithmetics, unless it is NULL in which case
55 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
56 ipa_get_jf_pass_through_result.
57 (propagate_vals_across_pass_through): Likewise.
58 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
59 is to propagate_vals_across_pass_through.
60 (propagate_constants_across_call): Pass PARM_TYPE to
61 propagate_scalar_across_jump_function.
62 (find_more_scalar_values_for_callers_subset): Pass parameter type to
64 (cgraph_edge_brings_all_scalars_for_node): Likewise.
65 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
66 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
67 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
68 target_type, pass it to ipa_value_from_jfunc.
69 (update_indirect_edges_after_inlining): Pass parameter type to
70 try_make_edge_direct_simple_call.
72 2017-11-28 Jeff Law <law@redhat.com>
74 * gimple-ssa-evrp-analyze.c
75 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
76 refine ranges if scev_initialized_p returns true.
77 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
79 2017-11-28 Julia Koval <julia.koval@intel.com>
81 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
82 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
83 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
84 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
85 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
86 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
87 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
88 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
89 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
90 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
91 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
92 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
93 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
94 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
95 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
96 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
97 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
98 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
99 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
100 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
101 __builtin_ia32_vpshrd_v2di_mask): New builtins.
102 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
104 2017-11-28 Julia Koval <julia.koval@intel.com>
106 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
107 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
108 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
109 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
110 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
111 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
112 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
113 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
114 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
115 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
116 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
117 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
118 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
119 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
120 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
121 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
122 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
123 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
124 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
125 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
126 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
127 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
128 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
129 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
130 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
131 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
132 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
133 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
134 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
135 __builtin_ia32_vpshld_v2di_mask): New builtins.
136 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
137 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
139 2017-11-28 Richard Biener <rguenther@suse.de>
141 PR tree-optimization/80776
142 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
144 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
146 (evrp_range_analyzer::record_ranges_from_incoming_edges):
147 If the incoming edge is an effective fallthru because the other
148 edge only reaches a __builtin_unreachable () then record ranges
149 derived from the controlling condition in SSA info.
150 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
151 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
153 2017-11-28 Olivier Hainque <hainque@adacore.com>
155 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
156 on mingw build hosts.
158 2017-11-28 Jakub Jelinek <jakub@redhat.com>
161 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
163 2017-11-28 Richard Biener <rguenther@suse.de>
166 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
167 copies generated from memcpy use a character array as reference
170 2017-11-28 Julia Koval <julia.koval@intel.com>
171 Sebastian Peryt <sebastian.peryt@intel.com>
173 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
174 c-family/c-cilkplus.o, c-family/array-notation-common.o,
175 cilk-common.o, cilk.h, cilk-common.c): Remove.
177 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
178 * builtins.c (is_builtin_name): Remove cilkplus condition.
179 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
180 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
181 cilk-builtins.def, cilkplus.def): Remove.
182 * cif-code.def (CILK_SPAWN): Remove.
183 * cilk-builtins.def: Delete.
184 * cilk-common.c: Ditto.
186 * cilkplus.def: Ditto.
187 * config/darwin.h (fcilkplus): Delete.
188 * cppbuiltin.c: Ditto.
189 * doc/extend.texi: Remove cilkplus doc.
190 * doc/generic.texi: Ditto.
191 * doc/invoke.texi: Ditto.
192 * doc/passes.texi: Ditto.
193 * gcc.c (fcilkplus): Remove.
194 * gengtype.c (cilk.h): Remove.
195 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
197 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
199 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
201 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
202 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
204 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
205 inline_read_section): Ditto.
206 * ipa-inline-analysis.c (cilk.h): Remove.
207 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
208 * lto-wrapper.c (merge_and_complain, append_compiler_options,
209 append_linker_options): Remove condition for fcilkplus.
210 * lto/lto-lang.c (cilk.h): Remove.
211 (lto_init): Remove condition for fcilkplus.
212 * omp-expand.c (expand_cilk_for_call): Delete.
213 (expand_omp_taskreg, expand_omp_for_static_chunk,
214 expand_omp_for): Remove cilkplus
216 (expand_cilk_for): Delete.
217 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
218 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
219 execute_lower_omp, diagnose_sb_0): Ditto.
220 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
221 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
222 * tree-nested.c: Ditto.
223 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
224 (dump_generic_node): Ditto.
225 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
226 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
228 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
230 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
232 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
233 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
234 for vector mode and !TARGET_SIMD.
236 2017-11-28 Jakub Jelinek <jakub@redhat.com>
238 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
240 * tree.h (SWITCH_LABELS): Remove.
241 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
242 assert SWITCH_BODY is non-NULL.
243 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
245 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
247 PR tree-optimization/80788
248 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
249 has TREE_OVERFLOW set, call drop_tree_overflow.
251 2017-11-28 Richard Biener <rguenther@suse.de>
253 PR tree-optimization/83158
254 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more
257 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
260 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
261 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
263 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
265 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
266 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
268 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
271 * varasm.c (compare_constant): Take the mode of the constants into
272 account when comparing floating point constants.
274 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
276 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
277 from explicit instantiation of debug_helper.
278 * vec.h (DEFINE_DEBUG_VEC): Ditto.
280 2017-11-27 Richard Biener <rguenther@suse.de>
282 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
285 2017-11-27 Richard Biener <rguenther@suse.de>
287 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
288 (type_hash_canon): Also clear int_cst_hash_table entry for
290 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
292 2017-11-27 Tamar Christina <tamar.christina@arm.com>
294 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
295 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
296 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
298 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
300 * hash-map.h (gt_cleare_cache): Avoid UB.
302 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
304 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
305 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
306 (decide_unroll_runtime_iterations): Likewise.
307 (decide_unroll_stupid): Likewise.
309 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
312 * config/i386/i386.md: Add a loop with incssp.
314 2017-11-27 Martin Jambor <mjambor@suse.cz>
316 PR tree-optimization/81248
317 * tree-sra.c (splice_param_accesses): Remove size check.
318 (decide_one_param_reduction): Fix size check.
319 * gimple-pretty-print.c (dump_profile): Silence warning.
320 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
322 2017-11-27 Jakub Jelinek <jakub@redhat.com>
325 * dbxout.c (lastlineno): New variable.
326 (dbx_debug_hooks): Use dbxout_switch_text_section as
327 switch_text_section debug hook.
328 (dbxout_function_end): Switch to current_function_section
329 rather than function_section. If crtl->has_bb_partition,
330 output just one N_FUN, depending on in_cold_section_p.
331 (dbxout_source_line): Remember last lineno in lastlineno.
332 (dbxout_switch_text_section): New function.
333 (dbxout_function_decl): Adjust dbxout_block caller.
334 (dbx_block_with_cold_children): New function.
335 (dbxout_block): Return true if any LBRAC/RBRAC have been
336 emitted. Use dbx_block_with_cold_children at depth == 0
337 in second partition. Add PARENT_BLOCKNUM argument, pass
338 it optionally adjusted to children. Output LBRAC/RBRAC
339 around recursive call only if the block is in the current
340 partition, if not and anything was output, emit empty
342 * final.c (final_scan_insn): Compute cold_function_name
343 before calling switch_text_section debug hook. Call
344 that hook even if dwarf2out_do_frame if not emitting
348 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
351 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
353 PR rtl-optimization/82488
354 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
356 2017-11-26 Julia Koval <julia.koval@intel.com>
358 * config/i386/i386.c (processor_target_table): Add skylake_cost for
360 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
363 2017-11-26 Julia Koval <julia.koval@intel.com>
365 * config/i386/driver-i386.c (host_detect_local_cpu):
366 Detect skylake-avx512.
368 2017-11-26 Julia Koval <julia.koval@intel.com>
370 * config.gcc: Add -march=cannonlake.
371 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
372 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
373 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
374 (PTA_CANNONLAKE): New.
375 (processor_target_table): Add cannonlake.
376 (ix86_option_override_internal): Ditto.
377 (fold_builtin_cpu): Ditto.
378 (get_builtin_code_for_version): Handle cannonlake.
379 (M_INTEL_COREI7_CANNONLAKE): New.
380 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
381 * doc/invoke.texi: Add -march=cannonlake.
383 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
385 * plugin.c (add_new_plugin): Use platform-specific library extensions.
386 (try_init_one_plugin): Alternative implementation for MinGW.
387 * Makefile.in (plugin_implib): New.
388 (gengtype-lex.c): Fix broken AIX workaround.
389 * configure: Regenerate.
390 * doc/plugins.texi: Document support for MinGW.
392 2017-11-25 Jakub Jelinek <jakub@redhat.com>
394 PR rtl-optimization/81553
395 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
396 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
397 is a shift where C1 has different mode than the whole shift, use C1's
398 mode for MULT rather than the shift's mode.
401 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
402 builtins not enabled in the currently selected ISA.
404 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
406 PR tree-optimization/71026
407 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
408 (insert_reciprocals): Change to insert reciprocals before a division
409 by a square and to insert the square of a reciprocal.
410 (execute_cse_reciprocals_1): Change to consider division by a square.
411 (register_division_in): Add importance parameter.
413 2017-11-24 Richard Biener <rguenther@suse.de>
415 PR tree-optimization/82402
416 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
417 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
419 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
421 * match.pd (0-ptr): New transformation.
423 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
426 * ipa-inline.c (inline_small_functions): Set current badnes correctly
427 when skipping checking.
429 2017-11-24 Richard Biener <rguenther@suse.de>
431 PR tree-optimization/83128
432 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
433 (vn_reference_lookup_3): Likewise.
435 2017-11-24 Jakub Jelinek <jakub@redhat.com>
438 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
439 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
442 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
443 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
444 check for the latter with gimple_call_builtin_p. Do not handle
445 BUILT_IN_STPNCPY_CHK which is not a pass through call.
447 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
449 * config/arm/arm_neon.h: Fix pragma GCC push_options before
452 2017-11-23 Julia Koval <julia.koval@intel.com>
454 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
455 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
456 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
457 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
458 _mm512_maskz_expandloadu_epi16): New intrinsics.
459 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
460 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
461 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
462 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
463 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
464 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
465 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
466 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
467 _mm256_maskz_expandloadu_epi8): New intrinsics.
468 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
469 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
470 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
471 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
472 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
473 * config/i386/sse.md (VI248_VLBW): New iterator.
474 (expand<mode>_mask, expand<mode>_maskz): New patterns.
476 2017-11-23 Julia Koval <julia.koval@intel.com>
478 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
479 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
480 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
481 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
482 _mm512_mask_compressstoreu_epi16): New.
483 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
484 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
485 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
486 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
487 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
488 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
489 _mm256_mask_compressstoreu_epi8): New.
490 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
491 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
492 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
493 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
494 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
495 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
496 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
497 __builtin_ia32_compresshi128_mask,
498 __builtin_ia32_compressstoreuqi512_mask,
499 __builtin_ia32_compressstoreuhi512_mask,
500 __builtin_ia32_compressstoreuqi256_mask,
501 __builtin_ia32_compressstoreuqi128_mask,
502 __builtin_ia32_compressstoreuhi256_mask,
503 __builtin_ia32_compressstoreuhi128_mask): New builtins.
504 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
506 (ix86_expand_special_args_builtin): Handle new types.
507 (s4fma_expand): Handle new builtin array.
508 * config/i386/immintrin.h: Include new headers.
509 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
510 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
512 2017-11-23 Jakub Jelinek <jakub@redhat.com>
514 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
515 fixes. Declare temp and g variables at the top in order to avoid
516 {} in most of the cases.
518 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
520 * match.pd (ptr-0): New transformation.
522 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
524 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
525 (struct addr_mode_cost_table): New.
526 (struct tune_params): Add field addr_mode_costs.
527 * config/arm/arm.c (generic_addr_mode_costs): New.
528 (arm_slowmul_tune): Initialise addr_mode_costs field.
529 (arm_fastmul_tune): Likewise.
530 (arm_strongarm_tune): Likewise.
531 (arm_xscale_tune): Likewise.
532 (arm_9e_tune): Likewise.
533 (arm_marvell_pj4_tune): Likewise.
534 (arm_v6t2_tune): Likewise.
535 (arm_cortex_tune): Likewise.
536 (arm_cortex_a8_tune): Likewise.
537 (arm_cortex_a7_tune): Likewise.
538 (arm_cortex_a15_tune): Likewise.
539 (arm_cortex_a35_tune): Likewise.
540 (arm_cortex_a53_tune): Likewise.
541 (arm_cortex_a57_tune): Likewise.
542 (arm_exynosm1_tune): Likewise.
543 (arm_xgene1_tune): Likewise.
544 (arm_cortex_a5_tune): Likewise.
545 (arm_cortex_a9_tune): Likewise.
546 (arm_cortex_a12_tune): Likewise.
547 (arm_cortex_a73_tune): Likewise.
548 (arm_v7m_tune): Likewise.
549 (arm_cortex_m7_tune): Likewise.
550 (arm_v6m_tune): Likewise.
551 (arm_fa726te_tune): Likewise.
552 (arm_mem_costs): Use table lookup to calculate cost of addressing
555 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
557 * config/arm/arm.c (arm_mem_costs): New function.
558 (arm_rtx_costs_internal): Use arm_mem_costs.
560 2017-11-23 Mark Wielaard <mark@klomp.org>
562 * dwarf2out.c (init_sections_and_labels): Use generation to create
563 unique ranges_section_label and ranges_base_label. Return generation.
564 (output_rnglists): Add generation argument. Use generation to create
565 unique ranges labels.
566 (dwarf2out_finish): Get generation from init_sections_and_labels
567 and pass generation to output_rnglists.
569 2017-11-23 Mike Stump <mikestump@comcast.net>
570 Eric Botcazou <ebotcazou@adacore.com>
572 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
573 * cfgloop.h (struct loop): Add unroll field.
574 * function.h (struct function): Add has_unroll bitfield.
575 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
576 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
577 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
578 (pass_rtl_unroll_loops::gate): Likewise.
579 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
580 for which loop->unroll==1.
581 (decide_unroll_constant_iterations): Use note for consistency and
582 take loop->unroll into account. Return early if loop->unroll is set.
583 Fix thinko in existing test.
584 (decide_unroll_runtime_iterations): Use note for consistency and
585 take loop->unroll into account.
586 (decide_unroll_stupid): Likewise.
587 * lto-streamer-in.c (input_cfg): Read loop->unroll.
588 * lto-streamer-out.c (output_cfg): Write loop->unroll.
589 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
591 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
592 (print_loop): Print loop->unroll if set.
593 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
594 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
595 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
597 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
598 loop->unroll is set and smaller than the trip count. Otherwise bypass
599 entirely the heuristics if loop->unroll is set. Remove dead note.
600 Fix off-by-one bug in other note.
601 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
602 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
604 (tree_unroll_loops_completely): Make static.
605 (pass_complete_unroll::execute): Use correct type for variable.
606 (pass_complete_unrolli::execute): Fix formatting.
607 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
609 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
611 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
612 enable when TARGET_PREFER_AVX128 is set.
614 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
616 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
618 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
619 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
621 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
623 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
624 frequencies. Fix estimation of aggregate parameters.
626 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
628 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
631 2017-11-23 Tom de Vries <tom@codesourcery.com>
633 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
635 2017-11-23 Tom de Vries <tom@codesourcery.com>
637 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
640 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
643 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
644 sibcall_value_pcrel_fdpic): Use local variable instead of
646 (calli_tbr_rel): Add missing operand 2.
647 (call_valuei_tbr_rel): Add missing operand 3.
649 2017-11-23 Jakub Jelinek <jakub@redhat.com>
652 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
653 bitpos/bitsize covering the whole destination, use store_expr only if
654 the complex mode is the same. Otherwise, use expand_normal and if
655 it returns CONCAT, subreg each part separately instead of trying to
656 subreg the whole result.
658 2017-11-23 Richard Biener <rguenther@suse.de>
660 PR tree-optimization/23094
661 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
662 come here from walking over backedges in the first iteration.
663 (vn_reference_lookup_3): Skip clobbers that store the same value.
665 2017-11-23 Richard Biener <rguenther@suse.de>
667 PR tree-optimization/81403
668 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
669 a block we need a leader relative to.
670 (phi_translate_1): For nary processing require a leader from
671 get_representative_for given we run expression simplification
672 using match-and-simplify. Remove previous fix.
674 2017-11-22 Jeff Law <law@redhat.com>
676 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
677 Use new method allocate_value_range rather than accessing the
678 vrp_value_range_pool data member directly.
679 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
680 to use extract_range_from_stmt method to avoid need for
681 extract_range_from_assignment method.
682 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
683 method rather than setting values_propgated data member directly.
684 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
685 and values propagated data members and extract_range_from_assignment
686 method. Reorder private data members to conform to standards.
687 Add new methods set_lattice_propagation_complete and
688 allocate_value_range.
690 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
692 PR rtl-optimization/83030
693 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
694 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
695 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
696 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
697 CROSSING_JUMP_P flag is already set before setting it.
698 * cfgrtl.c (fixup_partition_crossing): Likewise.
699 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
702 2017-11-22 Jakub Jelinek <jakub@redhat.com>
704 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
705 Handle the case where both arguments are using gen_const_vec_series.
707 2017-11-22 David Malcolm <dmalcolm@redhat.com>
710 * pretty-print.c (pp_format): Move quoting implementation to
711 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
712 to pass address of "quote" local.
713 (pp_begin_quote): New function.
714 (pp_end_quote): New function.
715 * pretty-print.h (printer_fn): Convert penultimate param from bool
717 (pp_begin_quote): New decl.
718 (pp_end_quote): New decl.
719 * tree-diagnostic.c (default_tree_printer): Convert penultimate
720 param from bool to bool *.
721 * tree-diagnostic.h (default_tree_printer): Likewise.
723 2017-11-22 Jeff Law <law@redhat.com>
725 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
726 Initialize vr_values.
727 (evrp_range_analyzer::try_find_new_range): Call methods attached to
728 vr_values via vr_values class instance rather than delegators.
729 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
730 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
731 (evrp_range_analyzer::push_value_range): Likewise.
732 (evrp_range_analyzer::pop_value_range): Likewise.
733 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
734 most delegators. Those remaining are exposed as public interfaces.
735 Make vr_values a pointer and private.
736 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
738 (evrp_range_analyzer::get_vr_value): New method.
739 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
740 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
741 (class evrp_dom_walker): Attach evrp_folder class, initialize
742 it in the ctor. Remove temporary delegators.
743 (evrp_dom_walker::before_dom_children): Call methods in attached
744 evrp_range_analyzer class via class instance pointer. Use
745 free value_range_constant_singleton to remove need for
746 op_with_constant_singleton_value delegator method. Do not
747 create a vrp_prop class instance for every call! Narrow
748 scope of a couple variables.
749 (evrp_dom_walker::cleanup): Call methods in attached
750 evrp_range_analyzer class via class instance pointer.
751 * vr-values.h (class vr_values): Privatize many methods and
754 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
756 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
757 * cfgexpand.c (expand_debug_expr): Remove handling for them.
758 * expr.c (expand_expr_real_2): Likewise.
759 * fold-const.c (const_unop): Likewise.
760 * optabs-tree.c (optab_for_tree_code): Likewise.
761 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
762 * tree-inline.c (estimate_operator_cost): Likewise.
763 * tree-pretty-print.c (dump_generic_node): Likewise.
764 (op_code_prio): Likewise.
765 (op_symbol_code): Likewise.
766 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
767 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
768 * internal-fn.c (direct_internal_fn_optab): New function.
769 (direct_internal_fn_array, direct_internal_fn_supported_p
770 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
771 * fold-const-call.c (fold_const_reduction): New function.
772 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
774 * tree-vect-loop.c: Include internal-fn.h.
775 (reduction_code_for_scalar_code): Rename to...
776 (reduction_fn_for_scalar_code): ...this and return an internal
778 (vect_model_reduction_cost): Take an internal_fn rather than
780 (vect_create_epilog_for_reduction): Likewise. Build calls rather
782 (vectorizable_reduction): Use internal functions rather than tree
783 codes for the reduction operation. Update calls to the functions
785 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
786 Use calls to internal functions rather than REDUC tree codes.
787 * config/aarch64/aarch64-simd.md: Update comment accordingly.
789 2017-11-22 Olivier Hainque <hainque@adacore.com>
791 * config/vxworks.c (vxworks_override_options): Pick default
792 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
793 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
794 DWARF_GNAT_ENCODINGS_DEFAULT.
795 * config/vxworksae.h: Likewise.
797 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
799 PR tree-optimization/83104
800 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
803 2017-11-22 Marek Polacek <polacek@redhat.com>
804 H.J. Lu <hongjiu.lu@intel.com>
805 Jason Merrill <jason@redhat.com>
810 * calls.c (initialize_argument_information): Call
811 warn_parameter_passing_abi target hook.
812 (store_one_arg): Use 0 for empty record size. Don't push 0 size
814 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
815 * common.opt: Update -fabi-version description.
816 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
817 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
819 (ix86_is_empty_record): New function.
820 (ix86_warn_parameter_passing_abi): New function.
821 (TARGET_EMPTY_RECORD_P): Redefine.
822 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
823 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
824 * doc/tm.texi: Regenerated.
825 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
826 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
827 * dwarf2out.c (get_ultimate_context): Move to tree.c.
828 * explow.c (hard_function_value): Call arg_int_size_in_bytes
829 instead of int_size_in_bytes.
830 * expr.c (copy_blkmode_to_reg): Likewise.
831 * function.c (aggregate_value_p): Return 0 for empty types.
832 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi target hook.
833 (locate_and_pad_parm): Call arg size_in_bytes instead
835 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
836 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
837 * target.def (empty_record_p, warn_parameter_passing_abi): New target
839 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
840 (std_gimplify_va_arg_expr): Skip empty records. Call
841 arg_size_in_bytes instead size_in_bytes.
842 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
843 * tree-core.h (tree_type_common): Add empty_flag.
844 (tree_decl_common): Update comments.
845 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
847 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
848 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
850 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
851 * tree.c (default_is_empty_type): New function.
852 (default_is_empty_record): New function.
853 (arg_int_size_in_bytes): New function.
854 (arg_size_in_bytes): New function.
855 (get_ultimate_context): New function.
856 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
857 TRANSLATION_UNIT_WARN_EMPTY_P.
858 (default_is_empty_record, arg_int_size_in_bytes,
859 arg_size_in_bytes, get_ultimate_context): Declare.
861 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
863 * config/arm/arm.c (cmse_clear_registers): New function.
864 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
865 code by call to cmse_clear_registers.
866 (cmse_nonsecure_entry_clear_before_return): Likewise.
868 2017-11-22 Tamar Christina <tamar.christina@arm.com>
870 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
871 (vdot_s32, vdotq_s32): New.
872 (vdot_lane_u32, vdotq_lane_u32): New.
873 (vdot_lane_s32, vdotq_lane_s32): New.
876 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
879 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
880 for unsigned values with fewer HWIs than the precision.
881 (test_overflow): New function.
882 (wide_int_cc_tests): Call it.
884 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
886 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
887 has the same mode class as Pmode.
889 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
891 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
892 padding_bits_to_clear_ptr.
893 (cmse_nonsecure_entry_clear_before_return): Likewise.
895 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
897 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
898 auto_sbitap instead of integer bitfield to control register needing
901 2017-11-22 Jakub Jelinek <jakub@redhat.com>
903 PR tree-optimization/83044
904 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
905 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
906 that need the upper bound. Subtract offset from
907 get_addr_base_and_unit_offset only if positive and subtract it
908 before division by eltsize rather than after it.
911 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
912 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
913 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
917 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
918 expand_binop with *mul_widen_optab, make sure at least one of the
919 operands doesn't have VOIDmode.
922 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
924 PR rtl-optimization/82044
925 PR tree-optimization/82042
926 * dse.c (record_store): Check for overflow.
927 (check_mem_read_rtx): Properly check for overflow if width == -1, call
928 add_wild_read instead of clear_rhs_from_active_local_stores on
929 overflow and log it into dump_file.
931 2017-11-22 Richard Biener <rguenther@suse.de>
933 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
934 fake edges to exit when looking for a place to insert.
935 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
937 (insert_into_preds_of_block): Commit edge insertion immediately,
938 assert that doesn't require new BBs.
939 (fini_pre): Release expressions.
940 (pass_pre::execute): Shuffle things around a bit, if the fn
941 is too large do not compute AVAIL either as this is really the
944 2017-11-22 Richard Biener <rguenther@suse.de>
946 PR tree-optimization/83089
947 * tree-if-conv.c (pass_if_conversion::execute): If anything
948 changed reset SCEV and free the number of iteration estimates.
950 2017-11-21 Martin Sebor <msebor@redhat.com>
952 PR tree-optimization/82945
953 * calls.h (warn_nonstring_bound): Remove unused function.
955 2017-11-21 Martin Sebor <msebor@redhat.com>
957 PR tree-optimization/82945
958 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
959 * calls.h (maybe_warn_nonstring_arg): Declare new function.
960 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
962 (initialize_argument_information): Call maybe_warn_nonstring_arg.
963 * calls.h (get_attr_nonstring_decl): Declare new function.
964 * doc/extend.texi (attribute nonstring): Update.
965 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
966 get_attr_nonstring_decl and handle it.
967 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
968 detection of nul-termination.
969 (strlen_to_stridx): Change to a pointer.
970 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
971 (pass_strlen::execute): Same.
973 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
975 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
976 for the new option -mprefer-vector-width=[none|128|256|512].
977 * config/i386/i386.c (ix86_target_string): remove old style options
978 -mprefer-avx256 and make -mprefer-avx128 as alias.
979 (ix86_option_override_internal): Apply defaults for the
980 -mprefer-vector-width=[128|256] option.
981 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
982 Implement macros to work with -mprefer-vector-width=.
983 * config/i386/i386.opt: Implemented option
984 -mprefer-vector-width=[none|128|256|512].
985 * doc/invoke.texi: Documentation for
986 -mprefer-vector-width=[none|128|256|512].
988 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
990 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
992 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
994 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
996 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
998 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
1000 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
1001 * expr.c (expand_expr_real_2): Likewise.
1002 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
1003 fold_binary_loc): Likewise.
1004 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
1005 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
1006 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
1007 MINUS_EXPR transformations.
1008 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
1009 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
1010 * tree-inline.c (estimate_operator_cost): Likewise.
1011 * tree-pretty-print.c (dump_generic_node, op_code_prio,
1012 op_symbol_code): Likewise.
1013 * tree-vect-stmts.c (vectorizable_operation): Likewise.
1014 * vr-values.c (extract_range_from_binary_expr): Likewise.
1015 * varasm.c (initializer_constant_valid_p_1): Likewise.
1016 * tree.def: New tree code POINTER_DIFF_EXPR.
1018 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
1020 * config/i386/i386.md (*bswap<mode>2_movbe): Add
1021 integer suffix to movbe mnemonic.
1022 (*bswaphi2_movbe): Ditto.
1023 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
1025 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1028 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
1029 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
1030 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
1032 2017-11-21 Martin Liska <mliska@suse.cz>
1034 * tree-inline.c (expand_call_inline): Remove not needed
1037 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
1038 Jakub Jelinek <jakub@redhat.com>
1041 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
1042 Remove static keyword from f variable.
1044 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1046 PR tree-optimization/83086
1047 * gimple-ssa-store-merging.c
1048 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
1049 rather than n.base_addr.
1051 2017-11-21 Martin Liska <mliska@suse.cz>
1053 PR rtl-optimization/82044
1054 PR tree-optimization/82042
1055 * dse.c (check_mem_read_rtx): Check for overflow.
1057 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
1059 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
1062 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
1064 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
1065 and its corresponding call arg location note.
1067 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
1068 Andrew Burgess <andrew.burgess@embecosm.com>
1070 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
1072 (arc_return_slot_offset): Likewise.
1073 (arc_eh_return_address_location): New declaration.
1074 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
1075 (MUST_SAVE_REGISTER): Add exception handler case.
1076 (MUST_SAVE_RETURN_ADDR): Likewise.
1077 (arc_frame_pointer_required): Likewise.
1078 (arc_frame_pointer_needed): New function.
1079 (arc_compute_frame_size): Changed.
1080 (arc_expand_prologue): Likewise.
1081 (arc_expand_epilogue): Likewise.
1082 (arc_initial_elimination_offset): Likewise.
1083 (arc_return_slot_offset): Delete.
1084 (arc_eh_return_address_location): New function.
1085 (arc_builtin_setjmp_frame_value): Likewise.
1086 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
1087 (EH_RETURN_STACKADJ_RTX): Define.
1088 (EH_RETURN_HANDLER_RTX): Likewise.
1089 * config/arc/arc.md (eh_return): Delete.
1091 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
1093 * print-tree.h (debug_vec_tree): Remove prototype.
1094 * gdbinit.in (pvt): Remove macro.
1096 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1098 PR tree-optimization/83047
1099 * gimple-ssa-store-merging.c
1100 (imm_store_chain_info::output_merged_store): If the loads with the
1101 same vuse are in different basic blocks, for load_gsi pick a load
1102 location that is dominated by the other loads.
1105 * config/i386/i386.c (ix86_memmodel_check): Start
1106 -Winvalid-memory-model diagnostics with lowercase letter.
1109 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
1110 set in_cold_section_p to the partition containing loc_list->first.
1111 When seeing loc_list->last_before_switch node, update secname and
1112 perform range_across_switch second partition handling only after that.
1115 * run-rtl-passes.c: Include debug.h.
1116 (run_rtl_passes): Call debug_hooks->assembly_start.
1117 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
1121 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
1122 OPTAB_DIRECT in calls to expand_simple_binop.
1124 2017-11-20 David Malcolm <dmalcolm@redhat.com>
1127 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
1129 2017-11-20 David Malcolm <dmalcolm@redhat.com>
1132 * spellcheck.h (best_match::blithely_get_best_candidate): New
1135 2017-11-20 Jakub Jelinek <jakub@redhat.com>
1137 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
1138 with lower case letter.
1140 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
1142 * config/i386/i386.md (bswaphi2): New expander.
1143 (*bswaphi2_movbe): New insn pattern.
1144 (bswaphi -> rorhi pepehole2): New peephole pattern.
1146 2017-11-20 Jeff Law <law@redhat.com>
1148 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
1149 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
1150 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
1151 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
1152 gimple-ssa-evrp-analyze.h.
1154 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
1155 set BB_VISITED here.
1156 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
1158 2017-11-20 Bin Cheng <bin.cheng@arm.com>
1160 * tree-predcom.c: Add general comment on Store-Store chains.
1161 (split_data_refs_to_components): Postpone clearing eliminate_store_p
1163 (get_chain_last_ref_at): Rename into...
1164 (get_chain_last_write_at): ...this.
1165 (get_chain_last_write_before_load): New function.
1166 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
1167 CT_STORE_STORE when write reference is added.
1168 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
1169 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
1170 (initialize_root_vars_store_elim_1): Ditto.
1171 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
1172 definition is created.
1173 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
1174 chain by replacing it with dominant stored value.
1176 2017-11-20 Bin Cheng <bin.cheng@arm.com>
1178 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
1180 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
1182 * vr-values.c (extract_range_from_binary_expr): Use a full range
1185 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
1187 * config/arm/arm.md (R4_REGNUM): Define constant.
1188 (nonsecure_call_internal): Remove r4 clobber.
1189 (nonsecure_call_value_internal): Likewise.
1190 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
1191 clobber and resequence match_operands.
1192 (nonsecure_call_value_reg_thumb1_v5): Likewise.
1193 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
1194 (nonsecure_call_value_reg_thumb2): Likewise.
1196 2017-11-20 Jakub Jelinek <jakub@redhat.com>
1198 PR tree-optimization/78821
1199 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
1200 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
1201 to the address of the base rather than the base itself.
1202 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
1203 (find_bswap_or_nop_finalize): New function.
1204 (find_bswap_or_nop): Use it.
1205 (bswap_replace): Return a tree rather than bool, change first
1206 argument from gimple * to gimple_stmt_iterator, allow inserting
1207 into an empty sequence, allow ins_stmt to be NULL - then emit
1208 all stmts into gsi. Fix up MEM_REF address gimplification.
1209 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
1210 (struct store_immediate_info): Add N and INS_STMT non-static
1212 (store_immediate_info::store_immediate_info): Initialize them
1213 from newly added ctor args.
1214 (merged_store_group::apply_stores): Formatting fixes. Sort by
1216 (stmts_may_clobber_ref_p): For stores call also
1217 refs_anti_dependent_p.
1218 (gather_bswap_load_refs): New function.
1219 (imm_store_chain_info::try_coalesce_bswap): New method.
1220 (imm_store_chain_info::coalesce_immediate_stores): Use it.
1221 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
1222 (imm_store_chain_info::output_merged_store): Fail if number of
1223 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
1224 and NOP_EXPR rhs_code.
1225 (pass_store_merging::process_store): Compute n and ins_stmt, if
1226 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
1227 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
1229 (pass_store_merging::execute): Calculate dominators.
1231 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
1232 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
1233 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
1234 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
1235 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
1236 class pass_optimize_bswap, bswap_replace,
1237 pass_optimize_bswap::execute): Moved to ...
1238 * gimple-ssa-store-merging.c: ... this file.
1239 Include optabs-tree.h.
1240 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
1241 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
1242 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
1243 anonymous namespace, remove static keywords.
1244 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
1245 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
1247 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
1250 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
1252 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
1254 * vec.h (debug_helper): New function.
1255 (DEFINE_DEBUG_VEC): New macro.
1256 * hash-set.h (debug_helper): New function.
1257 (DEFINE_DEBUG_HASH_SET): New macro.
1258 * cfg.c (debug_slim (edge)): New function.
1259 Call DEFINE_DEBUG_VEC for edges.
1260 Call DEFINE_DEBUG_HASH_SET for edges.
1261 * cfghooks.c (debug_slim (basic_block)): New function.
1262 Call DEFINE_DEBUG_VEC for basic blocks.
1263 Call DEFINE_DEBUG_HASH_SET for basic blocks.
1264 * print-tree.c (debug_slim): New function to handle trees.
1265 Call DEFINE_DEBUG_VEC for trees.
1266 Call DEFINE_DEBUG_HASH_SET for trees.
1267 (debug (vec<tree, va_gc>) &): Remove.
1268 (debug (<vec<tree, va_gc>) *): Remove.
1269 * print-rtl.c (debug_slim): New function to handle const_rtx.
1270 Call DEFINE_DEBUG_VEC for rtx_def.
1271 Call DEFINE_DEBUG_VEC for rtx_insn.
1272 Call DEFINE_DEBUG_HASH_SET for rtx_def.
1273 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
1274 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
1275 (debug (vec<rtx_insn *> *ptr): Remove.
1276 (debug_insn_vector): Remove.
1277 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
1279 2017-11-20 Tom de Vries <tom@codesourcery.com>
1281 PR rtl-optimization/82020
1282 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
1283 IF_THEN_ELSE condition.
1285 2017-11-19 Jeff Law <law@redhat.com>
1287 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1288 of degenerates resulting from ignoring an edge.
1290 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1293 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
1295 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1298 * profile-count.c (profile_count::to_sreal_scale): Fix return value
1299 for uninitialied counts.
1301 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1304 * tree-inline.c (estimate_num_insns): Set to 1 at least.
1306 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1309 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
1311 2017-11-19 Tom de Vries <tom@codesourcery.com>
1313 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
1316 2017-11-19 Tom de Vries <tom@codesourcery.com>
1318 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
1321 2017-11-19 Tom de Vries <tom@codesourcery.com>
1323 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1326 2017-11-19 Tom de Vries <tom@codesourcery.com>
1328 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
1330 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
1332 2017-11-19 Tom de Vries <tom@codesourcery.com>
1334 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
1336 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
1337 semicolon after MCORE_EXPORT_NAME call.
1339 2017-11-19 Tom de Vries <tom@codesourcery.com>
1342 * vmsdbgout.c (vmsdbgout_early_finish): New function.
1343 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
1345 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
1347 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
1348 (cgraph_edge::clone): Cleanup updating of profile.
1349 * ipa-cp.c (update_profiling_info): Likewise.
1350 * ipa-inline-transform.c (inline_transform): Likewise.
1351 * ipa-inline.c (inline_small_functions): Add missing space to dump.
1352 * ipa-split.c (execute_split_functions): Do not split when function
1354 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
1355 * profile-count.c (profile_count::dump): Add global0.
1356 (profile_count::to_cgraph_frequency): Do not ICE when entry is
1358 (profile_count::to_sreal_scale): Likewise.
1359 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
1360 (profile_count::combine_with_ipa_count): New function.
1361 * profile-count.h (profile_guessed_global0adjusted): New.
1362 (profile_count::adjusted_zero): New.
1363 (profile_count::global0adjusted): New.
1364 (profile_count::combine_with_ipa_count): New.
1365 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
1366 correct profile of return block of split functions.
1367 (copy_cfg_body): Remove unused profile_count.
1368 (copy_body): Likewise.
1369 (expand_call_inline): Update.
1370 (tree_function_versioning): Update.
1372 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
1374 * hash-set.h (hash_set::empty): New.
1375 * tree-ssa-threadbackward.h: Delete.
1376 * tree-ssa-threadbackward.c (class thread_jumps): New.
1377 Move max_threaded_paths into class.
1378 (fsm_find_thread_path): Remove arguments that are now in class.
1379 (profitable_jump_thread_path): Rename to...
1380 (thread_jumps::profitable_jump_thread_path): ...this.
1381 (convert_and_register_jump_thread_path): Rename to...
1382 (thread_jumps::convert_and_register_current_path): ...this.
1383 (check_subpath_and_update_thread_path): Rename to...
1384 (thread_jumps::check_subpath_and_update_thread_path): ...this.
1385 (register_jump_thread_path_if_profitable): Rename to...
1386 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
1387 (handle_phi): Rename to...
1388 (thread_jumps::handle_phi): ...this.
1389 (handle_assignment): Rename to...
1390 (thread_jumps::handle_assignment): ...this.
1391 (fsm_find_control_statement_thread_paths): Rename to...
1392 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
1393 (find_jump_threads_backwards): Rename to...
1394 (thread_jumps::find_jump_threads_backwards): ...this.
1395 Initialize path local data.
1396 (pass_thread_jumps::execute): Call find_jump_threads_backwards
1397 from within thread_jumps class.
1398 (pass_early_thread_jumps::execute): Same.
1400 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1402 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
1404 2017-11-17 Steve Ellcey <sellcey@cavium.com>
1406 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
1407 to canonical location.
1409 2017-11-17 Steve Ellcey <sellcey@cavium.com>
1412 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
1414 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
1416 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
1418 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
1419 rcrt1.o%s/grcrt1.o%s for -static-pie.
1421 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1423 * i386.c (ix86_multiplication_cost, ix86_division_cost,
1424 ix86_shift_rotate_cost): Break out from ...
1425 (ix86_rtx_costs): ... here.
1426 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
1429 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1431 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
1432 when entry block was promoted unlikely.
1433 (estimate_bb_frequencies): Increase frequency scale.
1434 * profile-count.h (profile_count): Export precision info.
1436 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1438 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
1439 disturbin profile of entry block.
1441 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1443 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
1446 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1448 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
1449 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
1451 2017-11-17 Jeff Law <law@redhat.com>
1453 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
1454 from evrp_dom_walker class. Various methods moved into new class.
1455 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
1456 (evrp_range_analyzer::enter): New method.
1457 (evrp_range_analyzer::leave): New method.
1458 (evrp_dom_walker): Remove delegators no longer needed by this class.
1459 Replace vr_values data member with evrp_range_analyzer
1461 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
1462 method extracted from evrp_dom_walker::before_dom_children.
1463 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
1464 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
1466 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
1467 Add private copy constructor and move assignment operators.
1468 Privatize methods and class data where trivially possible.
1469 (evrp_dom_walker::cleanup): New function, extracted from
1470 execute_early_vrp. Simplify access to class data.
1472 * vr-values.h (get_output_for_vrp): Prototype.
1473 * vr-values.c (get_output_for_vrp): New function extracted from
1474 vrp_visit_assignment_or_call and extract_range_from_stmt.
1475 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
1477 2017-11-17 Luis Machado <luis.machado@linaro.org>
1479 * config/aarch64/aarch64.c
1480 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
1481 (qdf24xx_tunings) <autoprefetcher_model>: Set to
1482 tune_params::AUTOPREFETCHER_WEAK.
1484 2017-11-17 Tamar Christina <tamar.christina@arm.com>
1487 * config/arm/arm.c (arm_valid_target_attribute_rec):
1488 Parse "arch=" and "+<ext>".
1489 (arm_valid_target_attribute_tree): Re-init global options.
1490 (arm_option_override): Make non-static.
1491 (arm_options_perform_arch_sanity_checks): Make errors fatal.
1492 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
1493 (__ARM_FEATURE_CRC32): Support undef.
1494 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
1495 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
1497 2017-11-17 David Malcolm <dmalcolm@redhat.com>
1499 * gdbinit.in (break-on-diagnostic): New command.
1501 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1503 * config/i386/i386.c (ix86_expand_epilogue): Change simple
1504 return to indirect jump for EH return if control-flow
1505 protection is enabled. Change explicit 'false' argument in
1506 pro_epilogue_adjust_stack with a value of
1508 * config/i386/i386.md (simple_return_indirect_internal):
1509 Remove SImode restriction to support 64-bit.
1511 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
1513 * combine.c (added_notes_insn): New.
1514 (try_combine): Handle added_notes_insn like added_links_insn.
1515 Rewrite return value code.
1516 (distribute_notes): Set added_notes_insn to the earliest insn we added
1519 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
1521 PR rtl-optimization/82621
1522 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
1523 dest of one of those SETs is unused.
1525 2017-11-17 Richard Biener <rguenther@suse.de>
1528 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
1529 * tree-pretty-print.c (dump_generic_node): Handle
1530 annot_expr_parallel_kind.
1531 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
1532 * gimplify.c (gimple_boolify): Likewise.
1534 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1536 * config.gcc (extra_headers): Add cet.h for x86 targets.
1537 * config/i386/cet.h: New file.
1538 * doc/install.texi: Add --enable-cet/--disable-cet.
1540 2017-11-17 Richard Biener <rguenther@suse.de>
1542 PR tree-optimization/83017
1543 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
1544 (gen_parallel_loop): Properly count iterations.
1545 (parallelize_loops): Handle loop->can_be_parallel independent
1546 of flag_loop_parallelize_all. Make static profitability test match
1548 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
1549 * invoke.texi (parloops-min-per-thread): Document.
1551 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
1553 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
1554 upstreaming review comments.
1556 2017-11-17 Tamar Christina <tamar.christina@arm.com>
1558 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
1559 with fast unaligned access.
1560 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
1562 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
1564 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
1565 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
1567 2017-11-17 Richard Biener <rguenther@suse.de>
1569 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
1570 folding of references.
1572 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
1575 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
1576 of replacing call to strncmp with corresponding call to strcmp when
1579 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
1581 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
1582 option prefer-avx256 for skylake-avx512 configuration.
1583 * config/i386/i386.c (ix86_option_override_internal): Ditto.
1584 (get_builtin_code_for_version): Ditto.
1586 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
1587 Monk Chiang <sh.chiang04@gmail.com>
1589 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
1590 (FIXED_REGISTERS): Reserve more register numbers.
1591 (CALL_USED_REGISTERS): Likewise.
1592 (REG_ALLOC_ORDER): Likewise.
1593 (REG_CLASS_CONTENTS): Likewise.
1594 (REGISTER_NAMES): Likewise.
1596 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
1597 Kito Cheng <kito.cheng@gmail.com>
1599 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
1601 * config/nds32/iterators.md: Add vector mode iterators and attributes.
1603 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
1605 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
1606 parameter list for vec_splats.
1608 2017-11-16 Joseph Myers <joseph@codesourcery.com>
1610 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
1612 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
1614 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
1617 * doc/install.texi: Document incompatibility of Perl >=5.6.26
1618 with the required version of automake 1.11.6.
1620 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
1622 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
1624 (DU_C2_3_power9): Correct reservation combinations.
1625 (FP_DIV_power9, VEC_DIV_power9): New.
1626 (power9-alu): Split out rotate/shift...
1627 (power9-rot): ...to here, correct dispatch resource.
1628 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
1630 (power9-fp): Correct latency.
1631 (power9-sdiv): Add div/sqrt resource.
1632 (power9-ddiv): Correct latency, add div/sqrt resource.
1633 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
1634 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
1636 (power9-qpdiv, power9-qpmul): Adjust resource usage.
1638 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1640 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
1641 switch statement mapping KF built-ins to TF built-ins if we don't
1642 have the proper ISA 3.0 assembler support.
1644 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1646 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
1647 (lower_emutls_function_body): Do not compute it.
1649 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1651 * ipa-split.c (split_bb_info): Turn time to sreal.
1652 (split_point): Likewise.
1653 (dump_split_point): Likewise.
1654 (fine_split_points): Likewise.
1655 (execute_split_functions): Only zero split_bbs; turn time to sreals.
1657 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1659 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
1662 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1664 * predict.c (combine_predictions_for_bb): Preserve zero predicted
1666 (expensive_function_p): Remove useless assert.
1667 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
1669 2017-11-16 Martin Sebor <msebor@redhat.com>
1671 PR tree-optimization/82588
1672 PR tree-optimization/82583
1673 * tree-vrp.c (check_array_ref): Handle flexible array members,
1674 string literals, and inner indices.
1675 (search_for_addr_array): Add detail to diagnostics.
1677 2017-11-16 Nathan Sidwell <nathan@acm.org>
1681 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
1682 (SET_DECL_ASSEMBLER_NAME): Forward to
1683 overwrite_decl_assembler_name.
1684 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
1685 (overwrite_decl_assembler_name): Declare.
1686 * tree.c (overwrite_decl_assembler_name): New.
1687 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
1688 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
1689 (LANG_HOOKS_INITIALIZER): Add it.
1690 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
1691 * langhooks.c (lhd_set_decl_assembler_name): Use
1692 SET_DECL_ASSEMBLER_NAME.
1693 (lhd_overwrite_decl_assembler_name): Default implementation.
1695 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
1696 Jackson Woodruff <jackson.woodruff@arm.com>
1698 PR tree-optimization/71026
1699 * match.pd: Canonicalize constant multiplies in division.
1701 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
1703 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
1704 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
1705 Remove OPT_fomit_frame_pointer entry.
1706 * common/config/arc/arc-common.c: Likewise.
1707 * common/config/arm/arm-common.c: Likewise.
1708 * common/config/avr/avr-common.c: Likewise.
1709 * common/config/c6x/c6x-common.c: Likewise.
1710 * common/config/cr16/cr16-common.c: Likewise.
1711 * common/config/cris/cris-common.c: Likewise.
1712 * common/config/epiphany/epiphany-common.c: Likewise.
1713 * common/config/fr30/fr30-common.c: Likewise.
1714 * common/config/frv/frv-common.c: Likewise.
1715 * common/config/ia64/ia64-common.c: Likewise.
1716 * common/config/iq2000/iq2000-common.c: Likewise.
1717 * common/config/lm32/lm32-common.c: Likewise.
1718 * common/config/m32r/m32r-common.c: Likewise.
1719 * common/config/mcore/mcore-common.c: Likewise.
1720 * common/config/microblaze/microblaze-common.c: Likewise.
1721 * common/config/mips/mips-common.c: Likewise.
1722 * common/config/mmix/mmix-common.c: Likewise.
1723 * common/config/mn10300/mn10300-common.c: Likewise.
1724 * common/config/nios2/nios2-common.c: Likewise.
1725 * common/config/pa/pa-common.c: Likewise.
1726 * common/config/pdp11/pdp11-common.c: Likewise.
1727 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
1728 * common/config/riscv/riscv-common.c: Likewise.
1729 * common/config/rs6000/rs6000-common.c: Likewise.
1730 * common/config/rx/rx-common.c: Likewise.
1731 * common/config/s390/s390-common.c: Likewise.
1732 * common/config/sh/sh-common.c: Likewise.
1733 * common/config/sparc/sparc-common.c: Likewise.
1734 * common/config/tilegx/tilegx-common.c: Likewise.
1735 * common/config/tilepro/tilepro-common.c: Likewise.
1736 * common/config/v850/v850-common.c: Likewise.
1737 * common/config/visium/visium-common.c: Likewise.
1738 * common/config/xstormy16/xstormy16-common.c: Likewise.
1739 * common/config/xtensa/xtensa-common.c: Likewise.
1740 * invoke.texi (-fomit-frame-pointer): Update documentation.
1742 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1744 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
1746 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1748 * cfg.c (scale_bbs_frequencies_int,
1749 cale_bbs_frequencies_gcov_type): Remove.
1750 * cfg.h (scale_bbs_frequencies_int,
1751 cale_bbs_frequencies_gcov_type): Remove.
1753 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1755 * tree-ssa-loop-manip.c
1756 (scale_dominated_blocks_in_loop): Update to profile counts.
1757 (tree_transform_and_unroll_loop): Likewise.
1759 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1761 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
1762 scale_bbs_frequencies_int.
1764 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1766 * final.c (compute_alignments): Use counts rather than frequencies.
1768 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1770 * cfgloopanal.c: Include sreal.h
1771 (average_num_loop_insns): Use counts and sreal for accounting.
1773 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1775 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
1778 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
1779 Kito Cheng <kito.cheng@gmail.com>
1781 * config/nds32/constraints.md: Provide more constraints.
1782 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
1783 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
1784 support constraints usage.
1786 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
1788 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
1790 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
1791 Kito Cheng <kito.cheng@gmail.com>
1793 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
1794 * config/nds32/nds32.opt: Refine the layout.
1795 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
1796 TARGET_EXT_STRING): Support new options.
1797 * config/nds32/nds32.h: Likewise.
1798 * config/nds32/nds32.md: Likewise.
1799 * config/nds32/nds32-predicates.c: Likewise.
1800 * config/nds32/constraints.md: Likewise.
1801 * common/config/nds32/nds32-common.c: Likewise.
1803 2017-11-16 Julia Koval <julia.koval@intel.com>
1806 * config/i386/gfniintrin.h: Add sse check.
1807 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
1809 2017-11-16 Julia Koval <julia.koval@intel.com>
1811 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
1812 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
1813 (ix86_handle_option): Handle -mavx512vbmi2.
1814 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
1815 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1816 * config/i386/i386-c.c (__AVX512VBMI2__): New.
1817 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
1818 (ix86_valid_target_attribute_inner_p): Ditto.
1819 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
1820 * config/i386/i386.opt (mavx512vbmi2): New option.
1821 * doc/invoke.texi: Add new option.
1823 2017-11-16 Julia Koval <julia.koval@intel.com>
1825 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
1826 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
1827 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
1828 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
1829 _mm512_gf2p8mul_epi8): New intrinsics.
1830 * config/i386/i386-builtin-types.def
1831 (V64QI_FTYPE_V64QI_V64QI): New type.
1832 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
1833 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
1834 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
1835 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
1836 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
1837 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1839 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
1841 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
1842 explicitly as a stream of bytes.
1844 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1846 * config/rs6000/altivec.h (vec_xst_be): New #define.
1847 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
1848 and externalize from *altivec_vperm_<mode>_internal.
1849 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
1851 (XL_BE_V8HI): Likewise.
1852 (XL_BE_V4SI): Likewise.
1853 (XL_BE_V4SI): Likewise.
1854 (XL_BE_V2DI): Likewise.
1855 (XL_BE_V4SF): Likewise.
1856 (XL_BE_V2DF): Likewise.
1857 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
1858 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
1859 all array entries with these keys: VSX_BUILTIN_VEC_XL,
1860 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
1861 VSX_BUILTIN_VEC_XST_BE.
1862 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
1863 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
1865 (altivec_init_builtins): Replace conditional calls to def_builtin
1866 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
1867 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
1868 with unconditional calls. Remove calls to def_builtin for
1869 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
1870 __builtin_vec_xst_be.
1871 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
1872 to define_expand, and add alternate RTL generation for P8.
1873 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
1874 vsx_ld_elemrev_v8hi.
1875 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
1876 add alternate RTL generation for P8.
1877 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
1878 vsx_ld_elemrev_v16qi.
1879 (vsx_st_elemrev_v8hi): Convert define_insn
1880 to define_expand, and add alternate RTL generation for P8.
1881 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
1882 vsx_st_elemrev_v8hi.
1883 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
1884 add alternate RTL generation for P8.
1885 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
1886 vsx_st_elemrev_v16qi.
1888 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
1891 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
1892 TARGET_AVX512ER check.
1893 (ix86_option_override_internal): Set MASK_VZEROUPPER if
1894 neither -mzeroupper nor -mno-zeroupper is used and
1895 TARGET_EMIT_VZEROUPPER is set.
1896 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
1897 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
1899 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
1901 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1902 folding of vector compares.
1903 (fold_build_vec_cmp): New helper function.
1904 (fold_compare_helper): New helper function.
1905 (builtin_function_type): Add compare builtins to the list of functions
1906 having unsigned arguments. Cosmetic updates to comment indentation.
1907 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
1908 the not+eq combination.
1910 2017-11-15 Bin Cheng <bin.cheng@arm.com>
1912 PR tree-optimization/82726
1913 PR tree-optimization/70754
1914 * tree-predcom.c (order_drefs_by_pos): New function.
1915 (combine_chains): Move code setting has_max_use_after to...
1916 (try_combine_chains): ...here. New parameter. Sort combined chains
1917 according to position information.
1918 (tree_predictive_commoning_loop): Update call to above function.
1919 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
1921 2017-11-15 Bin Cheng <bin.cheng@arm.com>
1923 PR tree-optimization/82726
1925 2017-01-23 Bin Cheng <bin.cheng@arm.com>
1927 PR tree-optimization/70754
1928 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
1929 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
1930 combined stmt before it if not NULL.
1931 (combine_chains): Process refs reversely and compute dominance point
1935 2017-02-23 Bin Cheng <bin.cheng@arm.com>
1937 PR tree-optimization/79663
1938 * tree-predcom.c (combine_chains): Process refs in reverse order
1939 only for ZERO length chains, and add explaining comment.
1941 2017-11-15 Tamar Christina <tamar.christina@arm.com>
1943 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
1944 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
1945 * doc/invoke.texi (ARM Options): Add armv8.3-a.
1947 2017-11-15 Tamar Christina <tamar.christina@arm.com>
1949 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
1951 2017-11-15 Martin Liska <mliska@suse.cz>
1953 * tree-cfg.c (pass_warn_function_return::execute):
1954 Compare warn_return_type for greater than zero.
1956 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
1960 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
1961 to return true on Xeon and not on Xeon Phi.
1962 (ix86_check_avx256_register): Changed to ...
1963 (ix86_check_avx_upper_register): ... this. Add extra check for
1964 VALID_AVX512F_REG_OR_XI_MODE.
1965 (ix86_avx_u128_mode_needed): Changed
1966 ix86_check_avx256_register to ix86_check_avx_upper_register.
1967 (ix86_check_avx256_stores): Changed to ...
1968 (ix86_check_avx_upper_stores): ... this. Changed
1969 ix86_check_avx256_register to ix86_check_avx_upper_register.
1970 (ix86_avx_u128_mode_after): Changed
1971 avx_reg256_found to avx_upper_reg_found. Changed
1972 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
1973 (ix86_avx_u128_mode_entry): Changed
1974 ix86_check_avx256_register to ix86_check_avx_upper_register.
1975 (ix86_avx_u128_mode_exit): Ditto.
1976 * config/i386/i386.h: (host_detect_local_cpu): New define.
1978 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1980 * config/arm/xgene1.md (xgene1): Split into automatons
1981 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
1982 (xgene1_f_load): Adjust reservations and/or types.
1983 (xgene1_f_store): Likewise.
1984 (xgene1_load_pair): Likewise.
1985 (xgene1_store_pair): Likewise.
1986 (xgene1_fp_load1): Likewise.
1987 (xgene1_load1): Likewise.
1988 (xgene1_store1): Likewise.
1989 (xgene1_move): Likewise.
1990 (xgene1_alu): Likewise.
1991 (xgene1_simd): Likewise.
1992 (xgene1_bfm): Likewise.
1993 (xgene1_neon_load1): Likewise.
1994 (xgene1_neon_store1): Likewise.
1995 (xgene1_neon_logic): Likewise.
1996 (xgene1_neon_st1): Likewise.
1997 (xgene1_neon_ld1r): Likewise.
1998 (xgene1_alu_cond): Added.
1999 (xgene1_shift_reg): Likwise.
2000 (xgene1_bfx): Likewise.
2001 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
2003 2017-11-15 Jakub Jelinek <jakub@redhat.com>
2006 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
2008 (can_widen_mult_without_libcall): New function.
2009 (expand_mul_overflow): If only checking unsigned mul overflow,
2010 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
2011 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
2012 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
2013 (expand_DIVMOD): Formatting fix.
2014 * expmed.h (expand_mult): Add NO_LIBCALL argument.
2015 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
2016 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
2018 PR tree-optimization/82977
2019 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
2020 constructed temporary to strlen_to_stridx.put.
2022 2017-11-15 Martin Liska <mliska@suse.cz>
2024 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
2025 * configure: Regenerate.
2027 2017-11-15 Martin Liska <mliska@suse.cz>
2030 * config/sh/sh-mem.cc: Use proper probability for
2033 2017-11-14 Jeff Law <law@redhat.com>
2035 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
2036 the red zone for stack_clash_protection_final_dynamic_probe targets
2037 when the total dynamic stack size is zero bytes.
2039 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
2040 blocks is post order.
2042 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
2044 * dumpfile.h (TDF_COMPARE_DEBUG): New.
2045 * final.c (rest_of_clean_state): Set it for the
2046 -fcompare-debug dump.
2047 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
2048 class when TDF_COMPARE_DEBUG is set.
2050 * dwarf2out.c (gen_producer_string): Discard
2053 2017-11-15 Joseph Myers <joseph@codesourcery.com>
2056 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
2057 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
2058 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
2059 (__TGMATH_REAL_2_3): Remove macros.
2060 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
2061 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
2063 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
2065 (remquo): Define using __TGMATH_REAL_3.
2067 2017-11-14 Jeff Law <law@redhat.com>
2069 * vr-values.c: New file with contents extracted from tree-vrp.c.
2070 * Makefile.in (OBJS): Add vr-values.o
2071 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
2072 (set_value_range, set_and_canonicalize_value_range): Likewise.
2073 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
2074 (value_range_constant_singleton, symbolic_range_p): Likewise.
2075 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
2076 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
2077 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
2078 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
2079 (find_case_label_range, find_case_label_index): Likewise.
2080 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
2081 (range_int_cst_singleton_p, value_inside_range): Likewise.
2082 (get_single_symbol): Likewise.
2083 (switch_update): Move structure definition here.
2084 (to_remove_edges, to_update_switch_stmts): Provide externs.
2085 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
2086 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
2087 (vrp_val_is_min, set_value_range): Likewise.
2088 (set_and_canonicalize_value_range, copy_value_range): Likewise.
2089 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
2090 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
2091 (range_is_nonnull, range_int_cst_p): Likewwise.
2092 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
2093 (get_single_symbol, operand_less_p): Likewise
2094 (compare_values_warnv, compare_values): Likewise.
2095 (value_inside_range, value_range_constant_singleton): Likewise.
2096 (zero_nonzero_bitgs_from_vr): Likewise.
2097 (extract_range_from_binary_expr_1): Likewise.
2098 (overflow_comparison_p): Likewise.
2099 (to_remove_edges, to_update_switch_stmts): Likewise.
2100 (find_case_label-index, find_case_label_range): Likewise.
2101 (switch_update, set_value_range_to_nonnegative): Remove.
2102 (set_value_range_to_truthvalue): Likewise.
2103 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
2104 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
2105 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
2106 (find_case_label_ranges, test_for_singularity): Likewise.
2107 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
2108 (x_vr_values): Move to its remaining use site.
2110 2017-11-10 Jeff Law <law@redhat.com>
2112 * vr-values.h (VR_INITIALIZER): Move #define here.
2113 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
2114 * Makefile.in (OBJS): Add tree-evrp.o
2115 * tree-vrp.h (assert_info): Move structure definition here.
2116 (set_value_range_to_varying): Prototype.
2117 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
2118 (infer_value_range, register_edge_assert_for): Likewise.
2119 (stmt_interesting_for_vrp): Likewise.
2120 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
2121 (set_value_range_to_varying): No longer static.
2122 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
2123 (infer_value_range, register_edge_assert_for): Likewise.
2125 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2127 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
2128 generate the XXBRD instruction.
2130 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
2131 (rs6000_builtin_type_compatible): Treat _Float128 and long double
2132 as being compatible if -mabi=ieeelongdouble.
2133 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
2134 to setup float128 built-ins with hardware support.
2135 (BU_FLOAT128_HW_2): Likewise.
2136 (BU_FLOAT128_HW_3): Likewise.
2137 (BU_FLOAT128_HW_VSX_1): Likewise.
2138 (BU_FLOAT128_HW_VSX_2): Likewise.
2139 (scalar_extract_expq): Change float128 built-in functions to
2140 accommodate having both KFmode and TFmode functions. Use the
2141 KFmode variant as the default.
2142 (scalar_extract_sigq): Likewise.
2143 (scalar_test_neg_qp): Likewise.
2144 (scalar_insert_exp_q): Likewise.
2145 (scalar_insert_exp_qp): Likewise.
2146 (scalar_test_data_class_qp): Likewise.
2147 (sqrtf128_round_to_odd): Delete processing the round to odd
2148 built-in functions as special built-in functions, and define them
2149 as float128 built-ins. Use the KFmode variant as the default.
2150 (truncf128_round_to_odd): Likewise.
2151 (addf128_round_to_odd): Likewise.
2152 (subf128_round_to_odd): Likewise.
2153 (mulf128_round_to_odd): Likewise.
2154 (divf128_round_to_odd): Likewise.
2155 (fmaf128_round_to_odd): Likewise.
2156 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
2157 support for KFmode and TFmode xststdcqp calls.
2158 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
2159 point, switch the built-in handlers for the get/set float128
2160 exponent, get float128 mantissa, float128 test built-ins, and the
2161 float128 round to odd built-in functions. Eliminate creating the
2162 float128 round to odd built-in functions as special built-ins.
2163 (rs6000_init_builtins): Eliminate special creation of the float128
2164 round to odd built-in functions.
2165 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
2166 function insns to support both TFmode and KFmode variants.
2167 (xsxsigqp_<mode>): Likewise.
2168 (xsiexpqpf_<mode>): Likewise.
2169 (xsiexpqp_<mode>): Likewise.
2170 (xststdcqp_<mode>): Likewise.
2171 (xststdcnegqp_<mode>): Likewise.
2172 (xststdcqp_<mode>): Likewise.
2174 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2176 * tree-ssa-threadupdate.c (compute_path_counts): Remove
2177 unused path_in_freq_ptr parameter.
2178 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
2180 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2182 * ipa-inline.c (edge_badness): Dump sreal frequency.
2183 (compute_inlined_call_time): Match natural implementaiton ...
2184 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
2185 forgotten division by CGRAPH_FREQ_BASE.
2187 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2189 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
2190 Solaris 11. Update comment.
2191 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
2193 * config/sol2.h (STARTFILE_SPEC): Likewise.
2194 * configure: Regenerate.
2196 2017-11-14 Carl Love <cel@us.ibm.com>
2198 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
2199 le_ and be_ prefixes to swap* variables. Remove
2200 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
2203 2017-11-14 Jason Merrill <jason@redhat.com>
2205 Support GTY((cache)) on hash_map.
2206 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
2207 (ggc_cache_remove): Override it instead of ggc_mx.
2208 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
2209 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
2210 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
2211 (simple_cache_map_traits): Override maybe_mx.
2212 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
2213 (hash_map): Friend gt_cleare_cache.
2214 (gt_cleare_cache): New.
2215 * tree.h (tree_cache_traits): New hash_map traits class.
2216 (tree_cache_map): New typedef.
2218 2017-11-14 Richard Biener <rguenther@suse.de>
2220 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
2221 paramter and handling.
2222 (cleanup_control_flow_bb): Likewise.
2223 (cleanup_control_flow_pre): New helper performing a DFS walk
2224 to call cleanup_control_flow_bb in PRE order.
2225 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
2226 via cleanup_control_flow_pre.
2228 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
2230 * config/aarch64/aarch64-simd.md
2231 (aarch64_simd_bsl<mode>_internal): Remove DImode.
2232 (*aarch64_simd_bsl<mode>_alt): Likewise.
2233 (aarch64_simd_bsldi_internal): New.
2234 (aarch64_simd_bsldi_alt): Likewise.
2236 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2238 * tracer.c (better_p): Do not compare frequencies.
2239 * reg-stack.c (better_edge): Likewise.
2240 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
2243 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2245 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
2246 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
2247 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
2248 * lto-streamer-in.c (input_function): Use update_max_bb_count.
2249 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
2250 * predict.c (maybe_hot_frequency_p): Inline to ...
2251 (maybe_hot_count_p): ... here; rewrite to counts.
2252 (counts_to_freqs): Rename to ...
2253 (update_max_bb_count): ... this one.
2254 (expensive_function_p): Use counts.
2255 (estimate_bb_frequencies): Update.
2256 (rebuild_frequencies): Update.
2257 * predict.h (counts_to_freqs): Rename to ...
2258 (update_max_bb_count): ... this one.
2259 * profile.c (compute_branch_probabilities): Add debug info
2260 * tree-inline.c (expand_call_inline): Update debug info.
2261 (optimize_inline_calls): Use update_max_bb_count..
2262 (tree_function_versioning): Use update_max_bb_count..
2263 * value-prof.c (gimple_value_profile_transformations):
2264 Do not use update_max_bb_count.
2266 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2268 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
2269 always use frequencies.
2271 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2273 * bb-reorder.c: Remove frequencies from comments.
2274 (better_edge_p): Use profile counts.
2275 (find_traces): Dump profile counts.
2276 (rotate_loop): Use profile counts.
2277 (find_traces_1_round): Likewise.
2278 (connect_better_edge_p): Use counts instead of probabilities for
2280 (copy_bb_p): Drop early check for non-0 frequency.
2281 (sanitize_hot_paths): Update comments.
2283 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2285 * ipa-split.c (struct split_point): Add count.
2286 (consider_split): Do not compute incoming frequency; compute incoming
2287 count and store it to split_point.
2288 (split_function): Set count of the call to split part correctly.
2290 2017-11-13 Carl Love <cel@us.ibm.com>
2292 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
2294 2017-11-13 Tom Tromey <tom@tromey.com>
2296 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
2298 2017-11-13 Carl Love <cel@us.ibm.com>
2300 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
2301 Add support for builtins:
2302 unsigned int vec_first_{,miss}_match_{,or_eos}index,
2303 vector {un,}signed {char,int,short},
2304 vector {un,}signed {char,int,short}) arguments.
2305 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
2306 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
2307 Add BU_P9V_AV_2 expansions for the builtins.
2308 * config/rs6000/altivec.h (vec_first_match_index,
2309 vec_first_mismatch_index, vec_first_match_or_eos_index,
2310 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
2311 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
2312 new extern declaration.
2313 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
2314 * config/rs6000/vsx.md (first_match_index_<mode>,
2315 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
2316 first_mismatch_or_eos_index_<mode>): Add define expand.
2317 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
2318 * doc/extend.texi: Update the built-in documenation file for the new
2321 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
2323 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
2324 into the min/max operations for _Float<N> and _Float<N>X types.
2326 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
2329 * dwarf2out.c (do_eh_frame): New static variable.
2330 (dwarf2out_begin_prologue): Set it.
2331 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
2333 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2335 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
2337 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
2339 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
2342 * debug.h (dwarf2out_do_eh_frame): Declare.
2343 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
2344 (dwarf2out_do_frame): Use it.
2345 (dwarf2out_do_cfi_asm): Likewise.
2346 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
2347 (dwarf2out_assembly_start): Likewise.
2348 (dwarf2out_begin_prologue): Fix comment.
2349 * toplev.c (compile_file): Always call dwarf2out_frame_finish
2350 if the target needs either debug or unwind DWARF2 info.
2351 * lto-opts.c (lto_write_options): Do not save -fexceptions,
2352 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
2353 -ftrapping-math, -ftrapv and -fwrapv.
2355 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2357 * cgraph.c (cgraph_edge::sreal_frequency): New function.
2358 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
2359 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
2360 (estimate_edge_size_and_time): Likewise.
2361 (ipa_merge_fn_summary_after_inlining): Likewise.
2362 * ipa-inline.c (cgraph_freq_base_rec): Remove.
2363 (compute_uninlined_call_time): Use sreal_frequency.
2364 (compute_inlined_call_time): Likewise.
2365 (ipa_inline): Do not initialize cgraph_freq_base_rec.
2366 * profile-count.c: Include sreal.h.
2367 (profile_count::to_sreal_scale): New.
2368 * profile-count.h: Forward declare sreal.
2369 (profile_count::to_sreal_scale): Declare.
2371 2017-11-13 Nathan Sidwell <nathan@acm.org>
2373 * diagnostic.c (maybe_line_and_column): New.
2374 (diagnostic_get_location_text): Use it.
2375 (diagnostic_report_current_module): Likewise.
2376 (test_diagnostic_get_location_text): Add tests.
2378 2017-11-13 Luis Machado <luis.machado@linaro.org>
2380 * doc/md.texi (Specifying processor pipeline description): Fix
2381 incorrect latency for the div instruction example.
2383 2017-11-13 Jakub Jelinek <jakub@redhat.com>
2385 PR tree-optimization/78821
2386 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
2387 that bit_not_p is the same.
2388 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
2389 (split_group): Count precisely bit_not_p bits in each statement.
2390 (invert_op): New function.
2391 (imm_store_chain_info::output_merged_store): Use invert_op to
2392 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
2393 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
2395 2017-11-13 Martin Liska <mliska@suse.cz>
2397 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
2398 (struct source_info): Likewise.
2399 (add_branch_counts): Likewise.
2400 (add_line_counts): Likewise.
2401 (function_summary): Likewise.
2402 (output_intermediate_line): Likewise.
2403 (generate_results): Likewise.
2405 2017-11-13 Martin Liska <mliska@suse.cz>
2407 * gcov.c (struct block_info): Remove typedef for block_t.
2408 (struct line_info): Likewise.
2409 (line_info::has_block): Likewise.
2410 (EXIT_BLOCK): Likewise.
2411 (unblock): Likewise.
2412 (circuit): Likewise.
2413 (get_cycles_count): Likewise.
2414 (process_file): Likewise.
2415 (read_graph_file): Likewise.
2416 (solve_flow_graph): Likewise.
2417 (find_exception_blocks): Likewise.
2418 (add_line_counts): Likewise.
2419 (accumulate_line_info): Likewise.
2420 (output_line_details): Likewise.
2422 2017-11-13 Martin Liska <mliska@suse.cz>
2424 * gcov.c (struct arc_info): Remove typedef for arc_t.
2425 (struct line_info): Likewise.
2426 (add_branch_counts): Likewise.
2427 (output_branch_count): Likewise.
2428 (function_info::~function_info): Likewise.
2429 (circuit): Likewise.
2430 (output_intermediate_line): Likewise.
2431 (read_graph_file): Likewise.
2432 (solve_flow_graph): Likewise.
2433 (find_exception_blocks): Likewise.
2434 (add_line_counts): Likewise.
2435 (accumulate_line_info): Likewise.
2436 (output_line_details): Likewise.
2437 (output_function_details): Likewise.
2439 2017-11-13 Martin Liska <mliska@suse.cz>
2441 * gcov.c (struct function_info): Remove typedef for function_t.
2442 (struct source_info): Likewise.
2443 (source_info::get_functions_at_location): Likewise.
2444 (solve_flow_graph): Likewise.
2445 (find_exception_blocks): Likewise.
2446 (add_line_counts): Likewise.
2447 (output_intermediate_file): Likewise.
2448 (process_file): Likewise.
2449 (generate_results): Likewise.
2450 (release_structures): Likewise.
2451 (read_graph_file): Likewise.
2452 (read_count_file): Likewise.
2453 (accumulate_line_counts): Likewise.
2454 (output_lines): Likewise.
2456 2017-11-13 Martin Liska <mliska@suse.cz>
2458 * gcov.c (function_info::function_info): Remove num_counts
2459 and add vector<gcov_type>.
2460 (function_info::~function_info): Use the vector.
2461 (process_file): Likewise.
2462 (read_graph_file): Likewise.
2463 (read_count_file): Likewise.
2464 (solve_flow_graph): Likewise.
2466 2017-11-13 Martin Liska <mliska@suse.cz>
2468 * gcov.c (function_info::is_artificial): New function.
2469 (process_file): Erase all artificial early.
2470 (generate_results): Skip as all artificial are already
2473 2017-11-13 Martin Liska <mliska@suse.cz>
2475 * gcov.c (read_graph_file): Store to global vector of functions.
2476 (read_count_file): Iterate the vector.
2477 (process_file): Likewise.
2478 (generate_results): Likewise.
2479 (release_structures): Likewise.
2481 2017-11-13 Jakub Jelinek <jakub@redhat.com>
2483 PR tree-optimization/82954
2484 * gimple-ssa-store-merging.c
2485 (imm_store_chain_info::coalesce_immediate_stores): If
2486 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
2488 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
2490 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
2491 Upddate call to ENDIAN_LANE_N.
2492 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
2493 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
2494 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
2495 and use aarch64_endian_lane_rtx.
2496 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
2498 2017-11-12 Tom de Vries <tom@codesourcery.com>
2500 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
2502 2017-11-12 Tom de Vries <tom@codesourcery.com>
2504 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
2506 2017-11-12 Tom de Vries <tom@codesourcery.com>
2508 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
2510 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
2511 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
2512 * defaults.h (ASM_OUTPUT_LABELREF): Same.
2514 2017-11-11 Martin Sebor <msebor@redhat.com>
2517 * doc/extend.texi (attribute nonstring): Remove spurious argument.
2520 * prefic.c (translate_name): Replace strncpy with memcpy to
2521 avoid -Wstringop-truncation.
2523 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2525 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
2527 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2529 * predict.c (maybe_hot_frequency_p): Do not use cfun.
2531 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2533 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
2536 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
2538 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
2539 enable generating XXBRH if the value is in a vector register.
2540 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
2541 value is in a vector register.
2542 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
2543 register to register bswap64's instead of doing the GPR sequence
2544 used on previous machines.
2545 (bswapdi2_xxbrd): New insn.
2546 (bswapdi2_reg): Disallow on ISA 3.0.
2547 (register to register bswap64 splitter): Do not split the insn on
2548 ISA 3.0 systems that use XXBRD.
2550 2017-11-10 Martin Sebor <msebor@redhat.com>
2553 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
2555 (find_subframework_file): Same.
2557 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2559 * auto-profile.c (afdo_indirect_call): Drop frequency.
2560 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
2561 (cgraph_node::create_edge): Drop frequency argument.
2562 (cgraph_node::create_indirect_edge): Drop frequency argument.
2563 (cgraph_edge::make_speculative): Drop frequency arguments.
2564 (cgraph_edge::resolve_speculation): Do not update frequencies
2565 (cgraph_edge::dump_edge_flags): Do not dump frequency.
2566 (cgraph_node::dump): Check consistency in IPA mode.
2567 (cgraph_edge::maybe_hot_p): Use IPA counter.
2568 (cgraph_edge::verify_count_and_frequency): Rename to ...
2569 (cgraph_edge::verify_count): ... this one; drop frequency checking.
2570 (cgraph_node::verify_node): Update.
2571 * cgraph.h (struct cgraph_edge): Drop frequency.
2572 (cgraph_edge::frequency): New function.
2573 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
2575 (cgraph_edge::rebuild_edges): Likewise.
2576 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
2577 (duplicate_thunk_for_node): Do not pass frequency.
2578 (cgraph_node::create_clone): Scale only counts.
2579 (cgraph_node::create_virtual_clone): Do not pass frequency.
2580 (cgraph_node::create_edge_including_clones): Do not pass frequency.
2581 (cgraph_node::create_version_clone): Do not pass frequency.
2582 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
2583 (cgraph_node::expand_thunk): Do not pass frequency.
2584 (cgraph_node::create_wrapper): Do not pass frequency.
2585 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
2587 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
2588 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
2589 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
2590 (gather_caller_stats): Use frequency function.
2591 (ipcp_cloning_candidate_p): Use frequency function.
2592 (ipcp_propagate_stage): Use frequency function.
2593 (get_info_about_necessary_edges): Use frequency function.
2594 (update_profiling_info): Update only IPA profile.
2595 (update_specialized_profile): Use frequency functoin.
2596 (perhaps_add_new_callers): Update only IPA profile.
2597 * ipa-devirt.c (ipa_devirt): Use IPA profile.
2598 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
2599 (dump_ipa_call_summary): Use frequency function.
2600 (estimate_edge_size_and_time): Use frequency function.
2601 (ipa_merge_fn_summary_after_inlining): Use frequency function.
2602 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
2603 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
2604 (update_noncloned_counts): ... ths one; scale counts only.
2605 (clone_inlined_nodes): Do not scale frequency.
2606 (inline_call): Do not pass frequency.
2607 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
2608 (compute_inlined_call_time): Use IPA profile.
2609 (want_inline_small_function_p): Use IPA profile.
2610 (want_inline_self_recursive_call_p): Use IPA profile.
2611 (edge_badness): Use IPA profile.
2612 (lookup_recursive_calls): Use IPA profile.
2613 (recursive_inlining): Do not pass frequency.
2614 (resolve_noninline_speculation): Do not update frequency.
2615 (inline_small_functions): Collect max of IPA profile.
2616 (dump_overall_stats): Dump IPA porfile.
2617 (dump_inline_stats): Dump IPA porfile.
2618 (ipa_inline): Collect IPA stats.
2619 * ipa-inline.h (clone_inlined_nodes): Update prototype.
2620 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
2621 (ipa_propagate_frequency): Use frequency function.
2622 (ipa_profile): Cleanup.
2623 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
2624 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
2625 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
2626 (input_node): Do not stream frequency.
2627 (input_edge): Do not stream frequency.
2628 (merge_profile_summaries): Scale only IPA profiles.
2629 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
2630 * predict.c (drop_profile): Do not recompute frequency.
2631 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
2632 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
2633 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
2634 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
2635 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
2636 * tree-inline.c (copy_bb): Do not scale frequency.
2637 (expand_call_inline): Do not scale frequency.
2638 (tree_function_versioning): Do not scale frequency.
2639 * ubsan.c (ubsan_create_edge): Do not pass frequency.
2641 2017-11-10 Julia Koval <julia.koval@intel.com>
2643 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
2644 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
2645 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
2646 (_mm256_maskz_gf2p8affine_epi64_epi8)
2647 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
2648 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
2649 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
2650 (__builtin_ia32_vgf2p8affineqb_v32qi)
2651 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
2652 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
2654 2017-11-10 Tamar Christina <tamar.christina@arm.com>
2658 (arm_option_override): Refactor.
2659 (arm_option_reconfigure_globals): New.
2660 (arm_options_perform_arch_sanity_checks): New.
2661 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
2663 (arm_options_perform_arch_sanity_checks): Likewise
2665 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
2667 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
2668 (power9-qpmul): New.
2669 * rs6000/rs6000.md ("type" attr): Add qmul.
2670 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
2671 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
2672 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
2674 2017-11-10 Martin Sebor <msebor@redhat.com>
2677 * builtins.c (compute_objsize): Handle arrays that
2678 compute_builtin_object_size likes to fail for. Make extern.
2679 * builtins.h (compute_objsize): Declare.
2680 (check_strncpy_sizes): New function.
2681 (expand_builtin_strncpy): Call check_strncpy_sizes.
2682 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
2683 -Wstringop-truncation.
2684 (gimple_fold_builtin_strncat): Same.
2685 * gimple.c (gimple_build_call_from_tree): Set call location.
2686 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
2687 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
2688 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
2689 (handle_builtin_strlen): Use strlen_to_stridx.
2690 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
2692 Use strlen_to_stridx.
2693 (pass_strlen::execute): Release strlen_to_stridx.
2694 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
2695 (-Wstringop-truncation): Document new option.
2697 2017-11-10 Martin Liska <mliska@suse.cz>
2699 PR gcov-profile/82702
2700 * gcov.c (main): Handle intermediate files in a different
2702 (get_gcov_intermediate_filename): New function.
2703 (output_gcov_file): Remove support of intermediate files.
2704 (generate_results): Allocate intermediate file.
2705 (release_structures): Clean-up properly fn_end.
2706 (output_intermediate_file): Start iterating with line 1.
2708 2017-11-10 Jakub Jelinek <jakub@redhat.com>
2710 PR tree-optimization/82929
2711 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
2712 ops_swapped_p non-static data member.
2713 (store_immediate_info::store_immediate_info): Clear it.
2714 (imm_store_chain_info::coalesce_immediate_stores): If swapping
2715 ops set ops_swapped_p.
2716 (count_multiple_uses): Handle ops_swapped_p.
2718 2017-11-10 Martin Liska <mliska@suse.cz>
2720 * coverage.c (coverage_init): Stream information about
2721 support of has_unexecuted_blocks.
2722 * doc/gcov.texi: Document that.
2723 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
2724 * gcov.c (read_graph_file): Likewise.
2725 (output_line_beginning): Fix a small issue with
2728 2017-11-10 Bin Cheng <bin.cheng@arm.com>
2730 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
2731 reference of trivial component.
2733 2017-11-10 Jakub Jelinek <jakub@redhat.com>
2736 * gimple-ssa-store-merging.c
2737 (pass_store_merging::terminate_all_aliasing_chains): For
2738 gimple_store_p stmts also call refs_output_dependent_p.
2740 PR rtl-optimization/82913
2741 * compare-elim.c (try_merge_compare): Punt if def_insn is not
2744 2017-11-09 Jeff Law <law@redhat.com>
2746 * vr-values.h: New file with vr_values class.
2747 * tree-vrp.c: Include vr-values.h
2748 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
2749 data objects into the vr_values class.
2750 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
2751 (get_value_range): Make it a member function within vr_values class.
2752 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
2753 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
2754 (op_with_constant_singleton_value_range): Likewise.
2755 (extract_range_for_var_from_comparison_expr): Likewise.
2756 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
2757 (extract_range_from_binary_expr): Likewise.
2758 (extract_range_from_unary_expr): Likewise.
2759 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
2760 (check_for_binary_op_overflow, extract_range_basic): Likewise.
2761 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
2762 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
2763 (compare_name_with_value, compare_names): Likewise.
2764 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
2765 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
2766 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
2767 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
2768 (extract_range_from_phi_node): Likewise.
2769 (simplify_truth_ops_using_ranges): Likewise.
2770 (simplify_div_or_mod_using_ranges): Likewise.
2771 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
2772 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
2773 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
2774 (simplify_float_conversion_using_ranges): Likewise.
2775 (simplify_internal_call_using_ranges): Likewise.
2776 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
2777 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
2778 poitner via x_vr_values for calls into gimple folder.
2779 (vrp_initialize_lattice): Make this the vr_values ctor.
2780 (vrp_free_lattice): Make this the vr_values dtor.
2781 (set_vr_value): New function.
2782 (class vrp_prop): Add vr_values data member. Add various member
2783 functions as well as member functions that delegate to vr_values.
2784 (check_array_ref): Make a member function within vrp_prop class.
2785 (search_for_addr_array, vrp_initialize): Likewise.
2786 (vrp_finalize): Likewise. Revamp to avoid direct access to
2787 vr_value, values_propagated, etc.
2788 (check_array_bounds): Extract vrp_prop class instance pointer from
2789 walk info structure. Use it to call member functions.
2790 (check_all_array_refs): Make a member function within vrp_prop class.
2791 Smuggle class instance pointer via walk info structure.
2792 (x_vr_values): New local static.
2793 (vrp_valueize): Use x_vr_values to get class instance.
2794 (vr_valueize_1): Likewise.
2795 (class vrp_folder): Add vr_values data member. Add various member
2796 functions as well as member functions that delegate to vr_values.
2797 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
2798 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
2799 class instance from vr_values. Use it to call member functions.
2800 (vrp_dom_walker): Add vr_values data member.
2801 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
2802 instance via x_vr_values.
2803 (identify_jump_threads): Accept vr_values as argument. Store
2804 it into the walker structure.
2805 (evrp_dom_walker): Add vr_values class data member. Add various
2807 (evrp_dom_walker::try_find_new_range): Use vr_values data
2808 member to access the memory allocator.
2809 (evrp_dom_walker::before_dom_children): Store vr_values class
2810 instance into the vrp_folder class.
2811 (evrp_dom_walker::push_value_range): Rework to avoid direct
2812 access to num_vr_values and vr_value.
2813 (evrp_dom_walker::pop_value_range): Likewise.
2814 (execute_early_vrp): Remove call to vrp_initialize_lattice.
2815 Use vr_values to get to dump_all_value_ranges member function.
2816 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
2817 and simplify_cond_using_ranges_2 via vrp_prop class instance.
2818 Pass vr_values class instance down to identify_jump_threads.
2819 Remove call to vrp_free_lattice.
2820 (debug_all_value_ranges): Remove.
2822 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
2823 (vrp_folder): Likewise.
2825 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
2826 Get it from the existing bitmap instead.
2827 (vrp_intersect_ranges_1): Likewise.
2829 2017-11-09 Jakub Jelinek <jakub@redhat.com>
2831 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
2833 (store_immediate_info::store_immediate_info): Add bitnotp argument,
2834 set bit_not_p to it.
2835 (imm_store_chain_info::coalesce_immediate_stores): Break group
2836 if bit_not_p is different.
2837 (count_multiple_uses, split_group,
2838 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
2839 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
2840 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
2841 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
2842 to store_immediate_info ctor.
2844 2017-11-09 Jim Wilson <jimw@sifive.com>
2846 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
2847 (scan_prog_file): Likewise.
2849 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
2851 * bb-reorder.c (max_entry_frequency): Remove.
2852 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
2853 connect_traces, push_to_next_round_p): Remove prototypes.
2854 (find_traces_1_round): Use counts only.
2855 (push_to_next_round_p): Likewise.
2856 (find_traces): Likewise.
2857 (rotate_loop): Likewise.
2858 (find_traces_1_round): Likewise.
2859 (connect_traces): Likewise.
2860 (edge_order): Likewise.
2862 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
2864 * config/arm/arm.c (output_return_instruction): Add comments to
2865 indicate requirement for cmse_nonsecure_entry return to account
2866 for the size of clearing instruction output here.
2867 (thumb_exit): Likewise.
2868 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
2869 return in hardfloat mode.
2871 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
2873 * config/rs6000/rs6000.c (machine_function): Add a bool,
2874 "toc_is_wrapped_separately".
2875 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
2876 if it wasn't explicitly set or unset, we are optimizing for speed, and
2877 doing separate shrink-wrapping.
2878 (rs6000_get_separate_components): Enable the TOC component if
2879 saving the TOC register in the prologue.
2880 (rs6000_components_for_bb): Handle the TOC component.
2881 (rs6000_emit_prologue_components): Store the TOC register where needed.
2882 (rs6000_set_handled_components): Mark TOC as handled, if handled.
2883 (rs6000_emit_prologue): Don't save the TOC if that is already done.
2885 2017-11-09 Martin Jambor <mjambor@suse.cz>
2887 * ipa-param-manipulation.c: New file.
2888 * ipa-param-manipulation.h: Likewise.
2889 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
2890 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
2891 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
2892 (ipa_parm_adjustment): Likewise.
2893 (ipa_parm_adjustment_vec): Likewise.
2894 (ipa_get_vector_of_formal_parms): Moved declaration to
2895 ipa-param-manipulation.h.
2896 (ipa_get_vector_of_formal_parm_types): Likewise.
2897 (ipa_modify_formal_parameters): Likewise.
2898 (ipa_modify_call_arguments): Likewise.
2899 (ipa_combine_adjustments): Likewise.
2900 (ipa_dump_param_adjustments): Likewise.
2901 (ipa_modify_expr): Likewise.
2902 (ipa_get_adjustment_candidate): Likewise.
2903 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
2904 ipa-param-manipulation.c.
2905 (ipa_get_vector_of_formal_parm_types): Likewise.
2906 (ipa_modify_formal_parameters): Likewise.
2907 (ipa_modify_call_arguments): Likewise.
2908 (ipa_modify_expr): Likewise.
2909 (get_ssa_base_param): Likewise.
2910 (ipa_get_adjustment_candidate): Likewise.
2911 (index_in_adjustments_multiple_times_p): Likewise.
2912 (ipa_combine_adjustments): Likewise.
2913 (ipa_dump_param_adjustments): Likewise.
2914 * tree-sra.c: Also include ipa-param-manipulation.h
2915 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
2918 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2919 Alan Hayward <alan.hayward@arm.com>
2920 David Sherwood <david.sherwood@arm.com>
2922 * doc/sourcebuild.texi (vect_masked_store): Document.
2924 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2925 Alan Hayward <alan.hayward@arm.com>
2926 David Sherwood <david.sherwood@arm.com>
2928 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
2930 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2931 Alan Hayward <alan.hayward@arm.com>
2932 David Sherwood <david.sherwood@arm.com>
2934 * doc/sourcebuild.texi (vect_variable_length): Document.
2936 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2937 Alan Hayward <alan.hayward@arm.com>
2938 David Sherwood <david.sherwood@arm.com>
2940 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
2942 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2943 Alan Hayward <alan.hayward@arm.com>
2944 David Sherwood <david.sherwood@arm.com>
2946 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
2948 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2949 Alan Hayward <alan.hayward@arm.com>
2950 David Sherwood <david.sherwood@arm.com>
2952 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
2953 previously undocumented selectors.
2954 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
2956 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2958 * doc/rtl.texi (const_vector): Say that elements can be
2959 const_wide_ints too.
2960 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
2961 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
2962 (gen_vec_duplicate): Use it instead of CONSTANT_P.
2963 * optabs.c (expand_vector_broadcast): Likewise.
2965 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2966 Alan Hayward <alan.hayward@arm.com>
2967 David Sherwood <david.sherwood@arm.com>
2969 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
2970 scaled index even if the unscaled address was invalid.
2971 Don't increase the complexity of using a scale in that case.
2973 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2974 Alan Hayward <alan.hayward@arm.com>
2975 David Sherwood <david.sherwood@arm.com>
2977 * doc/rtl.texi: Rewrite the subreg rules so that they partition
2978 the inner register into REGMODE_NATURAL_SIZE bytes rather than
2979 UNITS_PER_WORD bytes.
2980 * emit-rtl.c (validate_subreg): Divide subregs into blocks
2981 based on REGMODE_NATURAL_SIZE of the inner mode.
2982 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
2983 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
2984 * expmed.c (lowpart_bit_field_p): Divide the value up into
2985 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
2986 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
2987 whether something is likely to occupy more than one register.
2989 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
2992 * ipa-inline-transform.c (update_noncloned_frequencies): Use
2993 profile_count::adjust_for_ipa_scaling.
2994 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
2995 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
2997 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
2999 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3001 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
3002 (split_group): Add total_orig and total_new arguments, estimate the
3003 number of statements related to the store group without store merging
3004 and with store merging.
3005 (imm_store_chain_info::output_merged_store): Adjust split_group
3006 callers, punt if estimated number of statements with store merging
3007 is not smaller than estimated number of statements without it.
3009 (handled_load): Remove has_single_use checks.
3010 (pass_store_merging::process_store): Likewise.
3012 2017-11-09 Richard Biener <rguenther@suse.de>
3014 PR tree-optimization/82902
3015 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
3017 2017-11-09 Martin Liska <mliska@suse.cz>
3020 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
3023 2017-11-09 Martin Liska <mliska@suse.cz>
3025 PR tree-optimization/82669
3026 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
3028 2017-11-09 Martin Liska <mliska@suse.cz>
3030 PR gcov-profile/48463
3031 * coverage.c (coverage_begin_function): Output also end locus
3032 of a function and information whether the function is
3034 * gcov-dump.c (tag_function): Parse and print the information.
3035 * gcov.c (INCLUDE_MAP): Add include.
3036 (INCLUDE_SET): Likewise.
3037 (struct line_info): Move earlier in the source file because
3038 of vector<line_info> in function_info structure.
3039 (line_info::line_info): Likewise.
3040 (line_info::has_block): Likewise.
3041 (struct source_info): Add new member index.
3042 (source_info::get_functions_at_location): New function.
3043 (function_info::group_line_p): New function.
3044 (output_intermediate_line): New function.
3045 (output_intermediate_file): Use the mentioned function.
3046 (struct function_start): New.
3047 (struct function_start_pair_hash): Likewise.
3048 (process_file): Add code that identifies group functions.
3049 Assign lines either to global or function scope.
3050 (generate_results): Skip artificial functions.
3051 (find_source): Assign index for each source file.
3052 (read_graph_file): Read new flag artificial and end_line.
3053 (add_line_counts): Assign it either to global of function scope.
3054 (accumulate_line_counts): Isolate core of the function to
3055 accumulate_line_info and call it for both function and global
3057 (accumulate_line_info): New function.
3058 (output_line_beginning): Fix GNU coding style.
3059 (print_source_line): New function.
3060 (output_line_details): Likewise.
3061 (output_function_details): Likewise.
3062 (output_lines): Iterate both source (global) scope and function
3064 (struct function_line_start_cmp): New class.
3065 * doc/gcov.texi: Reflect changes in documentation.
3067 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3070 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
3071 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
3072 and similarly for not instead of neg.
3074 2017-11-08 Andi Kleen <ak@linux.intel.com>
3076 * config/i386/i386.opt: Add -mforce-indirect-call.
3077 * config/i386/predicates.md: Check for flag_force_indirect_call.
3078 * doc/invoke.texi: Document -mforce-indirect-call
3080 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
3082 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
3084 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
3085 config/riscv/riscv.c (predict.h): New include.
3086 (riscv_slow_unaligned_access_p): No longer static.
3087 (riscv_block_move_straight): Add require.
3088 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
3090 2017-11-08 Jakub Jelinek <jakub@redhat.com>
3093 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
3094 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
3095 nonimmediate_operand predicate for operand 1 instead of
3098 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3100 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
3102 * config/aarch64/constraints.md (Uml): New constraint.
3103 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
3106 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3108 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
3109 of two vec_duplicates into a vec_concat.
3111 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3113 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
3114 Simplify vec_merge of vec_duplicate and vec_concat.
3115 * config/aarch64/constraints.md (Utq): New constraint.
3116 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
3119 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3121 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
3122 Simplify vec_merge of vec_duplicate and const_vector.
3123 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
3125 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
3126 mode iterator. Update predicate on operand 1 to
3127 handle non-const_vec constants. Delete constraints.
3128 (*aarch64_combinez_be<mode>): Likewise for operand 2.
3130 2017-11-08 Jakub Jelinek <jakub@redhat.com>
3132 PR tree-optimization/78821
3133 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
3135 (store_operand_info::store_operand_info): Initialize it to false.
3136 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
3137 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
3138 store in the group, and if chain_info is non-NULL, to ignore altogether
3140 (compatible_load_p): Fail if bit_not_p does not match.
3141 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
3142 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
3143 (pass_store_merging::process_store): Adjust
3144 terminate_all_aliasing_chains calls to pass NULL in all current spots,
3145 call terminate_all_aliasing_chains newly when adding a store into
3146 a chain with non-NULL chain_info.
3148 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
3150 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
3152 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
3154 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3156 (aarch64_layout_frame): Initialise emit_frame_chain.
3157 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
3158 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
3160 2017-11-08 Martin Liska <mliska@suse.cz>
3162 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
3163 of gimple_call_internal_p.
3165 2017-11-07 Tom de Vries <tom@codesourcery.com>
3167 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
3169 2017-11-07 Tom de Vries <tom@codesourcery.com>
3171 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3174 2017-11-08 Martin Liska <mliska@suse.cz>
3177 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
3179 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
3181 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
3183 2017-11-07 Jakub Jelinek <jakub@redhat.com>
3186 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
3187 (*cmp<mode>_ccz_1): New insn with $k alternative.
3190 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
3191 RTX_COMM_COMPARE as commutative as well.
3192 (ix86_binary_operator_ok): Formatting fix.
3193 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
3194 *<code><mode>3<mask_name><round_saeonly_name>,
3195 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
3196 *<s>mul<mode>3_highpart<mask_name>,
3197 *vec_widen_umult_even_v16si<mask_name>,
3198 *vec_widen_umult_even_v8si<mask_name>,
3199 *vec_widen_umult_even_v4si<mask_name>,
3200 *vec_widen_smult_even_v16si<mask_name>,
3201 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
3202 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
3203 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
3204 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
3205 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
3206 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
3207 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
3208 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
3209 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
3210 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
3211 ix86_binary_operator_ok. Formatting fixes.
3212 (*<plusminus_insn><mode>3<mask_name><round_name>,
3213 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
3216 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
3218 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
3219 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
3220 eq and ne if TARGET_ISEL.
3221 (cmp): New code_iterator.
3222 (UNS, UNSU_, UNSIK): New code_attrs.
3223 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
3224 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
3226 ("eq<mode>3"): ... this.
3227 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
3229 ("ne<mode>3"): ... this.
3231 2017-11-07 Julia Koval <julia.koval@intel.com>
3234 * common/config/i386/i386-common.c
3235 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
3236 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
3237 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
3238 * config/i386/i386.opt: Ditto.
3239 * config/i386/i386.c (ix86_target_string): Ditto.
3240 (ix86_option_override_internal): Ditto.
3241 (ix86_init_mpx_builtins): Move MPX to args2.
3242 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
3243 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3244 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3245 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3246 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3247 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3248 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
3250 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
3253 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
3254 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
3255 (zero-extendsidi peephole2): Remove peephole.
3257 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
3260 * stor-layout.c: Include attribs.h.
3261 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
3262 explicit lookup of "aligned" attribute.
3264 2017-11-07 Andrew Waterman <andrew@sifive.com>
3266 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
3267 (riscv_expand_block_move): Likewise.
3268 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
3270 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
3271 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
3272 * config/riscv/riscv.c (riscv_block_move_straight): New function.
3273 (riscv_adjust_block_mem): Likewise.
3274 (riscv_block_move_loop): Likewise.
3275 (riscv_expand_block_move): Likewise.
3276 * config/riscv/riscv.md (movmemsi): New pattern.
3278 2017-11-07 Michael Clark <michaeljclark@mac.com>
3280 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
3281 (MUSL_DYNAMIC_LINKER): Likewise.
3283 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
3285 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
3286 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
3289 2017-11-07 Richard Biener <rguenther@suse.de>
3291 * match.pd: Fix build.
3293 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
3294 Jackson Woodruff <jackson.woodruff@arm.com>
3296 PR tree-optimization/71026
3297 * match.pd: Canonicalize negate in division.
3299 2017-11-07 Sudakshina Das <sudi.das@arm.com>
3302 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
3304 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
3306 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
3307 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
3309 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
3311 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
3312 non-scalar integral types.
3313 * match.pd (negate_expr_p): Handle MINUS_EXPR.
3314 (-(A-B), -(~A)): New transformations.
3316 2017-11-07 Tom de Vries <tom@codesourcery.com>
3318 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
3319 semicolon after "do {} while (0)".
3320 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3321 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3322 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3323 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3324 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3326 2017-11-07 Tom de Vries <tom@codesourcery.com>
3328 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
3329 after "do {} while (0)".
3330 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3331 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3332 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3333 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3334 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3336 2017-11-07 Tom de Vries <tom@codesourcery.com>
3339 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
3341 (arm_rtx_costs_internal): Add missing semicolon after
3342 HANDLE_NARROW_SHIFT_ARITH call.
3344 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
3346 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3347 disable isel if it was not set explicitly.
3349 2017-11-06 James Bowman <james.bowman@ftdichip.com>
3351 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
3352 (add_type_attribute) likewise.
3354 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
3356 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
3357 of struct ix86_frame.
3358 (ix86_initial_elimination_offset): Likewise.
3359 (ix86_expand_split_stack_prologue): Likewise.
3361 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
3363 * tree-vrp.h (enum value_range_type): Update stale comment.
3365 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3366 Alan Hayward <alan.hayward@arm.com>
3367 David Sherwood <david.sherwood@arm.com>
3369 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
3370 (aarch64_expand_vec_perm_const): Take the number of units too.
3371 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
3372 (aarch64_expand_vec_perm_const): Likewise.
3373 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
3374 (vec_perm<mode>): Update accordingly.
3376 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3377 Alan Hayward <alan.hayward@arm.com>
3378 David Sherwood <david.sherwood@arm.com>
3380 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
3381 Take the number of units too.
3382 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
3383 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
3384 but check for a vector mode before rather than after the call.
3385 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
3386 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
3387 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
3388 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
3389 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
3390 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
3391 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
3392 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
3393 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
3394 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
3395 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
3396 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
3397 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
3398 (aarch64_sqdmull2_n<mode>): Update accordingly.
3400 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3401 Alan Hayward <alan.hayward@arm.com>
3402 David Sherwood <david.sherwood@arm.com>
3404 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
3405 the number of units too.
3406 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
3407 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
3408 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
3409 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
3410 (vec_store_lanesxi<mode>): Update accordingly.
3412 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3413 Alan Hayward <alan.hayward@arm.com>
3414 David Sherwood <david.sherwood@arm.com>
3416 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
3417 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
3418 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
3419 of units rather than the mode.
3420 * config/aarch64/iterators.md (nunits): New mode attribute.
3421 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
3422 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
3423 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
3424 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
3425 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
3426 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
3427 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
3428 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
3429 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
3430 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
3431 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3432 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
3433 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
3434 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
3435 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
3436 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
3437 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
3438 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
3439 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
3440 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
3441 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
3442 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
3443 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
3444 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3445 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
3446 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
3447 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3448 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
3449 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
3450 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
3451 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
3452 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
3453 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
3454 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
3455 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
3456 (aarch64_simd_vec_setv2di): Likewise.
3458 2017-11-06 Carl Love <cel@us.ibm.com>
3460 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
3462 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
3463 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
3465 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
3467 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
3468 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
3469 Add power 8 macro expansions.
3470 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
3471 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
3472 power 8 instructions. (VSX_XXBR): Add iterator.
3474 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
3476 * config/arm/arm.md (predicable_short_it): Change default to "no",
3477 improve documentation, remove uses that are identical to the default.
3478 (enabled_for_depr_it): Rename to enabled_for_short_it.
3479 * gcc/config/arm/arm-fixed.md (predicable_short_it):
3480 Remove default uses.
3481 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
3482 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
3483 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
3484 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
3486 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
3489 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
3490 float128 helper macros, which are no longer used after deleting
3491 the old 'q' built-in functions, and moving the round to odd
3492 built-in functions to being special built-in functions.
3493 (BU_FLOAT128_2): Likewise.
3494 (BU_FLOAT128_1_HW): Likewise.
3495 (BU_FLOAT128_2_HW): Likewise.
3496 (BU_FLOAT128_3_HW): Likewise.
3497 (FABSQ): Delete old 'q' built-in functions.
3498 (COPYSIGNQ): Likewise.
3499 (SQRTF128_ODD): Move round to odd built-in functions to be
3500 special built-in functions, so that we can handle
3501 -mabi=ieeelongdouble.
3502 (TRUNCF128_ODD): Likewise.
3503 (ADDF128_ODD): Likewise.
3504 (SUBF128_ODD): Likewise.
3505 (MULF128_ODD): Likewise.
3506 (DIVF128_ODD): Likewise.
3507 (FMAF128_ODD): Likewise.
3508 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
3509 built-in names to 'f128'.
3510 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
3511 old 'q' built-in functions, as the machine independent code for
3512 'f128' built-in functions handles this.
3513 (rs6000_expand_builtin): Add expansion for float128 round to odd
3514 functions, keying off on -mabi=ieeelongdouble of whether to use
3515 the KFmode or TFmode variant.
3516 (rs6000_init_builtins): Initialize the _Float128 round to odd
3518 * doc/extend.texi (PowerPC Built-in Functions): Document the old
3519 _Float128 'q' built-in functions are now mapped into the new
3520 'f128' built-in functions.
3522 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
3524 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
3525 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
3527 2017-11-06 David Malcolm <dmalcolm@redhat.com>
3530 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
3531 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
3532 * toplev.c: Include "ipa-fnsummary.h".
3533 (toplev::finalize): Call ipa_fnsummary_c_finalize.
3535 2017-11-06 Jakub Jelinek <jakub@redhat.com>
3537 PR tree-optimization/82838
3538 * gimple-ssa-store-merging.c
3539 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
3540 on a separate gimple_seq which is then appended to seq.
3542 2017-11-06 Jeff Law <law@redhat.com>
3545 * config/i386/i386.c (PROBE_INTERVAL): Remove.
3546 (get_probe_interval): New functions.
3547 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
3548 (ix86_adjust_stack_and_probe): Likewise.
3549 (output_adjust_stack_and_probe): Likewise.
3550 (ix86_emit_probe_stack_range): Likewise.
3551 (ix86_expand_prologue): Likewise.
3553 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3555 PR tree-optimization/82816
3556 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
3557 if the modes of the two types are the same.
3558 (convert_plusminus_to_widen): Likewise.
3560 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3562 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
3564 (usadv16qi): New define_expand.
3565 (usadv8hi): New define_expand.
3567 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
3570 * ipa-inline-transform.c (update_noncloned_frequencies): Always
3572 (inline_transform): Likewise.
3573 * predict.c (counts_to_freqs): Remove useless conditional.
3574 * profile-count.h (profile_count::apply_scale): Move sanity check.
3575 * tree-inline.c (copy_bb): Always scale.
3576 (copy_cfg_body): Likewise.
3578 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
3581 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
3583 (*cmp_ite0): Add enabled_for_depr_it attribute.
3584 (*cmp_ite1): Likewise.
3586 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
3588 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
3591 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3593 * tree-vrp.c (vrp_int_const_binop): Return true on success and
3594 return the value by pointer.
3595 (extract_range_from_multiplicative_op_1): Update accordingly.
3596 Return as soon as an operation fails.
3598 2017-11-05 Tom de Vries <tom@codesourcery.com>
3601 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
3602 (DEF_SANITIZER_BUILTIN): ... here.
3603 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
3604 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
3606 2017-11-05 Tom de Vries <tom@codesourcery.com>
3609 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
3611 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
3612 ASM_OUTPUT_BEFORE_CASE_LABEL call.
3613 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
3615 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3616 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3617 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3619 2017-11-05 Tom de Vries <tom@codesourcery.com>
3622 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
3625 2017-11-04 Michael Clark <michaeljclark@mac.com>
3627 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
3628 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
3630 (*addsi3_extended): Likewise.
3631 (*addsi3_extended2): Likewise.
3632 (<optab>si3): Likewise.
3633 (<optab>di3): Likewise.
3634 (<optab><mode>3): Likewise.
3635 (<*optabe>si3_internal): Likewise.
3636 (zero_extendqi<SUPERQI:mode>2): Likewise.
3637 (*add<mode>hi3): Likewise.
3638 (*xor<mode>hi3): Likewise.
3639 (<optab>di3): Likewise.
3640 (*<optab>si3_extend): Likewise.
3641 (*sge<u>_<X:mode><GPR:mode>): Likewise.
3642 (*slt<u>_<X:mode><GPR:mode>): Likewise.
3643 (*sle<u>_<X:mode><GPR:mode>): Likewise.
3645 2017-11-04 Andrew Waterman <andrew@sifive.com>
3647 * config/riscv/riscv.c (riscv_option_override): Conditionally set
3648 TARGET_STRICT_ALIGN based upon -mtune argument.
3650 2017-11-04 Andrew Waterman <andrew@sifive.com>
3652 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
3654 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
3656 * config/i386/i386.c (choose_basereg): Use optional scratch
3657 register and add assertion.
3658 (x86_emit_outlined_ms2sysv_save): Use scratch register when
3659 needed, and don't allocate stack.
3660 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
3661 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
3662 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
3664 2017-11-03 Jeff Law <law@redhat.com>
3666 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
3667 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
3668 to probe at the start of a noreturn function.
3670 2017-11-03 Jakub Jelinek <jakub@redhat.com>
3672 PR tree-optimization/78821
3673 * gimple-ssa-store-merging.c: Update the file comment.
3674 (MAX_STORE_ALIAS_CHECKS): Define.
3675 (struct store_operand_info): New type.
3676 (store_operand_info::store_operand_info): New constructor.
3677 (struct store_immediate_info): Add rhs_code and ops data members.
3678 (store_immediate_info::store_immediate_info): Add rhscode, op0r
3679 and op1r arguments to the ctor, initialize corresponding data members.
3680 (struct merged_store_group): Add load_align_base and load_align
3682 (merged_store_group::merged_store_group): Initialize them.
3683 (merged_store_group::do_merge): Update them.
3684 (merged_store_group::apply_stores): Pick the constant for
3685 encode_tree_to_bitpos from one of the two operands, or skip
3686 encode_tree_to_bitpos if neither operand is a constant.
3687 (class pass_store_merging): Add process_store method decl. Remove
3688 bool argument from terminate_all_aliasing_chains method decl.
3689 (pass_store_merging::terminate_all_aliasing_chains): Remove
3690 var_offset_p argument and corresponding handling.
3691 (stmts_may_clobber_ref_p): New function.
3692 (compatible_load_p): New function.
3693 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
3694 if there is overlap and rhs_code is not INTEGER_CST. For
3695 non-overlapping stores terminate group if rhs is not mergeable.
3696 (get_alias_type_for_stmts): Change first argument from
3697 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
3698 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
3699 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
3701 (get_location_for_stmts): Change first argument from
3702 auto_vec<gimple *> & to vec<gimple *> &.
3703 (struct split_store): Remove orig_stmts data member, add orig_stores.
3704 (split_store::split_store): Create orig_stores rather than orig_stmts.
3705 (find_constituent_stmts): Renamed to ...
3706 (find_constituent_stores): ... this. Change second argument from
3707 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
3708 to info structures rather than the statements.
3709 (split_group): Rename ALLOW_UNALIGNED argument to
3710 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
3711 it. Adjust find_constituent_stores caller.
3712 (imm_store_chain_info::output_merged_store): Handle rhs_code other
3713 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
3714 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
3715 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
3716 (mem_valid_for_store_merging): New function.
3717 (handled_load): New function.
3718 (pass_store_merging::process_store): New method.
3719 (pass_store_merging::execute): Use process_store method. Adjust
3720 terminate_all_aliasing_chains caller.
3722 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3724 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
3725 Return true for more constants, symbols and label references.
3726 (aarch64_valid_floating_const): Remove unused function.
3728 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3731 * config/aarch64/aarch64.c (aarch64_layout_frame):
3732 Undo forcing of LR at bottom of frame.
3734 2017-11-03 Jeff Law <law@redhat.com>
3737 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
3738 for int_registers_saved.
3740 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
3741 extracted from tree-ssa-dom.c.
3742 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
3743 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
3744 (record_equivalences_from_incoming_edge): Add additional argument
3745 to single_pred_edge_ignoring_loop_edges call.
3746 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
3747 (uncprop_dom_walker::before_dom_children): Add additional argument
3748 to single_pred_edge_ignoring_loop_edges call.
3749 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
3750 single_pred_edge_ignoring_loop_edges rather than open coding.
3751 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
3753 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
3755 * match.pd (-(-A)): Rewrite.
3757 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3759 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
3760 (rs6000_emit_int_cmove): New declaration.
3761 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
3762 (rs6000_emit_sISEL): Delete.
3763 (rs6000_emit_int_cmove): Make non-static.
3764 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
3765 instead of rs6000_emit_sISEL.
3767 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
3769 * asan.c (create_cond_insert_point): Maintain profile.
3770 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
3772 * basic-block.h (struct basic_block_def): Remove frequency.
3773 (EDGE_FREQUENCY): Use to_frequency
3774 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
3776 (find_traces): Update to use to_frequency.
3777 (find_traces_1_round): Likewise; use only IPA counts.
3778 (bb_to_key): Likewise.
3779 (connect_traces): Use IPA counts only.
3780 (copy_bb_p): Update to use to_frequency.
3781 (fix_up_crossing_landing_pad): Likewise.
3782 (sanitize_hot_paths): Likewise.
3783 * bt-load.c (basic_block_freq): Likewise.
3784 * cfg.c (init_flow): Set count_max to uninitialized.
3785 (check_bb_profile): Remove frequencies; check counts.
3786 (dump_bb_info): Do not dump frequencies.
3787 (update_bb_profile_for_threading): Update counts only.
3788 (scale_bbs_frequencies_int): Likewise.
3789 (MAX_SAFE_MULTIPLIER): Remove.
3790 (scale_bbs_frequencies_gcov_type): Update counts only.
3791 (scale_bbs_frequencies_profile_count): Update counts only.
3792 (scale_bbs_frequencies): Update counts only.
3793 * cfg.h (struct control_flow_graph): Add count-max.
3794 (update_bb_profile_for_threading): Update prototype.
3795 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
3796 (find_many_sub_basic_blocks): Likewise.
3797 * cfgcleanup.c (try_forward_edges): Likewise.
3798 (try_crossjump_to_edge): Likewise.
3799 * cfgexpand.c (expand_gimple_cond): Likewise.
3800 (expand_gimple_tailcall): Likewise.
3801 (construct_init_block): Likewise.
3802 (construct_exit_block): Likewise.
3803 * cfghooks.c (verify_flow_info): Check consistency of counts.
3804 (dump_bb_for_graph): Do not dump frequencies.
3805 (split_block_1): Do not update frequencies.
3806 (split_edge): Do not update frequencies.
3807 (make_forwarder_block): Do not update frequencies.
3808 (duplicate_block): Do not update frequencies.
3809 (account_profile_record): Do not update frequencies.
3810 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
3811 for global heuristics.
3812 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
3813 (expected_loop_iterations_unbounded): Use counts only.
3814 * cfgloopmanip.c (scale_loop_profile): Simplify.
3815 (create_empty_loop_on_edge): Simplify
3817 (duplicate_loop_to_header_edge): Simplify
3818 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
3819 (update_br_prob_note): Take care of removing note when profile
3821 (relink_block_chain): Do not dump frequency.
3822 (rtl_account_profile_record): Use to_frequency.
3823 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
3824 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
3825 (cgraph_update_edges_for_call_stmt_node): Likewise.
3826 (cgraph_edge::verify_count_and_frequency): Update.
3827 (cgraph_node::verify_node): Temporarily disable frequency verification.
3828 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
3829 to_cgraph_frequency.
3830 (cgraph_edge::rebuild_edges): Convert to ipa counts.
3831 * cgraphunit.c (init_lowered_empty_function): Do not initialize
3833 (cgraph_node::expand_thunk): Update profile.
3834 * except.c (dw2_build_landing_pads): Do not update frequency.
3835 * final.c (compute_alignments): Use to_frequency.
3836 (dump_basic_block_info): Do not dump frequency.
3837 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
3838 (dump_gimple_bb_header): Do not dump frequency.
3839 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
3841 * gimple-streamer-in.c (input_bb): Do not stream frequency.
3842 * gimple-streamer-out.c (output_bb): Do not stream frequency.
3843 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
3844 (init_before_recovery): Do not update frequency.
3845 (sched_create_recovery_edges): Do not update frequency.
3846 * hsa-gen.c (convert_switch_statements): Do not update frequency.
3847 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
3848 (ipa_cp_c_finalize): Set max_count to uninitialized.
3849 * ipa-fnsummary.c (get_minimal_bb): Use counts.
3850 (param_change_prob): Use counts.
3851 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
3853 * ipa-split.c (consider_split): Use to_frequency.
3854 (split_function): Use to_frequency.
3855 * ira-build.c (loop_compare_func): Likewise.
3856 (mark_loops_for_removal): Likewise.
3857 (mark_all_loops_for_removal): Likewise.
3858 * loop-doloop.c (doloop_modify): Do not update frequency.
3859 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
3861 * lto-streamer-in.c (input_function): Update count_max.
3862 * omp-expand.c (expand_omp_taskreg): Update count_max.
3863 * omp-simd-clone.c (simd_clone_adjust): Update profile.
3864 * predict.c (maybe_hot_frequency_p): Use to_frequency.
3865 (maybe_hot_count_p): Use ipa counts only.
3866 (maybe_hot_bb_p): Simplify.
3867 (maybe_hot_edge_p): Simplify.
3868 (probably_never_executed): Do not take frequency argument.
3869 (probably_never_executed_bb_p): Do not pass frequency.
3870 (probably_never_executed_edge_p): Likewise.
3871 (combine_predictions_for_bb): Check that profile is nonzero.
3872 (propagate_freq): Do not set frequency.
3873 (drop_profile): Simplify.
3874 (counts_to_freqs): Simplify.
3875 (expensive_function_p): Use to_frequency.
3876 (propagate_unlikely_bbs_forward): Simplify.
3877 (determine_unlikely_bbs): Simplify.
3878 (estimate_bb_frequencies): Add hack to silence graphite issues.
3879 (compute_function_frequency): Use ipa counts.
3880 (pass_profile::execute): Update.
3881 (rebuild_frequencies): Use counts only.
3882 (force_edge_cold): Use counts only.
3883 * profile-count.c (profile_count::dump): Dump new count types.
3884 (profile_count::differs_from_p): Check compatiblity.
3885 (profile_count::to_frequency): New function.
3886 (profile_count::to_cgraph_frequency): New function.
3887 * profile-count.h (struct function): Declare.
3888 (enum profile_quality): Add profile_guessed_local and
3889 profile_guessed_global0.
3890 (class profile_proability): Decrease number of bits to 29;
3891 update from_reg_br_prob_note and to_reg_br_prob_note.
3892 (class profile_count: Update comment; decrease number of bits
3893 to 61. Check compatibility.
3894 (profile_count::compatible_p): New private member function.
3895 (profile_count::ipa_p): New member function.
3896 (profile_count::operator<): Handle global zero correctly.
3897 (profile_count::operator>): Handle global zero correctly.
3898 (profile_count::operator<=): Handle global zero correctly.
3899 (profile_count::operator>=): Handle global zero correctly.
3900 (profile_count::nonzero_p): New member function.
3901 (profile_count::force_nonzero): New member function.
3902 (profile_count::max): New member function.
3903 (profile_count::apply_scale): Handle IPA scalling.
3904 (profile_count::guessed_local): New member function.
3905 (profile_count::global0): New member function.
3906 (profile_count::ipa): New member function.
3907 (profile_count::to_frequency): Declare.
3908 (profile_count::to_cgraph_frequency): Declare.
3909 * profile.c (OVERLAP_BASE): Delete.
3910 (compute_frequency_overlap): Delete.
3911 (compute_branch_probabilities): Do not use compute_frequency_overlap.
3912 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
3913 * sched-ebb.c (rank): Use counts only.
3914 * shrink-wrap.c (handle_simple_exit): Use counts only.
3915 (try_shrink_wrapping): Use counts only.
3916 (place_prologue_for_one_component): Use counts only.
3917 * tracer.c (find_best_predecessor): Use to_frequency.
3918 (find_trace): Use to_frequency.
3919 (tail_duplicate): Use to_frequency.
3920 * trans-mem.c (expand_transaction): Do not update frequency.
3921 * tree-call-cdce.c: Do not update frequency.
3922 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
3923 (gimple_merge_blocks): Likewise.
3924 (gimple_split_edge): Likewise.
3925 (gimple_duplicate_sese_region): Likewise.
3926 (gimple_duplicate_sese_tail): Likewise.
3927 (move_sese_region_to_fn): Likewise.
3928 (gimple_account_profile_record): Likewise.
3929 (insert_cond_bb): Likewise.
3930 * tree-complex.c (expand_complex_div_wide): Likewise.
3931 * tree-eh.c (lower_resx): Update profile.
3932 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
3934 (initialize_cfun): Do not initialize frequencies
3935 (freqs_to_counts): Delete.
3936 (copy_cfg_body): Ignore count parameter.
3937 (copy_body): Update.
3938 (expand_call_inline): Update count_max.
3939 (optimize_inline_calls): Update count_max.
3940 (tree_function_versioning): Update count_max.
3941 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
3942 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
3944 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
3945 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
3946 (try_peel_loop): Likewise.
3947 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
3949 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
3950 (tree_transform_and_unroll_loop): Do not use frequencies
3951 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
3952 Use reliable prediction only.
3953 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
3954 * tree-ssa-sink.c (select_best_block): Use to_frequency.
3955 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
3956 probability scaling.
3957 * tree-ssa-threadupdate.c (create_block_for_threading): Do
3958 not update frequency
3959 (any_remaining_duplicated_blocks): Likewise.
3960 (update_profile): Likewise.
3961 (estimated_freqs_path): Delete.
3962 (freqs_to_counts_path): Delete.
3963 (clear_counts_path): Delete.
3964 (ssa_fix_duplicate_block_edges): Likewise.
3965 (duplicate_thread_path): Likewise.
3966 * tree-switch-conversion.c (gen_inbound_check): Use counts.
3967 * tree-tailcall.c (decrease_profile): Do not update frequency.
3968 (eliminate_tail_call): Likewise.
3969 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
3970 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
3971 (optimize_mask_stores): Likewise.
3972 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
3973 * ubsan.c (ubsan_expand_null_ifn): Update profile.
3974 (ubsan_expand_ptr_ifn): Update profile.
3975 * value-prof.c (gimple_ic): Simplify.
3976 * value-prof.h (gimple_ic): Update prototype.
3977 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
3978 * ipa-inline.c (compute_uninlined_call_time): Be sure that
3980 (want_inline_self_recursive_call_p): Likewise.
3981 (resolve_noninline_speculation): Only cummulate defined counts.
3982 (inline_small_functions): Use nonzero_p.
3983 (ipa_inline): Do not access freed node.
3985 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3987 * config/aarch64/aarch64.c (aarch64_override_options_internal):
3988 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
3990 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
3992 * config/riscv/riscv.c (riscv_legitimize_move): Handle
3993 non-legitimate address.
3995 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3997 * config/rs6000/rs6000.md (*lt0_disi): Delete.
3998 (*lt0_<mode>di, *lt0_<mode>si): New.
4000 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4002 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
4003 TARGET_PAIRED_FLOAT.
4005 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
4006 Jim Wilson <jim.wilson@linaro.org>
4008 * config/aarch64/aarch64-cores.def (saphira): New CPU.
4009 * config/aarch64/aarch64-tune.md: Regenerated.
4010 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
4011 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
4013 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
4015 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
4016 (arc_expand_prologue): Restore blink for millicode.
4017 * config/arc/linux.h (LINK_EH_SPEC): Defined.
4019 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
4022 * config/i386/i386.c (ix86_vector_duplicate_value): Use
4023 gen_vec_duplicate after forcing the scalar into a register.
4025 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
4027 * combine (try_combine): Print the insns input to try_combine to the
4030 2017-11-02 Steve Ellcey <sellcey@cavium.com>
4033 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
4034 Remove second argument from aarch64_process_target_attr call.
4035 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
4037 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
4039 (aarch64_handle_attr_arch): Remove second argument.
4040 (aarch64_handle_attr_cpu): Ditto.
4041 (aarch64_handle_attr_tune): Ditto.
4042 (aarch64_handle_attr_isa_flags): Ditto.
4043 (aarch64_process_one_target_attr): Ditto.
4044 (aarch64_process_target_attr): Ditto.
4045 (aarch64_option_valid_attribute_p): Remove second argument.
4046 on aarch64_process_target_attr call.
4048 2017-11-02 David Malcolm <dmalcolm@redhat.com>
4050 * diagnostic.c: Include "selftest-diagnostic.h".
4051 (selftest::assert_location_text): New function.
4052 (selftest::test_diagnostic_get_location_text): New function.
4053 (selftest::diagnostic_c_tests): Call it.
4055 2017-11-02 David Malcolm <dmalcolm@redhat.com>
4057 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
4058 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
4059 (class selftest::test_diagnostic_context): Move to...
4060 * selftest-diagnostic.c: New file.
4061 * selftest-diagnostic.h: New file.
4063 2017-11-02 James Bowman <james.bowman@ftdichip.com>
4065 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
4066 offset range for FT32B.
4067 * config/ft32/ft32.h: option "mcompress" enables relaxation.
4068 * config/ft32/ft32.md: Add TARGET_NOPM.
4069 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
4070 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
4072 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
4074 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
4076 2017-11-02 Jeff Law <law@redhat.com>
4078 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
4079 virtual keyword on FINAL OVERRIDE members.
4081 * tree-ssa-propagate.h (ssa_propagation_engine): Group
4082 virtuals together. Add virtual destructor.
4083 (substitute_and_fold_engine): Similarly.
4085 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
4087 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
4089 2017-11-02 Richard Biener <rguenther@suse.de>
4091 PR tree-optimization/82795
4092 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
4094 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4096 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
4098 (gcc_GAS_CHECK_FEATURE): Remove.
4099 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
4100 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
4101 * configure: Regenerate.
4103 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
4105 * config/arc/arc.c (hwloop_optimize): Account for empty
4108 2017-11-02 Richard Biener <rguenther@suse.de>
4111 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
4112 Truncate ARRAY_REF index and element size.
4114 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
4116 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
4118 2017-11-01 Jeff Law <law@redhat.com>
4120 * tree-ssa-ccp.c (ccp_folder): New class derived from
4121 substitute_and_fold_engine.
4122 (ccp_folder::get_value): New member function.
4123 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
4124 (ccp_fold_stmt): Remove prototype.
4125 (ccp_finalize): Call substitute_and_fold from the ccp_class.
4126 * tree-ssa-copy.c (copy_folder): New class derived from
4127 substitute_and_fold_engine.
4128 (copy_folder::get_value): Renamed from get_value.
4129 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
4130 * tree-vrp.c (vrp_folder): New class derived from
4131 substitute_and_fold_engine.
4132 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
4133 (vrp_folder::get_value): New member function.
4134 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
4135 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
4136 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
4137 provide a class interface to folder/substitute routines.
4138 (ssa_prop_fold_stmt_fn): Remove typedef.
4139 (ssa_prop_get_value_fn): Likewise.
4140 (subsitute_and_fold): Remove prototype.
4141 (replace_uses_in): Likewise.
4142 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
4143 Renamed from replace_uses_in. Call the virtual member function
4144 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
4145 (substitute_and_fold_dom_walker): Remove initialization of
4146 data member entries for calbacks. Add substitute_and_fold_engine
4147 member and initialize it.
4148 (substitute_and_fold_dom_walker::before_dom_children0: Use the
4149 member functions for get_value, replace_phi_args_in c
4150 replace_uses_in, and fold_stmt calls.
4151 (substitute_and_fold_engine::substitute_and_fold): Renamed from
4152 substitute_and_fold. Remove assert. Update ctor call.
4154 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
4155 (ssa_prop_visit_phi_fn): Likewise.
4156 (class ssa_propagation_engine): New class to provide an interface
4158 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
4160 (ssa_prop_visit_phi): Likewise.
4161 (ssa_propagation_engine::simulate_stmt): Moved into class.
4162 Call visit_phi/visit_stmt from the class rather than via
4163 file scoped static variables.
4164 (ssa_propagation_engine::simulate_block): Moved into class.
4165 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
4166 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
4167 set file scoped statics for the visit_stmt/visit_phi callbacks.
4168 * tree-complex.c (complex_propagate): New class derived from
4169 ssa_propagation_engine.
4170 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
4171 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
4172 (tree_lower_complex): Call ssa_propagate via the complex_propagate
4174 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
4175 ssa_propagation_engine.
4176 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
4177 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
4178 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
4179 * tree-ssa-copy.c (copy_prop): New class derived from
4180 ssa_propagation_engine.
4181 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
4182 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
4183 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
4184 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
4185 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
4186 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
4187 (execute_vrp): Call ssa_propagate from the vrp_prop class.
4189 2017-11-01 Jakub Jelinek <jakub@redhat.com>
4191 PR rtl-optimization/82778
4192 PR rtl-optimization/82597
4193 * compare-elim.c (struct comparison): Add in_a_setter field.
4194 (find_comparison_dom_walker::before_dom_children): Remove killed
4195 bitmap and df_simulate_find_defs call, instead walk the defs.
4196 Compute last_setter and initialize in_a_setter. Merge definitions
4197 with first initialization for a few variables.
4198 (try_validate_parallel): Use insn_invalid_p instead of
4199 recog_memoized. Return insn rather than just the pattern.
4200 (try_merge_compare): Fix up comment. Don't uselessly test if
4201 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
4203 (execute_compare_elim_after_reload): Remove df_chain_add_problem
4206 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4207 Alan Hayward <alan.hayward@arm.com>
4208 David Sherwood <david.sherwood@arm.com>
4210 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
4211 aarch64_hard_regno_nregs to get the number of registers
4214 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4215 Alan Hayward <alan.hayward@arm.com>
4216 David Sherwood <david.sherwood@arm.com>
4218 * config/aarch64/constraints.md (Upl): Rename to...
4220 * config/aarch64/aarch64.md
4221 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
4224 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4225 Alan Hayward <alan.hayward@arm.com>
4226 David Sherwood <david.sherwood@arm.com>
4228 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
4229 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
4232 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4233 Alan Hayward <alan.hayward@arm.com>
4234 David Sherwood <david.sherwood@arm.com>
4236 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
4237 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
4238 (aarch64_evpc_dup): Generate rtl direcly, rather than using
4240 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
4241 of a single element.
4242 * config/aarch64/iterators.md: Add a comment above the permute
4243 unspecs to say that they are generated directly by
4244 aarch64_expand_vec_perm_const.
4245 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
4247 2017-11-01 Nathan Sidwell <nathan@acm.org>
4249 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
4251 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
4253 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
4254 and medany code models, and describe what they do.
4256 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4258 Revert accidental duplicate:
4260 * combine.c (can_change_dest_mode): Reject changes in
4261 REGMODE_NATURAL_SIZE.
4263 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
4265 PR rtl-optimization/64682
4266 PR rtl-optimization/69567
4267 PR rtl-optimization/69737
4268 PR rtl-optimization/82683
4269 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
4270 register mentioned in the note, drop the note, unless it came from I3,
4271 in which case it should go to I3 again.
4273 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4275 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
4276 and return false if not.
4277 (clear_bytes_written_by, live_bytes_read): Update accordingly.
4279 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4281 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
4282 range is known to be empty.
4284 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4285 Alan Hayward <alan.hayward@arm.com>
4286 David Sherwood <david.sherwood@arm.com>
4288 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
4289 and CONST_VECTOR_NUNITS instead of computing the number of units from
4290 the byte sizes of the vector and element.
4291 (simplify_binary_operation_1): Likewise.
4292 (simplify_const_binary_operation): Likewise.
4293 (simplify_ternary_operation): Likewise.
4295 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4296 Alan Hayward <alan.hayward@arm.com>
4297 David Sherwood <david.sherwood@arm.com>
4299 * var-tracking.c (INT_MEM_OFFSET): Replace with...
4300 (int_mem_offset): ...this new function.
4301 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
4302 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
4303 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
4306 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4307 Alan Hayward <alan.hayward@arm.com>
4308 David Sherwood <david.sherwood@arm.com>
4310 * lower-subreg.c (interesting_mode_p): New function.
4311 (compute_costs, find_decomposable_subregs, decompose_register)
4312 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
4313 (resolve_clobber, dump_choices): Use it.
4315 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4316 Alan Hayward <alan.hayward@arm.com>
4317 David Sherwood <david.sherwood@arm.com>
4319 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
4321 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4322 Alan Hayward <alan.hayward@arm.com>
4323 David Sherwood <david.sherwood@arm.com>
4325 * alias.c (find_base_value, find_base_term): Only process integer
4326 truncations. Check the precision rather than the size.
4328 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4329 Alan Hayward <alan.hayward@arm.com>
4330 David Sherwood <david.sherwood@arm.com>
4332 * machmode.h (is_narrower_int_mode): New function
4333 * optabs.c (expand_float, expand_fix): Use it.
4334 * dwarf2out.c (rotate_loc_descriptor): Likewise.
4336 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4337 Alan Hayward <alan.hayward@arm.com>
4338 David Sherwood <david.sherwood@arm.com>
4340 * rtl.h (narrower_subreg_mode): New function.
4341 * ira-color.c (update_costs_from_allocno): Use it.
4343 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4344 Alan Hayward <alan.hayward@arm.com>
4345 David Sherwood <david.sherwood@arm.com>
4347 * optabs-query.h (convert_optab_p): New function, split out from...
4348 (convert_optab_handler): ...here.
4349 (widening_optab_handler): Delete.
4350 (find_widening_optab_handler): Remove permit_non_widening parameter.
4351 (find_widening_optab_handler_and_mode): Likewise. Provide an
4352 override that operates on mode class wrappers.
4353 * optabs-query.c (widening_optab_handler): Delete.
4354 (find_widening_optab_handler_and_mode): Remove permit_non_widening
4355 parameter. Assert that the two modes are the same class and that
4356 the "from" mode is narrower than the "to" mode. Use
4357 convert_optab_handler instead of widening_optab_handler.
4358 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
4359 instead of widening_optab_handler.
4360 * expr.c (expand_expr_real_2): Update calls to
4361 find_widening_optab_handler.
4362 * optabs.c (expand_widen_pattern_expr): Likewise.
4363 (expand_binop_directly): Take the insn_code as a parameter.
4364 (expand_binop): Only call find_widening_optab_handler for
4365 conversion optabs; use optab_handler otherwise. Update calls
4366 to find_widening_optab_handler and expand_binop_directly.
4367 Use convert_optab_handler instead of widening_optab_handler.
4368 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
4369 find_widening_optab_handler and use scalar_mode rather than
4371 (convert_plusminus_to_widen): Likewise.
4373 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4374 Alan Hayward <alan.hayward@arm.com>
4375 David Sherwood <david.sherwood@arm.com>
4377 * machmode.h (fixed_size_mode): New class.
4378 * rtl.h (get_pool_mode): Return fixed_size_mode.
4379 * gengtype.c (main): Add fixed_size_mode.
4380 * target.def (get_raw_result_mode): Return a fixed_size_mode.
4381 (get_raw_arg_mode): Likewise.
4382 * doc/tm.texi: Regenerate.
4383 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
4384 * targhooks.c (default_get_reg_raw_mode): Likewise.
4385 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
4386 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
4387 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
4388 (msp430_get_raw_result_mode): Likewise.
4389 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
4390 * dbxout.c (dbxout_parms): Require fixed-size modes.
4391 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
4392 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
4393 * omp-low.c (lower_oacc_reductions): Likewise.
4394 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
4395 (simplify_subreg): Update accordingly.
4396 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
4397 (force_const_mem): Update accordingly. Return NULL_RTX for modes
4398 that aren't fixed-size.
4399 (get_pool_mode): Return a fixed_size_mode.
4400 (output_constant_pool_2): Take a fixed_size_mode.
4402 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4403 Alan Hayward <alan.hayward@arm.com>
4404 David Sherwood <david.sherwood@arm.com>
4406 * doc/rtl.texi (vec_series): Document.
4407 (const): Say that the operand can be a vec_series.
4408 * rtl.def (VEC_SERIES): New rtx code.
4409 * rtl.h (const_vec_series_p_1): Declare.
4410 (const_vec_series_p): New function.
4411 * emit-rtl.h (gen_const_vec_series): Declare.
4412 (gen_vec_series): Likewise.
4413 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
4414 (gen_vec_series): Likewise.
4415 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
4416 * simplify-rtx.c (simplify_unary_operation): Handle negations
4418 (simplify_binary_operation_series): New function.
4419 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
4420 (test_vector_ops_series): New function.
4421 (test_vector_ops): Call it.
4422 * config/powerpcspe/altivec.md (altivec_lvsl): Use
4423 gen_const_vec_series.
4424 (altivec_lvsr): Likewise.
4425 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
4427 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4428 Alan Hayward <alan.hayward@arm.com>
4429 David Sherwood <david.sherwood@arm.com>
4431 * doc/rtl.texi (const): Update description of address constants.
4432 Say that vector constants are allowed too.
4433 * common.md (E, F): Use CONSTANT_P instead of checking for
4435 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
4436 checking for CONST_VECTOR.
4437 * expmed.c (make_tree): Use build_vector_from_val for a CONST
4439 * expr.c (expand_expr_real_2): Check for vector modes instead
4440 of checking for CONST_VECTOR.
4441 * rtl.h (const_vec_p): New function.
4442 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
4443 (unwrap_const_vec_duplicate): Handle them here too.
4445 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4446 David Malcolm <dmalcolm@redhat.com>
4447 Alan Hayward <alan.hayward@arm.com>
4448 David Sherwood <david.sherwood@arm.com>
4450 * rtl.h (vec_duplicate_p): New function.
4451 * selftest-rtl.c (assert_rtx_eq_at): New function.
4452 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
4453 (assert_rtx_eq_at): Declare.
4454 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
4455 * selftest-run-tests.c (selftest::run_tests): Call it.
4456 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
4457 (simplify_unary_operation_1): Recursively handle vector duplicates.
4458 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
4460 (simplify_subreg): Handle subregs of vector duplicates.
4461 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
4462 (selftest::simplify_rtx_c_tests): New functions.
4464 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4465 Alan Hayward <alan.hayward@arm.com>
4466 David Sherwood <david.sherwood@arm.com>
4468 * emit-rtl.h (gen_const_vec_duplicate): Declare.
4469 (gen_vec_duplicate): Likewise.
4470 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
4472 (gen_const_vector): ...here.
4473 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
4474 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
4475 whose elements are all equal.
4476 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
4477 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
4478 (simplify_relational_operation): Likewise.
4479 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
4481 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
4482 (aarch64_expand_vector_init): Likewise.
4483 * config/arm/arm.c (neon_vdup_constant): Likewise.
4484 (neon_expand_vector_init): Likewise.
4485 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
4486 (arm_block_set_unaligned_vect): Likewise.
4487 (arm_block_set_aligned_vect): Likewise.
4488 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
4489 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
4490 (expand_vec_perm_even_odd_pack): Likewise.
4491 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
4492 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
4493 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
4494 gen_const_vec_duplicate.
4495 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
4496 * config/mips/mips.c (mips_gen_const_int_vector): Use
4497 gen_const_vec_duplicate.
4498 (mips_expand_vector_init): Use CONST0_RTX.
4499 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
4500 (define_split): Use gen_const_vec_duplicate.
4501 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
4502 (define_split): Use gen_const_vec_duplicate.
4503 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
4504 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
4505 * config/spu/spu.c (spu_const): Likewise.
4507 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4508 Alan Hayward <alan.hayward@arm.com>
4509 David Sherwood <david.sherwood@arm.com>
4511 * combine.c (can_change_dest_mode): Reject changes in
4512 REGMODE_NATURAL_SIZE.
4514 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
4516 * configure.ac (--enable-libssp): New.
4517 (gcc_cv_libc_provides_ssp): Check for explicit setting before
4518 trying to determine target-specific default. Adjust indentation.
4519 * configure: Regenerated.
4520 * doc/install.texi (Configuration): Expand --disable-libssp
4523 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
4525 config/i386/i386.c (ix86_expand_epilogue): Correct stack
4528 2017-10-31 Martin Jambor <mjambor@suse.cz>
4531 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
4533 2017-10-31 David Malcolm <dmalcolm@redhat.com>
4535 * auto-profile.c (autofdo_source_profile::read): Use
4536 UNKNOWN_LOCATION rather than 0.
4537 * diagnostic-core.h (warning_at_rich_loc): Rename to...
4538 (warning_at): ...this overload.
4539 (warning_at_rich_loc_n): Rename to...
4540 (warning_n): ...this overload.
4541 (error_at_rich_loc): Rename to...
4542 (error_at): ...this overload.
4543 (pedwarn_at_rich_loc): Rename to...
4544 (pedwarn): ...this overload.
4545 (permerror_at_rich_loc): Rename to...
4546 (permerror): ...this overload.
4547 (inform_at_rich_loc): Rename to...
4548 (inform): ...this overload.
4549 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
4550 (diagnostic_n_impl_richloc): Rename to...
4551 (diagnostic_n_impl): ...this rich_location *-based decl.
4552 (inform_at_rich_loc): Rename to...
4553 (inform): ...this, and add an assertion.
4554 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
4555 (warning_at_rich_loc): Rename to...
4556 (warning_at): ...this, and add an assertion.
4557 (warning_at_rich_loc_n): Rename to...
4558 (warning_n): ...this, and add an assertion.
4559 (warning_n): Update location_t-based implementation for removal of
4560 location_t-based diagnostic_n_impl.
4561 (pedwarn_at_rich_loc): Rename to...
4562 (pedwarn): ...this, and add an assertion.
4563 (permerror_at_rich_loc): Rename to...
4564 (permerror): ...this, and add an assertion.
4565 (error_n): Update for removal of location_t-based diagnostic_n_impl.
4566 (error_at_rich_loc): Rename to...
4567 (error_at): ...this, and add an assertion.
4568 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
4569 (driver::do_spec_on_infiles): Likewise.
4570 * substring-locations.c (format_warning_va): Update for renaming
4571 of inform_at_rich_loc.
4573 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
4575 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
4576 _Float<N>X built-in functions so that the variant without the
4577 "__builtin_" prefix is only enabled for the GNU C and Objective C
4578 languages when they are in non-strict ANSI/ISO mode.
4579 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
4580 * target.def (floatn_builtin_p): Add a target hook to control
4581 whether _Float<N> and _Float<N>X built-in functions without the
4582 "__builtin_" prefix are enabled, and return true for C and
4583 Objective C in the default hook. Include langhooks.h in
4585 * targhooks.h (default_floatn_builtin_p): Likewise.
4586 * targhooks.c (default_floatn_builtin_p): Likewise.
4587 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
4588 floatn_builtin_p target hook.
4589 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
4591 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
4592 Eric Botcazou <ebotcazou@adacore.com>
4594 PR rtl-optimization/81803
4595 * lra-constraints.c (curr_insn_transform): Also reload the whole
4596 register for a strict subreg no wider than a word if this is for
4597 a WORD_REGISTER_OPERATIONS target.
4599 2017-10-31 Jason Merrill <jason@redhat.com>
4601 * gdbinit.in: Skip over inlines from timevar.h.
4603 2017-10-31 Martin Liska <mliska@suse.cz>
4605 * doc/gcov.texi: Document new option.
4606 * gcov.c (print_usage): Likewise print it.
4607 (process_args): Support the argument.
4608 (format_count): New function.
4609 (format_gcov): Use the function.
4611 2017-10-31 Martin Liska <mliska@suse.cz>
4613 * gcov.c (struct name_map): do not use typedef.
4614 Define operator== and operator<.
4615 (name_search): Remove.
4616 (name_sort): Remove.
4617 (main): Do not allocate names.
4618 (process_file): Add vertical space.
4619 (generate_results): Use std::find.
4620 (release_structures): Do not release memory.
4621 (find_source): Use std::find.
4623 2017-10-31 Martin Liska <mliska@suse.cz>
4625 * gcov.c (struct line_info): Remove it's typedef.
4626 (line_info::line_info): Add proper ctor.
4627 (line_info::has_block): Do not use a typedef.
4628 (struct source_info): Do not use typedef.
4629 (circuit): Likewise.
4630 (get_cycles_count): Likewise.
4631 (output_intermediate_file): Iterate via vector iterator.
4632 (add_line_counts): Use std::vector methods.
4633 (accumulate_line_counts): Likewise.
4634 (output_lines): Likewise.
4636 2017-10-31 Martin Liska <mliska@suse.cz>
4638 * gcov.c (struct source_info): Remove typedef.
4639 (source_info::source_info): Add proper ctor.
4640 (accumulate_line_counts): Use struct, not it's typedef.
4641 (output_gcov_file): Likewise.
4642 (output_lines): Likewise.
4643 (main): Do not allocate an array.
4644 (output_intermediate_file): Use size of vector container.
4645 (process_file): Resize the vector.
4646 (generate_results): Do not preallocate, use newly added vector
4648 (release_structures): Do not release sources.
4649 (find_source): Use vector methods.
4650 (add_line_counts): Do not use typedef.
4652 2017-10-31 Martin Liska <mliska@suse.cz>
4654 * doc/gcov.texi: Document that.
4655 * gcov.c (add_line_counts): Mark lines with a non-executed
4657 (output_line_beginning): Handle such lines.
4658 (output_lines): Pass new argument.
4659 (output_intermediate_file): Print it in intermediate format.
4661 2017-10-31 Martin Liska <mliska@suse.cz>
4663 * color-macros.h: New file.
4664 * diagnostic-color.c: Factor out color related to macros to
4666 * doc/gcov.texi: Document -k option.
4667 * gcov.c (INCLUDE_STRING): Include string.h.
4668 (print_usage): Add -k option.
4669 (process_args): Parse it.
4670 (pad_count_string): New function.
4671 (output_line_beginning): Likewise.
4672 (DEFAULT_LINE_START): New macro.
4673 (output_lines): Support color output.
4675 2017-10-31 Martin Liska <mliska@suse.cz>
4677 PR gcov-profile/82633
4678 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
4679 their interaction with GCOV infrastructure.
4680 * configure.ac: Add -fkeep-{inline,static}-functions to
4682 * configure: Regenerate.
4684 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
4687 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
4689 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
4692 * config/rs6000/rs6000.md (allocate_stack): Force update interval
4693 into a register if it does not fit into an immediate offset field.
4695 2017-10-31 Olivier Hainque <hainque@adacore.com>
4697 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
4699 2017-10-31 Julia Koval <julia.koval@intel.com>
4701 * config.gcc: Add gfniintrin.h.
4702 * config/i386/gfniintrin.h: New.
4703 * config/i386/i386-builtin-types.def
4704 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
4705 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
4706 __builtin_ia32_vgf2p8affineinvqb_v32qi,
4707 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
4708 __builtin_ia32_vgf2p8affineinvqb_v16qi,
4709 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
4710 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
4711 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
4712 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
4713 V64QI_FTYPE_V64QI_V64QI_INT): New types.
4714 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4715 * config/i386/immintrin.h: Include gfniintrin.h.
4716 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
4718 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
4720 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
4722 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
4724 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
4725 (arm_ashldi3_1bit): Remove pattern.
4726 (ashrdi3): Remove shift by 1 expansion.
4727 (arm_ashrdi3_1bit): Remove pattern.
4728 (lshrdi3): Remove shift by 1 expansion.
4729 (arm_lshrdi3_1bit): Remove pattern.
4730 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
4731 cost of ashldi3 by 1.
4732 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
4733 (<shift>di3_neon): Likewise.
4735 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4737 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
4738 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
4739 and (*aarch64_simd_mov<VQ:mode>).
4740 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
4741 pattern alternative.
4742 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
4743 attributes to match pattern alternative.
4745 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
4747 * config.gcc (powerpc*-*-*): Add emmintrin.h.
4748 * config/rs6000/emmintrin.h: New file.
4749 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
4751 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
4753 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
4754 * (movdi_vfp_cortexa8): Remove pattern.
4756 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4758 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
4760 (Specific, alpha*-dec-osf5.1): Remove.
4761 (Specific, mips-sgi-irix5): Remove.
4762 (Specific, mips-sgi-irix6): Remove.
4764 2017-10-30 Jakub Jelinek <jakub@redhat.com>
4767 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
4768 arguments to clear_bit_region_be.
4770 2017-10-30 Jim Wilson <wilson@tuliptree.org>
4772 * gimplify.c: Include memmodel.h.
4774 2017-10-30 Martin Jambor <mjambor@suse.cz>
4776 * omp-grid.c (grid_attempt_target_gridification): Also insert a
4777 condition whether loop should be executed at all.
4779 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
4781 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4782 gimple folding of vec_madd() intrinsics.
4783 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
4784 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
4785 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
4787 2017-10-30 Richard Biener <rguenther@suse.de>
4789 PR tree-optimization/82762
4791 2017-10-23 Richard Biener <rguenther@suse.de>
4793 PR tree-optimization/82129
4795 2017-08-01 Richard Biener <rguenther@suse.de>
4797 PR tree-optimization/81181
4798 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4799 (compute_antic): ... end of iteration here.
4801 2017-10-30 Joseph Myers <joseph@codesourcery.com>
4803 * doc/invoke.texi (C Dialect Options): Document -std=c17,
4804 -std=iso9899:2017 and -std=gnu17.
4805 * doc/standards.texi (C Language): Document C17 support.
4806 * doc/cpp.texi (Overview): Mention -std=c17.
4807 (Standard Predefined Macros): Document C11 and C17 values of
4808 __STDC_VERSION__. Do not refer to C99 support as incomplete.
4809 * doc/extend.texi (Inline): Do not list individual options for
4810 standards newer than C99.
4811 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
4813 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
4816 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
4818 * asan.c (asan_finish_file): Align asan globals array by shadow
4821 2017-10-30 Jakub Jelinek <jakub@redhat.com>
4824 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
4825 (struct store_immediate_info): Add bitregion_start and bitregion_end
4827 (store_immediate_info::store_immediate_info): Add brs and bre
4828 arguments and initialize bitregion_{start,end} from those.
4829 (struct merged_store_group): Add bitregion_start, bitregion_end,
4830 align_base and mask fields. Drop unnecessary struct keyword from
4831 struct store_immediate_info. Add do_merge method.
4832 (clear_bit_region_be): Use memset instead of loop storing zeros.
4833 (merged_store_group::do_merge): New method.
4834 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
4835 stores as long as the surrounding bitregions have no gaps.
4836 (merged_store_group::merge_overlapping): Use do_merge.
4837 (merged_store_group::apply_stores): Test that bitregion_{start,end}
4838 is byte aligned, rather than requiring that start and width are
4839 byte aligned. Drop unnecessary struct keyword from
4840 struct store_immediate_info. Allocate and populate also mask array.
4841 Make start of the arrays relative to bitregion_start rather than
4842 start and size them according to bitregion_{end,start} difference.
4843 (struct imm_store_chain_info): Drop unnecessary struct keyword from
4844 struct store_immediate_info.
4845 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
4846 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
4847 struct keyword from struct store_immediate_info.
4848 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
4849 between stores as long as the surrounding bitregions have no gaps.
4851 (struct split_store): Add orig non-static data member.
4852 (split_store::split_store): Initialize orig to false.
4853 (find_constituent_stmts): Return store_immediate_info *, non-NULL
4854 if there is exactly a single original stmt. Change stmts argument
4855 to pointer from reference, if NULL, don't push anything to it. Add
4856 first argument, use it to optimize skipping over orig stmts that
4857 are known to be before bitpos already. Simplify.
4858 (split_group): Return unsigned int count how many stores are or
4859 would be needed rather than a bool. Add allow_unaligned argument.
4860 Change split_stores argument from reference to pointer, if NULL,
4861 only do a dry run computing how many stores would be produced.
4862 Rewritten algorithm to use both alignment and misalign if
4863 !allow_unaligned and handle bitfield stores with gaps.
4864 (imm_store_chain_info::output_merged_store): Set start_byte_pos
4865 from bitregion_start instead of start. Compute allow_unaligned
4866 here, if true, do 2 split_group dry runs to compute which one
4867 produces fewer stores and prefer aligned if equal. Punt if
4868 new count is bigger or equal than original before emitting any
4869 statements, rather than during that. Remove no longer needed
4870 new_ssa_names tracking. Replace num_stmts with
4871 split_stores.length (). Use 32-bit stack allocated entries
4872 in split_stores auto_vec. Try to reuse original store lhs/rhs1
4873 if possible. Handle bitfields with gaps.
4874 (pass_store_merging::execute): Ignore bitsize == 0 stores.
4875 Compute bitregion_{start,end} for the stores and construct
4876 store_immediate_info with that. Formatting fixes.
4878 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
4881 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
4882 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
4884 2017-10-29 Jim Wilson <wilson@tuliptree.org>
4886 * gimplify.c: Include tm_p.h.
4888 * common.opt (gcoff): Re-add as ignored option.
4889 (gcoff1, gcoff2, gcoff3): Likewise.
4891 * Makefile.in (OBJS): Delete sdbout.o.
4892 (GTFILES): Delete $(srcdir)/sdbout.c.
4893 * debug.h: Delete sdb_debug_hooks.
4894 * final.c: Delete sdbout.h include.
4895 (final_scan_insn): Delete SDB_DEBUG check.
4896 (rest_of_clean_state): Likewise.
4897 * output.h: Delete sdb_begin_function_line.
4900 * toplev.c: Delete sdbout.h include.
4901 (process_options): Delete SDB_DEBUG check.
4902 * tree-core.h (tree_type_common): Delete pointer field of
4904 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
4905 TYPE_SYMTAB_POINTER.
4906 * tree.h (TYPE_SYMTAB_POINTER): Delete.
4907 (TYPE_SYMTAB_IS_POINTER): Delete.
4908 (TYPE_SYMTAB_IS_DIE): Renumber.
4909 * xcoffout.c: Refer to former sdbout.c file.
4910 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
4912 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
4913 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
4914 (Debugging Options): Delete -gcoff.
4915 (-gstabs): Delete SDB reference.
4917 (-gcoff@var{level}): Delete.
4918 * doc/passes.texi (Debugging information output): Delete SDB and
4919 sdbout.c references.
4920 * doc/tm.texi: Regenerate.
4921 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
4922 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
4924 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
4925 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
4926 (SDB_DEBUGGING_INFO): Delete.
4927 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
4928 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
4929 * target.def (output_source_filename): Delete COFF reference.
4931 * common.opt (gcoff): Delete.
4932 (gxcoff+): Update Negative chain.
4933 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
4935 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
4936 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
4937 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
4939 (expand_function_start): Change sdb reference to past tense.
4940 (expand_function_end): Change sdb reference to past tense.
4941 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
4942 * opts.c (debug_type_names): Delete coff entry.
4943 (common_handle_option): Delete OPT_gcoff case.
4944 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
4946 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
4947 * config/cris/cris.h: Delete SDB reference in comment.
4948 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
4949 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
4950 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
4951 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
4953 (ix86_expand_prologue): Likewise.
4954 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
4955 * config/ia64/ia64.h: Likewise.
4956 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
4957 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
4959 * config/mmix/mmix.h: Likewise.
4960 * config/nds32/nds32.c: Likewise.
4961 * config/stormy/storym16.h: Likewise.
4962 * config/visium/visium.h: Likewise.
4963 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
4965 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
4967 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
4968 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
4969 FRAME_POINTER_REGNUM point at high end of local var area.
4971 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
4973 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
4974 Move comment around. Do not reset best_edge for a copiable
4975 destination if the copy would cause a partition change.
4976 (better_edge_p): Remove redundant check.
4978 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
4980 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
4982 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4984 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
4985 for math functions that have _Float<N> and _Float<N>X variants.
4986 (mathfn_built_in_2): Add support for math functions that have
4987 _Float<N> and _Float<N>X variants.
4988 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
4989 (expand_builtin_mathfn_ternary): Add support for fma with
4990 _Float<N> and _Float<N>X variants.
4991 (expand_builtin): Likewise.
4992 (fold_builtin_3): Likewise.
4993 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
4994 create math function _Float<N> and _Float<N>X variants as external
4996 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
4997 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
4998 the __builtin_ prefix and if not strict ansi, without the prefix.
4999 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
5000 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
5001 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
5002 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
5003 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
5004 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
5005 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
5006 function signatures for fma _Float<N> and _Float<N>X variants.
5007 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
5008 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
5009 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
5010 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
5011 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
5012 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
5013 * gencfn-macros.c (print_case_cfn): Add support for math functions
5014 that have _Float<N> and _Float<N>X variants.
5015 (print_define_operator_list): Likewise.
5016 (fltfn_suffixes): Likewise.
5018 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
5019 for math functions that have _Float<N> and _Float<N>X variants.
5020 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
5021 and _Float<N>X variants.
5022 (COPYSIGN): Likewise.
5025 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
5026 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
5028 (integer_valued_read_call_p): Likewise.
5029 * fold-const-call.c (fold_const_call_ss): Likewise.
5030 (fold_const_call_sss): Add support for copysign, fmin, and fmax
5031 _Float<N> and _Float<N>X variants.
5032 (fold_const_call_ssss): Add support for fma _Float<N> and
5033 _Float<N>X variants.
5034 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
5035 support for copysign and fma _Float<N> and _Float<N>X variants.
5036 (backprop::process_builtin_call_use): Likewise.
5037 * tree-call-cdce.c (can_test_argument_range); Add support for
5038 sqrt _Float<N> and _Float<N>X variants.
5039 (edom_only_function): Likewise.
5040 (get_no_error_domain): Likewise.
5041 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
5042 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
5043 copysign _Float<N> and _Float<N>X variants.
5044 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
5045 handled by machine independent code.
5046 (FMAF128): Likewise.
5047 * doc/cpp.texi (Common Predefined Macros): Document defining
5048 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
5049 fma _Float<N> and _Float<N>X variants.
5051 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
5054 * config/i386/i386-modes.def (CCFPU): Remove definition.
5055 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
5056 (ix86_cc_modes_compatible): Ditto.
5057 (ix86_expand_carry_flag_compare): Ditto.
5058 (ix86_expand_int_movcc): Ditto.
5059 (ix86_expand_int_addcc): Ditto.
5060 (ix86_reverse_condition): Ditto.
5061 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
5062 Return true/false for unordered/ordered fp comparisons.
5063 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
5064 (ix86_prepare_fp_compare_args): Update for rename.
5065 (ix86_expand_fp_compare): Update for rename. Generate unordered
5066 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
5067 (ix86_expand_sse_compare_and_jump): Ditto.
5068 * config/i386/predicates.md (fcmov_comparison_operator):
5069 Remove CCFPU mode handling.
5070 (ix86_comparison_operator): Ditto.
5071 (ix86_carry_flag_operator): Ditto.
5072 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
5073 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
5074 (*cmpu<mode>_cc_i387): Ditto.
5075 (FPCMP): Remove mode iterator.
5076 (unord): Remove mode attribute.
5077 (unord_subst): New define_subst transformation
5078 (unord): New define_subst attribute.
5080 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
5081 (*cmpi<unord>xf_i387): Ditto.
5082 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
5083 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
5084 using unord_subst transformation.
5085 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
5086 (round_saeonly): Also handle CCFP mode.
5087 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
5088 Remove UNSPEC_SAHF unspec handling.
5090 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
5092 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
5094 2017-10-27 Jeff Law <law@redhat.com>
5096 * gimple-ssa-sprintf.c: Include domwalk.h.
5097 (class sprintf_dom_walker): New class, derived from dom_walker.
5098 (sprintf_dom_walker::before_dom_children): New function.
5099 (struct call_info): Moved into sprintf_dom_walker class
5100 (compute_formath_length, handle_gimple_call): Likewise.
5101 (sprintf_length::execute): Call the dominator walker rather
5102 than walking the statements.
5104 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
5105 gimple statement locations.
5106 (check_array_bounds): Corresponding changes. Get the statement's
5107 location directly from wi->stmt.
5109 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
5112 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
5114 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
5116 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
5117 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
5119 2017-10-27 Jakub Jelinek <jakub@redhat.com>
5122 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
5123 * config/i386/i386.c (maybe_get_pool_constant): Removed.
5124 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
5125 maybe_get_pool_constant.
5126 * config/i386/predicates.md (zero_extended_scalar_load_operand):
5129 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5131 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
5132 2.26 caveat. Update gas and gld versions.
5133 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
5136 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5138 * cgraph.h (set_malloc_flag): Declare.
5139 * cgraph.c (set_malloc_flag_1): New function.
5140 (set_malloc_flag): Likewise.
5141 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
5142 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
5144 (read_ipa_call_summary): Add support for reading is_return_callee.
5145 (write_ipa_call_summary): Stream is_return_callee.
5146 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
5147 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
5148 ipa-prop.h, ipa-fnsummary.h.
5149 (pure_const_names): Change to static.
5150 (malloc_state_e): Define.
5151 (malloc_state_names): Define.
5152 (funct_state_d): Add field malloc_state.
5153 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
5154 (check_retval_uses): New function.
5155 (malloc_candidate_p): Likewise.
5156 (analyze_function): Add support for malloc attribute.
5157 (pure_const_write_summary): Stream malloc_state.
5158 (pure_const_read_summary): Add support for reading malloc_state.
5159 (dump_malloc_lattice): New function.
5160 (propagate_malloc): New function.
5161 (warn_function_malloc): New function.
5162 (ipa_pure_const::execute): Call propagate_malloc and
5163 ipa_free_fn_summary.
5164 (pass_local_pure_const::execute): Add support for malloc attribute.
5165 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
5166 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
5168 2017-10-27 Martin Liska <mliska@suse.cz>
5170 PR gcov-profile/82457
5171 * doc/invoke.texi: Document that one needs a non-strict ISO mode
5172 for fork-like functions to be properly instrumented.
5174 2017-10-27 Richard Biener <rguenther@suse.de>
5177 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
5178 info when we redirected EH.
5180 2017-10-26 Michael Collison <michael.collison@arm.com>
5182 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
5184 (<optab>_trunchf<GPI:mode>2: New pattern.
5185 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
5186 * config/aarch64/iterators.md (wv): New mode attribute.
5187 (vf, VF): New mode attributes.
5188 (vgp, VGP): New mode attributes.
5189 (s): Update attribute with SImode and DImode prefixes.
5191 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
5193 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
5194 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
5195 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
5196 (nios2_option_overide): Initialize it. Don't allow R0-relative
5197 addressing with PIC.
5198 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
5199 (nios2_symbolic_constant_p): Likewise.
5200 (nios2_legitimate_address_p): Likewise.
5201 (nios2_r0rel_section_name_p): New.
5202 (nios2_symbol_ref_in_r0rel_data_p): New.
5203 (nios2_emit_move_sequence): Handle r0rel_constant_p.
5204 (r0rel_constant_p): New.
5205 (nios2_print_operand_address): Handle r0rel_constant_p.
5206 (nios2_cdx_narrow_form_p): Likewise.
5207 * config/nios2/nios2.opt (mr0rel-sec=): New option.
5208 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
5209 (Nios II Options): Document -mr0rel-sec.
5211 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
5213 * config/nios2/nios2.c: Include xregex.h.
5214 (nios2_gprel_sec_regex): New.
5215 (nios2_option_overide): Initialize it. Don't allow GP-relative
5216 addressing with PIC.
5217 (nios2_small_section_name_p): Check for regex match.
5218 * config/nios2/nios2.opt (mgprel-sec=): New option.
5219 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
5220 (Nios II Options): Document -mgprel-sec.
5222 2017-10-26 Jim Wilson <wilson@tuliptree.org>
5224 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
5226 2017-10-26 Tom de Vries <tom@codesourcery.com>
5228 PR tree-optimization/82707
5229 * gimple.c (gimple_copy): Fix unsharing of
5230 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
5232 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
5234 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
5235 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
5236 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
5237 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
5238 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
5239 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
5240 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
5241 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
5242 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
5243 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
5244 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
5245 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
5246 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
5247 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
5248 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
5249 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
5250 _mm512_mask_cmpunord_ps_mask): New intrinsics.
5252 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5254 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
5256 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
5257 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
5258 warning to rs6000.c. Remove the Undocumented flag, since it has
5260 (-mabi=ibmlongdouble): Likewise.
5261 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
5262 already set, set the default format for long double.
5263 (rs6000_debug_reg_global): Print whether long double is IBM or
5265 (rs6000_option_override_internal): Rework setting long double
5266 format. Only warn if the user is changing the long double default
5267 and they did not use -Wno-psabi.
5268 * doc/invoke.texi (PowerPC options): Update the documentation for
5269 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
5271 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5272 Alan Hayward <alan.hayward@arm.com>
5273 David Sherwood <david.sherwood@arm.com>
5275 * rtl.h (wider_subreg_mode): New function.
5276 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
5277 rather than an unsigned int *.
5278 * ira-color.c (regno_max_ref_width): Replace with...
5279 (regno_max_ref_mode): ...this new variable.
5280 (coalesced_pseudo_reg_slot_compare): Update accordingly.
5281 Use wider_subreg_mode.
5282 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
5283 rather than an unsigned int *.
5284 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
5285 (process_alt_operands): Likewise.
5286 (invariant_p): Likewise.
5287 * lra-spills.c (assign_mem_slot): Likewise.
5288 (add_pseudo_to_slot): Likewise.
5289 * lra.c (collect_non_operand_hard_regs): Likewise.
5290 (add_regs_to_insn_regno_info): Likewise.
5291 * reload1.c (regno_max_ref_width): Replace with...
5292 (regno_max_ref_mode): ...this new variable.
5293 (reload): Update accordingly. Update call to
5294 ira_sort_regnos_for_alter_reg.
5295 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
5296 (init_eliminable_invariants): Update to use regno_max_ref_mode.
5297 (scan_paradoxical_subregs): Likewise.
5299 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5301 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
5302 (aarch64_frame): Add emit_frame_chain boolean.
5303 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5304 Move eh_return case to aarch64_layout_frame.
5305 (aarch64_layout_frame): Initialize emit_frame_chain.
5306 (aarch64_expand_prologue): Use emit_frame_chain.
5308 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5310 * config/aarch64/aarch64.c (aarch64_layout_frame):
5311 Ensure LR is always stored at the bottom of the callee-saves.
5312 Remove rarely used frame layout which saves callee-saves at top of
5313 frame, so the store of LR can be used as a valid probe in all cases.
5315 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5317 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
5318 Improve unaligned TImode/TFmode base/offset split.
5320 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5321 Alan Hayward <alan.hayward@arm.com>
5322 David Sherwood <david.sherwood@arm.com>
5324 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
5325 * combine.c (find_single_use_1): Likewise.
5326 (expand_field_assignment): Likewise.
5327 (move_deaths): Likewise.
5328 * lra-constraints.c (simplify_operand_subreg): Likewise.
5329 (curr_insn_transform): Likewise.
5330 * lra.c (collect_non_operand_hard_regs): Likewise.
5331 (add_regs_to_insn_regno_info): Likewise.
5332 * rtlanal.c (reg_referenced_p): Likewise.
5333 (covers_regno_no_parallel_p): Likewise.
5335 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5337 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
5338 Don't print any bits outside the precision of the value.
5339 * wide-int.cc (test_printing): Add some new tests.
5341 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5343 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
5344 supports -xbrace_comment option.
5345 * configure: Regenerate.
5346 * config.in: Regenerate.
5347 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
5348 (ASM_CPU_SPEC): Use it.
5350 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5352 * target.def (static_rtx_alignment): New hook.
5353 * targhooks.h (default_static_rtx_alignment): Declare.
5354 * targhooks.c (default_static_rtx_alignment): New function.
5355 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
5356 * doc/tm.texi: Regenerate.
5357 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
5358 instead of targetm.constant_alignment. Remove call to
5360 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5361 (cris_preferred_mininum_alignment): New function, split out from...
5362 (cris_constant_alignment): ...here.
5363 (cris_static_rtx_alignment): New function.
5364 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
5366 (ix86_constant_alignment): ...here.
5367 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5368 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5369 (mmix_static_rtx_alignment): New function.
5370 * config/spu/spu.c (spu_static_rtx_alignment): New function.
5371 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5373 2017-10-26 Tamar Christina <tamar.christina@arm.com>
5376 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
5377 Add flag_trapping_math and flag_fp_int_builtin_inexact.
5379 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
5381 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
5382 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
5383 mark as a sign-extending load.
5384 (local_pic_load_u): Define.
5386 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
5389 * fold-const.c (operand_equal_for_comparison_p): Also return true
5390 if ARG0 is a simple variant of ARG1 with narrower precision.
5391 (fold_ternary_loc): Always pass unstripped operands to the predicate.
5393 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
5395 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
5397 * i386.h (processor_costs): Add gather_static, gather_per_elt,
5398 scatter_static, scatter_per_elt.
5399 * x86-tune-costs.h: Add new cost entries.
5401 2017-10-25 Richard Biener <rguenther@suse.de>
5403 * tree-ssa-sccvn.h (vn_eliminate): Declare.
5404 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
5405 class pass_fre): Move to ...
5406 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
5407 class pass_fre): ... here and adjust for statistics.
5409 2017-10-25 Jakub Jelinek <jakub@redhat.com>
5412 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
5413 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
5414 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
5415 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
5418 2017-10-25 Richard Biener <rguenther@suse.de>
5420 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
5421 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
5422 eliminate_push_avail, eliminate_insert): Move inside...
5423 (class eliminate_dom_walker): ... this class in preparation
5425 (fini_eliminate): Remove by merging with ...
5426 (eliminate): ... this function. Adjust for class changes.
5427 (pass_pre::execute): Remove fini_eliminate call.
5428 (pass_fre::execute): Likewise.
5430 2017-10-24 Jakub Jelinek <jakub@redhat.com>
5433 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
5434 (VPERMI2, VPERMI2I): New mode iterators.
5435 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
5436 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
5438 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
5439 mode iterator. Remove 3 old define_insn patterns.
5440 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
5441 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
5442 VPERMI2 mode iterator, remove the other two expanders.
5443 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
5444 to use VPERMI2 mode iterator, add another alternative for vpermi2*
5445 instructions, remove the other two patterns.
5446 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
5447 mode iterator, remove the other two patterns.
5448 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
5449 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
5450 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
5451 and adjust argument order accordingly.
5452 (ix86_expand_vec_perm): Adjust caller.
5453 (expand_vec_perm_1): Likewise.
5454 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
5455 (expand_vec_perm_vpermt2_vpshub2): ... this.
5456 (ix86_expand_vec_perm_const_1): Adjust caller.
5457 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
5460 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
5461 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
5462 (vec_shl_<mode>): Remove unused expander.
5463 (avx512bw_<shift_insn><mode>3): New define_insn.
5464 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
5465 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
5467 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
5470 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
5473 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
5475 PR rtl-optimization/82396
5476 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
5477 (autopref_multipass_init): Simplify initialization.
5478 (autopref_rank_data): Simplify sort order.
5479 * gcc/sched-int.h (autopref_multipass_data_): Remove
5480 multi_mem_insn_p, min_offset and max_offset.
5482 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
5485 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5486 Check special value of flag_omit_frame_pointer.
5487 (aarch64_can_eliminate): Likewise.
5488 (aarch64_override_options_after_change_1): Simplify handling of
5489 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
5491 2017-10-24 Richard Biener <rguenther@suse.de>
5493 PR tree-optimization/82697
5494 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
5495 zero for conditional load and unconditional store.
5497 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
5499 * doc/install.texi: Document bootstrap-cet.
5501 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
5504 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
5505 ENDBR instruction at function entrance if function is only
5508 2017-10-24 Jakub Jelinek <jakub@redhat.com>
5511 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
5512 patterns to better describe from which operation the CF is computed.
5513 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
5514 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
5515 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
5516 is 0, use _0 suffixed expanders instead of emitting a comparison
5519 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
5521 * config/i386/i386.md(*movsf_internal, *movdf_internal):
5522 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
5524 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
5527 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
5528 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
5529 * loop-iv.c (iv_get_reaching_def): Likewise.
5530 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
5531 variable is promoted and the partition contains undefined values.
5533 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5535 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
5537 (nios2_address_cost): Define.
5538 (nios2_legitimize_address): Recognize (exp + constant) directly.
5539 (TARGET_ADDRESS_COST): Define.
5541 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5543 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
5544 (nios2_symbolic_memory_operand_p): Declare.
5545 (nios2_split_large_constant): Declare.
5546 (nios2_split_symbolic_memory_operand): Declare.
5547 * config/nios2/nios2.c: Adjust includes.
5548 (nios2_symbolic_constant_allowed): New.
5549 (nios2_symbolic_constant_p): New.
5550 (nios2_plus_symbolic_constant_p): New.
5551 (nios2_valid_addr_expr_p): Recognize addresses involving
5553 (nios2_legitimate_address_p): Likewise, also LO_SUM.
5554 (nios2_symbolic_memory_operand_p): New.
5555 (nios2_large_constant_p): New.
5556 (nios2_split_large_constant): New.
5557 (nios2_split_plus_large_constant): New.
5558 (nios2_split_symbolic_memory_operand): New.
5559 (nios2_legitimize_address): Code refactoring. Handle addresses
5560 involving symbolic constants.
5561 (nios2_emit_move_sequence): Likewise.
5562 (nios2_print_operand): Improve error output.
5563 (nios2_print_operand_address): Handle LO_SUM.
5564 (nios2_cdx_narrow_form_p): Likewise.
5565 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
5566 operands involving symbolic constants.
5567 (movhi_internal, movsi_internal): Likewise.
5568 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
5569 (extendhisi2, extendqi<mode>2): Likewise.
5571 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5573 * tree-pass.h (PROP_rtl_split_insns): Define.
5574 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
5576 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5578 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
5580 2017-10-23 Jakub Jelinek <jakub@redhat.com>
5583 * target.def (const_not_ok_for_debug_p): Default to
5584 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
5585 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
5586 * targhooks.c (default_const_not_ok_for_debug_p): New function.
5587 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
5588 which targetm.const_not_ok_for_debug_p returned true.
5589 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
5591 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
5593 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
5594 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
5595 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
5597 (ix86_const_not_ok_for_debug_p): New function.
5598 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
5599 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
5601 2017-10-23 David Malcolm <dmalcolm@redhat.com>
5604 * system.h: Conditionally include "unique-ptr.h" if
5605 INCLUDE_UNIQUE_PTR is defined.
5606 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
5607 of defining INCLUDE_UNIQUE_PTR before including "system.h".
5609 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
5611 * config/rl78/rl78.md: New define_expand "subdi3".
5613 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
5616 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
5617 DF_REF_INSN if DF_REF_INSN_INFO is false.
5619 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
5621 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
5622 xmm_move instead of sse_move.
5623 (sse_store_index): New function.
5624 (ix86_register_move_cost): Be more sensible about mismatch stall;
5625 model AVX moves correctly; make difference between sse->integer and
5627 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
5628 moves; make difference between SSE and AVX.
5629 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
5630 and zmm_move. Increase size of sse load and store tables;
5631 add unaligned load and store tables; add ssemmx_to_integer.
5632 * x86-tune-costs.h: Update all entries according to real
5633 move latencies from Agner Fog's manual and chip documentation.
5635 2017-10-23 Jakub Jelinek <jakub@redhat.com>
5638 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
5639 * config/i386/constraints.md (Wf): New constraint.
5640 * config/i386/i386.md (UNSPEC_SBB): New unspec.
5641 (cmp<dwi>_doubleword): Removed.
5642 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
5643 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
5644 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
5645 expand with cmp<dwi>_doubleword. For LTU and GEU use
5646 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
5648 * common.opt (gcolumn-info): Enable by default.
5649 * doc/invoke.texi (gcolumn-info): Document new default.
5651 2017-10-23 Richard Biener <rguenther@suse.de>
5653 PR tree-optimization/82672
5654 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
5655 Fold the stmt if we propagated into it.
5657 2017-10-23 Richard Biener <rguenther@suse.de>
5659 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
5660 (bitmap_remove_expr_from_set): ... this. All callers call this
5661 for non-constant values.
5662 (bitmap_set_subtract): Rename to...
5663 (bitmap_set_subtract_expressions): ... this. Adjust and
5665 (bitmap_set_contains_value): Remove superfluous check.
5666 (bitmap_set_replace_value): Inline into single caller ...
5667 (bitmap_value_replace_in_set): ... here and simplify.
5668 (dependent_clean): Merge into ...
5669 (clean): ... this using an overload. Adjust.
5670 (prune_clobbered_mems): Adjust.
5671 (compute_antic_aux): Likewise.
5672 (compute_partial_antic_aux): Likewise.
5674 2017-10-23 Richard Biener <rguenther@suse.de>
5676 PR tree-optimization/82129
5678 2017-08-01 Richard Biener <rguenther@suse.de>
5680 PR tree-optimization/81181
5681 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5682 (compute_antic): ... end of iteration here.
5684 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5686 * target.def (starting_frame_offset): New hook.
5687 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
5688 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
5689 * doc/tm.texi.in: Regenerate.
5690 * hooks.h (hook_hwi_void_0): Declare.
5691 * hooks.c (hook_hwi_void_0): New function.
5692 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
5693 STARTING_FRAME_OFFSET.
5694 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
5695 * reload1.c (reload): Likewise.
5696 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
5697 instead of STARTING_FRAME_OFFSET.
5698 * function.c (try_fit_stack_local): Likewise.
5699 (assign_stack_local_1): Likewise
5700 (instantiate_virtual_regs): Likewise.
5701 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
5702 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
5703 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
5704 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
5705 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
5706 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
5707 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
5708 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
5709 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
5710 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
5711 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
5712 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
5713 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
5714 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
5715 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
5716 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
5717 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
5718 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
5719 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
5720 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
5721 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
5722 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
5723 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
5724 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
5725 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
5726 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
5727 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
5728 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
5729 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
5730 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
5731 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
5732 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
5733 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
5734 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
5735 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
5736 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
5737 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
5738 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
5739 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
5740 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
5741 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
5742 * config/avr/avr.c (avr_starting_frame_offset): Make static and
5743 return a HOST_WIDE_INT.
5744 (avr_builtin_setjmp_frame_value): Use it instead of
5745 STARTING_FRAME_OFFSET.
5746 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5747 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
5748 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
5750 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5751 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
5752 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
5753 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5754 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
5755 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
5756 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5757 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
5758 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
5759 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5760 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
5761 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
5763 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5764 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
5765 * config/mips/mips.c (mips_compute_frame_info): Refer to
5766 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
5767 (mips_starting_frame_offset): New function.
5768 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5769 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
5770 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
5771 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
5772 and return a HOST_WIDE_INT.
5773 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5774 (mmix_initial_elimination_offset): Refer to
5775 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
5776 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
5777 * config/pa/pa.c (pa_starting_frame_offset): New function.
5778 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
5779 (pa_expand_prologue): Likewise.
5780 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5781 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
5782 !FRAME_GROWS_DOWNWARD handling to...
5783 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5784 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
5785 !FRAME_GROWS_DOWNWARD handling to...
5786 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5787 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
5788 !FRAME_GROWS_DOWNWARD handling to...
5789 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5790 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
5792 (rs6000_starting_frame_offset): New function.
5793 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
5794 !FRAME_GROWS_DOWNWARD handling to...
5795 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5796 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
5797 !FRAME_GROWS_DOWNWARD handling to...
5798 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5799 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
5800 !FRAME_GROWS_DOWNWARD handling to...
5801 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5802 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
5803 (rs6000_starting_frame_offset): New function.
5804 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
5805 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
5806 * config/vax/vax.c (vax_starting_frame_offset): New function.
5807 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
5808 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5809 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
5810 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
5811 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5812 * system.h (STARTING_FRAME_OFFSET): Poison.
5814 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5816 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
5817 SCALAR_TYPE_MODE instead of TYPE_MODE.
5819 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5820 Alan Hayward <alan.hayward@arm.com>
5821 David Sherwood <david.sherwood@arm.com>
5823 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
5825 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5826 Alan Hayward <alan.hayward@arm.com>
5827 David Sherwood <david.sherwood@arm.com>
5829 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
5831 2017-10-23 Richard Biener <rguenther@suse.de>
5833 PR tree-optimization/82129
5834 * tree-ssa-pre.c (bitmap_set_and): Remove.
5835 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
5836 canonicalizing expressions in the set to those with lowest
5837 ID rather than taking that from the first edge.
5839 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5841 * combine.c (rtx_equal_for_field_assignment_p): Use
5842 byte_lowpart_offset.
5844 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5845 Alan Hayward <alan.hayward@arm.com>
5846 David Sherwood <david.sherwood@arm.com>
5848 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
5849 to a SUBREG_PROMOTED_VAR.
5851 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5852 Alan Hayward <alan.hayward@arm.com>
5853 David Sherwood <david.sherwood@arm.com>
5855 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
5856 (expand_debug_source_expr): Likewise.
5857 * combine.c (combine_simplify_rtx): Likewise.
5858 * cse.c (fold_rtx): Likewise.
5859 * optabs.c (expand_float): Likewise.
5860 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5861 (simplify_binary_operation_1): Likewise.
5863 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5864 Alan Hayward <alan.hayward@arm.com>
5865 David Sherwood <david.sherwood@arm.com>
5867 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
5868 (record_promoted_value): Likewise.
5869 * expr.c (expand_expr_real_2): Likewise.
5870 * ree.c (update_reg_equal_equiv_notes): Likewise.
5871 (combine_set_extension): Likewise.
5872 * rtlanal.c (low_bitmask_len): Likewise.
5873 * simplify-rtx.c (neg_const_int): Likewise.
5874 (simplify_binary_operation_1): Likewise.
5876 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5877 Alan Hayward <alan.hayward@arm.com>
5878 David Sherwood <david.sherwood@arm.com>
5880 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
5881 * regcprop.c (maybe_mode_change): Likewise.
5882 * reload1.c (alter_reg): Likewise.
5884 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5886 * inchash.h (inchash::hash::add_wide_int): New function.
5887 * lto-streamer-out.c (hash_tree): Use it.
5889 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5891 * inchash.h (inchash::hash::add_wide_int): Rename to...
5892 (inchash::hash::add_hwi): ...this.
5893 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
5894 (polymorphic_call_target_hasher::hash): Likewise.
5895 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
5896 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
5897 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
5898 * lto-streamer-out.c (hash_tree): Likewise.
5899 * optc-save-gen.awk: Likewise.
5900 * tree.c (add_expr): Likewise.
5902 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
5905 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
5906 for ordered inequality comparisons even with TARGET_IEEE_FP.
5908 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
5911 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
5912 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5913 Expand with cmp<dwi>_doubleword.
5915 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5917 * extend.texi: Add x86 specific to 'nocf_check' attribute.
5918 List CET intrinsics.
5919 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
5920 specific to -fcf-protection option.
5922 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5924 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
5925 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
5926 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
5927 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
5928 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
5929 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
5930 (extra_objs): Add cet.o for Linux/x86 targets.
5931 (tmake_file): Add i386/t-cet for Linux/x86 targets.
5932 * config/i386/cet.c: New file.
5933 * config/i386/cetintrin.h: Likewise.
5934 * config/i386/t-cet: Likewise.
5935 * config/i386/cpuid.h (bit_SHSTK): New.
5936 (bit_IBT): Likewise.
5937 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
5938 pass IBT and SHSTK bits.
5939 * config/i386/i386-builtin-types.def
5940 (VOID_FTYPE_UNSIGNED_PVOID): New.
5941 (VOID_FTYPE_UINT64_PVOID): Likewise.
5942 * config/i386/i386-builtin.def: Add CET intrinsics.
5943 * config/i386/i386-c.c (ix86_target_macros_internal): Add
5944 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
5945 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
5946 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
5948 * config/i386/i386.c (rest_of_insert_endbranch): New.
5949 (pass_data_insert_endbranch): Likewise.
5950 (pass_insert_endbranch): Likewise.
5951 (make_pass_insert_endbranch): Likewise.
5952 (ix86_notrack_prefixed_insn_p): Likewise.
5953 (ix86_target_string): Add -mibt, -mshstk flags.
5954 (ix86_option_override_internal): Add flag_cf_protection
5956 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
5957 (ix86_print_operand): Add 'notrack' prefix output.
5958 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
5959 (ix86_expand_builtin): Expand CET intrinsics.
5960 (x86_output_mi_thunk): Add 'endbranch' instruction.
5961 * config/i386/i386.h (TARGET_IBT): New.
5962 (TARGET_IBT_P): Likewise.
5963 (TARGET_SHSTK): Likewise.
5964 (TARGET_SHSTK_P): Likewise.
5965 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
5966 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
5967 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
5968 (builtin_setjmp_setup): New pattern.
5969 (builtin_longjmp): Likewise.
5970 (rdssp<mode>): Likewise.
5971 (incssp<mode>): Likewise.
5972 (saveprevssp): Likewise.
5973 (rstorssp): Likewise.
5974 (wrss<mode>): Likewise.
5975 (wruss<mode>): Likewise.
5976 (setssbsy): Likewise.
5977 (clrssbsy): Likewise.
5978 (nop_endbr): Likewise.
5979 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
5981 * config/i386/immintrin.h: Include <cetintrin.h>.
5982 * config/i386/linux-common.h
5983 (file_end_indicate_exec_stack_and_cet): New prototype.
5984 (TARGET_ASM_FILE_END): New.
5986 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
5988 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
5989 latencies instead of having separate table; make difference between
5990 integer and float costs.
5991 * i386.h (processor_costs): Remove scalar_stmt_cost,
5992 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
5993 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
5995 * x86-tune-costs.h: Remove entries which has been removed in
5996 procesor_costs from all tables; make cond_taken_branch_cost
5997 and cond_not_taken_branch_cost COST_N_INSNS based.
5999 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
6001 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
6003 2017-10-20 Jakub Jelinek <jakub@redhat.com>
6005 * config/i386/i386.md (isa): Remove fma_avx512f.
6006 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
6007 <avx512>_fmadd_<mode>_mask3<round_name>,
6008 <avx512>_fmsub_<mode>_mask<round_name>,
6009 <avx512>_fmsub_<mode>_mask3<round_name>,
6010 <avx512>_fnmadd_<mode>_mask<round_name>,
6011 <avx512>_fnmadd_<mode>_mask3<round_name>,
6012 <avx512>_fnmsub_<mode>_mask<round_name>,
6013 <avx512>_fnmsub_<mode>_mask3<round_name>,
6014 <avx512>_fmaddsub_<mode>_mask<round_name>,
6015 <avx512>_fmaddsub_<mode>_mask3<round_name>,
6016 <avx512>_fmsubadd_<mode>_mask<round_name>,
6017 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
6018 (*vec_widen_umult_even_v16si<mask_name>,
6019 *vec_widen_smult_even_v16si<mask_name>): Likewise.
6020 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
6022 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6024 * extend.texi: Add 'nocf_check' documentation.
6025 * gimple.texi: Add second parameter to
6026 gimple_build_call_from_tree.
6027 * invoke.texi: Add -fcf-protection documentation.
6028 * rtl.texi: Add REG_CALL_NOTRACK documenation.
6030 2017-10-20 Richard Biener <rguenther@suse.de>
6032 PR tree-optimization/82473
6033 * tree-vect-loop.c (vectorizable_reduction): Properly get at
6034 the largest input type.
6036 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6038 * c-attribs.c (handle_nocf_check_attribute): New function.
6039 (c_common_attribute_table): Add 'nocf_check' handling.
6040 * gimple-parser.c: Add second argument NULL to
6041 gimple_build_call_from_tree.
6042 * attrib.c (comp_type_attributes): Check nocf_check attribute.
6043 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
6045 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
6046 * common.opt: Add fcf-protection flag.
6047 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
6048 * flag-types.h: Add enum cf_protection_level.
6049 * gimple.c (gimple_build_call_from_tree): Add second parameter.
6050 Add 'nocf_check' attribute propagation to gimple call.
6051 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
6052 (gimple_build_call_from_tree): Update prototype.
6053 (gimple_call_nocf_check_p): New function.
6054 (gimple_call_set_nocf_check): Likewise.
6055 * gimplify.c: Add second argument to gimple_build_call_from_tree.
6056 * ipa-icf.c: Add nocf_check attribute in statement hash.
6057 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
6058 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
6059 * toplev.c (process_options): Add flag_cf_protection handling.
6061 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6063 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
6065 2017-10-20 Richard Biener <rguenther@suse.de>
6067 PR tree-optimization/82603
6068 * tree-if-conv.c (predicate_mem_writes): Make sure to only
6069 remove false predicated stores.
6071 2017-10-20 Richard Biener <rguenther@suse.de>
6073 * graphite-isl-ast-to-gimple.c
6074 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
6075 Remove return value and simplify, dump copied stmt after lhs
6077 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
6078 Reduce dump verbosity.
6079 (gsi_insert_earliest): Likewise.
6080 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
6081 * graphite.c (print_global_statistics): Adjust dumping.
6082 (print_graphite_scop_statistics): Likewise.
6083 (print_graphite_statistics): Do not dump loops here.
6084 (graphite_transform_loops): But here.
6086 2017-10-20 Nicolas Roche <roche@adacore.com>
6088 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
6089 * configure: Regenerate.
6091 2017-10-20 Jakub Jelinek <jakub@redhat.com>
6094 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
6095 functions when optimizing replace GIMPLE_RETURN stmts with
6096 calls to __builtin_unreachable ().
6099 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
6100 for -fsanitize=thread link of executables.
6101 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
6102 link of executables.
6105 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
6107 (<shift_insn><mode>3<mask_name>): Change the last of the 3
6108 define_insns for logical vector shifts to use VI248_AVX512BW
6109 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
6110 condition, useless isa and prefix attributes. Change the first
6111 2 of these define_insns to ...
6112 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
6113 define_insn for avx512vl.
6114 (<shift_insn><mode>3): ... and this, new define_insn without
6115 masking for non-avx512vl.
6118 * config/i386/sse.md (*andnot<mode>3,
6119 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
6120 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
6121 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
6122 not applied use empty suffix even for TARGET_AVX512VL.
6123 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
6124 is applied, supply evex,evex or evex,evex,evex instead of just
6127 2017-10-20 Julia Koval <julia.koval@intel.com>
6129 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
6130 (OPTION_MASK_ISA_GFNI_UNSET): New.
6131 (ix86_handle_option): Handle OPT_mgfni.
6132 * config/i386/cpuid.h (bit_GFNI): New.
6133 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
6134 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
6135 * config/i386/i386.c (ix86_target_string): Add -mgfni.
6136 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
6137 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
6138 * config/i386/i386.opt: Add mgfni.
6140 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
6142 * config/msp430/msp430.c (msp430_option_override): Disable
6143 -fdelete-null-pointer-checks.
6144 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
6146 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6148 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
6149 of x87 and SSE instructions.
6151 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6153 * asan.c (create_cond_insert_point): Do not update edge count.
6154 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
6155 (afdo_propagate_circuit): Likewise.
6156 (afdo_calculate_branch_prob): Likewise.
6157 (afdo_annotate_cfg): Likewise.
6158 * basic-block.h (struct edge_def): Remove count.
6159 (edge_def::count): New accessor.
6160 * bb-reorder.c (rotate_loop): Update.
6161 (find_traces_1_round): Update.
6162 (connect_traces): Update.
6163 (sanitize_hot_paths): Update.
6164 * cfg.c (unchecked_make_edge): Update.
6165 (make_single_succ_edge): Update.
6166 (check_bb_profile): Update.
6167 (dump_edge_info): Update.
6168 (update_bb_profile_for_threading): Update.
6169 (scale_bbs_frequencies_int): Update.
6170 (scale_bbs_frequencies_gcov_type): Update.
6171 (scale_bbs_frequencies_profile_count): Update.
6172 (scale_bbs_frequencies): Update.
6173 * cfganal.c (connect_infinite_loops_to_exit): Update.
6174 * cfgbuild.c (compute_outgoing_frequencies): Update.
6175 (find_many_sub_basic_blocks): Update.
6176 * cfgcleanup.c (try_forward_edges): Update.
6177 (try_crossjump_to_edge): Update
6178 * cfgexpand.c (expand_gimple_cond): Update
6179 (expand_gimple_tailcall): Update
6180 (construct_exit_block): Update
6181 * cfghooks.c (verify_flow_info): Update
6182 (redirect_edge_succ_nodup): Update
6183 (split_edge): Update
6184 (make_forwarder_block): Update
6185 (duplicate_block): Update
6186 (account_profile_record): Update
6187 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6188 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6189 * cfgloopmanip.c (scale_loop_profile): Update.
6191 (lv_adjust_loop_entry_edge): Update.
6192 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
6193 (force_nonfallthru_and_redirect): Update.
6194 (purge_dead_edges): Update.
6195 (rtl_flow_call_edges_add): Update.
6196 * cgraphunit.c (init_lowered_empty_function): Update.
6197 (cgraph_node::expand_thunk): Update.
6198 * gimple-pretty-print.c (dump_probability): Update.
6199 (dump_edge_probability): Update.
6200 * gimple-ssa-isolate-paths.c (isolate_path): Update.
6201 * haifa-sched.c (sched_create_recovery_edges): Update.
6202 * hsa-gen.c (convert_switch_statements): Update.
6203 * ifcvt.c (dead_or_predicable): Update.
6204 * ipa-inline-transform.c (inline_transform): Update.
6205 * ipa-split.c (split_function): Update.
6206 * ipa-utils.c (ipa_merge_profiles): Update.
6207 * loop-doloop.c (add_test): Update.
6208 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
6209 * lto-streamer-in.c (input_cfg): Update.
6210 (input_function): Update.
6211 * lto-streamer-out.c (output_cfg): Update.
6212 * modulo-sched.c (sms_schedule): Update.
6213 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6214 * predict.c (maybe_hot_edge_p): Update.
6215 (unlikely_executed_edge_p): Update.
6216 (probably_never_executed_edge_p): Update.
6217 (dump_prediction): Update.
6218 (drop_profile): Update.
6219 (propagate_unlikely_bbs_forward): Update.
6220 (determine_unlikely_bbs): Update.
6221 (force_edge_cold): Update.
6222 * profile.c (compute_branch_probabilities): Update.
6223 * reg-stack.c (better_edge): Update.
6224 * shrink-wrap.c (handle_simple_exit): Update.
6225 * tracer.c (better_p): Update.
6226 * trans-mem.c (expand_transaction): Update.
6227 (split_bb_make_tm_edge): Update.
6228 * tree-call-cdce.c: Update.
6229 * tree-cfg.c (gimple_find_sub_bbs): Update.
6230 (gimple_split_edge): Update.
6231 (gimple_duplicate_sese_region): Update.
6232 (gimple_duplicate_sese_tail): Update.
6233 (gimple_flow_call_edges_add): Update.
6234 (insert_cond_bb): Update.
6235 (execute_fixup_cfg): Update.
6236 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
6237 * tree-complex.c (expand_complex_div_wide): Update.
6238 * tree-eh.c (lower_resx): Update.
6239 (unsplit_eh): Update.
6240 (cleanup_empty_eh_move_lp): Update.
6241 * tree-inline.c (copy_edges_for_bb): Update.
6242 (freqs_to_counts): Update.
6243 (copy_cfg_body): Update.
6244 * tree-ssa-dce.c (remove_dead_stmt): Update.
6245 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6246 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
6247 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
6248 (unloop_loops): Update.
6249 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6250 * tree-ssa-loop-split.c (connect_loops): Update.
6251 (split_loop): Update.
6252 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6253 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
6254 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
6255 * tree-ssa-reassoc.c (branch_fixup): Update.
6256 * tree-ssa-tail-merge.c (replace_block_by): Update.
6257 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
6258 (compute_path_counts): Update.
6259 (update_profile): Update.
6260 (recompute_probabilities): Update.
6261 (update_joiner_offpath_counts): Update.
6262 (estimated_freqs_path): Update.
6263 (freqs_to_counts_path): Update.
6264 (clear_counts_path): Update.
6265 (ssa_fix_duplicate_block_edges): Update.
6266 (duplicate_thread_path): Update.
6267 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
6268 (case_bit_test_cmp): Update.
6269 (collect_switch_conv_info): Update.
6270 (gen_inbound_check): Update.
6271 (do_jump_if_equal): Update.
6272 (emit_cmp_and_jump_insns): Update.
6273 * tree-tailcall.c (decrease_profile): Update.
6274 (eliminate_tail_call): Update.
6275 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6276 (vect_do_peeling): Update.
6277 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6278 * ubsan.c (ubsan_expand_null_ifn): Update.
6279 (ubsan_expand_ptr_ifn): Update.
6280 * value-prof.c (gimple_divmod_fixed_value): Update.
6281 (gimple_mod_pow2): Update.
6282 (gimple_mod_subtract): Update.
6283 (gimple_ic): Update.
6284 (gimple_stringop_fixed_value): Update.
6286 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
6289 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
6291 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
6293 PR rtl-optimization/82395
6294 * ira-color.c (allocno_priority_compare_func): Fix comparison step
6295 based on non_spilled_static_chain_regno_p.
6297 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
6299 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
6300 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
6301 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
6303 2017-10-19 Martin Sebor <msebor@redhat.com>
6305 PR tree-optimization/82596
6306 * tree.c (array_at_struct_end_p): Handle STRING_CST.
6308 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
6310 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
6311 (get_mem_refs_of_builtin_call): Likewise.
6312 * builtins.c (expand_builtin_apply): Adjust call to
6313 allocate_dynamic_stack_space.
6314 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
6315 the third argument to allocate_dynamic_stack_space, otherwise -1.
6316 (expand_builtin): Deal with all alloca variants.
6317 (is_inexpensive_builtin): Likewise.
6318 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
6319 * calls.c (special_function_p): Deal with all alloca variants.
6320 (initialize_argument_information): Adjust call to
6321 allocate_dynamic_stack_space.
6322 (expand_call): Likewise.
6323 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
6324 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
6325 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
6326 use it for the stack usage computation.
6327 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
6328 * function.c (gimplify_parameters): Call build_alloca_call_expr.
6329 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
6330 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
6331 (in_loop_p): Remove first argument and useless check.
6332 (pass_walloca::execute): Remove useless test and adjust call to above.
6333 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
6334 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
6335 (gimplify_call_expr): Deal with all alloca variants.
6336 * hsa-gen.c (gen_hsa_alloca): Likewise.
6337 (gen_hsa_insns_for_call): Likewise.
6338 * ipa-pure-const.c (special_builtin_state): Likewise.
6339 * tree-chkp.c (chkp_build_returned_bound): Likewise.
6340 * tree-object-size.c (alloc_object_size): Likewise.
6341 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
6342 (call_may_clobber_ref_p_1): Likewise.
6343 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
6344 (ccp_fold_stmt): Likewise.
6345 (optimize_stack_restore): Likewise.
6346 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
6347 (mark_all_reaching_defs_necessary_1): Likewise.
6348 (propagate_necessity): Likewise.
6349 (eliminate_unnecessary_stmts): Likewise.
6350 * tree.c (build_common_builtin_nodes): Build
6351 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
6352 (build_alloca_call_expr): New function.
6353 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
6354 (CASE_BUILT_IN_ALLOCA): Likewise.
6355 (build_alloca_call_expr): Declare.
6356 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
6358 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
6361 * dwarf2out.c (new_die_raw): New static inline function.
6362 (new_die): Use it to create the DIE.
6363 (add_AT_external_die_ref): Likewise.
6364 (clone_die): Likewise.
6365 (clone_as_declaration): Likewise.
6366 (dwarf2out_vms_debug_main_pointer): Likewise.
6367 (base_type_die): Likewise. Remove early return for corner cases.
6368 Do not call add_pubtype on the DIE here.
6369 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
6370 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
6371 typedefs for base types with DW_AT_endianity. Make sure a DIE with
6372 native order exists for base types, attach the DIE manually and call
6373 add_pubtype on it. Do not equate a reverse order DIE to the type.
6375 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
6377 * config/arm/arm.c (align_ok_ldrd_strd): New function.
6378 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
6380 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
6382 2017-10-19 Jakub Jelinek <jakub@redhat.com>
6384 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
6385 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
6386 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
6387 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
6388 * opts.c (sanitizer_opts): Add builtin.
6389 * ubsan.c (instrument_builtin): New function.
6390 (pass_ubsan::execute): Call it.
6391 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
6392 * doc/invoke.texi: Document -fsanitize=builtin.
6394 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
6395 builtins, store max (log2 (align), 0) into uchar field instead of
6396 align into uptr field.
6397 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
6398 store uchar 0 field instead of uptr 0 field.
6399 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
6400 instead of passing one address of struct with 2 locations pass
6401 two addresses of structs with 1 location each.
6402 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
6403 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
6404 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
6405 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
6406 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
6407 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
6408 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
6409 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
6411 2017-10-19 Martin Liska <mliska@suse.cz>
6414 * file-find.c (remove_prefix): Remove.
6415 * file-find.h (remove_prefix): Likewise.
6416 * gcc-ar.c: Remove smartness of lookup.
6418 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
6420 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
6421 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
6422 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
6424 2017-10-19 Jakub Jelinek <jakub@redhat.com>
6427 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
6428 (setcc + and to xor + setcc): New peephole2.
6430 2017-10-19 Tom de Vries <tom@codesourcery.com>
6432 * doc/sourcebuild.texi (Test Directives, Variants of
6433 dg-require-support): Add dg-require-stack-size.
6435 2017-10-19 Martin Liska <mliska@suse.cz>
6438 * gimplify.c (gimplify_decl_expr): Do not instrument variables
6439 that have a large alignment.
6440 (gimplify_target_expr): Likewise.
6442 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
6444 PR rtl-optimization/82602
6445 * ira.c (rtx_moveable_p): Return false for volatile asm.
6447 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
6450 * config/i386/i386-modes.def (CCGZ): New CC mode.
6451 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
6452 * config/i386/predicates.md (ix86_comparison_operator):
6454 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
6455 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
6456 with double-word subtraction.
6457 (put_condition_code): Handle CCGZmode.
6459 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
6461 * wide-int.cc (debug (const wide_int &)): New.
6462 (debug (const wide_int *)): New.
6463 (debug (const widest_int &)): New.
6464 (debug (const widest_int *)): New.
6466 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
6469 * lra-constraints.c (curr_insn_transform): Use non-input operand
6470 instead of output one for matched reload.
6472 2017-10-18 Bin Cheng <bin.cheng@arm.com>
6474 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
6475 (tree-ssa-loop-ivopts.h): New header file.
6476 (struct builtin_info): New fields.
6477 (classify_builtin_1): Compute and record base and offset parts for
6478 memset builtin partition by calling strip_offset.
6479 (offset_cmp, fuse_memset_builtins): New functions.
6480 (finalize_partitions): Fuse adjacent memset partitions by calling
6482 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
6483 Expose the interface.
6484 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
6486 2017-10-18 Bin Cheng <bin.cheng@arm.com>
6488 PR tree-optimization/82574
6489 * tree-loop-distribution.c (find_single_drs): New parameter. Check
6490 that data reference must be executed exactly once per iteration
6491 against the outermost loop in nest.
6492 (classify_partition): Update call to above function.
6494 2017-10-18 Richard Biener <rguenther@suse.de>
6496 PR tree-optimization/82591
6497 * graphite.c (graphite_transform_loops): Move code gen message
6499 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6500 Here. Handle scop_to_isl_ast failing.
6501 (scop_to_isl_ast): Limit the number of ISL operations.
6503 2017-10-18 Richard Biener <rguenther@suse.de>
6505 * graphite-isl-ast-to-gimple.c
6506 (translate_isl_ast_to_gimple::set_rename): Simplify.
6507 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
6508 (graphite_copy_stmts_from_block): ... here.
6509 (copy_bb_and_scalar_dependences): Simplify.
6510 (add_parameters_to_ivs_params): Canonicalize.
6511 (generate_entry_out_of_ssa_copies): Simplify.
6512 * graphite-sese-to-poly.c (extract_affine_name): Simplify
6513 by passing in ISL dimension.
6514 (parameter_index_in_region_1): Rename to ...
6515 (parameter_index_in_region): ... this.
6516 (extract_affine): Adjust assert, pass down parameter index.
6517 (add_param_constraints): Use range-info when available.
6518 (build_scop_context): Adjust.
6519 * sese.c (new_sese_info): Adjust.
6520 (free_sese_info): Likewise.
6521 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
6522 Remove unused typedefs.
6523 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
6525 2017-10-18 Martin Liska <mliska@suse.cz>
6527 * combine.c (simplify_compare_const): Add gcc_fallthrough.
6529 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
6531 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
6532 (s390_sched_init): Do not reset s390_sched_state if we entered the
6533 current basic block via a fallthru edge and all others are unlikely.
6535 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
6537 * config/s390/s390.c (NUM_SIDES): New variable.
6538 (LONGRUNNING_THRESHOLD): New variable.
6539 (LATENCY_FACTOR): New variable.
6540 (s390_sched_score): Decrease score for long-running instructions on
6542 (s390_sched_variable_issue): Perform bookkeeping for long-running
6545 2017-10-18 Richard Biener <rguenther@suse.de>
6547 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
6548 Simplify with removal of the parameter rename map.
6549 (set_rename): Likewise.
6550 (should_copy_to_new_region): Likewise.
6551 (graphite_copy_stmts_from_block): Likewise.
6552 (copy_bb_and_scalar_dependences): Remove initialization of
6553 unused copied_bb_map.
6555 (copy_internal_parameters): Likewise.
6556 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
6557 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
6558 Use INTEGRAL_TYPE_P.
6559 (parameter_index_in_region_1): Rename to ...
6560 (assign_parameter_index_in_region): ... this. Assert we have
6561 a parameter we handle.
6562 (scan_tree_for_params): Adjust.
6563 * sese.h (parameter_rename_map_t): Remove.
6564 (struct sese_info_t): Remove unused parameter_rename_map and
6565 copied_bb_map members.
6566 * sese.c (new_sese_info): Adjust.
6567 (free_sese_info): Likewise.
6569 2017-10-18 Martin Liska <mliska@suse.cz>
6572 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
6573 on an abnormal edge.
6575 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
6577 * doc/invoke.texi (ffunction-sections and fdata-sections):
6580 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
6582 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
6583 the use statement can throw internally.
6585 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
6587 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
6588 any RTX present on the RHS of a SET.
6589 * compare-elim.c (try_eliminate_compare): Restore comment.
6591 2017-10-17 Jakub Jelinek <jakub@redhat.com>
6593 * langhooks.h (struct lang_hooks): Document that tree_size langhook
6594 may be also called on tcc_type nodes.
6595 * langhooks.c (lhd_tree_size): Likewise.
6597 2017-10-17 David Malcolm <dmalcolm@redhat.com>
6599 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
6600 format_warning_at_substring.
6601 (maybe_warn): Convert source_range * param to a location_t. Pass
6602 UNKNOWN_LOCATION rather than NULL to fmtwarn.
6603 (format_directive): Remove code to extract source_ranges and
6604 source_range * in favor of just a location_t.
6605 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
6607 * substring-locations.c (format_warning_va): Convert
6608 source_range * param to a location_t.
6609 (format_warning_at_substring): Likewise.
6610 * substring-locations.h (format_warning_va): Likewise.
6611 (format_warning_at_substring): Likewise.
6613 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
6615 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
6617 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
6618 and scatter/gather ops.
6620 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
6621 vec_gather_load and vec_scatter_store.
6622 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
6623 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
6624 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
6625 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
6626 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
6627 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
6629 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
6631 * reg-stack.c (compare_for_stack_reg): Add bool argument.
6632 Detect FTST instruction and handle its register pops. Only pop
6633 second operand if can_pop_second_op is true.
6634 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
6635 set can_pop_second_op to false in the compare_for_stack_reg call.
6637 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
6638 output_fp_compare for stack register operands.
6639 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
6640 instructions here. Do not emit stack register pops here. Assert
6641 that FCOMPP pops next to top stack register. Rewrite function.
6643 2017-10-17 Nathan Sidwell <nathan@acm.org>
6646 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
6647 use DECL_ASSEMBLER_NAME_RAW.
6650 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
6653 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
6654 Wilco Dijkstra <wilco.dijkstra@arm.com>
6656 * builtins.c (expand_builtin_update_setjmp_buf): Add a
6657 converstion to Pmode from the buf_addr.
6659 2017-10-17 Richard Biener <rguenther@suse.de>
6661 * graphite-dependences.c (scop_get_reads_and_writes): Change
6662 output parameters to references.
6664 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
6666 PR 71026/tree-optimization
6667 * fold-const.c (distribute_real_division): Removed.
6668 (fold_binary_loc): Remove calls to distribute_real_divison.
6670 2017-10-17 Richard Biener <rguenther@suse.de>
6672 * graphite-scop-detection.c
6673 (scop_detection::stmt_has_simple_data_refs_p): Always use
6674 the full nest as region.
6675 (try_generate_gimple_bb): Likewise.
6676 * sese.c (scalar_evolution_in_region): Simplify now that
6677 SCEV can handle instantiation in regions.
6678 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
6679 in the non-loop part of a function if requested.
6681 2017-10-17 Richard Biener <rguenther@suse.de>
6683 PR tree-optimization/82563
6684 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
6686 (graphite_regenerate_ast_isl): Call it.
6687 * graphite-scop-detection.c (build_scops): Remove entry edge split.
6689 2017-10-17 Jakub Jelinek <jakub@redhat.com>
6691 PR tree-optimization/82549
6692 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
6693 Formatting fixes. Instead of calling make_bit_field_ref with negative
6696 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
6698 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
6699 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
6700 _mm_maskz_reduce_ss): New.
6701 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
6702 __builtin_ia32_reducess_mask): Ditto..
6703 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
6704 * config/i386/sse.md (reduces<mode>): Renamed to ...
6705 (reduces<mode><mask_scalar_name>): ... this.
6706 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
6708 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
6709 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
6711 2017-10-16 David Malcolm <dmalcolm@redhat.com>
6713 * Makefile.in (OBJS): Add unique-ptr-tests.o.
6714 * selftest-run-tests.c (selftest::run_tests): Call
6715 selftest::unique_ptr_tests_cc_tests.
6716 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
6717 * unique-ptr-tests.cc: New file.
6719 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
6722 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6724 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6725 (make_hard_regno_born, make_hard_regno_dead): Update
6726 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
6728 2017-10-16 Jeff Law <law@redhat.com>
6730 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
6732 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
6734 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
6736 2017-10-16 Olivier Hainque <hainque@adacore.com>
6738 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
6739 with_cpu if we were configured for an e500v2 target cpu name.
6741 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
6743 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
6744 * doc/invoke.texi: Document +nodsp as a valid extension for
6747 2017-10-16 Martin Liska <mliska@suse.cz>
6749 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
6750 (test_set_range): Likewise.
6751 (test_range_functions): Rename to ...
6752 (test_bit_in_range): ... this.
6753 (sbitmap_c_tests): Add new test.
6755 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6757 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
6759 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
6760 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
6762 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6764 * config/aarch64/aarch64-builtins.c
6765 (aarch64_types_quadopu_lane_qualifiers): New.
6766 (TYPES_QUADOPU_LANE): New.
6767 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
6768 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
6769 (aarch64_<sur>dot_laneq<vsi2qi>): New.
6770 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
6771 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
6772 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
6773 (Vdottype, DOTPROD): New.
6774 (sur): Add SDOT and UDOT.
6776 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6778 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
6779 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
6780 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
6782 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
6783 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
6784 Enable TARGET_DOTPROD.
6785 (cortex-a75.cortex-a55): Likewise.
6786 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
6788 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6790 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
6791 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
6793 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
6795 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
6796 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
6797 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
6798 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
6799 * config/arm/types.md (neon_dot, neon_dot_q): New.
6800 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
6802 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6804 * config/arm/arm.h (TARGET_DOTPROD): New.
6805 * config/arm/arm.c (arm_arch_dotprod): New.
6806 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
6807 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
6808 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
6809 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
6810 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
6811 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
6812 * doc/invoke.texi (armv8.2-a): Document dotprod
6814 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
6816 * i386.c (ix86_vec_cost): New function.
6817 (ix86_rtx_costs): Handle vector operations better.
6818 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
6819 * x86-tune-costs.h: Add new costs to all tables.
6821 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
6823 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
6825 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
6826 divsd, sqrtss and sqrtsd
6827 * x86-tune-costs.h: Add new entries to all costs.
6828 (znver1_cost): Fix to match real instruction latencies.
6830 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6831 Michael Collison <michael.collison@arm.com>
6833 * compare-elim.c: Include emit-rtl.h.
6834 (can_merge_compare_into_arith): New function.
6835 (try_validate_parallel): Likewise.
6836 (try_merge_compare): Likewise.
6837 (try_eliminate_compare): Call the above when no previous clobber
6839 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
6842 2017-10-14 Jakub Jelinek <jakub@redhat.com>
6846 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
6847 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
6851 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
6852 to be any operand_equal_p operands. For & (B - 1) require
6853 B to be power of 2. Recognize
6854 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
6856 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6859 * optabs.c (expand_memory_blockage): Fix call of
6860 targetm.have_memory_blockage.
6862 2017-10-14 Jakub Jelinek <jakub@redhat.com>
6865 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
6866 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
6867 objects to extra_objs instead of overwriting it.
6869 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6871 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
6872 Use any_fp_register_operand as operand[3] predicate. Simplify
6873 equality test for operands[2] and operands[4] memory location.
6874 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
6875 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
6876 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
6877 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
6878 any_fp_register_operand as operand[1] predicate. Simplify
6879 equality test for operands[0] and operands[3] memory location.
6880 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
6881 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
6882 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
6884 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6886 * target-insns.def: Add memory_blockage.
6887 * optabs.c (expand_memory_blockage): New function.
6888 (expand_asm_memory_barrier): Rename ...
6889 (expand_asm_memory_blockage): ... to this.
6890 (expand_mem_thread_fence): Call expand_memory_blockage
6891 instead of expand_asm_memory_barrier.
6892 (expand_mem_singnal_fence): Ditto.
6893 (expand_atomic_load): Ditto.
6894 (expand_atomic_store): Ditto.
6895 * doc/md.texi (Standard Pattern Names For Generation):
6896 Document memory_blockage instruction pattern.
6898 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
6900 * config/rl78/rl78.c (rl78_emit_libcall): New function.
6901 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
6902 * config/rl78/rl78.md: New define_expand "adddi3".
6904 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
6906 * cfghooks.c (verify_flow_info): Disable check that all probabilities
6909 2017-10-13 Jeff Law <law@redhat.com>
6911 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
6913 2017-10-13 Jakub Jelinek <jakub@redhat.com>
6916 * internal-fn.c (expand_mul_overflow): If both operands have
6917 the same highpart of -1 or 0 and the topmost bit of lowpart
6918 is different, overflow is if res <= 0 rather than res < 0.
6920 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
6922 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
6923 TARGET_P9_VECTOR code for unaligned_load case.
6925 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
6927 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
6929 2017-10-13 Nathan Sidwell <nathan@acm.org>
6931 * tree-core.h (tree_contains_struct): Make bool.
6932 * tree.c (tree_contains_struct): Likewise.
6933 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
6934 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
6935 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
6936 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
6937 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
6939 2017-10-13 Richard Biener <rguenther@suse.de>
6941 * graphite-isl-ast-to-gimple.c
6942 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
6943 parameters and dominance check.
6944 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
6945 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
6946 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
6947 Do not update SSA form here or do intermediate IL verification.
6948 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
6949 (graphite_initialize): Remove check on the number of loops in
6950 the function and inline into graphite_transform_loops.
6951 (graphite_finalize): Inline into graphite_transform_loops.
6952 (graphite_transform_loops): Perform SSA update and IL verification
6954 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
6956 2017-10-13 Richard Biener <rguenther@suse.de>
6958 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
6959 graphite_expression_type_precision): Avoid global constructor
6961 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
6962 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
6963 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
6964 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
6965 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
6966 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
6968 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
6971 * config/i386/i386.h (ix86_red_zone_size): New.
6972 * config/i386/i386.md (push peephole2s): Replace
6973 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
6975 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6976 Alan Hayward <alan.hayward@arm.com>
6977 David Sherwood <david.sherwood@arm.com>
6979 * combine.c (can_change_dest_mode): Reject changes in
6980 REGMODE_NATURAL_SIZE.
6982 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6983 Alan Hayward <alan.hayward@arm.com>
6984 David Sherwood <david.sherwood@arm.com>
6986 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
6987 (expand_debug_source_expr): Likewise.
6988 * combine.c (combine_simplify_rtx): Likewise.
6989 * cse.c (fold_rtx): Likewise.
6990 * fwprop.c (canonicalize_address): Likewise.
6991 * targhooks.c (default_shift_truncation_mask): Likewise.
6993 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6994 Alan Hayward <alan.hayward@arm.com>
6995 David Sherwood <david.sherwood@arm.com>
6997 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
6998 (widened_mode): Likewise.
6999 (expand_unop): Likewise.
7000 * ree.c (transform_ifelse): Likewise.
7001 (merge_def_and_ext): Likewise.
7002 (combine_reaching_defs): Likewise.
7003 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7005 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7006 Alan Hayward <alan.hayward@arm.com>
7007 David Sherwood <david.sherwood@arm.com>
7009 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
7010 * combine.c (gen_lowpart_for_combine): Likewise.
7011 * dwarf2out.c (rtl_for_decl_location): Likewise.
7012 * final.c (alter_subreg): Likewise.
7013 * rtlhooks.c (gen_lowpart_general): Likewise.
7014 (gen_lowpart_if_possible): Likewise.
7016 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7017 Alan Hayward <alan.hayward@arm.com>
7018 David Sherwood <david.sherwood@arm.com>
7020 * calls.c (expand_call): Use subreg_lowpart_offset.
7021 * cse.c (cse_insn): Likewise.
7022 * regcprop.c (copy_value): Likewise.
7023 (copyprop_hardreg_forward_1): Likewise.
7025 2017-10-13 Jakub Jelinek <jakub@redhat.com>
7028 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
7029 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
7030 =Q constraints to +Q and into insn condition add check
7031 that operands[0] and operands[1] are equal.
7032 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
7033 =Q constraints to +Q and into insn condition add check
7034 that operands[0] is equal to either operands[1] or operands[2].
7037 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
7038 instead of handling MINUS_EXPR twice (once for each argument),
7039 canonicalize operand order and handle just once, use rtype where
7040 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
7043 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
7044 any values of __C while still being pattern recognizable as a simple
7047 2017-10-13 Richard Biener <rguenther@suse.de>
7049 PR tree-optimization/82451
7051 2017-10-02 Richard Biener <rguenther@suse.de>
7053 PR tree-optimization/82355
7054 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7055 a mapping for the enclosing loop but avoid generating one for
7057 (copy_bb_and_scalar_dependences): Remove premature codegen
7058 error on PHIs in blocks duplicated into multiple places.
7059 * graphite-scop-detection.c
7060 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7061 in the region use it as loop and nest to analyze the DR in.
7062 (try_generate_gimple_bb): Likewise.
7063 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7064 (add_loop_constraints): For blocks in a loop not in the region
7065 create a dimension with a single iteration.
7066 * sese.h (gbb_loop_at_index): Remove assert.
7068 * cfgloop.c (loop_preheader_edge): For the loop tree root
7069 return the single successor of the entry block.
7070 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7071 Reset the SCEV hashtable and niters.
7072 * graphite-scop-detection.c
7073 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
7074 assert that we only have POLYNOMIAL_CHREC that vary in loops
7075 contained in the region.
7076 (scop_detection::graphite_can_represent_expr): Adjust.
7077 (scop_detection::stmt_has_simple_data_refs_p): For loops
7078 not in the region set loop to NULL. The nest is now the
7079 entry edge to the region.
7080 (try_generate_gimple_bb): Likewise.
7081 * sese.c (scalar_evolution_in_region): Adjust for
7082 instantiate_scev change.
7083 * tree-data-ref.h (graphite_find_data_references_in_stmt):
7084 Make nest parameter the edge into the region.
7085 (create_data_ref): Likewise.
7086 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
7087 entry edge into a region and adjust instantiate_scev calls.
7088 (create_data_ref): Likewise.
7089 (graphite_find_data_references_in_stmt): Likewise.
7090 (find_data_references_in_stmt): Pass the loop preheader edge
7091 from the nest argument.
7092 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
7093 parameter the edge into the region.
7094 (instantiate_parameters): Use the loop preheader edge as entry.
7095 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
7097 (get_instantiated_value_entry): Make instantiate_below parameter
7098 the edge into the region.
7099 (instantiate_scev_name): Likewise. Adjust dominance checks,
7100 when we cannot use loop-based instantiation instantiate by
7101 walking use-def chains.
7102 (instantiate_scev_poly): Adjust.
7103 (instantiate_scev_binary): Likewise.
7104 (instantiate_scev_convert): Likewise.
7105 (instantiate_scev_not): Likewise.
7106 (instantiate_array_ref): Remove.
7107 (instantiate_scev_3): Likewise.
7108 (instantiate_scev_2): Likewise.
7109 (instantiate_scev_1): Likewise.
7110 (instantiate_scev_r): Do not blindly handle N-operand trees.
7111 Do not instantiate array-refs. Handle all constants and invariants.
7112 (instantiate_scev): Make instantiate_below parameter
7113 the edge into the region.
7114 (resolve_mixers): Use the loop preheader edge for the region
7115 parameter to instantiate_scev_r.
7116 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
7118 2017-10-13 Richard Biener <rguenther@suse.de>
7120 PR tree-optimization/82525
7121 * graphite-isl-ast-to-gimple.c
7122 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
7124 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
7125 Fail code generation when we cannot represent the isl integer.
7126 (binary_op_to_tree): Elide modulo operations that are no-ops
7127 in the type we code generate. Remove now superfluous code
7130 2017-10-13 Richard Biener <rguenther@suse.de>
7132 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
7133 (scop_detection::harmful_loop_in_region): Remove premature
7134 IV type restriction.
7135 (scop_detection::graphite_can_represent_scev): We can handle
7136 pointer IVs just fine.
7138 2017-10-13 Alan Modra <amodra@gmail.com>
7140 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
7141 "Clobbers and Scratch Registers". Add paragraph on
7142 alternative to clobbers for scratch registers and OpenBLAS
7145 2017-10-13 Alan Modra <amodra@gmail.com>
7147 * doc/extend.texi (Clobbers): Correct vax example. Delete old
7148 example of a memory input for a string of known length. Move
7149 commentary out of table. Add a number of new examples
7150 covering array memory inputs.
7152 2017-10-12 Martin Liska <mliska@suse.cz>
7154 PR tree-optimization/82493
7155 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
7156 (test_range_functions): New function.
7157 (sbitmap_c_tests): Likewise.
7158 * selftest-run-tests.c (selftest::run_tests): Run new tests.
7159 * selftest.h (sbitmap_c_tests): New function.
7161 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
7163 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7165 * config/rs6000/amo.h: Fix spacing issue.
7167 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7170 * config/i386/i386.md (*ashl<mode>3_mask_1,
7171 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
7172 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
7175 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
7177 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
7178 (profile_probability): Set max_probability
7179 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
7181 * profile-count.c (profile_probability::differs_from_p): Do not
7182 rely on max_probaiblity == 10000
7184 2017-10-12 Jeff Law <law@redhat.com>
7186 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
7189 2017-10-12 Martin Sebor <msebor@redhat.com>
7193 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
7194 (handle_alias_pairs): Call it.
7195 * common.opt (-Wattribute-alias): New option.
7196 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
7197 * doc/invoke.texi (-Wattribute-alias): Document.
7199 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
7202 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
7204 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7206 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7207 (make_hard_regno_born, make_hard_regno_dead): Update
7208 bb_killed_pseudos and bb_gen_pseudos.
7210 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
7212 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
7214 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
7216 * config/alpha/alpha.c (alpha_split_conditional_move):
7217 Use std::swap instead of manually swapping.
7218 (alpha_stdarg_optimize_hook): Ditto.
7219 (alpha_canonicalize_comparison): Ditto.
7221 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7223 * tree-loop-distribution.c (struct builtin_info): New struct.
7224 (struct partition): Refactor fields into struct builtin_info.
7225 (partition_free): Free struct builtin_info.
7226 (build_size_arg_loc, build_addr_arg_loc): Delete.
7227 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
7228 information from struct builtin_info.
7229 (find_single_drs): New function refactored from classify_partition.
7230 Also moved builtin validity checks to this function.
7231 (compute_access_range, alloc_builtin): New functions.
7232 (classify_builtin_st, classify_builtin_ldst): New functions.
7233 (classify_partition): Refactor code into functions find_single_drs,
7234 classify_builtin_st and classify_builtin_ldst.
7235 (distribute_loop): Don't do runtime alias check when distributing
7237 (find_seed_stmts_for_distribution): New function.
7238 (pass_loop_distribution::execute): Refactor code finding seed
7239 stmts into above function. Support distribution for the innermost
7240 two-level loop nest. Adjust dump information.
7242 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7244 * tree-loop-distribution.c: Adjust the general comment.
7245 (NUM_PARTITION_THRESHOLD): New macro.
7246 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
7247 (classify_partition): Skip builtin pattern of loop nest's inner loop.
7248 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
7249 in call to build_partition_graph.
7250 (finalize_partitions): New parameter. Make loop distribution more
7251 conservative by fusing more partitions.
7252 (distribute_loop): Don't do runtime alias check in case of loop nest
7254 (find_seed_stmts_for_distribution): New function.
7255 (prepare_perfect_loop_nest): New function.
7256 (pass_loop_distribution::execute): Refactor code finding seed stmts
7257 and loop nest into above functions. Support loop nest distribution.
7258 Adjust dump information accordingly.
7260 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7262 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
7263 and set PTYPE_SEQUENTIAL for merged partition.
7265 2017-10-12 Richard Biener <rguenther@suse.de>
7267 PR tree-optimization/69728
7269 2017-09-19 Richard Biener <rguenther@suse.de>
7271 PR tree-optimization/69728
7272 * graphite-sese-to-poly.c (schedule_error): New global.
7273 (add_loop_schedule): Handle empty domain by failing the
7275 (build_original_schedule): Handle schedule_error.
7277 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
7278 domain by returning an unchanged schedule.
7280 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7282 * genrecog.c (validate_pattern): For VEC_SELECT verify that
7283 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
7285 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
7287 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
7290 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7293 * expr.c (store_field): Don't optimize away bitsize == 0 store
7294 from CALL_EXPR with addressable return type.
7296 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
7298 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
7299 * config/rs6000/rs6000.md (sel): Delete mode attribute.
7300 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
7301 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
7302 TARGET_ISEL instead of TARGET_ISEL<sel>.
7304 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
7306 * config/rs6000/rs6000.c
7307 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
7309 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
7311 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
7313 (reg_or_cint_operand): Fix comment.
7314 (reg_or_zero_operand): New predicate.
7315 * config/rs6000/rs6000-protos.h (output_isel): Delete.
7316 * config/rs6000/rs6000.c (output_isel): Delete.
7317 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
7318 instead of reg_or_cint_operand. Output instruction directly (not via
7320 (isel_unsigned_<mode>): Ditto.
7321 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
7322 gpc_reg_operand. Add an instruction alternative for this. Output
7323 instruction directly.
7324 (*isel_reversed_unsigned_<mode>): Ditto.
7326 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7328 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
7329 (TARGET_CANONICALIZE_COMPARISON): Define.
7331 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
7334 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
7335 Check whether the dest is REG before adding REG_EQUIV note.
7337 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
7340 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7342 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7343 (make_hard_regno_born, make_hard_regno_dead): Update
7344 bb_killed_pseudos and bb_gen_pseudos.
7346 2017-10-11 Nathan Sidwell <nathan@acm.org>
7348 * incpath.h (enum incpath_kind): Name enum, prefix values.
7349 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
7350 * incpath.c (heads, tails): Use INC_MAX.
7351 (add_env_var_paths, add_standard_paths): Use incpath_kind.
7352 (merge_include_chains, split_quote_chain,
7353 register_include_chains): Update incpath_kind names.
7354 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
7355 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
7357 (add_framework_path, darwin_register_objc_includes): Likewise.
7358 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
7360 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7362 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
7363 Do not use float_operator operator predicate.
7364 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
7365 * config/i386/predicates.md (float_operator): Remove predicate.
7367 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7369 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
7370 (*jccxf_i387): Ditto.
7371 (*jcc<mode>_i387): Ditto.
7372 (*jccu<mode>_i387): Ditto.
7373 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
7374 (*jcc_*_i387 splitters): Remove.
7375 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
7376 * config/i386/i386.c (ix86_split_fp_branch): Remove.
7377 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
7380 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
7382 * profile-count.h (slow_safe_scale_64bit): New function.
7383 (safe_scale_64bit): New inline.
7384 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
7385 * profile-count.c: Include wide-int.h
7386 (slow_safe_scale_64bit): New.
7388 2017-10-11 Nathan Sidwell <nathan@acm.org>
7390 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
7391 HAS_DECL_ASSEMBLER_NAME_P.
7392 * gimple-expr.c (gimple_decl_printable_name: Check
7393 HAS_DECL_ASSEMBLER_NAME_P too.
7394 * ipa-utils.h (type_in_anonymous_namespace_p): Check
7395 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
7396 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
7397 * passes.c (rest_of_decl_compilation): Check
7398 HAS_DECL_ASSEMBLER_NAME_P too.
7399 * recog.c (verify_changes): Likewise.
7400 * tree-pretty-print.c (dump_decl_name): Likewise.
7401 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
7403 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
7404 (SET_DECL_ASSEMBLER_NAME): Use it.
7405 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
7406 (COPY_DECL_ASSEMBLER_NAME): Likewise.
7407 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
7409 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
7411 * config.gcc (i386, x86_64): Add extra objects.
7412 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
7413 (ix86_min_insn_size): Declare.
7414 (ix86_issue_rate): Declare.
7415 (ix86_adjust_cost): Declare.
7416 (ia32_multipass_dfa_lookahead): Declare.
7417 (ix86_macro_fusion_p): Declare.
7418 (ix86_macro_fusion_pair_p): Declare.
7419 (ix86_bd_has_dispatch): Declare.
7420 (ix86_bd_do_dispatch): Declare.
7421 (ix86_core2i7_init_hooks): Declare.
7422 (ix86_atom_sched_reorder): Declare.
7423 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
7424 (COSTS_N_BYTES): Move to x86-tune-costs.h.
7425 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
7426 (rip_relative_addr_p): Rename to ...
7427 (ix86_rip_relative_addr_p): ... this one; export.
7428 (memory_address_length): Update.
7429 (ix86_issue_rate): Move to x86-tune-sched.c.
7430 (ix86_flags_dependent): Move to x86-tune-sched.c.
7431 (ix86_agi_dependent): Move to x86-tune-sched.c.
7432 (exact_dependency_1): Move to x86-tune-sched.c.
7433 (exact_store_load_dependency): Move to x86-tune-sched.c.
7434 (ix86_adjust_cost): Move to x86-tune-sched.c.
7435 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
7436 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
7437 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
7438 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
7439 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
7440 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
7441 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
7442 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
7443 (min_insn_size): Rename to ...
7444 (ix86_min_insn_size): ... this one; export.
7445 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
7446 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
7447 (core2i7_first_cycle_multipass_backtrack): Move to
7448 x86-tune-sched-core.c.
7449 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
7450 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
7451 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
7452 (ix86_avoid_jump_mispredicts): Update.
7453 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
7454 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
7455 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
7456 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
7457 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
7458 (MAX_INSN): Move to ix86-tune-sched-bd.c.
7459 (MAX_IMM): Move to ix86-tune-sched-bd.c.
7460 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
7461 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
7462 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
7463 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
7464 (MAX_STORE): Move to ix86-tune-sched-bd.c.
7465 (BIG): Move to ix86-tune-sched-bd.c.
7466 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
7467 (enum insn_path): Move to ix86-tune-sched-bd.c.
7468 (get_mem_group): Move to ix86-tune-sched-bd.c.
7469 (is_cmp): Move to ix86-tune-sched-bd.c.
7470 (dispatch_violation): Move to ix86-tune-sched-bd.c.
7471 (is_branch): Move to ix86-tune-sched-bd.c.
7472 (is_prefetch): Move to ix86-tune-sched-bd.c.
7473 (init_window): Move to ix86-tune-sched-bd.c.
7474 (allocate_window): Move to ix86-tune-sched-bd.c.
7475 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
7476 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
7477 (process_end_window): Move to ix86-tune-sched-bd.c.
7478 (allocate_next_window): Move to ix86-tune-sched-bd.c.
7479 (find_constant): Move to ix86-tune-sched-bd.c.
7480 (get_num_immediates): Move to ix86-tune-sched-bd.c.
7481 (has_immediate): Move to ix86-tune-sched-bd.c.
7482 (get_insn_path): Move to ix86-tune-sched-bd.c.
7483 (get_insn_group): Move to ix86-tune-sched-bd.c.
7484 (count_num_restricted): Move to ix86-tune-sched-bd.c.
7485 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
7486 (add_insn_window): Move to ix86-tune-sched-bd.c.
7487 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
7488 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
7489 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
7490 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
7491 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
7492 (do_dispatch): Move to ix86-tune-sched-bd.c.
7493 (has_dispatch): Move to ix86-tune-sched-bd.c.
7494 * config/i386/t-i386: Add new object files.
7495 * config/i386/x86-tune-costs.h: New file.
7496 * config/i386/x86-tune-sched-atom.c: New file.
7497 * config/i386/x86-tune-sched-bd.c: New file.
7498 * config/i386/x86-tune-sched-core.c: New file.
7499 * config/i386/x86-tune-sched.c: New file.
7501 2017-10-11 Liu Hao <lh_mouse@126.com>
7503 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
7504 the generic version below instead.
7505 (should_colorize): Recognize Windows consoles as terminals
7507 * pretty-print.c [__MINGW32__] (write_all): New function.
7508 [__MINGW32__] (find_esc_head): Likewise.
7509 [__MINGW32__] (find_esc_terminator): Likewise.
7510 [__MINGW32__] (eat_esc_sequence): Likewise.
7511 [__MINGW32__] (mingw_ansi_fputs): New function that handles
7513 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
7516 2017-10-11 Richard Biener <rguenther@suse.de>
7518 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
7519 Properly call analyze_scalar_evolution with the loop of the stmt.
7521 2017-10-11 Richard Biener <rguenther@suse.de>
7523 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
7524 * tree-core.h (tree_base): Add chrec_var union member.
7525 * tree.h (CHREC_VAR): Remove.
7526 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
7527 * tree-chrec.h (build_polynomial_chrec): Adjust.
7528 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
7529 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
7531 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
7533 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
7534 * match.pd: ... here.
7535 ((T) X == (T) Y): Relax condition.
7537 2017-10-11 Bin Cheng <bin.cheng@arm.com>
7539 PR tree-optimization/82472
7540 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
7542 (break_alias_scc_partitions): Update postorder number.
7544 2017-10-11 Martin Liska <mliska@suse.cz>
7547 * opts.c (parse_no_sanitize_attribute): Do not use error_value
7549 * opts.h (parse_no_sanitize_attribute): Remove last argument.
7551 2017-10-11 Martin Liska <mliska@suse.cz>
7553 * print-rtl.c (print_insn): Move declaration of idbuf
7554 to same scope as name.
7556 2017-10-11 Martin Liska <mliska@suse.cz>
7561 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
7564 2017-10-11 Martin Liska <mliska@suse.cz>
7567 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
7570 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
7572 * config.gcc (default_gnu_indirect_function): Default to yes for
7573 arm*-*-linux* with glibc.
7575 2017-10-11 Richard Biener <rguenther@suse.de>
7577 * tree-scalar-evolution.c (get_scalar_evolution): Handle
7578 default-defs and types we do not want to analyze.
7579 (interpret_loop_phi): Replace unreachable code with an assert.
7580 (compute_scalar_evolution_in_loop): Remove and inline ...
7581 (analyze_scalar_evolution_1): ... here, replacing condition with
7582 what makes the intent clearer. Remove handling of cases
7583 get_scalar_evolution now handles.
7585 2017-10-10 Jim Wilson <wilson@tuliptree.org>
7587 PR rtl-optimization/81434
7588 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
7589 comment for main loop. In sched_group_found if, also add checks for
7590 pass and min_cost_group.
7592 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
7594 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
7595 (rs6000_insn_cost): New function.
7596 * config/rs6000/rs6000.md (cost): New attribute.
7598 2017-10-10 Jakub Jelinek <jakub@redhat.com>
7599 H.J. Lu <hongjiu.lu@intel.com>
7603 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
7604 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
7605 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
7606 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
7607 to OPTION_MASK_ISA_AVX512VL - builtins that have both
7608 OPTION_MASK_ISA_MMX and some other bit set require both
7609 mmx and the ISAs without the mmx bit.
7610 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
7611 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
7612 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
7613 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
7614 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
7615 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
7616 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
7617 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
7618 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
7619 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
7620 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
7621 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
7622 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
7623 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
7624 Add OPTION_MASK_ISA_MMX.
7626 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
7628 * config.gcc (armv7*-*-freebsd*): New target.
7629 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
7631 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
7633 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
7634 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
7637 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
7639 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
7640 a template parameter.
7641 (WIDE_INT_REF_FOR): Update accordingly.
7642 * tree.h (wi::int_traits <const_tree>): Delete.
7643 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
7644 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
7645 (wi::tree_to_wide_ref): New typedef.
7646 (wi::to_wide): New function.
7647 * calls.c (get_size_range): Use wi::to_wide when operating on
7649 * cgraph.c (cgraph_node::create_thunk): Likewise.
7650 * config/i386/i386.c (ix86_data_alignment): Likewise.
7651 (ix86_local_alignment): Likewise.
7652 * dbxout.c (stabstr_O): Likewise.
7653 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
7654 * expr.c (const_vector_from_tree): Likewise.
7655 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
7656 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
7657 (fold_negate_expr_1, int_const_binop_1, const_binop)
7658 (fold_convert_const_int_from_real, optimize_bit_field_compare)
7659 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
7660 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
7661 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
7662 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
7663 (fold_not_const, round_up_loc): Likewise.
7664 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
7665 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
7666 (alloca_call_type): Likewise.
7667 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
7668 * godump.c (go_output_typedef): Likewise.
7669 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
7670 * internal-fn.c (get_min_precision): Likewise.
7671 * ipa-cp.c (ipcp_store_vr_results): Likewise.
7672 * ipa-polymorphic-call.c
7673 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
7674 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
7675 (ipa_modify_call_arguments): Likewise.
7676 * match.pd: Likewise.
7677 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
7678 * print-tree.c (print_node_brief, print_node): Likewise.
7679 * stmt.c (expand_case): Likewise.
7680 * stor-layout.c (layout_type): Likewise.
7681 * tree-affine.c (tree_to_aff_combination): Likewise.
7682 * tree-cfg.c (group_case_labels_stmt): Likewise.
7683 * tree-data-ref.c (dr_analyze_indices): Likewise.
7684 (prune_runtime_alias_test_list): Likewise.
7685 * tree-dump.c (dequeue_and_dump): Likewise.
7686 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
7687 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
7688 * tree-pretty-print.c (dump_generic_node): Likewise.
7689 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
7690 (simple_iv_with_niters): Likewise.
7691 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
7692 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
7693 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
7694 * tree-ssa-loop-niter.c (split_to_var_and_offset)
7695 (refine_value_range_using_guard, number_of_iterations_ne_max)
7696 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
7697 (get_cst_init_from_scev, record_nonwrapping_iv)
7698 (scev_var_range_cant_overflow): Likewise.
7699 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
7700 * tree-ssa-pre.c (compute_avail): Likewise.
7701 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
7702 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
7703 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
7704 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
7705 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
7706 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
7707 (dump_case_nodes, try_switch_expansion): Likewise.
7708 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
7709 (vect_do_peeling): Likewise.
7710 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
7711 * tree-vect-stmts.c (vectorizable_load): Likewise.
7712 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
7713 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
7714 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
7715 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
7716 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
7717 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
7718 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
7719 (evrp_dom_walker::before_dom_children): Likewise.
7720 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
7721 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
7722 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
7723 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
7724 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
7725 (get_range_pos_neg): Likewise.
7726 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
7727 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
7728 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
7729 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
7730 * config/avr/avr.c (avr_fold_builtin): Likewise.
7731 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
7732 * config/msp430/msp430.c (msp430_attr): Likewise.
7733 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
7734 * config/powerpcspe/powerpcspe-c.c
7735 (altivec_resolve_overloaded_builtin): Likewise.
7736 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
7737 (rs6000_expand_ternop_builtin): Likewise.
7738 * config/rs6000/rs6000-c.c
7739 (altivec_resolve_overloaded_builtin): Likewise.
7740 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
7741 (rs6000_expand_ternop_builtin): Likewise.
7742 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
7744 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7746 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
7747 when copying loop nest with only one inner loop.
7749 2017-10-10 Richard Biener <rguenther@suse.de>
7751 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
7752 blocks if SCEV is active.
7753 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
7755 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
7756 (scev_initialize): Assert we are not yet initialized.
7758 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7760 * tree-loop-distribution.c (generate_loops_for_partition): Remove
7761 inner loop's exit stmt by making it always exit the loop, otherwise
7762 we would generate an infinite empty loop.
7764 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7766 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
7767 renaming variables in new preheader if it's deleted.
7769 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7771 * tree-loop-distribution.c (struct partition): Remove unused field
7772 loops of the structure.
7773 (partition_alloc, partition_free): Ditto.
7774 (build_rdg_partition_for_vertex): Ditto.
7776 2017-10-09 Jeff Law <law@redhat.com>
7778 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
7779 return type to match prototype and documentation.
7781 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7783 * config/rs6000/rs6000.c (processor_costs): Move to ...
7784 * config/rs6000/rs6000.h: ... here.
7785 (rs6000_cost): Declare.
7787 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
7789 * except.c (setjmp_fn): New global variable.
7790 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
7791 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
7792 if DONT_USE_BUILTIN_SETJMP is defined.
7794 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7796 * target.def (insn_cost): New hook.
7797 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
7798 * doc/tm.texi: Regenerate.
7799 * rtlanal.c (insn_cost): Use the new hook.
7801 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7803 * combine.c (combine_validate_cost): Compute the new insn_cost,
7804 not just pattern_cost.
7805 (try_combine): Adjust comment.
7807 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7809 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
7811 * combine.c (uid_insn_cost): Adjust comment.
7812 (combine_validate_cost): Adjust comment. Use pattern_cost instead
7814 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
7815 * dse.c (find_shift_sequence): Ditto.
7816 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
7817 (bb_valid_for_noce_process_p): Use pattern_cost.
7818 * rtl.h (insn_rtx_cost): Delete.
7819 (pattern_cost): New prototype.
7820 (insn_cost): New prototype.
7821 * rtlanal.c (insn_rtx_cost): Rename to...
7822 (pattern_cost): ... this.
7825 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
7827 * config/i386/i386.md (*jcc_2): Remove insn pattern.
7828 (*jcc<mode>_0_r_i387): Ditto.
7829 (*jccxf_r_i387): Ditto.
7830 (*jcc<mode>_r_i387): Ditto.
7831 (*jccu<mode>_r_i387): Ditto.
7832 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
7833 (*jcc): Rename from *jcc_1.
7835 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7837 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
7838 deferred rescans after the lvx/stvx recombination pre-pass.
7840 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
7842 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
7843 memory operation instruction support.
7844 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
7845 (rs6000-ibm-aix[789]*): Likewise.
7846 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
7847 Document new functions.
7849 2017-10-09 Richard Biener <rguenther@suse.de>
7851 PR tree-optimization/82397
7852 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
7853 equality only for semantically equal trees.
7855 2017-10-09 Richard Biener <rguenther@suse.de>
7857 PR tree-optimization/82449
7858 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
7859 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
7860 allow constant addresses.
7861 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
7864 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7866 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
7869 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7872 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
7875 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7878 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
7880 2017-10-09 Jakub Jelinek <jakub@redhat.com>
7883 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
7884 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
7886 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
7888 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
7889 (WI_BINARY_PREDICATE_RESULT): Likewise.
7890 (wi::binary_traits::operator_result): New type.
7891 (wi::binary_traits::predicate_result): Likewise.
7892 (generic_wide_int::operator~, unary generic_wide_int::operator-)
7893 (generic_wide_int::operator==, generic_wide_int::operator!=)
7894 (generic_wide_int::operator&, generic_wide_int::and_not)
7895 (generic_wide_int::operator|, generic_wide_int::or_not)
7896 (generic_wide_int::operator^, generic_wide_int::operator+
7897 (binary generic_wide_int::operator-, generic_wide_int::operator*):
7899 (operator~, unary operator-, operator==, operator!=, operator&)
7900 (operator|, operator^, operator+, binary operator-, operator*): New
7902 * expr.c (get_inner_reference): Use wi::bit_and_not.
7903 * fold-const.c (fold_binary_loc): Likewise.
7904 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
7905 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
7906 (bit_value_binop): Likewise.
7907 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
7908 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
7909 (extract_range_from_binary_expr_1): Likewise.
7910 (masked_increment): Likewise.
7911 (simplify_bit_ops_using_ranges): Likewise.
7913 2017-10-09 Martin Jambor <mjambor@suse.cz>
7916 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
7917 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
7918 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
7920 (hsa_fixup_mov_insn_type): New function.
7921 (hsa_op_with_type::get_in_type): Use it.
7922 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
7923 immediates in an assert.
7924 (hsa_op_with_type::extend_int_to_32bit): New method.
7925 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
7926 types. Convert to dest type if necessary.
7927 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
7928 (reg_for_gimple_ssa): Pass false as min32int to
7929 hsa_type_for_scalar_tree_type.
7930 (gen_hsa_addr): Fixup type when creating addresable temporary.
7931 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
7932 (gen_hsa_unary_operation): Extend operands and convert to dest type if
7933 necessary. Call hsa_fixup_mov_insn_type.
7934 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
7935 extend operands and convert to dest type if necessary.
7936 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
7937 to dest type if necessary.
7938 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
7939 if conversion nt necessary and size matches.
7940 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
7941 to dest type if necessary.
7942 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
7943 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
7945 (gen_hsa_clrsb): Likewise.
7946 (gen_hsa_ffs): Likewise.
7947 (gen_hsa_divmod): Extend operands and convert to dest type if
7949 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
7951 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
7953 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
7954 Remove empty default arguments. Use a brace block as output
7956 (conditional return): Ditto.
7958 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
7959 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
7961 (group_ending_nop): Ditto.
7962 (doloop_end): Ditto.
7963 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
7964 (splitters for those): Ditto.
7966 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
7968 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
7969 a conditional jump (and the compare for it) so that pc_rtx is the
7971 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
7972 for the deleted and renamed ctr<mode>_internal[234] patterns.
7973 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
7974 Delete second conditional return pattern.
7975 (ctr<mode>_internal2): Delete this second bdnz pattern.
7976 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
7977 (ctr<mode>_internal4): Delete this second bdz pattern.
7979 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
7981 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
7982 (always_initialized_rtx_for_ssa_name_p): New predicate.
7983 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
7984 (finish_out_of_ssa): Free new field of SA.
7985 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
7986 * tree-ssa-coalesce.c: Include tree-ssa.h.
7987 (get_parm_default_def_partitions): Remove extern keyword.
7988 (get_undefined_value_partitions): New function.
7989 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
7990 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
7992 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
7994 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
7996 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
7998 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
8000 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
8001 for targets that preffer 128bit.
8003 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
8005 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
8007 2017-10-08 Olivier Hainque <hainque@adacore.com>
8009 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
8010 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
8012 (thumb_set_return_address): Likewise.
8014 2017-10-08 Olivier Hainque <hainque@adacore.com>
8016 * common/config/arm/arm-common.c (arm_except_unwind_info):
8017 Handle DWARF2_UNWIND_INFO.
8019 2017-10-07 Michael Collison <michael.collison@arm.com>
8021 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
8024 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
8026 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
8027 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
8028 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
8029 defined, force the creation of a new block for a dispatch label.
8031 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
8033 * invoke.texi (Wsuggest-attribute=cold): Document.
8034 * common.opt (Wsuggest-attribute=cold): New
8035 * ipa-pure-const.c (warn_function_cold): New function.
8036 * predict.c (compute_function_frequency): Use it.
8037 * predict.h (warn_function_cold): Declare.
8039 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
8041 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
8044 2017-10-06 Martin Liska <mliska@suse.cz>
8046 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
8047 keyword for member functions.
8048 (struct sanopt_tree_couple): New struct.
8049 (struct sanopt_tree_couple_hash): New function.
8050 (struct sanopt_ctx): Add new hash_map.
8051 (has_dominating_ubsan_ptr_check): New function.
8052 (record_ubsan_ptr_check_stmt): Likewise.
8053 (maybe_optimize_ubsan_ptr_ifn): Likewise.
8054 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
8055 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
8057 2017-10-06 Sudakshina Das <sudi.das@arm.com>
8060 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
8061 aarch64_simd_valid_immediate on CONST_VECTORs.
8062 (aarch64_reg_or_bic_imm): Likewise.
8064 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
8066 PR rtl-optimization/82396
8067 * haifa-sched.c (ready_sort_real): Disable qsort checking.
8069 2017-10-06 Sebastian Pop <sebpop@gmail.com>
8071 * graphite-dependences.c (scop_get_reads): Move code to...
8072 (scop_get_must_writes): Move code to...
8073 (scop_get_may_writes): Move code to...
8074 (scop_get_reads_and_writes): ... here.
8075 (scop_get_dependences): Call scop_get_reads_and_writes.
8077 2017-10-06 Jakub Jelinek <jakub@redhat.com>
8079 PR tree-optimization/82434
8080 * fold-const.h (can_native_encode_type_p,
8081 can_native_encode_string_p): Remove.
8082 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
8083 don't encode anything, just return what would be otherwise returned.
8084 (native_encode_fixed, native_encode_complex, native_encode_vector):
8086 (native_encode_string): Likewise. Inline by hand
8087 can_native_encode_string_p.
8088 (can_native_encode_type_p): Remove.
8089 (can_native_encode_string_p): Remove.
8090 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
8091 STRING_CSTs using can_native_encode_string_p, test all
8092 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
8093 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
8094 argument from native_encode_expr.
8095 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
8096 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
8099 2017-10-06 Richard Biener <rguenther@suse.de>
8101 PR tree-optimization/82397
8102 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
8103 operand_equal_p but rely on data_ref_compare_tree for detecting
8105 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
8106 to match up with dr_group_sort_cmp.
8108 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8111 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
8113 * config/s390/s390-builtin-types.def: Regenerate.
8115 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8118 * config/s390/s390-builtin-types.def: Regenerate.
8119 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
8120 Change flag from B_VXE to B_VX.
8121 (s390_vec_min_dbl): Remove B_VXE flag.
8123 2017-10-06 Richard Biener <rguenther@suse.de>
8125 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
8126 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
8127 translate_isl_ast_to_gimple::is_valid_rename,
8128 translate_isl_ast_to_gimple::get_rename,
8129 translate_isl_ast_to_gimple::get_def_bb_for_const,
8130 translate_isl_ast_to_gimple::get_new_name,
8131 translate_isl_ast_to_gimple::collect_all_ssa_names,
8132 translate_isl_ast_to_gimple::copy_loop_phi_args,
8133 translate_isl_ast_to_gimple::collect_all_ssa_names,
8134 translate_isl_ast_to_gimple::copy_loop_phi_args,
8135 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
8136 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
8137 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
8138 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
8139 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
8140 translate_isl_ast_to_gimple::copy_cond_phi_args,
8141 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
8142 translate_isl_ast_to_gimple::edge_for_new_close_phis,
8143 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
8144 translate_isl_ast_to_gimple::rename_uses,
8145 translate_isl_ast_to_gimple::rename_all_uses): Remove.
8146 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
8147 (set_rename_for_each_def): Likewise.
8148 (graphite_copy_stmts_from_block): Handle debug stmt resetting
8149 here. Handle rewriting SCEV analyzable uses here.
8150 (copy_bb_and_scalar_dependences): Generate code for PHI
8152 (graphite_regenerate_ast_isl): Adjust.
8153 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
8154 (add_write, add_read): New functions.
8155 (build_cross_bb_scalars_def): Use it and simplify.
8156 (build_cross_bb_scalars_use): Likewise.
8157 (graphite_find_cross_bb_scalar_vars): Inline into...
8158 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
8159 simulating out-of-SSA. Compute liveout and add dependencies.
8160 (build_scops): Force an empty entry block.
8161 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
8163 (sese_build_liveouts): Declare.
8164 (sese_trivially_empty_bb_p): Likewise.
8165 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
8166 compute liveout and debug_liveout.
8167 (sese_bad_liveouts_use): Remove.
8168 (sese_reset_debug_liveouts_bb): Likewise.
8169 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
8170 (sese_build_liveouts): Build liveout and debug_liveout and store
8172 (new_sese_info): Adjust.
8173 (free_sese_info): Likewise.
8174 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
8175 do not build liveout here.
8176 (move_sese_in_condition): Adjust region entry.
8177 (scev_analyzable_p): Match up with chrec_apply requirements.
8178 (sese_trivially_empty_bb_p): New.
8179 * tree-into-ssa.c (get_reaching_def): Properly support generating
8180 default-defs for incremental rewrite of anonymous names.
8182 2017-10-06 Richard Biener <rguenther@suse.de>
8184 * graphite-sese-to-poly.c (extract_affine): For casts increasing
8185 precision do not perform modulo reduction.
8187 2017-10-06 Richard Biener <rguenther@suse.de>
8189 PR tree-optimization/82436
8190 * tree-vect-slp.c (vect_supported_load_permutation_p): More
8191 conservatively choose the vectorization factor when checking
8192 whether we can perform the required load permutation.
8193 (vect_transform_slp_perm_load): Assert when we may not fail.
8195 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
8197 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
8198 message for incompatible -msdata=* and -mcall-* options.
8200 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8202 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
8203 rate for post-reload scheduling.
8205 2017-10-05 Tamar Christina <tamar.christina@arm.com>
8207 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
8209 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8211 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
8212 to improve monte carlo in scimark.
8214 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8216 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
8217 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
8218 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
8219 pentium4_cost, nocona_cost): Set reassociation width to 1.
8220 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
8221 width to 2 for fp operations and 1 otherwise.
8222 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
8224 (atom_cost): Set reassociation width to 2.
8225 (slm_cost, generic_cost): Set fp reassociation width
8226 to 2 and 1 otherwise.
8227 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
8228 (core_cost): Set fp reassociation width to 4 and vector to 2.
8229 (ix86_reassociation_width): Rewrite using cost table; special case
8230 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
8231 and TARGET_AVX128_OPTIMAL.
8232 * config/i386/i386.h (processor_costs): Add
8233 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
8234 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
8235 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
8236 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
8237 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
8238 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
8240 2017-10-05 Nathan Sidwell <nathan@acm.org>
8242 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
8244 2017-10-05 Tamar Christina <tamar.christina@arm.com>
8246 * config/arm/arm.c (arm_test_fpu_data): New.
8247 (arm_run_selftests): Call arm_test_fpu_data.
8249 2017-10-04 Nathan Sidwell <nathan@acm.org>
8251 * toplev.c (toplev::main): Remove excess parens on pretty_printer
8253 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
8255 2017-10-04 Sudakshina Das <sudi.das@arm.com>
8257 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
8258 check type for aarch64_simd_valid_immediate.
8259 (aarch64_output_simd_mov_immediate): Update prototype.
8260 (aarch64_simd_valid_immediate): Update prototype.
8261 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
8262 support for ORR-immediate.
8263 (and<mode>3): modified pattern to add support for BIC-immediate.
8264 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
8265 now checks for valid immediate for BIC and ORR based on new enum
8267 (aarch64_output_simd_mov_immediate): Function now used to output
8268 BIC/ORR imm as well based on new enum argument.
8269 * config/aarch64/constraints.md (Do): New vector immediate constraint.
8271 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
8272 (aarch64_reg_or_bic_imm): Likewise.
8274 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8276 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
8277 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
8279 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
8283 PR rtl-optimization/82396
8284 * haifa-sched.c (autopref_multipass_init): Simplify
8286 (autopref_rank_data): Simplify sort order.
8287 * sched-int.h (autopref_multipass_data_): Remove
8288 multi_mem_insn_p, min_offset and max_offset.
8290 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8292 * doc/sourcebuild.texi: Document vect_peeling_profitable.
8294 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8296 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
8299 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8301 * doc/sourcebuild.texi: Document vect_long_mult.
8303 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
8305 PR tree-optimization/82413
8306 * fold-const.c (build_range_check): Use widest_int when comparing
8307 the maximum ETYPE value with HIGH.
8309 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
8311 PR rtl-optimization/82396
8312 * haifa-sched.c (autopref_multipass_init): Simplify
8314 (autopref_rank_data): Simplify sort order.
8315 * sched-int.h (autopref_multipass_data_): Remove
8316 multi_mem_insn_p, min_offset and max_offset.
8318 2017-10-04 Jakub Jelinek <jakub@redhat.com>
8320 PR tree-optimization/82381
8321 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
8322 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
8325 PR tree-optimization/82374
8326 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
8327 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
8328 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
8329 current_function_decl to the new decl.
8331 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
8333 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
8334 helper macro for IEEE float128 hardware built-in functions.
8335 (SQRTF128_ODD): Add built-in functions with the round-to-odd
8337 (TRUNCF128_ODD): Likewise.
8338 (ADDF128_ODD): Likewise.
8339 (SUBF128_ODD): Likewise.
8340 (MULF128_ODD): Likewise.
8341 (DIVF128_ODD): Likewise.
8342 (FMAF128_ODD): Likewise.
8343 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
8344 UNSPEC_TRUNC_ROUND_TO_ODD.
8345 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
8346 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
8347 floating point round to odd instructions.
8348 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
8349 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
8350 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
8351 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
8352 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
8353 (trunc<mode>sf2_hw): Change the truncate with round to odd
8354 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
8355 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
8356 to odd hardware instructions.
8357 (sub<mode>3_odd): Likewise.
8358 (mul<mode>3_odd): Likewise.
8359 (div<mode>3_odd): Likewise.
8360 (sqrt<mode>2_odd): Likewise.
8361 (fma<mode>4_odd): Likewise.
8362 (fms<mode>4_odd): Likewise.
8363 (nfma<mode>4_odd): Likewise.
8364 (nfms<mode>4_odd): Likewise.
8365 (trunc<mode>df2_odd): Change the truncate with round to odd
8366 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
8368 * doc/extend.texi (PowerPC built-in functions): Update documentation
8369 for existing IEEE float128-bit built-in functions. Add built-in
8370 functions that generate the IEEE 128-bit floating point round to
8373 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
8375 PR rtl-optimization/77729
8376 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
8377 to (X&(C1&~C2))|C2 transformations.
8379 2017-10-03 Martin Jambor <mjambor@suse.cz>
8381 PR tree-optimization/82363
8382 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
8383 mismatch, mark lacc written regardless of racc.
8385 2017-10-03 Jakub Jelinek <jakub@redhat.com>
8387 PR tree-optimization/82381
8388 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
8389 stmt_to_insert nor wheather SSA_NAMEs are default defs.
8390 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
8391 fallthrough into reassoc_stmt_dominates_stmt_p.
8394 * combine.c (combine_instructions): Don't combine in unreachable
8397 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
8400 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
8401 function to not use the have_cpu variable. Do not set cpu_index,
8402 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
8404 (rs6000_valid_attribute_p): Remove duplicate initializations of
8405 old_optimize and func_optimize.
8406 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
8407 (rs6000_activate_target_options): Make global.
8408 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
8411 2017-10-02 Jakub Jelinek <jakub@redhat.com>
8413 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
8414 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
8415 Set *poffset to 0 and *psize and *pmax_size to -1 if
8416 *poffset + *psize overflows in HOST_WIDE_INT.
8418 PR tree-optimization/82387
8419 PR tree-optimization/82388
8420 PR tree-optimization/82389
8421 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
8422 instead of live_bytes non-NULL.
8424 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
8427 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
8430 2017-10-02 Richard Biener <rguenther@suse.de>
8432 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
8433 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
8434 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
8436 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
8438 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
8439 requested precision matches the type's.
8440 * calls.c (alloc_max_size): Calculate the new candidate size as
8441 a widest_int and use wi::to_widest when comparing it with the
8442 current candidate size.
8443 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
8444 zero rather than integer_zero_node.
8445 * match.pd: Check for a no-op conversion before using wi::add
8446 rather than after. Use tree_to_uhwi when summing small shift
8447 counts into an unsigned int.
8449 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
8450 Alan Hayward <alan.hayward@arm.com>
8451 David Sherwood <david.sherwood@arm.com>
8454 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
8455 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
8456 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
8457 POINTER_AND_FP_REGS.
8459 2017-10-02 Richard Biener <rguenther@suse.de>
8461 PR tree-optimization/82355
8462 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8463 a mapping for the enclosing loop but avoid generating one for
8465 (copy_bb_and_scalar_dependences): Remove premature codegen
8466 error on PHIs in blocks duplicated into multiple places.
8467 * graphite-scop-detection.c
8468 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8469 in the region use it as loop and nest to analyze the DR in.
8470 (try_generate_gimple_bb): Likewise.
8471 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8472 (add_loop_constraints): For blocks in a loop not in the region
8473 create a dimension with a single iteration.
8474 * sese.h (gbb_loop_at_index): Remove assert.
8476 2017-10-01 Kevin Buettner <kevinb@redhat.com>
8478 * omp-expand.c (adjust_context_scope): New function.
8479 (expand_parallel_call): Call adjust_context_scope.
8481 2017-10-01 Jeff Law <law@redhat.com>
8483 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
8484 dom_opt_dom_walker class with direct access to private members.
8485 Add comments. Call test_for_singularity.
8486 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
8487 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
8488 m_dummy_cond anymore.
8489 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
8491 (pass_dominator:execute): Build the dummy_cond here and pass it
8492 to the dom_opt_dom_walker ctor.
8493 (test_for_singularity): New function.
8495 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
8496 Maya Rashish <coypu@sdf.org>
8498 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
8499 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
8500 (alpha*-*-netbsd*) Use nbsd_tm_file.
8501 (arm*-*-netbsdelf*) Likewise.
8502 (i[34567]86-*-netbsdelf*) Likewise.
8503 (x86_64-*-netbsd*) Likewise.
8504 (mips*-*-netbsd*) Likewise.
8505 (powerpc-*-netbsd*) Likewise.
8506 (sh*-*-netbsd*) Likewise.
8507 (sparc-*-netbsdelf*) Likewise.
8508 (sparc64-*-netbsd*) Likewise.
8509 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
8511 (vax-*-netbsdelf*) Likewise.
8512 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
8513 (UINT_FAST8_TYPE) Likewise.
8514 (INT_FAST16_TYPE) Check CHAR_FAST16.
8515 (UINT_FAST16_TYPE) Likewise.
8517 2017-09-30 Jakub Jelinek <jakub@redhat.com>
8520 * config/i386/i386.md
8521 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
8522 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
8523 *divmodsi4_zext_2): New define_insn_and_split.
8524 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
8525 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
8526 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
8527 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
8528 New define_insn_and_split.
8529 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
8530 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
8531 operands[1] having DImode when mode is SImode.
8533 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
8534 always SImode for DIV and MOD in REG_EQUAL notes.
8536 2017-09-29 Yury Gribov <tetra2005@gmail.com>
8539 * match.pd: Fix handling of NaNs in pattern.
8541 2017-09-29 Jeff Law <law@redhat.com>
8543 * sbitmap.c (bitmap_bit_in_range_p): New function.
8544 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
8545 * tree-ssa-dse.c (live_bytes_read): New function.
8546 (dse_classify_store): Ignore reads of dead bytes.
8548 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
8549 typos and whitespace errors.
8550 * config/i386/predicates.md (address_no_seg_operand): Likewise.
8551 * config/s390/s390.c (s390_emit_prologue): Likewise.
8553 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
8556 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
8557 with a symbol for LRA.
8559 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
8561 PR rtl-optimization/82338
8562 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
8564 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
8566 * genmodes.c (calc_wider_mode): Suppress qsort macro.
8567 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
8568 (qsort_chk): Declare.
8569 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
8570 (qsort_chk): New function.
8572 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8574 PR tree-optimization/82337
8575 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
8576 phi definition if the PHI result appears in an abnormal PHI.
8577 (find_basis_for_base_expr): Don't record a basis if the LHS of the
8578 basis appears in an abnormal PHI.
8580 2017-09-29 Richard Biener <rguenther@suse.de>
8582 * graphite-isl-ast-to-gimple.c
8583 (translate_isl_ast_to_gimple::set_codegen_error): New function.
8584 (binary_op_to_tree): Use it.
8585 (get_rename_from_scev): Likewise.
8586 (copy_loop_phi_nodes): Likewise.
8587 (copy_bb_and_scalar_dependences): Likewise.
8588 (translate_pending_phi_nodes): Likewise.
8590 2017-09-29 Jakub Jelinek <jakub@redhat.com>
8593 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
8594 for movabsq $(i32 << shift), r64.
8596 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
8598 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
8599 index when encoding %esp as %rsp to avoid 0x67 prefix.
8601 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
8603 * config/i386/i386.md (*movsf_internal, *movdf_internal):
8604 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
8606 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
8608 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
8609 Extensions with more than 16 double VFP registers.
8610 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
8611 to_clear_mask and all code related to it. Replace the remaining
8612 entry by a sbitmap and adapt code accordingly.
8614 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
8616 * brig-builtins.def: Change pure attributes to const.
8618 2017-09-28 Joseph Myers <joseph@codesourcery.com>
8620 * config.gcc (default_gnu_indirect_function): Default to yes for
8621 sparc*-*-linux* with glibc.
8623 2017-09-28 Joseph Myers <joseph@codesourcery.com>
8625 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
8626 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
8627 when creating .init_array and .fini_array sections with priority
8630 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
8633 * config/aarch64/aarch64.c
8634 (aarch64_builtin_support_vector_misalignment): Always return false
8635 when misalignment is unknown.
8637 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
8639 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
8640 this function to return false if the definition used by the swap
8641 instruction is artificial, or if the memory address from which the
8642 constant value is loaded is not represented by a base address held
8643 in a register or if the base address register is a frame or stack
8644 pointer. Additionally, return false if the base address of the
8645 loaded constant is a SYMBOL_REF but is not considered to be a
8647 (replace_swapped_load_constant): New function.
8648 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
8649 loaded constant vector with a load of a swapped constant vector.
8651 2017-09-27 Carl Love <cel@us.ibm.com>
8653 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
8654 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
8655 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
8658 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
8660 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
8661 first, always call autopref_rank_data otherwise.
8663 2017-09-27 Richard Biener <rguenther@suse.de>
8665 * graphite-scop-detection.c (find_scop_parameters): Move
8666 loop bound handling ...
8667 (gather_bbs::before_dom_children): ... here, avoiding the need
8668 to build scop_info->loop_nest.
8669 (record_loop_in_sese): Remove.
8670 * sese.h (sese_info_t::loop_nest): Remove.
8671 * sese.c (new_sese_info): Do not allocate loop_nest.
8672 (free_sese_info): Do not free loop_nest.
8674 2017-09-27 Jakub Jelinek <jakub@redhat.com>
8677 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
8678 lhs from calls if the lhs has addressable type.
8680 2017-09-27 Richard Biener <rguenther@suse.de>
8682 * graphite.h (scop::max_alias_set): New member.
8683 * graphite-scop-detection.c: Remove references to non-existing
8684 --param in comments.
8685 (build_alias_sets): Record the maximum alias set used for drs.
8686 (build_scops): Support zero as unlimited for
8687 --param graphite-max-arrays-per-scop.
8688 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
8690 (build_poly_sr_1): ... here. Compute alias set based on the
8691 maximum alias set used for drs rather than
8692 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
8694 2017-09-27 Richard Biener <rguenther@suse.de>
8696 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
8697 --param loop-block-tile-size=0 to disable tiling.
8699 2017-09-27 Richard Biener <rguenther@suse.de>
8701 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
8702 (graphite-max-nb-scop-params): Document special value zero.
8703 * domwalk.h (dom_walker::STOP): New symbolical constant.
8704 (dom_walker::dom_walker): Add optional parameter for bb to
8706 (dom_walker::~dom_walker): Declare.
8707 (dom_walker::before_dom_children): Document STOP return value.
8708 (dom_walker::m_user_bb_to_rpo): New member.
8709 (dom_walker::m_bb_to_rpo): Likewise.
8710 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
8711 mapping here if not provided by the user.
8712 (dom_walker::~dom_walker): Free bb to RPO mapping if not
8713 provided by the user.
8714 (dom_walker::STOP): Define.
8715 (dom_walker::walk): Do not compute bb to RPO mapping here.
8716 Support STOP return value from before_dom_children to stop
8718 * graphite-optimize-isl.c (optimize_isl): If the schedule
8719 is the same still generate code if -fgraphite-identity
8720 or -floop-parallelize-all are given.
8721 * graphite-scop-detection.c: Include cfganal.h.
8722 (gather_bbs::gather_bbs): Get and pass through bb to RPO
8724 (gather_bbs::before_dom_children): Return STOP for BBs
8726 (build_scops): Compute bb to RPO mapping and pass it to
8727 the domwalk. Treat --param graphite-max-nb-scop-params=0
8728 as not limiting the number of params.
8729 * graphite.c (graphite_initialize): Remove limit on the
8730 number of basic-blocks in a function.
8731 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
8732 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
8733 default value of 10.
8735 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8737 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
8738 Adjust code to eliminate needing to do the shift right 32-bits
8739 operation after XSCVDPSPN.
8741 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8743 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
8744 ((X / Y) != 0 -> X >= Y): Likewise.
8746 2017-09-26 Carl Love <cel@us.ibm.com>
8748 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
8749 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
8750 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
8751 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
8752 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
8753 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
8754 definitions and overloading.
8755 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
8756 statement for P9V_BUILTIN_XST_LEN_R.
8757 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
8758 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
8759 define_expand and define_insn for the instructions and builtins.
8760 * doc/extend.texi: Update the built-in documentation file for the new
8762 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
8763 define_insn for the instructions
8765 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
8768 * gcc/config/netbsd-protos.h: New file.
8769 * gcc/config/netbsd.c: New file.
8770 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
8771 * gcc/config/t-netbsd: New file.
8772 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
8773 (tmake_file) Add t-netbsd.
8774 (extra_objs) Add netbsd.o.
8776 2017-09-26 Janus Weil <janus@gcc.gnu.org>
8780 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
8782 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8784 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
8785 sign extension from a vector register to a GPR by doing a 32-bit
8786 direct move and then an EXTSW.
8787 (extendsi<mode>2 splitter): Likewise.
8788 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
8789 right or vector extract after doing XSCVDPSPN. Use
8790 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
8792 (movdi_from_sf_zero_ext): Likewise.
8793 (reload_gpr_from_vsxsf): Likewise.
8794 (p8_mfvsrd_4_disf): Delete, no longer used.
8795 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
8796 then needing to move the SFmode to a GPR to use the XSCVDPSP
8797 instruction instead of FRSP and XSCVDPSPN.
8798 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
8799 it is adjacent to the other XSCVSPDP insns.
8800 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
8801 SFmode to be in traditional Altivec registers.
8802 (vsx_xscvdpspn): Eliminate useless alternative constraint.
8803 (vsx_xscvspdpn): Likewise.
8804 (vsx_xscvspdpn_scalar): Likewise.
8806 2017-09-26 Martin Jambor <mjambor@suse.cz>
8808 * tree-sra.c (compare_access_positions): Put integral types first,
8809 stabilize sorting of integral types, remove conditions putting
8810 non-full-precision integers last.
8811 (sort_and_splice_var_accesses): Disable scalarization if a
8812 non-integert would be represented by a non-full-precision integer.
8814 2017-09-26 Joseph Myers <joseph@codesourcery.com>
8816 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
8817 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
8818 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
8819 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
8820 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
8821 conditionals inside the function instead of around it. Call
8822 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
8823 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
8825 2017-09-26 Richard Biener <rguenther@suse.de>
8827 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
8829 (scop_detection::build_scop_breadth): ... this. Removed.
8830 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
8831 (scop_detection::harmful_stmt_in_bb): Likewise.
8832 (scop_detection::graphite_can_represent_stmt): Likewise.
8833 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
8834 (scop_detection::can_represent_loop): Remove recursion, fold in ...
8835 (scop_detection::can_represent_loop_1): ... this. Removed.
8836 (scop_detection::harmful_loop_in_region): Simplify after inlining
8837 the above and remove more quadraticness.
8838 (build_scops): Adjust.
8839 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
8842 2017-09-26 Jakub Jelinek <jakub@redhat.com>
8845 * config/i386/i386.c (ix86_print_operand_address_as): Only test
8846 REGNO (base) == SP_REG if base is a REG.
8849 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
8850 if it is different SSA_NAME.
8851 (optimize_range_tests_cmp_bitwise): New function.
8852 (optimize_range_tests): Call it.
8854 2017-09-26 Richard Biener <rguenther@suse.de>
8856 PR tree-optimization/82321
8857 * graphite.c (canonicalize_loop_closed_ssa): Properly check
8858 for the def being inside the loop.
8860 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8862 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
8864 * config/s390/s390-builtins.def: Fix constraint on op4.
8866 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8868 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
8869 independent expanders.
8870 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
8871 ("vec_ordered", "vec_unordered"): New expanders.
8873 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8875 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
8878 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8880 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
8881 vec_unpacks_lo_v16qi.
8882 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
8884 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8886 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
8887 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
8888 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
8890 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8892 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
8894 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
8896 ("*vec_slb<mode>"): New insn pattern.
8897 ("vec_shr_<mode>"): New expander.
8898 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
8899 and force the shift count operand to V16QImode.
8900 ("vec_srb<mode>"): Set shift count mode to V16QI.
8902 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8904 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
8905 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
8906 ("vec_widen_smult_hi_<mode>"): New expander definitions.
8908 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
8911 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
8913 2017-09-26 Richard Biener <rguenther@suse.de>
8915 PR tree-optimization/82320
8916 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
8919 2017-09-25 Jeff Law <law@redhat.com>
8921 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
8922 prototype for new argument.
8923 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
8924 mostly extracted from rs6000_emit_allocate_stack.
8925 (rs6000_emit_probe_stack_range_stack_clash): New function.
8926 (rs6000_emit_allocate_stack): Call
8927 rs6000_emit_probe_stack_range_stack_clash as needed.
8928 (rs6000_emit_probe_stack_range): Add additional argument
8929 to call to gen_probe_stack_range{si,di}.
8930 (output_probe_stack_range): New.
8931 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
8932 (output_probe_stack_range_stack_clash): New.
8933 (rs6000_emit_prologue): Emit notes into dump file as requested.
8934 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
8935 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
8936 Add additional operand and pass it to output_probe_stack_range.
8938 2017-09-25 Bin Cheng <bin.cheng@arm.com>
8940 PR tree-optimization/82163
8941 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
8942 (checking_verify_loop_closed_ssa): New parameter.
8943 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
8944 (check_loop_closed_ssa_stmt): Delete.
8945 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
8946 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
8947 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
8950 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
8952 * brig-builtins.def: Treat HSAIL barrier builtins as
8953 setjmp/longjump style functions.
8955 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
8957 * target.def (constant_alignment): New hook.
8958 * defaults.h (CONSTANT_ALIGNMENT): Delete.
8959 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
8960 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
8961 * doc/tm.texi: Regenerate.
8962 * targhooks.h (default_constant_alignment): Declare.
8963 (constant_alignment_word_strings): Likewise.
8964 * targhooks.c (default_constant_alignment): New function.
8965 (constant_alignment_word_strings): Likewise.
8966 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
8967 instead of CONSTANT_ALIGNMENT.
8968 * varasm.c (align_variable, get_variable_align, build_constant_desc)
8969 (force_const_mem): Likewise.
8970 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
8971 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
8972 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
8973 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8974 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
8976 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
8977 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8978 constant_alignment_word_strings.
8979 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
8980 (CONSTANT_ALIGNMENT): Likewise.
8981 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8982 (arm_constant_alignment): New function.
8983 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
8984 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8985 constant_alignment_word_strings.
8986 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
8987 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8988 constant_alignment_word_strings.
8989 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
8990 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8991 (cris_constant_alignment): New function.
8992 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
8993 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8994 (epiphany_constant_alignment): New function.
8995 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
8996 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8997 constant_alignment_word_strings.
8998 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
8999 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9000 constant_alignment_word_strings.
9001 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
9002 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9003 constant_alignment_word_strings.
9004 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
9005 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
9006 * config/i386/i386.c (ix86_constant_alignment): Make static.
9007 Use the same interface as the target hook.
9008 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9009 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
9010 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9011 constant_alignment_word_strings.
9012 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
9013 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
9014 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9015 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
9016 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9017 constant_alignment_word_strings.
9018 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
9019 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9020 constant_alignment_word_strings.
9021 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
9022 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9023 constant_alignment_word_strings.
9024 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
9025 * config/microblaze/microblaze.c (microblaze_constant_alignment):
9027 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9028 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
9029 * config/mips/mips.c (mips_constant_alignment): New function.
9030 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9031 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
9032 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
9033 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9034 (mmix_constant_alignment): Make static. Use the same interface
9036 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
9037 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9038 constant_alignment_word_strings.
9039 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
9040 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9041 constant_alignment_word_strings.
9042 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
9043 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9044 constant_alignment_word_strings.
9045 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
9046 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9047 (rs6000_constant_alignment): New function.
9048 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
9049 * config/riscv/riscv.c (riscv_constant_alignment): New function.
9050 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9051 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
9052 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9053 (rs6000_constant_alignment): New function.
9054 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
9055 * config/s390/s390.c (s390_constant_alignment): New function.
9056 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9057 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
9058 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9059 constant_alignment_word_strings.
9060 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
9061 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9062 (sparc_constant_alignment): New function.
9063 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
9064 * config/spu/spu.c (spu_constant_alignment): New function.
9065 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9066 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
9067 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9068 constant_alignment_word_strings.
9069 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
9070 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9071 constant_alignment_word_strings.
9072 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
9073 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9074 constant_alignment_word_strings.
9075 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
9076 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9077 (visium_constant_alignment): New function.
9078 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
9079 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9080 (xtensa_constant_alignment): New function.
9081 * system.h (CONSTANT_ALIGNMENT): Poison.
9083 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
9085 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9086 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
9087 (rs6000_builtin_valid_without_lhs): New helper function.
9088 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9089 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
9091 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
9093 * target.h (vec_perm_indices): Use unsigned short rather than
9095 (auto_vec_perm_indices): Likewise.
9096 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
9097 Use unsigned int rather than unsigned char.
9098 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
9100 2017-09-25 Richard Biener <rguenther@suse.de>
9102 * cfgloop.h (sort_sibling_loops): Declare.
9103 * cfgloop.c (sort_sibling_loops_cmp): New helper.
9104 (sort_sibling_loops): New function sorting the sibling loop list
9106 * graphite.c (graphite_transform_loops): Sort sibling loops.
9108 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
9110 * target.def (vec_perm_const_ok): Change sel parameter to
9112 * optabs-query.c (can_vec_perm_p): Update accordingly.
9113 * doc/tm.texi: Regenerate.
9114 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
9115 auto_vec_perm_indices and remove separate nelt field.
9116 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
9117 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
9118 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
9119 (aarch64_expand_vec_perm_const): Update accordingly.
9120 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
9121 to vec_perm_indices.
9122 * config/arm/arm.c (expand_vec_perm_d): Change perm to
9123 auto_vec_perm_indices and remove separate nelt field.
9124 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
9125 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
9126 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
9128 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
9129 to vec_perm_indices.
9130 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
9131 sel to vec_perm_indices.
9132 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
9133 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
9134 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
9136 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
9139 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
9142 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
9143 on the FUNCTION_DECL function context if it has a DIE that is a
9146 2017-09-25 Richard Biener <rguenther@suse.de>
9148 PR tree-optimization/82285
9149 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
9152 2017-09-25 Tom de Vries <tom@codesourcery.com>
9156 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
9159 2017-09-25 Richard Biener <rguenther@suse.de>
9161 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
9162 ISL errors other than isl_error_quota happen. Dump if the
9163 schedule is the same.
9164 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
9165 errors instead of aborting inside ISL.
9167 2017-09-25 Iain Sandoe <iain@codesourcery.com>
9170 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
9171 of libgcc_eh for m64.
9172 * config/i386/darwin64.h: Likewise.
9174 2017-09-25 Richard Biener <rguenther@suse.de>
9177 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
9178 attribute for incomplete types nor twice for complete ones.
9180 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
9183 * config/i386/i386.c (ix86_print_operand_address_as): Encode
9184 %esp as %rsp to avoid 0x67 prefix if there is no index or base
9187 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
9190 * config/i386/i386.opt (mprefer-avx256): Use
9191 ix86_target_flags variable.
9192 * config/i386/i386.c (ix86_target_string): Move
9193 -mprefer-avx256 to flag2_opts.
9195 2017-09-22 Jakub Jelinek <jakub@redhat.com>
9198 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
9199 and x != -1 | y != -1 into (x & y) != -1.
9201 2017-09-22 Steve Ellcey <sellcey@cavium.com>
9203 * config.gcc: Add new case statement to set
9204 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
9205 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
9206 s390x-*-linux* case statements. Added aarch64 to the list of
9207 supported architectures.
9209 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9211 PR tree-optimization/82289
9212 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
9213 STMT_VINFO_RELEVANT_P.
9215 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9216 Alan Hayward <alan.hayward@arm.com>
9217 David Sherwood <david.sherwood@arm.com>
9219 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
9220 for VR_RANGE only; don't allow VR_ANTI_RANGE.
9221 (extract_range_from_binary_expr_1): Don't call
9222 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
9224 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9225 Alan Hayward <alan.hayward@arm.com>
9226 David Sherwood <david.sherwood@arm.com>
9228 * target.def (preferred_vector_alignment): New hook.
9229 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
9231 * doc/tm.texi: Regenerate.
9232 * targhooks.h (default_preferred_vector_alignment): Declare.
9233 * targhooks.c (default_preferred_vector_alignment): New function.
9234 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
9236 (DR_TARGET_ALIGNMENT): New macro.
9237 (aligned_access_p): Update commentary.
9238 (vect_known_alignment_in_bytes): New function.
9239 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
9241 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
9242 Calculate the misalignment based on the target alignment rather than
9244 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
9245 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
9246 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
9247 the target alignment, rather than masking the element misalignment
9248 with the number of elements in a vector. Also use the target
9249 alignment when calculating the maximum number of peels.
9250 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
9251 instead of TYPE_ALIGN_UNIT.
9252 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
9253 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
9254 (vect_create_addr_base_for_vector_ref): Update call accordingly.
9255 (vect_create_data_ref_ptr): Likewise.
9256 (vect_setup_realignment): Realign by ANDing with
9257 -DR_TARGET_MISALIGNMENT.
9258 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
9259 the number of peels based on DR_TARGET_ALIGNMENT.
9260 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
9261 with the guaranteed alignment boundary when deciding whether
9263 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
9264 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
9265 (ensure_base_align): Remove stmt_info parameter. Get the
9266 target base alignment from DR_TARGET_ALIGNMENT.
9267 (vectorizable_store): Update call accordingly. Interpret
9268 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
9270 (vectorizable_load): Likewise.
9272 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9273 Alan Hayward <alan.hayward@arm.com>
9274 David Sherwood <david.sherwood@arm.com>
9276 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
9277 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
9278 (vect_enhance_data_refs_alignment): Likewise.
9280 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
9282 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9283 error. Only quit immediately if parsing is complete.
9284 (BEGIN): Initialize fatal_err and parse_done.
9285 (begin fpu, end fpu): Check number of arguments.
9286 (begin arch, end arch): Likewise.
9287 (begin cpu, end cpu): Likewise.
9288 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9289 (optalias): Likewise.
9291 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
9293 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9294 * config/arm/arm-isa.h: Delete. Move definitions to ...
9295 * arm-cpus.in: ... here. Use new feature and fgroup values.
9296 * config/arm/arm.c (arm_option_override): Use lower case for feature
9298 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9299 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9300 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9302 (print_isa_bits_for): New function.
9303 (gen_isa): New function.
9304 (gen_comm_data): Use print_isa_bits_for.
9305 (define feature): New keyword.
9306 (define fgroup): New keyword.
9307 * config/arm/t-arm (TM_H): Remove.
9308 (GTM_H): Add arm-isa.h.
9309 (arm-isa.h): Add rule to generate file.
9310 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9311 case for feature bit names.
9313 2017-09-22 Richard Biener <rguenther@suse.de>
9315 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
9317 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
9318 print of no dependency loops ...
9319 * graphite.c (graphite_transform_loops): ... here.
9320 (canonicalize_loop_closed_ssa_form): Work from inner to outer
9322 (same_close_phi_node, remove_duplicate_close_phi,
9323 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
9324 (canonicalize_loop_closed_ssa): ... here and simplify.
9325 * graphite-optimize-isl.c: Include tree-vectorizer.h.
9326 (optimize_isl): Use dump_printf_loc to tell when we stopped
9327 optimizing because of an ISL timeout.
9329 2017-09-22 Richard Biener <rguenther@suse.de>
9331 PR tree-optimization/82291
9332 * tree-if-conv.c (predicate_mem_writes): Make sure to
9333 remove writes in blocks predicated with false.
9335 2017-09-22 Richard Biener <rguenther@suse.de>
9337 * sese.c: Include cfganal.h.
9338 (if_region_set_false_region): Remove.
9339 (create_if_region_on_edge): Likewise.
9340 (move_sese_in_condition): Re-implement without destroying
9343 2017-09-22 Richard Biener <rguenther@suse.de>
9345 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
9346 Verify both BBs contain loop PHI nodes before dispatching to
9348 (graphite_regenerate_ast_isl): Do not recompute dominators,
9349 do not verify three times. Restructure for clarity.
9350 * graphite-scop-detection.c (same_close_phi_node,
9351 remove_duplicate_close_phi, make_close_phi_nodes_unique,
9352 defined_in_loop_p, canonicalize_loop_closed_ssa,
9353 canonicalize_loop_closed_ssa_form): Simplify, remove excess
9354 checking and SSA rewrite, move to ...
9355 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
9356 (graphite_initialize): Do not pass in ctx, do not reset the
9357 SCEV cache, compute only dominators.
9358 (graphite_transform_loops): Allocate ISL ctx after
9359 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
9360 Maintain post-dominators only around build_scops.
9361 * sese.c (if_region_set_false_region): Make static. Free
9362 and recompute dominators.
9363 (move_sese_in_condition): Assert we don't get called with
9364 post-dominators computed.
9365 * sese.h (if_region_set_false_region): Remove.
9367 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
9369 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
9370 mode attribute for TARGET_AVX512VL.
9372 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
9374 * config/i386/i386.opt (mprefer-avx256): New option.
9375 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
9377 (ix86_preferred_simd_mode): Return 256-bit AVX modes
9378 for TARGET_PREFER_AVX256.
9379 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
9381 2017-09-21 Jeff Law <law@redhat.com>
9383 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
9384 Fix dump output if the only stack space is for pushed registers.
9386 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9388 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
9391 2017-09-21 Martin Sebor <msebor@redhat.com>
9394 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
9395 code (in C++) or code that triggers warnings.
9397 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
9399 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
9401 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
9403 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
9404 * sched-rgn.c: Ditto.
9405 * sel-sched-ir.c: Ditto.
9407 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
9409 * toplev.h (set_random_seed): Adjust return type.
9410 * toplev.c (init_local_tick): Move eager initialization of random_seed
9411 to get_random_seed. Adjust comment.
9412 (init_random_seed): Inline to get_random_seed, delete.
9413 (get_random_seed): Initialize random_seed lazily.
9414 (set_random_seed): Do not return previous value.
9415 (print_switch_value): Do not call get_random_seed.
9417 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
9419 * cgraph.c (delete_function_version): New, broken out from...
9420 (cgraph_node::delete_function_version): ...here. Rename to
9421 cgraph_node::delete_function_version_by_decl. Update all uses.
9422 (cgraph_node::remove): Call delete_function_version.
9424 2017-09-21 Jakub Jelinek <jakub@redhat.com>
9427 * tree-inline.c (expand_call_inline): Emit clobber stmts for
9428 VAR_DECLs to which addressable non-volatile parameters are mapped
9429 and for id->retvar after the return value assignment. Clear
9430 id->retval and id->retbnd after inlining.
9432 2017-09-21 Richard Biener <rguenther@suse.de>
9434 PR tree-optimization/82276
9435 PR tree-optimization/82244
9436 * tree-vrp.c (build_assert_expr_for): Set
9437 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
9439 (remove_range_assertions): Revert earlier change.
9441 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
9444 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
9446 2017-09-21 Richard Biener <rguenther@suse.de>
9448 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9449 Restore valid IL after code generation errors.
9450 * graphite.c (graphite_transform_loops): Diagnose code
9451 generation issues as MSG_MISSED_OPTIMIZATION and continue
9452 with processing SCOPs.
9454 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9455 Alan Hayward <alan.hayward@arm.com>
9456 David Sherwood <david.sherwood@arm.com>
9458 * calls.c (compute_argument_addresses): Use simplify_gen_binary
9459 rather than choosing between plus_constant and gen_rtx_<CODE>.
9460 * expr.c (emit_push_insn): Likewise.
9461 (expand_expr_real_2): Likewise.
9463 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9464 Alan Hayward <alan.hayward@arm.com>
9465 David Sherwood <david.sherwood@arm.com>
9467 * loop-unroll.c (split_iv): Call copy_rtx on the step.
9469 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9470 Alan Hayward <alan.hayward@arm.com>
9471 David Sherwood <david.sherwood@arm.com>
9473 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
9474 calling tree_to_uhwi.
9476 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9477 Alan Hayward <alan.hayward@arm.com>
9478 David Sherwood <david.sherwood@arm.com>
9480 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
9481 INTEGER_CST rather than a negative test for ADDR_EXPR.
9483 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9484 Alan Hayward <alan.hayward@arm.com>
9485 David Sherwood <david.sherwood@arm.com>
9487 * tree-vrp.c (extract_range_from_binary_expr_1): Check
9488 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
9490 2017-09-21 Richard Biener <rguenther@suse.de>
9492 PR tree-optimization/71351
9493 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
9494 graphite_create_new_loop_guard): Remove, fold remaining parts
9496 (translate_isl_ast_node_for): ... here and simplify.
9498 2017-09-21 Jakub Jelinek <jakub@redhat.com>
9501 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
9502 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
9503 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
9504 alternative always use QI mode, for -Os imov (=R,R) alternative
9505 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
9508 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9509 Jeff Law <law@redhat.com>
9511 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
9512 (allocate_stack_space): New function, partially extracted from
9514 (s390_emit_prologue): Track offset to most recent stack probe.
9515 Code to allocate space moved into allocate_stack_space.
9516 Dump actions when no stack is allocated.
9517 (s390_prologue_plus_offset): New function.
9518 (s390_emit_stack_probe): Likewise.
9520 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
9522 * common.opt (Wa, Wl, Wp, g, gz=): Add
9524 (gno-column-info): Remove.
9525 (gcolumn-info): Drop RejectNegative.
9527 (gno-record-gcc-switches): Remove.
9528 (grecord-gcc-switches): Drop RejectNegative.
9529 (gno-split-dwarf): Remove.
9530 (gsplit-dwarf): Drop RejectNegative.
9531 (gno-strict-dwarf): Remove.
9532 (gstrict-dwarf): Drop RejectNegative.
9533 * config/darwin.opt (gfull, gused): Add RejectNegative.
9534 * dwarf2out.c (gen_producer_string): Drop
9535 gno-record-gcc-switches handler.
9536 * optc-gen.awk: Add g to prefixes with negative forms.
9537 * opts-common.c (remapping_prefix_p): New.
9538 (find_opt): Check it.
9539 (generate_canonical_option): Test g prefix.
9540 (option_map): Add -gno- mapping.
9541 (add_misspelling_candidates): Check remapping_prefix_p.
9543 2017-09-20 Jeff Law <law@redhat.com>
9545 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
9546 thinko in stack clash protection support.
9548 * explow.c (compute_stack_clash_protection_loop_data): Use
9549 CONST_INT_P instead of explicit test. Verify object is a
9550 CONST_INT_P before looking at INTVAL.
9551 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
9552 instead of explicit test.
9554 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
9557 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
9558 address instead of to r1 and r11.
9560 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
9562 * config.gcc: Support "knm".
9563 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
9564 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
9566 * config/i386/i386.c (m_KNM): Define.
9567 (processor_target_table): Add "knm".
9569 (ix86_option_override_internal): Add "knm".
9570 (ix86_issue_rate): Add PROCESSOR_KNM.
9571 (ix86_adjust_cost): Ditto.
9572 (ia32_multipass_dfa_lookahead): Ditto.
9573 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
9574 (fold_builtin_cpu): Add M_INTEL_KNM.
9575 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
9576 (processor_type): Add PROCESSOR_KNM.
9577 * config/i386/x86-tune.def: Add m_KNM.
9578 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
9580 2017-09-20 Richard Biener <rguenther@suse.de>
9582 PR tree-optimization/80213
9583 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
9584 are allowed in empty BBs as well.
9585 (canonicalize_loop_closed_ssa): Also look for other complex
9587 (scop_detection::get_sese): Include the loop-closed PHI block
9589 (scop_detection::merge_sese): Remove code adding extra blocks.
9590 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
9591 (build_scops): Assert the final returned scop is invalid.
9593 2017-09-20 Richard Biener <rguenther@suse.de>
9595 PR tree-optimization/82264
9596 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
9598 (vn_phi_lookup): Likewise.
9599 (vn_phi_insert): Likewise.
9601 2017-09-20 Jakub Jelinek <jakub@redhat.com>
9603 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
9604 that fits into uhwi or shwi, add DW_AT_const_value regardless
9605 of early_dwarf without going through RTL, using add_AT_unsigned
9608 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
9609 (DEBUG_LTO_ABBREV_SECTION): Likewise.
9610 (DEBUG_LTO_MACINFO_SECTION): Likewise.
9611 (DEBUG_MACRO_SECTION): Likewise.
9612 (DEBUG_LTO_MACRO_SECTION): Likewise.
9613 (DEBUG_STR_DWO_SECTION): Likewise.
9614 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
9615 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
9616 (DEBUG_LTO_DWO_LINE_SECTION): Define.
9617 (DEBUG_LTO_LINE_STR_SECTION): Define.
9618 (init_sections_and_labels): Initialize debug_line_str_section
9619 variable. Initialize debug_loc_section for -gdwarf-5 to
9620 DEBUG_LOCLISTS_SECTION. Formatting fixes.
9622 2017-09-20 Richard Biener <rguenther@suse.de>
9624 * graphite-sese-to-poly.c (extract_affine): Properly handle
9625 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
9627 2017-09-20 Richard Biener <rguenther@suse.de>
9629 PR tree-optimization/81373
9630 * graphite-scop-detection.c (build_cross_bb_scalars_def):
9631 Force SESE live-out defs to be handled even if they are
9634 2017-09-19 Jeff Law <law@redhat.com>
9636 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
9637 nothing for stack adjustments with REG_STACK_CHECK.
9638 * sched-deps.c (parse_add_or_inc): Reject insns with
9639 REG_STACK_CHECK from dependency breaking.
9640 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
9641 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
9642 * reg-notes.def (STACK_CHECK): New note.
9644 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
9645 (ix86_expand_prologue): Dump stack clash info as needed.
9646 Call ix86_adjust_stack_and_probe_stack_clash as needed.
9648 * function.c (dump_stack_clash_frame_info): New function.
9649 * function.h (dump_stack_clash_frame_info): Prototype.
9650 (enum stack_clash_probes): New enum.
9652 * config/alpha/alpha.c (alpha_expand_prologue): Also check
9653 flag_stack_clash_protection.
9654 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
9655 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
9656 (arm_frame_pointer_required): Likewise.
9657 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
9658 (ia64_expand_prologue): Likewise.
9659 * config/mips/mips.c (mips_expand_prologue): Likewise.
9660 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
9661 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
9662 (sparc_flat_expand_prologue): Likewise.
9663 * config/spu/spu.c (spu_expand_prologue): Likewise.
9665 * explow.c: Include "params.h".
9666 (anti_adjust_stack_and_probe_stack_clash): New function.
9667 (get_stack_check_protect): Likewise.
9668 (compute_stack_clash_protection_loop_data): Likewise.
9669 (emit_stack_clash_protection_loop_start): Likewise.
9670 (emit_stack_clash_protection_loop_end): Likewise.
9671 (allocate_dynamic_stack_space): Use get_stack_check_protect.
9672 Use anti_adjust_stack_and_probe_stack_clash.
9673 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
9674 (emit_stack_clash_protection_loop_start): Likewise.
9675 (emit_stack_clash_protection_loop_end): Likewise.
9676 * rtl.h (get_stack_check_protect): Prototype.
9677 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
9678 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
9679 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
9681 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
9683 * doc/tm.texi: Rebuilt.
9684 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
9685 get_stack_check_protect.
9686 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
9687 * config/arm/arm.c (arm_expand_prologue): Likewise.
9688 (arm_frame_pointer_required): Likewise.
9689 * config/i386/i386.c (ix86_expand_prologue): Likewise.
9690 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
9691 * config/mips/mips.c (mips_expand_prologue): Likewise.
9692 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
9693 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
9694 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
9695 (sparc_flat_expand_prologue): Likewise.
9697 * common.opt (-fstack-clash-protection): New option.
9698 * flag-types.h (enum stack_check_type): Note difference between
9699 -fstack-check= and -fstack-clash-protection.
9700 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
9701 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
9702 * toplev.c (process_options): Issue warnings/errors for cases
9703 not handled with -fstack-clash-protection.
9704 * doc/invoke.texi (-fstack-clash-protection): Document new option.
9705 (-fstack-check): Note additional problem with -fstack-check=generic.
9706 Note that -fstack-check is primarily for Ada and refer users
9707 to -fstack-clash-protection for stack-clash-protection.
9708 Document new params for stack clash protection.
9710 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
9712 * config/i386/i386.c (ix86_split_long_move): Do not handle
9713 address used for LEA in a special way.
9715 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
9717 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
9719 2017-09-19 Martin Sebor <msebor@redhat.com>
9722 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
9723 of incompatible types.
9725 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
9727 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9728 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
9729 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9730 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
9732 2017-09-19 Richard Biener <rguenther@suse.de>
9734 PR tree-optimization/82244
9735 * tree-vrp.c (remove_range_assertions): Do not propagate
9736 a constant to abnormals but replace the assert with a copy.
9738 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
9740 PR rtl-optimization/57878
9741 PR rtl-optimization/68988
9742 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
9743 avoidance test involving non_reload_pseudos. Move frequency test
9744 below the general fragmentation avoidance test.
9746 2017-09-19 Richard Biener <rguenther@suse.de>
9748 PR tree-optimization/69728
9749 * graphite-sese-to-poly.c (schedule_error): New global.
9750 (add_loop_schedule): Handle empty domain by failing the
9752 (build_original_schedule): Handle schedule_error.
9754 2017-09-19 Richard Biener <rguenther@suse.de>
9756 * graphite-scop-detection.c (scop_detection::can_represent_loop):
9757 Do not iterate to sibling loops but only to siblings of inner
9760 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
9763 * config/m68k/m68k.md (moveq feeding equality comparison): Check
9764 that the registers are different.
9766 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
9768 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
9769 to processor_model and "amdfam17h" to arch_names_table.
9770 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
9772 2017-09-18 Jakub Jelinek <jakub@redhat.com>
9775 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
9777 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9778 Alan Hayward <alan.hayward@arm.com>
9779 David Sherwood <david.sherwood@arm.com>
9781 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
9783 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
9785 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
9786 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
9787 vect_schedule_slp_instance.
9788 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
9789 Update call to vect_slp_analyze_node_operations. Simplify return
9791 (vect_slp_analyze_bb_1): Update call accordingly.
9792 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
9793 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
9794 (vect_schedule_slp): Update call accordingly.
9796 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9797 Alan Hayward <alan.hayward@arm.com>
9798 David Sherwood <david.sherwood@arm.com>
9800 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
9801 with types that aren't in fact scalar.
9803 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9805 * tree-vect-slp.c (vect_record_max_nunits): New function,
9807 (vect_build_slp_tree_1): ...here.
9808 (vect_build_slp_tree_2): Call it for phis too.
9810 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9812 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
9813 to vect_get_vec_def_for_operand when getting the mask operand.
9815 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9816 Alan Hayward <alan.hayward@arm.com>
9817 David Sherwood <david.sherwood@arm.com>
9819 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
9822 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9823 Alan Hayward <alan.hayward@arm.com>
9824 David Sherwood <david.sherwood@arm.com>
9826 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
9827 calculation for vector booleans.
9829 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9830 Alan Hayward <alan.hayward@arm.com>
9831 David Sherwood <david.sherwood@arm.com>
9833 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
9835 (vect_transform_stmt): ...here.
9836 (vect_analyze_stmt): Use it instead of calling
9837 vectorizable_live_operation directly.
9839 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
9841 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
9842 non-SIMT targets in acc vector loops.
9844 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
9846 * configure.ac: Add arc and check if assembler supports gdwarf2.
9847 * configure: Regenerate.
9849 2017-09-18 Richard Biener <rguenther@suse.de>
9851 PR tree-optimization/82220
9852 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
9853 epilogue niters from the min_profitable_iters compute.
9855 2017-09-18 Jakub Jelinek <jakub@redhat.com>
9858 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
9859 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
9860 (ix86_init_pic_reg): Revert 2017-09-01 changes.
9862 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
9865 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
9866 switching to a new text section.
9868 2017-09-18 Richard Biener <rguenther@suse.de>
9870 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
9872 (build_alias_set): Reject aliases with no access function.
9874 2017-09-18 Richard Biener <rguenther@suse.de>
9876 PR tree-optimization/79622
9877 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
9879 (build_cross_bb_scalars_use): Likewise.
9881 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
9883 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
9885 2017-09-18 Alan Modra <amodra@gmail.com>
9888 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
9889 stack_pointer_rtx for count 0. Update comments. Break up
9890 large rtl expression.
9892 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
9894 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
9895 Increase to 20 bytes.
9896 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
9897 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
9898 or avx version of the stub.
9900 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
9903 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
9904 compute the minimum stack alignment. Also update preferred stack
9905 boundary for leaf functions.
9907 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
9909 PR tree-optimization/82228
9910 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
9913 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
9915 * common/config/nds32/nds32-common.c
9916 (nds32_option_optimization_table): Refine formatting.
9917 (nds32_option_optimization_table): Use -fsched-pressure and
9918 -fomit-frame-pointer for specific optimization level.
9920 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
9922 * config/nds32/nds32.c: Refine formatting and comments.
9923 * config/nds32/nds32.h: Likewise.
9924 * config/nds32/nds32.md: Likewise.
9925 * config/nds32/nds32-cost.c: Likewise.
9926 * config/nds32/nds32-isr.c: Likewise.
9927 * config/nds32/nds32-md-auxiliary.c: Likewise.
9928 * config/nds32/nds32-multiple.md: Likewise.
9929 * config/nds32/nds32-predicates.c: Likewise.
9931 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
9932 Jakub Jelinek <jakub@redhat.com>
9934 Add support for -std=c++2a.
9935 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
9937 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
9939 2017-09-15 Steve Ellcey <sellcey@cavium.com>
9942 * doc/extend.texi (Common Function Attributes): Add
9943 references to ARM, AArch64, and S/390 specific attributes.
9944 (Function Specific Option Pragmas): Add AArch64 and S/390
9945 to list of back ends that support the target pragma.
9947 2017-09-15 Nathan Sidwell <nathan@acm.org>
9949 * doc/standards.texi: Fix C++17 description. Update URLs for
9952 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
9954 * common.opt (Wcast-align=strict): New warning option.
9955 * doc/invoke.texi: Document -Wcast-align=strict.
9957 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
9959 * cgraph.h (cgraph_thunk_info): Add comments.
9960 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
9961 assert for VIRTUAL_* arguments stricter.
9963 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
9965 PR tree-optimization/71026
9966 * match.pd: Move RDIV patterns from fold-const.c
9967 * fold-const.c (distribute_real_division): Removed.
9968 (fold_binary_loc): Remove calls to distribute_real_divison.
9970 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9972 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
9973 c++1z and gnu++1z as deprecated. Change other references to
9974 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
9975 Change -Wc++1z-compat to -Wc++17-compat.
9976 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
9977 * dwarf2out.c (highest_c_language): Handle C++17.
9978 (gen_compile_unit_die): Likewise.
9980 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9982 PR rtl-optimization/82192
9983 * combine.c (make_extraction): Don't look through non-paradoxical
9984 SUBREGs or TRUNCATE if pos + len is or might be bigger than
9987 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
9988 Alan Hayward <alan.hayward@arm.com>
9989 David Sherwood <david.sherwood@arm.com>
9991 * target.def (function_arg_offset): New hook.
9992 * targhooks.h (default_function_arg_offset): Declare.
9993 * targhooks.c (default_function_arg_offset): New function.
9994 * function.c (locate_and_pad_parm): Use
9995 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
9996 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
9997 (TARGET_FUNCTION_ARG_OFFSET): ...this.
9998 * doc/tm.texi: Regenerate.
9999 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
10000 * config/spu/spu.c (spu_function_arg_offset): New function.
10001 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
10002 * system.h (FUNCTION_ARG_OFFSET): Poison.
10004 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10005 Alan Hayard <alan.hayward@arm.com>
10006 David Sherwood <david.sherwood@arm.com>
10008 * target.def (truly_noop_truncation): New hook.
10009 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
10010 than TRULY_NOOP_TRUNCATION.
10011 * hooks.h (hook_bool_uint_uint_true): Declare.
10012 * hooks.c (hook_bool_uint_uint_true): New function.
10013 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
10014 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
10015 * doc/tm.texi: Regenerate.
10016 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
10017 rather than TRULY_NOOP_TRUNCATION in comments.
10018 (simplify_comparison): Likewise.
10019 (record_truncated_value): Likewise.
10020 * expmed.c (extract_bit_field_1): Likewise.
10021 (extract_split_bit_field): Likewise.
10022 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
10023 instead of TRULY_NOOP_TRUNCATION.
10024 * function.c (assign_parm_setup_block): Likewise.
10025 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
10026 * rtlhooks.c: Include target.h.
10027 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
10028 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
10029 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
10030 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
10031 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
10032 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
10033 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
10034 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
10035 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
10036 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
10037 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
10038 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
10039 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
10040 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
10041 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
10042 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
10043 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
10044 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
10045 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
10046 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
10047 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
10048 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
10049 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
10050 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
10051 * config/mips/mips.c (mips_truly_noop_truncation): New function.
10052 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10053 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10054 rather than TRULY_NOOP_TRUNCATION in comments.
10055 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
10056 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
10057 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
10058 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
10059 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
10060 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
10061 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
10062 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
10063 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
10064 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
10065 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
10066 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10067 rather than TRULY_NOOP_TRUNCATION in comments.
10068 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
10069 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
10070 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
10071 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
10072 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
10073 TRULY_NOOP_TRUNCATION condition.
10074 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
10075 (TRULY_NOOP_TRUNCATION): Delete.
10076 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
10077 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
10078 * config/spu/spu.c (spu_truly_noop_truncation): New function.
10079 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10080 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
10081 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
10082 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
10083 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10084 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10085 rather than TRULY_NOOP_TRUNCATION in comments.
10086 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
10087 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
10088 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
10089 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
10090 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
10091 * system.h (TRULY_NOOP_TRUNCATION): Poison.
10093 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
10096 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
10097 (*cmp_ior): Likewise.
10098 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
10099 (*ior_scc_scc_cmp): Likewise.
10100 (*and_scc_scc): Likewise.
10101 (*and_scc_scc_cmp): Likewise.
10103 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10104 Alan Hayard <alan.hayward@arm.com>
10105 David Sherwood <david.sherwood@arm.com>
10107 * target.def (can_change_mode_class): New hook.
10108 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
10109 (hard_regno_nregs): Likewise.
10110 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
10111 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
10112 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
10113 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
10114 (LOAD_EXTEND_OP): Update accordingly.
10115 * doc/tm.texi: Regenerate.
10116 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
10117 CANNOT_CHANGE_MODE_CLASS.
10118 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
10119 (REG_CAN_CHANGE_MODE_P): ...this new macro.
10120 * combine.c (simplify_set): Update accordingly.
10121 * emit-rtl.c (validate_subreg): Likewise.
10122 * recog.c (general_operand): Likewise.
10123 * regcprop.c (mode_change_ok): Likewise.
10124 * reload1.c (choose_reload_regs): Likewise.
10125 (inherit_piecemeal_p): Likewise.
10126 * rtlanal.c (simplify_subreg_regno): Likewise.
10127 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
10128 instead of CANNOT_CHANGE_MODE_CLASS.
10129 (reload_cse_simplify_operands): Likewise.
10130 * reload.c (push_reload): Use targetm.can_change_mode_class
10131 instead of CANNOT_CHANGE_MODE_CLASS.
10132 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
10133 REG_CANNOT_CHANGE_MODE_P.
10134 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10135 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
10136 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10137 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10138 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10139 (arm_can_change_mode_class): New function.
10140 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
10141 than CANNOT_CHANGE_MODE_CLASS in comments.
10142 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10143 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
10144 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
10145 (ix86_can_change_mode_class): ...this new function, inverting the
10146 sense of the return value.
10147 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10148 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10149 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10150 (ia64_can_change_mode_class): New function.
10151 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10152 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
10153 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
10154 (m32c_can_change_mode_class): ...this new function, inverting the
10155 sense of the return value.
10156 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10157 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10158 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
10159 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
10160 (mips_can_change_mode_class): ...this new function, inverting the
10161 sense of the return value.
10162 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10163 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10164 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10165 (msp430_can_change_mode_class): New function.
10166 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10167 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
10168 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10169 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10170 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10171 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
10172 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10173 (pa_cannot_change_mode_class): Replace with...
10174 (pa_can_change_mode_class): ...this new function, inverting the
10175 sense of the return value.
10176 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
10177 than CANNOT_CHANGE_MODE_CLASS in comments.
10178 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10179 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
10180 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10181 (pdp11_cannot_change_mode_class): Replace with...
10182 (pdp11_can_change_mode_class): ...this new function, inverting the
10183 sense of the return value.
10184 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10185 * config/powerpcspe/powerpcspe-protos.h
10186 (rs6000_cannot_change_mode_class_ptr): Delete.
10187 * config/powerpcspe/powerpcspe.c
10188 (rs6000_cannot_change_mode_class_ptr): Delete.
10189 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10190 (rs6000_option_override_internal): Assign to
10191 targetm.can_change_mode_class instead of
10192 rs6000_cannot_change_mode_class_ptr.
10193 (rs6000_cannot_change_mode_class): Replace with...
10194 (rs6000_can_change_mode_class): ...this new function, inverting the
10195 sense of the return value.
10196 (rs6000_debug_cannot_change_mode_class): Replace with...
10197 (rs6000_debug_can_change_mode_class): ...this new function.
10198 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10199 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
10200 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10201 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10202 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
10204 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
10205 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10206 (rs6000_option_override_internal): Assign to
10207 targetm.can_change_mode_class instead of
10208 rs6000_cannot_change_mode_class_ptr.
10209 (rs6000_cannot_change_mode_class): Replace with...
10210 (rs6000_can_change_mode_class): ...this new function, inverting the
10211 sense of the return value.
10212 (rs6000_debug_cannot_change_mode_class): Replace with...
10213 (rs6000_debug_can_change_mode_class): ...this new function.
10214 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10215 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
10216 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
10217 (s390_can_change_mode_class): ...this new function, inverting the
10218 sense of the return value.
10219 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10220 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10221 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
10222 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10223 (sh_cannot_change_mode_class): Replace with...
10224 (sh_can_change_mode_class): ...this new function, inverting the
10225 sense of the return value.
10226 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10227 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10228 (sparc_can_change_mode_class): New function.
10229 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10230 * config/spu/spu.c (spu_can_change_mode_class): New function.
10231 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10232 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10233 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10234 (visium_can_change_mode_class): New function.
10235 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
10237 2017-09-15 Richard Biener <rguenther@suse.de>
10239 PR tree-optimization/82217
10240 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
10241 but not undefined case.
10243 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10246 * postreload.c (reload_cse_simplify_operands): Skip
10247 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
10249 2017-09-15 Richard Biener <rguenther@suse.de>
10251 PR tree-optimization/68823
10252 * graphite-scop-detection.c (build_alias_set): If we have a
10253 possible dependence check whether we can handle them by just
10254 looking at the DRs DR_ACCESS_FNs.
10255 (build_scops): If build_alias_set fails, fail the SCOP.
10257 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
10259 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
10260 to support float128 built-in functions that require the ISA 3.0
10262 (BU_FLOAT128_3_HW): Likewise.
10263 (SQRTF128): Add support for the IEEE 128-bit square root and fma
10264 built-in functions.
10265 (FMAF128): Likewise.
10267 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
10268 support for built-in functions that need the ISA 3.0 IEEE 128-bit
10269 floating point instructions.
10270 (rs6000_invalid_builtin): Likewise.
10271 (rs6000_builtin_mask_names): Likewise.
10272 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
10273 (RS6000_BTM_FLOAT128_HW): Likewise.
10274 (RS6000_BTM_COMMON): Likewise.
10275 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
10277 * doc/extend.texi (RS/6000 built-in functions): Document the
10278 IEEE 128-bit floating point square root and fused multiply-add
10279 built-in functions.
10281 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
10283 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
10284 reg (r2) isn't in the set of registers defined in the prologue.
10286 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10287 Alan Hayward <alan.hayward@arm.com>
10288 David Sherwood <david.sherwood@arm.com>
10290 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
10291 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
10292 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
10293 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
10294 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
10296 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
10297 max_vectorization_factor.
10298 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
10300 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10301 Alan Hayward <alan.hayward@arm.com>
10302 David Sherwood <david.sherwood@arm.com>
10304 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
10305 (vect_worthwhile_without_simd_p): Declare.
10306 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
10307 (vectorizable_reduction): Use it.
10308 * tree-vect-stmts.c (vectorizable_shift): Likewise.
10309 (vectorizable_operation): Likewise.
10311 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10312 Alan Hayward <alan.hayward@arm.com>
10313 David Sherwood <david.sherwood@arm.com>
10315 * tree-vectorizer.h (vect_get_num_copies): New function.
10316 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
10317 * tree-vect-loop.c (vectorizable_reduction): Likewise.
10318 (vectorizable_induction): Likewise.
10319 (vectorizable_live_operation): Likewise.
10320 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
10321 (vectorizable_bswap): Likewise.
10322 (vectorizable_call): Likewise.
10323 (vectorizable_conversion): Likewise.
10324 (vectorizable_assignment): Likewise.
10325 (vectorizable_shift): Likewise.
10326 (vectorizable_operation): Likewise.
10327 (vectorizable_store): Likewise.
10328 (vectorizable_load): Likewise.
10329 (vectorizable_condition): Likewise.
10330 (vectorizable_comparison): Likewise.
10331 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
10333 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10334 Alan Hayward <alan.hayward@arm.com>
10335 David Sherwood <david.sherwood@arm.com>
10337 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
10338 of vect_init_vector.
10340 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10341 Alan Hayward <alan.hayward@arm.com>
10342 David Sherwood <david.sherwood@arm.com>
10344 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
10345 an inline wrapper that provides a location.
10346 (gimple_build_vector): Likewise.
10347 * gimple-fold.c (gimple_build_vector_from_val): New function.
10348 (gimple_build_vector): Likewise.
10349 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
10350 functions to build the initial value. Always return a gimple value.
10351 (get_initial_defs_for_reduction): Likewise. Only compute
10353 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
10354 vect_init_vector on the results from get_initial_def(s)_for_reduction.
10355 (vectorizable_induction): Use gimple_build_vector rather than
10358 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10359 Alan Hayward <alan.hayward@arm.com>
10360 David Sherwood <david.sherwood@arm.com>
10362 * target.h (vec_perm_indices): New typedef.
10363 (auto_vec_perm_indices): Likewise.
10364 * optabs-query.h: Include target.h
10365 (can_vec_perm_p): Take a vec_perm_indices *.
10366 * optabs-query.c (can_vec_perm_p): Likewise.
10367 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
10368 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
10369 * tree-vect-generic.c (lower_vec_perm): Likewise.
10370 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
10371 (vect_grouped_load_supported): Likewise.
10372 (vect_shift_permute_load_chain): Likewise.
10373 (vect_permute_store_chain): Use auto_vec_perm_indices.
10374 (vect_permute_load_chain): Likewise.
10375 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
10376 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
10377 Update uses of can_vec_perm_p.
10378 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
10379 mode with a number of elements. Take a vec_perm_indices *.
10380 (vect_create_epilog_for_reduction): Update accordingly.
10381 Use auto_vec_perm_indices.
10382 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
10383 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
10384 (vect_transform_slp_perm_load): Likewise.
10385 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
10386 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
10387 (vect_gen_perm_mask_checked): Likewise.
10388 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
10389 (vect_gen_perm_mask_checked): Likewise.
10390 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
10391 (vectorizable_store): Likewise.
10392 (vectorizable_load): Likewise.
10393 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
10394 (vectorizable_bswap): Likewise.
10396 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10397 Alan Hayward <alan.hayward@arm.com>
10398 David Sherwood <david.sherwood@arm.com>
10400 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
10401 * tree.c (build_vector): Likewise.
10402 (build_vector_from_ctor): Update accordingly.
10403 (build_vector_from_val): Likewise.
10404 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
10405 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
10406 * tree-vect-generic.c (add_rshift): Likewise.
10407 (expand_vector_divmod): Likewise.
10408 (optimize_vector_constructor): Likewise.
10409 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
10410 (vect_transform_slp_perm_load): Likewise.
10411 (vect_schedule_slp_instance): Likewise.
10412 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
10413 (vectorizable_call): Likewise.
10414 (vect_gen_perm_mask_any): Likewise. Add elements in order.
10415 * expmed.c (make_tree): Likewise.
10416 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
10417 a vector passed to build_vector.
10418 (fold_convert_const): Likewise.
10419 (exact_inverse): Likewise.
10420 (fold_ternary_loc): Likewise.
10421 (fold_relational_const): Likewise.
10422 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
10423 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
10424 (const_unop): Likewise. Store the reduction accumulator in a
10425 variable rather than an array.
10426 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
10427 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
10428 the new vector, rather than constructing it after the input arrays.
10429 (native_interpret_vector): Use auto_vec<tree> when building
10430 a vector passed to build_vector. Add elements in order.
10431 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
10432 auto_vec<tree> when building a vector passed to build_vector.
10433 (vect_create_epilog_for_reduction): Likewise.
10434 (vectorizable_induction): Likewise.
10435 (get_initial_def_for_reduction): Likewise. Fix indentation of
10437 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
10439 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
10440 passed to build_vector.
10442 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10443 Alan Hayward <alan.hayward@arm.com>
10444 David Sherwood <david.sherwood@arm.com>
10446 * tree-core.h (tree_base::u): Add an "nelts" field.
10447 (tree_vector): Use VECTOR_CST_NELTS as the length.
10448 * tree.c (tree_size): Likewise.
10449 (make_vector): Initialize VECTOR_CST_NELTS.
10450 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
10451 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
10452 TYPE_VECTOR_SUBPARTS.
10453 * expr.c (const_vector_mask_from_tree): Consistently use "units"
10454 as the number of units, setting it from VECTOR_CST_NELTS.
10455 (const_vector_from_tree): Likewise.
10456 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
10457 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
10458 (fold_negate_expr_1): Likewise.
10459 (fold_convert_const): Likewise.
10460 (const_binop): Likewise. Differentiate the number of output and
10462 (const_unop): Likewise.
10463 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
10464 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
10465 in cases that did the opposite.
10467 2017-09-14 Richard Biener <rguenther@suse.de>
10469 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
10472 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
10474 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
10476 2017-09-14 Jakub Jelinek <jakub@redhat.com>
10479 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
10480 if and where to split a bb, except for splitting before debug insn
10481 sequences followed by non-label real insn. Delete debug insns
10482 in between basic blocks.
10484 * combine.c (make_compound_operation_int): Formatting fixes.
10486 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
10487 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10488 * config/netbsd.h (LINK_EH_SPEC): Likewise.
10489 * config/sol2.h (LINK_EH_SPEC): Likewise.
10490 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10491 * config/s390/linux.h (LINK_SPEC): Likewise.
10492 * config/freebsd.h (LINK_EH_SPEC): Likewise.
10493 * config/openbsd.h (LINK_EH_SPEC): Likewise.
10494 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10495 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
10496 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
10497 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10498 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
10499 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
10501 2017-09-13 Jakub Jelinek <jakub@redhat.com>
10503 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
10505 (ENDFILE_LINUX_SPEC): Likewise.
10506 (LINK_EH_SPEC): Likewise.
10507 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
10508 (LINK_OS_LINUX_SPEC32): Likewise.
10509 (LINK_OS_LINUX_SPEC64): Likewise.
10510 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
10511 (LINK_OS_LINUX_SPEC): Likewise.
10513 2017-09-13 Martin Liska <mliska@suse.cz>
10515 PR middle-end/82154
10516 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
10517 CASE_HIGH is NULL_TREE.
10519 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
10520 Alan Hayward <alan.hayward@arm.com>
10521 David Sherwood <david.sherwood@arm.com>
10523 * target.def (secondary_memory_needed): New hook.
10524 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
10525 instead of SECONDARY_MEMORY_NEEDED.
10526 (secondary_memory_needed_mode): Likewise.
10527 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
10528 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
10529 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
10530 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
10531 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
10532 * doc/tm.texi: Regenerate.
10533 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
10534 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
10535 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10536 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
10537 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
10538 * config/i386/i386.c (inline_secondary_memory_needed): Put the
10539 mode argument first and change the reg_class arguments to reg_class_t.
10540 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
10541 Make static. Update the call to inline_secondary_memory_needed.
10542 (ix86_register_move_cost): Update the call to
10543 inline_secondary_memory_needed.
10544 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10545 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
10547 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
10548 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
10550 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
10551 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
10552 * config/mips/mips.c (mips_secondary_memory_needed): Make static
10553 and match hook interface. Add comment from mips.h.
10554 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10555 * config/mmix/mmix.md (truncdfsf2): Refer to
10556 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
10558 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
10559 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
10560 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10561 (pa_secondary_memory_needed): New function.
10562 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
10563 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
10564 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10565 (pdp11_secondary_memory_needed): Make static and match hook interface.
10566 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
10567 * config/powerpcspe/powerpcspe-protos.h
10568 (rs6000_secondary_memory_needed_ptr): Delete.
10569 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
10571 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10572 (rs6000_option_override_internal): Assign to
10573 targetm.secondary_memory_needed rather than
10574 rs6000_secondary_memory_needed_ptr.
10575 (rs6000_secondary_memory_needed): Match hook interface.
10576 (rs6000_debug_secondary_memory_needed): Likewise.
10577 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
10578 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
10579 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
10580 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10581 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
10582 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
10584 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
10585 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10586 (rs6000_option_override_internal): Assign to
10587 targetm.secondary_memory_needed rather than
10588 rs6000_secondary_memory_needed_ptr.
10589 (rs6000_secondary_memory_needed): Match hook interface.
10590 (rs6000_debug_secondary_memory_needed): Likewise.
10591 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
10592 * config/s390/s390.c (s390_secondary_memory_needed): New function.
10593 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10594 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
10595 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10596 (sparc_secondary_memory_needed): New function.
10597 * lra-constraints.c (check_and_process_move): Refer to
10598 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
10600 (curr_insn_transform): Likewise.
10601 (process_alt_operands): Use targetm.secondary_memory_needed
10602 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10603 (check_secondary_memory_needed_p): Likewise.
10604 (choose_split_class): Likewise.
10605 * reload.c: Unconditionally include code that was previously
10606 conditional on SECONDARY_MEMORY_NEEDED.
10607 (push_secondary_reload): Use targetm.secondary_memory_needed
10608 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10609 (push_reload): Likewise.
10610 * reload1.c: Unconditionally include code that was previously
10611 conditional on SECONDARY_MEMORY_NEEDED.
10612 (choose_reload_regs): Use targetm.secondary_memory_needed
10613 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10614 (gen_reload): Likewise.
10615 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
10617 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
10618 Alan Hayward <alan.hayward@arm.com>
10619 David Sherwood <david.sherwood@arm.com>
10621 * target.def (secondary_memory_needed_mode): New hook:
10622 * targhooks.c (default_secondary_memory_needed_mode): Declare.
10623 * targhooks.h (default_secondary_memory_needed_mode): New function.
10624 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
10625 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
10626 * doc/tm.texi: Regenerate.
10627 * lra-constraints.c (check_and_process_move): Use
10628 targetm.secondary_memory_needed_mode instead of
10629 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
10630 (curr_insn_transform): Likewise.
10631 * reload.c (get_secondary_mem): Likewise.
10632 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10633 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
10635 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10636 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10637 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
10638 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10639 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
10641 * config/powerpcspe/powerpcspe-protos.h
10642 (rs6000_secondary_memory_needed_mode): Delete.
10643 * config/powerpcspe/powerpcspe.c
10644 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10645 (rs6000_secondary_memory_needed_mode): Make static.
10646 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10647 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
10649 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
10651 (rs6000_secondary_memory_needed_mode): Make static.
10652 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10653 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
10654 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10655 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10656 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
10658 (sparc_secondary_memory_needed_mode): New function.
10659 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
10661 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
10663 * config/aarch64/constraints.md (Umq): New constraint.
10664 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
10666 (mov<mode>): Update condition.
10668 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10670 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
10671 when bitposition is the same.
10673 2017-09-13 Richard Biener <rguenther@suse.de>
10675 * dwarf2out.c (output_die_symbol): Remove.
10676 (output_die): Do not output a DIEs symbol.
10678 2017-09-13 Richard Biener <rguenther@suse.de>
10680 PR middle-end/82128
10681 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
10682 default-def to avoid breaking iterator update with the weird
10683 interaction with cgraph_update_edges_for_call_stmt_node.
10685 2017-09-13 Richard Biener <rguenther@suse.de>
10687 * tree-cfg.c (verify_gimple_assign_binary): Add verification
10688 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
10689 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
10690 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
10692 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
10694 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
10695 Disable pc relative literal load irrespective of
10696 TARGET_FIX_ERR_A53_84341 for default.
10698 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
10700 * config/sparc/sparc.c (output_return): Output the source location of
10701 the insn in the delay slot, if any.
10702 (output_sibcall): Likewise.
10704 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
10707 * common.opt (-static-pie): New alias.
10708 (shared): Negate static-pie.
10709 (-no-pie): Update help text.
10711 (static-pie): New option.
10712 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
10713 -static-pie support.
10714 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
10715 (LINK_EH_SPEC): Likewise.
10716 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10717 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
10718 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
10719 * gcc.c (LINK_COMMAND_SPEC): Likewise.
10720 (init_gcc_specs): Likewise.
10721 (init_spec): Likewise.
10722 (display_help): Update help message for -pie.
10723 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
10726 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
10728 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
10729 (movdi_aarch64): Likewise.
10730 (movti_aarch64): Likewise.
10732 2017-09-12 Simon Wright <simon@pushface.org>
10735 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
10736 calculation of the minor version, always output as 0.
10738 2017-09-12 Jakub Jelinek <jakub@redhat.com>
10741 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
10742 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
10743 on it early, rather than manual conversion late. For
10744 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
10745 instead of performing manual conversion.
10747 2017-09-12 Carl Love <cel@us.ibm.com>
10749 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
10750 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
10752 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10753 VMULOSW): Add definitions.
10754 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10755 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10756 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10757 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10758 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10760 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
10762 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
10764 (movti_aarch64): Likewise.
10765 (movdf_aarch64): Likewise.
10766 (movtf_aarch64): Likewise.
10767 (load_pairdi): Likewise.
10768 (store_pairdi): Likewise.
10769 (load_pairdf): Likewise.
10770 (store_pairdf): Likewise.
10771 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
10772 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
10773 (ldr_got_small_<mode>): Likewise.
10774 (ldr_got_small_28k_<mode>): Likewise.
10775 (ldr_got_tiny): Likewise.
10776 * config/aarch64/iterators.md (ldst_sz): New.
10777 (ldpstp_sz): Likewise.
10778 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
10780 (thunderx_load): Split load_8 to load_16.
10781 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
10783 (thunderx2t99_storepair_basic): Split store_8 to store_16.
10784 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
10785 (xgene1_store_pair): Split store_8 to store_16.
10786 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
10787 (falkor_st_0_st_sd): Split store_8 to store_16.
10789 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
10791 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
10792 and store1/2/3/4 to store_4/8/12/16.
10793 * config/aarch64/aarch64.md: Update for rename.
10794 * config/arm/arm.md: Likewise.: Likewise.
10795 * config/arm/arm.c: Likewise.
10796 * config/arm/thumb1.md: Likewise.
10797 * config/arm/thumb2.md: Likewise.
10798 * config/arm/vfp.md: Likewise.
10799 * config/arm/arm-generic.md: Likewise.
10800 * config/arm/arm1020e.md: Likewise.
10801 * config/arm/arm1026ejs.md: Likewise.
10802 * config/arm/arm1136jfs.md: Likewise.
10803 * config/arm/arm926ejs.md: Likewise.
10804 * config/arm/cortex-a15.md: Likewise.
10805 * config/arm/cortex-a17.md: Likewise.
10806 * config/arm/cortex-a5.md: Likewise.
10807 * config/arm/cortex-a53.md: Likewise.
10808 * config/arm/cortex-a57.md: Likewise.
10809 * config/arm/cortex-a7.md: Likewise.
10810 * config/arm/cortex-a8.md: Likewise.
10811 * config/arm/cortex-a9.md: Likewise.
10812 * config/arm/cortex-m4.md: Likewise.
10813 * config/arm/cortex-m7.md: Likewise.
10814 * config/arm/cortex-r4.md: Likewise.
10815 * config/arm/exynos-m1.md: Likewise.
10816 * config/arm/fa526.md: Likewise.
10817 * config/arm/fa606te.md: Likewise.
10818 * config/arm/fa626te.md: Likewise.
10819 * config/arm/fa726te.md: Likewise.
10820 * config/arm/fmp626.md: Likewise.
10821 * config/arm/iwmmxt.md: Likewise.
10822 * config/arm/ldmstm.md: Likewise.
10823 * config/arm/marvell-pj4.md: Likewise.
10824 * config/arm/xgene1.md: Likewise.
10825 * config/aarch64/thunderx.md: Likewise.
10826 * config/aarch64/thunderx2t99.md: Likewise.
10827 * config/aarch64/falkor.md: Likewise.
10829 2017-09-12 Martin Liska <mliska@suse.cz>
10831 * attribs.c (private_lookup_attribute): New function.
10832 * attribs.h (private_lookup_attribute): Declared here.
10833 (lookup_attribute): Called from this place.
10835 2017-09-12 Richard Biener <rguenther@suse.de>
10837 PR tree-optimization/82157
10838 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
10839 stmts with side-effects.
10841 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10842 Alan Hayward <alan.hayward@arm.com>
10843 David Sherwood <david.sherwood@arm.com>
10845 * target.def (hard_regno_nregs): New hook.
10846 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
10847 * targhooks.h (default_hard_regno_nregs): Declare.
10848 * targhooks.c (default_hard_regno_nregs): New function.
10849 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
10850 (TARGET_HARD_REGNO_NREGS): ...this hook.
10851 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
10852 (CLASS_MAX_NREGS): Likewise.
10853 * doc/tm.texi: Regenerate.
10854 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
10855 instead of HARD_REGNO_NREGS.
10856 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
10857 HARD_REGNO_NREGS in the comment.
10858 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
10859 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
10860 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
10861 Return an unsigned int.
10862 (TARGET_HARD_REGNO_NREGS): Redefine.
10863 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
10864 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
10865 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
10866 (arc_hard_regno_nregs): New function.
10867 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
10868 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
10869 (arm_hard_regno_nregs): New function.
10870 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
10871 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
10872 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
10873 (TARGET_HARD_REGNO_NREGS): Redefine.
10874 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
10875 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
10876 (HARD_REGNO_NREGS): Delete.
10877 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
10878 (cr16_hard_regno_nregs): New function.
10879 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
10880 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
10881 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
10882 (cris_hard_regno_nregs): New function.
10883 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
10884 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
10885 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
10886 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
10887 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
10888 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
10889 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
10890 (frv_hard_regno_nregs): Make static. Take and return an
10892 (frv_class_max_nregs): Remove outdated copy of documentation.
10893 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
10894 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
10895 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
10896 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
10897 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
10898 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
10899 (TARGET_HARD_REGNO_NREGS): Redefine.
10900 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
10901 (CLASS_MAX_NREGS): Update comment.
10902 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
10903 (ia64_hard_regno_nregs): New function.
10904 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
10905 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
10906 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
10907 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
10908 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
10910 (m32c_hard_regno_nregs): Likewise. Make static.
10911 (TARGET_HARD_REGNO_NREGS): Redefine.
10912 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
10913 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
10914 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
10915 (m68k_hard_regno_nregs): New function.
10916 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
10917 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
10918 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
10919 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
10920 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
10921 Take and return an unsigned int.
10922 (TARGET_HARD_REGNO_NREGS): Redefine.
10923 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
10924 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
10925 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
10926 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
10927 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
10928 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
10929 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
10930 (msp430_hard_regno_nregs): Make static. Take and return an
10932 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
10933 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
10934 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
10935 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
10936 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
10937 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
10938 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
10939 (TARGET_HARD_REGNO_NREGS): Redefine.
10940 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
10941 (PA_HARD_REGNO_NREGS): ...this.
10942 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
10943 (PA_HARD_REGNO_NREGS): ...this.
10944 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
10945 (pa_hard_regno_nregs): New function.
10946 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
10947 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
10948 (pdp11_hard_regno_nregs): New function.
10949 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
10950 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
10951 (rs6000_hard_regno_nregs_hook): New function.
10952 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
10953 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
10954 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
10955 Take and return an unsigned int. Move earlier in file.
10956 (TARGET_HARD_REGNO_NREGS): Redefine.
10957 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
10958 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
10959 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
10960 (rl78_hard_regno_nregs): Make static. Take and return an
10962 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
10963 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
10964 (rs6000_hard_regno_nregs_hook): New function.
10965 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
10966 * config/rx/rx.c (rx_hard_regno_nregs): New function.
10967 (TARGET_HARD_REGNO_NREGS): Redefine.
10968 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
10969 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
10970 instead of HARD_REGNO_NREGS.
10971 (s390_hard_regno_nregs): New function.
10972 (s390_hard_regno_mode_ok): Add comment from s390.h.
10973 (TARGET_HARD_REGNO_NREGS): Redefine.
10974 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
10975 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
10976 (sh_hard_regno_nregs): New function.
10977 (sh_pass_in_reg_p): Use it.
10978 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
10979 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
10980 (sparc_hard_regno_nregs): New function.
10981 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
10982 * config/spu/spu.c (spu_hard_regno_nregs): New function.
10983 (spu_function_arg_advance): Use it, supplying a valid register number.
10984 (TARGET_HARD_REGNO_NREGS): Redefine.
10985 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
10986 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
10987 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
10988 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
10989 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
10990 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
10991 (CLASS_MAX_NREGS): Remove copy of old documentation.
10992 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
10993 (visium_hard_regno_nregs): New function.
10994 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
10995 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
10996 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
10997 xtensa_hard_regno_nregs): New function.
10998 * system.h (HARD_REGNO_NREGS): Poison.
11000 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11002 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
11003 hard_regno_nregs instead of HARD_REGNO_NREGS.
11004 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
11005 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
11006 (c6x_expand_epilogue): Likewise.
11007 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
11008 (frv_read_iacc_argument): Likewise.
11009 * config/sh/sh.c: Include regs.h.
11010 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
11011 (regs_used): Likewise.
11012 (output_stack_adjust): Likewise.
11013 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
11014 * expmed.c: Include regs.h.
11015 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
11016 * ree.c: Include regs.h.
11017 (combine_reaching_defs): Use hard_regno_nregs instead of
11019 (add_removable_extension): Likewise.
11021 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11023 * regs.h (hard_regno_nregs): Turn into a function.
11024 (end_hard_regno): Update accordingly.
11025 * caller-save.c (setup_save_areas): Likewise.
11026 (save_call_clobbered_regs): Likewise.
11027 (replace_reg_with_saved_mem): Likewise.
11028 (insert_restore): Likewise.
11029 (insert_save): Likewise.
11030 * combine.c (can_change_dest_mode): Likewise.
11031 (move_deaths): Likewise.
11032 (distribute_notes): Likewise.
11033 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
11034 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
11035 (rs6000_split_multireg_move): Likewise.
11036 (rs6000_register_move_cost): Likewise.
11037 (rs6000_memory_move_cost): Likewise.
11038 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
11039 (rs6000_split_multireg_move): Likewise.
11040 (rs6000_register_move_cost): Likewise.
11041 (rs6000_memory_move_cost): Likewise.
11042 * cselib.c (cselib_reset_table): Likewise.
11043 (cselib_lookup_1): Likewise.
11044 * emit-rtl.c (set_mode_and_regno): Likewise.
11045 * function.c (aggregate_value_p): Likewise.
11046 * ira-color.c (setup_profitable_hard_regs): Likewise.
11047 (check_hard_reg_p): Likewise.
11048 (calculate_saved_nregs): Likewise.
11049 (assign_hard_reg): Likewise.
11050 (improve_allocation): Likewise.
11051 (calculate_spill_cost): Likewise.
11052 * ira-emit.c (modify_move_list): Likewise.
11053 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
11054 (ira_hard_reg_in_set_p): Likewise.
11055 * ira.c (setup_reg_mode_hard_regset): Likewise.
11056 (clarify_prohibited_class_mode_regs): Likewise.
11057 (check_allocation): Likewise.
11058 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11059 (lra_setup_reg_renumber): Likewise.
11060 (setup_try_hard_regno_pseudos): Likewise.
11061 (spill_for): Likewise.
11062 (assign_hard_regno): Likewise.
11063 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
11064 * lra-constraints.c (in_class_p): Likewise.
11065 (lra_constraint_offset): Likewise.
11066 (simplify_operand_subreg): Likewise.
11067 (lra_constraints): Likewise.
11068 (split_reg): Likewise.
11069 (split_if_necessary): Likewise.
11070 (invariant_p): Likewise.
11071 (inherit_in_ebb): Likewise.
11072 * lra-lives.c (process_bb_lives): Likewise.
11073 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
11074 (get_hard_regs): Likewise.
11075 (do_remat): Likewise.
11076 * lra-spills.c (assign_spill_hard_regs): Likewise.
11077 * mode-switching.c (create_pre_exit): Likewise.
11078 * postreload.c (reload_combine_recognize_pattern): Likewise.
11079 * recog.c (peep2_find_free_register): Likewise.
11080 * regcprop.c (kill_value_regno): Likewise.
11081 (set_value_regno): Likewise.
11082 (copy_value): Likewise.
11083 (maybe_mode_change): Likewise.
11084 (find_oldest_value_reg): Likewise.
11085 (copyprop_hardreg_forward_1): Likewise.
11086 * regrename.c (check_new_reg_p): Likewise.
11087 (regrename_do_replace): Likewise.
11088 * reload.c (push_reload): Likewise.
11089 (combine_reloads): Likewise.
11090 (find_dummy_reload): Likewise.
11091 (operands_match_p): Likewise.
11092 (find_reloads): Likewise.
11093 (find_equiv_reg): Likewise.
11094 (reload_adjust_reg_for_mode): Likewise.
11095 * reload1.c (count_pseudo): Likewise.
11096 (count_spilled_pseudo): Likewise.
11097 (find_reg): Likewise.
11098 (clear_reload_reg_in_use): Likewise.
11099 (free_for_value_p): Likewise.
11100 (allocate_reload_reg): Likewise.
11101 (choose_reload_regs): Likewise.
11102 (reload_adjust_reg_for_temp): Likewise.
11103 (emit_reload_insns): Likewise.
11104 (delete_output_reload): Likewise.
11105 * rtlanal.c (subreg_get_info): Likewise.
11106 * sched-deps.c (sched_analyze_reg): Likewise.
11107 * sel-sched.c (init_regs_for_mode): Likewise.
11108 (mark_unavailable_hard_regs): Likewise.
11109 (choose_best_reg_1): Likewise.
11110 (verify_target_availability): Likewise.
11111 * valtrack.c (dead_debug_insert_temp): Likewise.
11112 * var-tracking.c (track_loc_p): Likewise.
11113 (emit_note_insn_var_location): Likewise.
11114 * varasm.c (make_decl_rtl): Likewise.
11115 * reginfo.c (choose_hard_reg_mode): Likewise.
11116 (init_reg_modes_target): Refer directly to
11117 this_target_regs->x_hard_regno_nregs.
11119 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11121 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
11122 instead of hard_regno_nregs.
11124 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11126 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
11127 end_hard_regno instead of hard_regno_nregs.
11128 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
11129 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
11130 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
11131 * ira-color.c (improve_allocation): Likewise.
11132 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11133 * lra-lives.c (mark_regno_live): Likewise.
11134 (mark_regno_dead): Likewise.
11135 * lra-remat.c (operand_to_remat): Likewise.
11136 * lra.c (collect_non_operand_hard_regs): Likewise.
11137 * postreload.c (reload_combine_note_store): Likewise.
11138 (move2add_valid_value_p): Likewise.
11139 * reload.c (regno_clobbered_p): Likewise.
11141 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11143 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
11145 * config/v850/v850.c (v850_reorg): Likewise.
11146 * reload.c (refers_to_regno_for_reload_p): Likewise.
11147 (find_equiv_reg): Likewise.
11148 * reload1.c (reload_reg_reaches_end_p): Likewise.
11150 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11152 * caller-save.c (add_used_regs): Use REG_NREGS instead of
11154 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
11155 * config/arm/arm.c (output_move_neon): Likewise.
11156 (arm_attr_length_move_neon): Likewise.
11157 (neon_split_vcombine): Likewise.
11158 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
11159 (c6x_mark_reg_written): Likewise.
11160 (c6x_dwarf_register_span): Likewise.
11161 * config/i386/i386.c (ix86_save_reg): Likewise.
11162 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
11163 (rws_access_reg): Likewise.
11164 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
11165 * mode-switching.c (create_pre_exit): Likewise.
11166 * ree.c (combine_reaching_defs): Likewise.
11167 (add_removable_extension): Likewise.
11168 * regcprop.c (find_oldest_value_reg): Likewise.
11169 (copyprop_hardreg_forward_1): Likewise.
11170 * reload.c (reload_inner_reg_of_subreg): Likewise.
11171 (push_reload): Likewise.
11172 (combine_reloads): Likewise.
11173 (find_dummy_reload): Likewise.
11174 (reload_adjust_reg_for_mode): Likewise.
11175 * reload1.c (find_reload_regs): Likewise.
11176 (forget_old_reloads_1): Likewise.
11177 (reload_reg_free_for_value_p): Likewise.
11178 (reload_adjust_reg_for_temp): Likewise.
11179 (emit_reload_insns): Likewise.
11180 (delete_output_reload): Likewise.
11181 * sel-sched.c (choose_best_reg_1): Likewise.
11182 (choose_best_pseudo_reg): Likewise.
11184 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11185 Alan Hayward <alan.hayward@arm.com>
11186 David Sherwood <david.sherwood@arm.com>
11188 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
11189 * target.def (slow_unaligned_access): New hook.
11190 * targhooks.h (default_slow_unaligned_access): Declare.
11191 * targhooks.c (default_slow_unaligned_access): New function.
11192 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
11193 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
11194 * doc/tm.texi: Regenerate.
11195 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
11196 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
11197 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
11199 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
11200 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
11202 (rs6000_slow_unaligned_access): New function.
11203 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
11204 (expand_block_compare): Likewise.
11205 (expand_strn_compare): Likewise.
11206 (rs6000_rtx_costs): Likewise.
11207 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
11208 (riscv_slow_unaligned_access): Likewise.
11209 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
11210 (riscv_slow_unaligned_access_p): ...this and make static.
11211 (riscv_option_override): Update accordingly.
11212 (riscv_slow_unaligned_access): New function.
11213 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
11214 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
11215 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
11216 (rs6000_slow_unaligned_access): New function.
11217 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
11218 (rs6000_rtx_costs): Likewise.
11219 * config/rs6000/rs6000-string.c (expand_block_compare)
11220 (expand_strn_compare): Use targetm.slow_unaligned_access instead
11221 of SLOW_UNALIGNED_ACCESS.
11222 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
11223 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
11224 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
11225 of SLOW_UNALIGNED_ACCESS.
11226 * expmed.c (simple_mem_bitfield_p): Likewise.
11227 * expr.c (alignment_for_piecewise_move): Likewise.
11228 (emit_group_load_1): Likewise.
11229 (emit_group_store): Likewise.
11230 (copy_blkmode_from_reg): Likewise.
11231 (emit_push_insn): Likewise.
11232 (expand_assignment): Likewise.
11233 (store_field): Likewise.
11234 (expand_expr_real_1): Likewise.
11235 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
11236 * lra-constraints.c (simplify_operand_subreg): Likewise.
11237 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11238 * gimple-ssa-store-merging.c: Likewise in block comment at start
11240 * tree-ssa-strlen.c: Include target.h.
11241 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
11242 of SLOW_UNALIGNED_ACCESS.
11243 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
11245 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11247 PR rtl-optimization/82185
11248 * expmed.c (emit_store_flag_int): Only test tem if it has been
11251 2017-09-12 Richard Biener <rguenther@suse.de>
11253 PR middle-end/82149
11254 * match.pd ((FTYPE) N CMP CST): Fix typo.
11256 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
11258 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
11260 (mips_near_type_p): Add 'short_call' attribute as a synonym
11262 * doc/extend.texi (short_call): Document new function attribute.
11264 2017-09-12 Jakub Jelinek <jakub@redhat.com>
11267 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
11268 assertion check that in the condition.
11269 (get_atomic_generic_size): Likewise. Before testing if parameter
11270 has pointer type, if it has array type, call for C++
11271 default_conversion to perform array-to-pointer conversion.
11273 2017-09-12 Richard Biener <rguenther@suse.de>
11275 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
11276 for operations we cannot scalarize.
11278 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
11280 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
11281 vectors heap vectors. Clean up comments.
11282 Make visited_bbs a reference.
11283 (profitable_jump_thread_path): Make GC
11284 vectors heap vectors. Clean up comments.
11286 (convert_and_register_jump_thread_path): Make GC vectors heap
11288 (check_subpath_and_update_thread_path): Same. Clean up comments.
11289 Make visited_bbs a reference.
11290 (handle_phi): Abstract common code to to
11291 register_jump_thread_path_if_profitable.
11292 Rename VAR_BB to DEF_BB.
11294 Make GC vectors heap vectors.
11295 Make visited_bbs a reference.
11296 (handle_assignment): Same.
11297 (register_jump_thread_path_if_profitable): New.
11298 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
11300 Make GC vectors heap vectors. Clean up comments.
11301 Make visited_bbs a reference.
11302 (find_jump_threads_backwards): Make visited_bbs live in the stack.
11303 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
11306 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
11309 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
11310 words of E_DImode object are reachable by xtensa_uimm8x4 access.
11312 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
11314 Revert r251800 and r251799.
11316 2017-09-11 Martin Jambor <mjambor@suse.cz>
11319 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
11320 arguments in advance.
11321 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
11322 use it to find predecessor edges.
11323 (naive_outof_ssa): Collect vector of predecessors.
11325 2017-09-08 Jason Merrill <jason@redhat.com>
11327 PR c++/70029 - ICE with ref-qualifier and -flto
11328 * langhooks.h (struct lang_hooks_for_types): Add
11329 copy_lang_qualifiers.
11330 * attribs.c (build_type_attribute_qual_variant): Use it.
11331 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
11333 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
11334 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
11336 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
11339 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
11340 (*mulsi3_sp64): New instruction.
11341 (mulsi3): New expander.
11343 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
11345 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
11347 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11349 * sancov.c: Include memmodel.h.
11351 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
11354 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
11357 2017-09-07 Carl Love <cel@us.ibm.com>
11359 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
11360 the sldi instruction.
11362 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
11364 * sancov.c: Include tm_p.h.
11366 2017-09-07 Jakub Jelinek <jakub@redhat.com>
11369 * output.h (switch_to_other_text_partition): New declaration.
11370 * varasm.c (switch_to_other_text_partition): New function.
11371 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
11372 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
11373 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
11374 to the other text partition before emitting LCL label and switch back
11375 after emitting the word after it.
11377 2017-09-07 Richard Biener <rguenther@suse.de>
11379 * passes.def (pass_split_crit_edges): Remove instance before PRE.
11380 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
11381 critical edges here, after loop init.
11382 (pass_data_pre): Remove PROP_no_crit_edges flags.
11383 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
11384 for valueization of call args to avoid leaking VN_TOP.
11385 (visit_use): Assert we do not visit default defs.
11386 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
11387 Use error_mark_node to more easily detect leaking VN_TOP.
11388 All default-defs are varying, not VN_TOP. Mark them visited.
11389 (run_scc_vn): Make code match comment.
11391 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11393 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
11394 OPTION_MASK_FLOAT128_KEYWORD.
11395 (POWERPC_MASKS): Likewise.
11396 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
11397 support for the -mfloat128-type option, and make -mfloat128
11398 default on PowerPC Linux systems. Define or undefine
11399 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
11400 Define __float128 to be __ieee128 if IEEE 128-bit support is
11401 enabled, or undefine it.
11402 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
11403 Delete defining __FLOAT128_TYPE__.
11404 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
11405 -mfloat128-type option and make -mfloat128 default on PowerPC
11407 (TARGET_FLOAT128_TYPE): Likewise.
11408 (-mfloat128-type): Likewise.
11409 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11410 Delete the -mfloat128-type option and make -mfloat128 default on
11411 PowerPC Linux systems. Always use __ieee128 to be the keyword for
11412 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
11413 128-bit floating point is enabled. Change tests from using
11414 -mfloat128-type to -mfloat128.
11415 (rs6000_mangle_type): Use the correct mangling for the __float128
11416 type even if normal long double is restricted to 64-bits.
11417 (floatn_mode): Enable the _Float128 type by default on VSX Linux
11419 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
11420 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
11421 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
11423 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
11424 documentation for -mfloat128.
11426 2017-09-06 Olivier Hainque <hainque@adacore.com>
11428 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
11430 2017-09-06 Wish Wu <wishwu007@gmail.com>
11431 Jakub Jelinek <jakub@redhat.com>
11433 * asan.c (initialize_sanitizer_builtins): Add
11434 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
11435 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
11436 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
11437 BT_FN_VOID_UINT64_PTR variables.
11438 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
11439 (BT_FN_VOID_UINT16_UINT16): Likewise.
11440 (BT_FN_VOID_UINT32_UINT32): Likewise.
11441 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
11442 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
11443 (BT_FN_VOID_UINT64_PTR): Likewise.
11444 * common.opt (flag_sanitize_coverage): New variable.
11445 (fsanitize-coverage=trace-pc): Remove.
11446 (fsanitize-coverage=): Add.
11447 * flag-types.h (enum sanitize_coverage_code): New enum.
11448 * fold-const.c (fold_range_test): Disable non-short-circuit
11449 optimization if flag_sanitize_coverage.
11450 (fold_truth_andor): Likewise.
11451 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
11452 * opts.c (COVERAGE_SANITIZER_OPT): Define.
11453 (coverage_sanitizer_opts): New array.
11454 (get_closest_sanitizer_option): Add OPTS argument, handle also
11455 OPT_fsanitize_coverage_.
11456 (parse_sanitizer_options): Adjusted to also handle
11457 OPT_fsanitize_coverage_.
11458 (common_handle_option): Add OPT_fsanitize_coverage_.
11459 * sancov.c (instrument_comparison, instrument_switch): New function.
11460 (sancov_pass): Add trace-cmp support.
11461 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
11462 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
11463 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
11464 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
11465 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
11466 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
11467 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
11468 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
11469 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
11470 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
11472 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
11474 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11475 error. Only quit immediately if parsing is complete.
11476 (BEGIN): Initialize fatal_err and parse_done.
11477 (begin fpu, end fpu): Check number of arguments.
11478 (begin arch, end arch): Likewise.
11479 (begin cpu, end cpu): Likewise.
11480 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
11481 (optalias): Likewise.
11483 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
11485 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
11486 * config/arm/arm-isa.h: Delete. Move definitions to ...
11487 * arm-cpus.in: ... here. Use new feature and fgroup values.
11488 * config/arm/arm.c (arm_option_override): Use lower case for feature
11490 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
11491 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
11492 * config/arm/parsecpu.awk (END): Add new command 'isa'.
11494 (print_isa_bits_for): New function.
11495 (gen_isa): New function.
11496 (gen_comm_data): Use print_isa_bits_for.
11497 (define feature): New keyword.
11498 (define fgroup): New keyword.
11499 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
11500 (arm-isa.h): Add rule to generate file.
11501 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11502 case for feature bit names.
11504 2017-09-06 Richard Biener <rguenther@suse.de>
11506 * tree-ssa-pre.c (NECESSARY): Remove.
11507 (create_expression_by_pieces): Do not touch pass-local flags.
11508 (insert_into_preds_of_block): Likewise.
11509 (do_pre_regular_insertion): Likewise.
11510 (eliminate_insert): Likewise.
11511 (eliminate_dom_walker::before_dom_children): Likewise.
11512 (fini_eliminate): Do not look at inserted_exprs.
11513 (mark_operand_necessary): Remove.
11514 (remove_dead_inserted_code): Replace with simple work-list
11515 algorithm based on inserted_exprs and SSA uses.
11516 (pass_pre::execute): Re-order fini_eliminate and
11517 remove_dead_inserted_code.
11519 2017-09-06 Olivier Hainque <hainque@adacore.com>
11521 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
11522 for VxWorks 7. Adjust surrounding comments.
11524 2017-09-06 Richard Biener <rguenther@suse.de>
11526 * gimple-ssa-strength-reduction.c
11527 (find_candidates_dom_walker::before_dom_children): Also allow
11530 2017-09-06 Richard Biener <rguenther@suse.de>
11532 PR tree-optimization/82108
11533 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
11534 for gap in the non-permutation SLP case.
11536 2017-09-06 Martin Jambor <mjambor@suse.cz>
11538 PR tree-optimization/82078
11539 * tree-sra.c (sort_and_splice_var_accesses): Move call to
11540 add_access_to_work_queue...
11541 (build_accesses_from_assign): ...here.
11542 (propagate_all_subaccesses): Make sure racc is the group
11543 representative, if there is one.
11545 2017-09-06 Jakub Jelinek <jakub@redhat.com>
11547 PR middle-end/82095
11548 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
11551 2017-09-06 Richard Biener <rguenther@suse.de>
11553 * gimple-ssa-strength-reduction.c
11554 (find_candidates_dom_walker::before_doom_children): Use a
11555 type and not a mode check.
11557 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
11560 * config/arm/predicates.md (arm_general_adddi_operand): Create new
11562 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
11564 2017-09-05 Jeff Law <law@redhat.com>
11566 PR tree-optimization/64910
11567 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
11568 cases where we have 3 or more operands.
11570 2017-09-05 Jakub Jelinek <jakub@redhat.com>
11572 PR middle-end/81768
11573 * omp-low.c (lower_omp_for): Recompute tree invariant if
11574 gimple_omp_for_initial/final is ADDR_EXPR.
11576 PR middle-end/81768
11577 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
11578 into gimple val before gimplification fo the COND_EXPR.
11580 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
11582 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
11583 REGION_COPY argument.
11584 (thread_through_all_blocks): Remove unused argument to
11585 duplicate_thread_path.
11587 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11588 Alan Hayward <alan.hayward@arm.com>
11589 David Sherwood <david.sherwood@arm.com>
11591 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
11592 Take a scalar_mode rather than a machine_mode.
11593 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11594 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
11595 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11596 (aarch64_gen_adjusted_ldpstp): Likewise.
11597 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
11599 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11600 Alan Hayward <alan.hayward@arm.com>
11601 David Sherwood <david.sherwood@arm.com>
11603 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
11604 Take a scalar_int_mode instead of a machine_mode.
11605 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
11606 (aarch64_output_scalar_simd_mov_immediate): Likewise.
11607 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
11608 (aarch64_simd_attr_length_rglist): Delete.
11609 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
11610 a scalar_int_mode instead of a machine_mode.
11611 (aarch64_add_offset): Likewise.
11612 (aarch64_internal_mov_immediate): Likewise
11613 (aarch64_add_constant_internal): Likewise.
11614 (aarch64_add_constant): Likewise.
11615 (aarch64_movw_imm): Likewise.
11616 (aarch64_rtx_arith_op_extract_p): Likewise.
11617 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
11618 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
11619 Remove assert that the mode isn't a vector.
11620 (aarch64_output_scalar_simd_mov_immediate): Likewise.
11621 (aarch64_expand_mov_immediate): Update calls after above changes.
11622 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
11623 (aarch64_and_bitmask_imm): Check for scalar integer modes.
11624 (aarch64_move_imm): Likewise.
11625 (aarch64_can_const_movi_rtx_p): Likewise.
11626 (aarch64_strip_extend): Likewise.
11627 (aarch64_extr_rtx_p): Likewise.
11628 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
11629 a CONST_INT when the mode parameter is VOIDmode.
11630 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
11632 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11634 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
11635 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11636 (bitwise_type_for_mode): Update accordingly.
11638 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11640 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
11641 * stor-layout.c (mode_for_size_tree): Likewise.
11642 (mode_for_array): Update accordingly.
11643 (layout_decl): Likewise.
11644 (compute_record_mode): Likewise. Only set the mode once.
11646 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11648 * target.def (get_mask_mode): Change return type to opt_mode.
11650 * doc/tm.texi: Regenerate.
11651 * targhooks.h (default_get_mask_mode): Return an opt_mode.
11652 * targhooks.c (default_get_mask_mode): Likewise.
11653 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11654 * optabs-query.c (can_vec_mask_load_store_p): Update use of
11655 targetm.get_mask_mode.
11656 * tree.c (build_truth_vector_type): Likewise.
11658 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11660 * machmode.h (mode_for_vector): Return an opt_mode.
11661 * stor-layout.c (mode_for_vector): Likewise.
11662 (mode_for_int_vector): Update accordingly.
11663 (layout_type): Likewise.
11664 * config/i386/i386.c (emit_memmov): Likewise.
11665 (ix86_expand_set_or_movmem): Likewise.
11666 (ix86_expand_vector_init): Likewise.
11667 (ix86_get_mask_mode): Likewise.
11668 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
11670 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
11671 * expmed.c (extract_bit_field_1): Likewise.
11672 * expr.c (expand_expr_real_2): Likewise.
11673 * optabs-query.c (can_vec_perm_p): Likewise.
11674 (can_vec_mask_load_store_p): Likewise.
11675 * optabs.c (expand_vec_perm): Likewise.
11676 * targhooks.c (default_get_mask_mode): Likewise.
11677 * tree-vect-stmts.c (vectorizable_store): Likewise.
11678 (vectorizable_load): Likewise.
11679 (get_vectype_for_scalar_type_and_size): Likewise.
11681 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11683 * machmode.h (mode_for_int_vector): New function.
11684 * stor-layout.c (mode_for_int_vector): Likewise.
11685 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
11686 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
11687 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11688 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11689 (s390_expand_vcond): Likewise.
11691 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11693 * machmode.h (opt_machine_mode): New type.
11694 (opt_mode<T>): Allow construction from anything that can be
11696 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
11697 (mode_for_size): Return an opt_machine_mode.
11698 * stor-layout.c (mode_for_size): Likewise.
11699 (mode_for_size_tree): Update call accordingly.
11700 (bitwise_mode_for_mode): Likewise.
11701 (make_fract_type): Likewise.
11702 (make_accum_type): Likewise.
11703 * caller-save.c (replace_reg_with_saved_mem): Update call
11705 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11706 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11707 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11708 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11709 * expmed.c (extract_bit_field_1): Likewise.
11710 * reload.c (get_secondary_mem): Likewise.
11711 * varasm.c (assemble_integer): Likewise.
11712 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
11715 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11717 * machmode.h (decimal_float_mode_for_size): New function.
11718 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
11719 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
11720 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
11721 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
11722 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
11723 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
11725 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11727 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
11728 (get_builtin_sync_mode): Likewise.
11729 (expand_ifn_atomic_compare_exchange): Likewise.
11730 (expand_builtin_atomic_clear): Likewise.
11731 (expand_builtin_atomic_test_and_set): Likewise.
11732 (fold_builtin_atomic_always_lock_free): Likewise.
11733 * calls.c (compute_argument_addresses): Likewise.
11734 (emit_library_call_value_1): Likewise.
11735 (store_one_arg): Likewise.
11736 * combine.c (combine_instructions): Likewise.
11737 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
11738 * config/arm/arm.c (arm_function_value): Likewise.
11739 (aapcs_allocate_return_reg): Likewise.
11740 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
11741 * config/i386/i386.c (construct_container): Likewise.
11742 (ix86_gimplify_va_arg): Likewise.
11743 (ix86_expand_sse_cmp): Likewise.
11744 (emit_memmov): Likewise.
11745 (emit_memset): Likewise.
11746 (expand_small_movmem_or_setmem): Likewise.
11747 (ix86_expand_pextr): Likewise.
11748 (ix86_expand_pinsr): Likewise.
11749 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
11750 * config/microblaze/microblaze.c (microblaze_block_move_straight):
11752 * config/mips/mips.c (mips_function_value_1) Likewise.
11753 (mips_block_move_straight): Likewise.
11754 (mips_expand_ins_as_unaligned_store): Likewise.
11755 * config/powerpcspe/powerpcspe.c
11756 (rs6000_darwin64_record_arg_advance_flush): Likewise.
11757 (rs6000_darwin64_record_arg_flush): Likewise.
11758 * config/rs6000/rs6000.c
11759 (rs6000_darwin64_record_arg_advance_flush): Likewise.
11760 (rs6000_darwin64_record_arg_flush): Likewise.
11761 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
11762 (sparc_function_value_1): Likewise.
11763 * config/spu/spu.c (adjust_operand): Likewise.
11764 (spu_emit_branch_or_set): Likewise.
11765 (arith_immediate_p): Likewise.
11766 * emit-rtl.c (gen_lowpart_common): Likewise.
11767 * expr.c (expand_expr_real_1): Likewise.
11768 * function.c (assign_parm_setup_block): Likewise.
11769 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
11770 * reload1.c (alter_reg): Likewise.
11771 * stor-layout.c (mode_for_vector): Likewise.
11772 (layout_type): Likewise.
11774 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11776 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
11777 (spu_convert_move): Likewise.
11778 * lower-subreg.c (resolve_simple_move): Likewise.
11780 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11783 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
11784 define_insn to a define_expand.
11785 (altivec_vsum2sws_direct): New define_insn.
11786 (altivec_vsumsws): Convert from a define_insn to a define_expand.
11788 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
11790 * config/arm/arm.c (arm_option_params_internal): Improve setting of
11793 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
11798 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
11799 realign stack if stack alignment needed is less than incoming
11802 2017-09-05 Marek Polacek <polacek@redhat.com>
11805 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
11808 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
11810 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
11812 2017-09-05 Richard Biener <rguenther@suse.de>
11814 PR tree-optimization/82084
11815 * fold-const.c (can_native_encode_string_p): Handle wide characters.
11817 2017-09-05 Richard Biener <rguenther@suse.de>
11819 PR tree-optimization/82102
11820 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
11822 2017-09-05 Martin Liska <mliska@suse.cz>
11824 PR tree-optimization/82032
11825 * tree-cfg.c (generate_range_test): New function.
11826 * tree-cfg.h (generate_range_test): Declared here.
11827 * tree-cfgcleanup.c (convert_single_case_switch): New function.
11828 (cleanup_control_expr_graph): Use it.
11829 * tree-switch-conversion.c (try_switch_expansion): Remove
11831 (emit_case_nodes): Use generate_range_test.
11833 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
11836 * config/i386/i386.md (*<btsc><mode>_mask): Add
11837 TARGET_USE_BT to insn constraint.
11838 (*btr<mode>_mask): Ditto.
11840 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
11842 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
11843 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
11845 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
11848 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
11849 TARGET_NEON and TARGET_IWMMXT.
11850 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
11851 TARGET_NEON and TARGET_IWMMXT.
11852 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
11854 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
11856 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
11857 (ix86_rewrite_tls_address): Ditto.
11858 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
11859 (ix86_rewrite_tls_address_1): Ditto.
11860 (ix86_rewrite_tls_address): Ditto.
11861 * config/i386/predicates.md (tls_address_pattern): New predicate.
11862 * config/i386/i386.md (TLS address splitter): New splitter.
11864 2017-09-04 Richard Biener <rguenther@suse.de>
11866 PR tree-optimization/82084
11867 * fold-const.h (can_native_encode_string_p): Declare.
11868 * fold-const.c (can_native_encode_string_p): Factor out from ...
11869 (native_encode_string): ... here.
11870 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
11871 vectorizing stores from constants we later cannot handle.
11873 2017-09-04 Marek Polacek <polacek@redhat.com>
11876 * doc/invoke.texi: Update -Wtautological-compare documentation.
11878 2017-09-04 Jeff Law <law@redhat.com>
11880 PR tree-optimization/64910
11881 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
11882 swap the first and last operand if the last is a constant.
11884 2017-09-04 Marek Polacek <polacek@redhat.com>
11887 * convert.c (do_narrow): When sanitizing signed integer overflows,
11888 bail out for signed types.
11889 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
11891 2017-09-04 Richard Biener <rguenther@suse.de>
11893 PR tree-optimization/82060
11894 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
11895 Move devirtualization after stmt folding and before EH/AB/noreturn
11896 cleanup to get the stmt refs canonicalized. Use a bool instead
11897 of gimple_modified_p since that doesn't work for NOPs. Schedule
11898 NOPs generated by folding for removal.
11900 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11901 Alan Hayward <alan.hayward@arm.com>
11902 David Sherwood <david.sherwood@arm.com>
11904 * coretypes.h (pad_direction): New enum.
11905 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
11906 (FUNCTION_ARG_PADDING): Likewise.
11907 * target.def (function_arg_padding): New hook.
11908 * targhooks.h (default_function_arg_padding): Declare.
11909 * targhooks.c (default_function_arg_padding): New function.
11910 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
11911 (TARGET_FUNCTION_ARG_PADDING): ...this.
11912 * doc/tm.texi: Regenerate.
11913 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
11914 instead of direction.
11915 (compute_argument_addresses): Likewise.
11916 (load_register_parameters): Likewise.
11917 (emit_library_call_value_1): Likewise.
11918 (store_one_arg): Use targetm.calls.function_arg_padding instead
11919 of FUNCTION_ARG_PADDING.
11920 (must_pass_in_stack_var_size_or_pad): Likewise.
11921 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
11922 (emit_group_store): Likewise.
11923 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
11924 instead of FUNCTION_ARG_PADDING.
11925 (emit_push_insn): Likewise, and propagate enum change throughout
11927 * function.h (direction): Delete.
11928 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
11930 * function.c (assign_parm_find_stack_rtl): Likewise.
11931 (assign_parm_setup_block_p): Likewise.
11932 (assign_parm_setup_block): Likewise.
11933 (gimplify_parameters): Likewise.
11934 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
11935 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
11937 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
11938 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11939 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
11940 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
11941 (aarch64_function_arg_padding): ...this new function.
11942 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
11943 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11944 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
11945 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11946 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
11947 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11948 (arm_pad_arg_upward): Replace with...
11949 (arm_function_arg_padding): ...this new function.
11950 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
11952 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
11953 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
11954 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11955 (ia64_hpux_function_arg_padding): Replace with...
11956 (ia64_function_arg_padding): ...this new function. Use pad_direction
11957 instead of direction. Check for TARGET_HPUX.
11958 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
11959 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11960 (iq2000_function_arg_padding): New function.
11961 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
11962 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
11963 (mips_function_arg_padding): ...this new function.
11964 (mips_pad_reg_upward): Update accordingly.
11965 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11966 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
11967 targetm.calls.function_arg_padding.
11968 (FUNCTION_ARG_PADDING): Delete.
11969 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11970 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
11971 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11972 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
11973 (nios2_block_reg_padding): Return pad_direction instead of direction.
11974 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
11975 instead of direction.
11976 (nios2_function_arg_padding): Likewise. Make static.
11977 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11978 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
11979 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
11980 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
11981 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11982 (pa_function_arg_padding): Make static. Return pad_direction instead
11984 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
11985 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11986 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
11987 instead of direction. Use targetm.calls.function_arg_padding.
11988 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
11989 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
11990 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
11991 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
11992 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
11994 (function_arg_padding): Rename to...
11995 (rs6000_function_arg_padding): ...this. Make static. Return
11996 pad_direction instead of direction.
11997 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
11998 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
11999 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12000 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
12001 instead of direction. Use targetm.calls.function_arg_padding.
12002 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
12003 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
12004 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
12005 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
12006 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12007 (function_arg_padding): Rename to...
12008 (rs6000_function_arg_padding): ...this. Make static. Return
12009 pad_direction instead of direction.
12010 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
12011 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
12012 * config/s390/s390.c (s390_function_arg_padding): New function.
12013 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12014 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
12015 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
12016 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12017 (function_arg_padding): Rename to...
12018 (sparc_function_arg_padding): ...this. Make static. Return
12019 pad_direction instead of direction.
12020 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
12021 * config/spu/spu.c (spu_function_arg_padding): New function.
12022 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12023 * system.h (FUNCTION_ARG_PADDING): Poison.
12025 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12026 Alan Hayward <alan.hayward@arm.com>
12027 David Sherwood <david.sherwood@arm.com>
12029 * target.def (modes_tieable_p): New hook.
12030 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
12031 (TARGET_MODES_TIEABLE_P): ...this.
12032 * doc/tm.texi.in: Regenerate.
12033 * hooks.h (hook_bool_mode_mode_true): Declare.
12034 * hooks.c (hook_bool_mode_mode_true): New function.
12035 * combine.c (subst): Use targetm.modes_tieable_p instead of
12037 * dse.c (find_shift_sequence): Likewise.
12038 * expmed.c (extract_low_bits): Likewise.
12039 * lower-subreg.c: Include target.h.
12040 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
12042 * rtlanal.c (rtx_cost): Likewise.
12043 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
12044 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
12045 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
12046 (TARGET_MODES_TIEABLE_P): Redefine.
12047 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
12048 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
12049 (TARGET_MODES_TIEABLE_P): Redefine.
12050 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
12051 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
12052 (arc_modes_tieable_p): New function.
12053 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
12054 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
12055 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
12056 (arm_modes_tieable_p): Make static.
12057 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
12058 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
12059 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
12060 (TARGET_MODES_TIEABLE_P): Redefine.
12061 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
12062 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
12063 (TARGET_MODES_TIEABLE_P): Redefine.
12064 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
12065 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
12066 (cr16_modes_tieable_p): New function.
12067 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
12068 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
12069 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
12070 (TRULY_NOOP_TRUNCATION): Update comment.
12071 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
12072 (TRULY_NOOP_TRUNCATION): Update comment.
12073 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
12074 (frv_modes_tieable_p): New function.
12075 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
12076 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
12077 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
12078 (TARGET_MODES_TIEABLE_P): Redefine.
12079 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
12080 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
12081 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
12082 (TARGET_MODES_TIEABLE_P): Redefine.
12083 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
12084 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
12085 (ia64_modes_tieable_p): New function.
12086 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
12087 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
12088 (iq2000_modes_tieable_p): New function.
12089 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
12090 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
12091 (lm32_modes_tieable_p): New function.
12092 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
12093 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
12094 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
12095 (TARGET_MODES_TIEABLE_P): Redefine.
12096 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
12097 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
12098 (m32r_modes_tieable_p): New function.
12099 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
12100 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
12101 (m68k_modes_tieable_p): New function.
12102 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
12103 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
12104 (mcore_modes_tieable_p): New function.
12105 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
12106 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
12108 (TARGET_MODES_TIEABLE_P): Redefine.
12109 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
12110 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
12111 * config/mips/mips.c (mips_modes_tieable_p): Make static.
12112 (TARGET_MODES_TIEABLE_P): Redefine.
12113 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
12114 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
12115 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
12116 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
12117 (mn10300_modes_tieable_p): ...this and make static.
12118 (TARGET_MODES_TIEABLE_P): Redefine.
12119 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
12120 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
12121 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
12122 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
12123 (msp430_modes_tieable_p): Make static.
12124 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
12125 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
12126 (TARGET_MODES_TIEABLE_P): Redefine.
12127 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
12128 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
12129 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
12130 (TARGET_MODES_TIEABLE_P): Redefine.
12131 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
12132 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
12133 * config/pa/pa.c (pa_modes_tieable_p): Make static.
12134 (TARGET_MODES_TIEABLE_P): Redefine.
12135 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
12136 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
12137 (pdp11_modes_tieable_p): New function.
12138 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
12139 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
12140 (rs6000_modes_tieable_p): New function.
12141 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
12142 * config/powerpcspe/powerpcspe.md: Update comment.
12143 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
12144 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
12145 (TARGET_MODES_TIEABLE_P): Redefine.
12146 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
12147 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
12148 (rl78_modes_tieable_p): New function.
12149 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
12150 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
12151 (rs6000_modes_tieable_p): New function.
12152 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
12153 * config/rs6000/rs6000.md: Update comment.
12154 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
12155 * config/rx/rx.c (rx_modes_tieable_p): New function.
12156 (TARGET_MODES_TIEABLE_P): Redefine.
12157 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
12158 * config/s390/s390.c (s390_modes_tieable_p): New function.
12159 (TARGET_MODES_TIEABLE_P): Redefine.
12160 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
12161 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
12162 (sh_modes_tieable_p): New function.
12163 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
12164 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
12165 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
12166 (sparc_modes_tieable_p): Make static.
12167 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
12168 * config/spu/spu.c (spu_modes_tieable_p): New function.
12169 (TARGET_MODES_TIEABLE_P): Redefine.
12170 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
12171 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
12172 (TARGET_MODES_TIEABLE_P): Redefine.
12173 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
12174 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
12175 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
12176 * config/v850/v850.c (v850_modes_tieable_p): New function.
12177 (TARGET_MODES_TIEABLE_P): Redefine.
12178 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
12179 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
12180 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
12181 (visium_modes_tieable_p): New function.
12182 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
12183 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
12184 (xtensa_modes_tieable_p): New function.
12185 * system.h (MODES_TIEABLE_P): Poison.
12187 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12188 Alan Hayward <alan.hayward@arm.com>
12189 David Sherwood <david.sherwood@arm.com>
12191 * target.def (hard_regno_mode_ok): New hook.
12192 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
12193 (TARGET_HARD_REGNO_MODE_OK): ...this.
12194 * doc/tm.texi.in: Regenerate.
12195 * hooks.h (hook_bool_uint_mode_true): Declare.
12196 * hooks.c (hook_bool_uint_mode_true): New function.
12197 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
12198 HARD_REGNO_MODE_OK.
12199 * genpreds.c (write_insn_preds_c): Add an include of target.h.
12200 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
12201 instead of HARD_REGNO_MODE_OK.
12202 * caller-save.c: Include target.h.
12203 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
12204 HARD_REGNO_MODE_OK.
12205 * combine.c (can_combine_p): Likewise.
12206 (combinable_i3pat): Likewise.
12207 (can_change_dest_mode): Likewise.
12208 * expr.c (init_expr_target): Likewise.
12209 (convert_move): Likewise.
12210 (convert_modes): Likewise.
12211 * ira.c (setup_prohibited_class_mode_regs): Likewise.
12212 (setup_prohibited_mode_move_regs): Likewise.
12213 * ira.h (target_ira): Likewise.
12214 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12215 * lra-constraints.c (process_alt_operands): Likewise.
12216 (split_reg): Likewise.
12217 * recog.c (peep2_find_free_register): Likewise.
12218 * ree.c (combine_reaching_defs): Likewise.
12219 * regcprop.c (maybe_mode_change): Likewise.
12220 * reginfo.c (init_reg_sets_1): Likewise.
12221 (choose_hard_reg_mode): Likewise.
12222 (simplifiable_subregs): Likewise.
12223 * regrename.c (check_new_reg_p): Likewise.
12224 * reload.c (find_valid_class): Likewise.
12225 (find_valid_class_1): Likewise.
12226 (reload_inner_reg_of_subreg): Likewise.
12227 (push_reload): Likewise.
12228 (combine_reloads): Likewise.
12229 (find_dummy_reload): Likewise.
12230 (find_reloads): Likewise.
12231 * reload1.c (find_reg): Likewise.
12232 (set_reload_reg): Likewise.
12233 (allocate_reload_reg): Likewise.
12234 (choose_reload_regs): Likewise.
12235 (reload_adjust_reg_for_temp): Likewise.
12236 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
12237 (simplify_subreg_regno): Likewise.
12238 * sel-sched.c (init_regs_for_mode): Likewise.
12239 * varasm.c (make_decl_rtl): Likewise.
12240 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
12241 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
12242 HARD_REGNO_MODE_OK.
12243 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
12244 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
12245 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12246 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
12247 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
12248 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12249 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
12250 (arc_mode_class): Delete.
12251 (HARD_REGNO_MODE_OK): Delete.
12252 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12253 (arc_hard_regno_mode_ok): Rename old array to...
12254 (arc_hard_regno_mode_ok_modes): ...this.
12255 (arc_conditional_register_usage): Update accordingly.
12256 (arc_mode_class): Make static.
12257 (arc_hard_regno_mode_ok): New function.
12258 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
12259 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
12260 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12261 (arm_hard_regno_mode_ok): Make static.
12262 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
12263 HARD_REGNO_MODE_OK.
12264 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
12265 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
12266 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
12268 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12269 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
12270 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
12271 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
12272 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
12273 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12274 * config/bfin/predicates.md (valid_reg_operand): Use
12275 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
12276 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
12277 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
12278 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12279 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
12280 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
12281 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12282 (cr16_hard_regno_mode_ok): Make static and return a bool.
12283 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
12284 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12285 (cris_hard_regno_mode_ok): New function.
12286 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
12287 (epiphany_mode_class): Delete.
12288 (HARD_REGNO_MODE_OK): Delete.
12289 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
12290 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12291 (hard_regno_mode_ok): Rename to...
12292 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
12293 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
12294 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
12295 HARD_REGNO_MODE_OK.
12296 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
12297 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
12298 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12299 (frv_hard_regno_mode_ok): Make static and return a bool.
12300 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
12301 HARD_REGNO_MODE_OK.
12302 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
12303 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
12304 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
12305 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
12307 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12308 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
12309 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
12310 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
12312 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12313 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
12314 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12315 (ia64_hard_regno_mode_ok): New function.
12316 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
12317 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12318 (iq2000_hard_regno_mode_ok): New function.
12319 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
12320 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12321 (lm32_hard_regno_mode_ok): New function.
12322 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
12323 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
12324 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
12325 instead of HARD_REGNO_MODE_OK.
12326 (m32c_hard_regno_ok): Rename to...
12327 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
12328 (m32c_cannot_change_mode_class): Update accordingly.
12329 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12330 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
12331 (m32r_mode_class): Delete.
12332 (HARD_REGNO_MODE_OK): Delete.
12333 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12334 (m32r_hard_regno_mode_ok): Rename to...
12335 (m32r_hard_regno_modes): ...this.
12336 (m32r_mode_class): Make static.
12337 (m32r_hard_regno_mode_ok): New function.
12338 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
12339 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
12340 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12341 (m68k_hard_regno_mode_ok): Make static.
12342 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
12343 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12344 (mcore_hard_regno_mode_ok): New function.
12345 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
12346 (HARD_REGNO_MODE_OK): Delete.
12347 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
12349 (microblaze_hard_regno_mode_ok_p): ...this and make static.
12350 (microblaze_hard_regno_mode_ok): New function.
12351 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12352 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
12353 (mips_hard_regno_mode_ok): Delete.
12354 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
12355 (mips_hard_regno_mode_ok_p): ...this and make static.
12356 (mips_hard_regno_mode_ok_p): Rename to...
12357 (mips_hard_regno_mode_ok_uncached): ...this.
12358 (mips_hard_regno_mode_ok): New function.
12359 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
12360 of HARD_REGNO_MODE_OK.
12361 (mips_option_override): Update after above name changes.
12362 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12363 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
12364 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
12365 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
12366 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
12367 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12368 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
12369 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
12370 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
12371 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12372 (msp430_hard_regno_mode_ok): Make static and return a bool.
12373 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
12374 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
12375 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
12377 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12378 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
12379 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
12380 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
12381 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
12382 (PA_HARD_REGNO_MODE_OK): ...this
12383 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
12384 (PA_HARD_REGNO_MODE_OK): ...this.
12385 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12386 (pa_hard_regno_mode_ok): New function.
12387 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
12388 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12389 (pdp11_hard_regno_mode_ok): New function.
12390 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
12391 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
12393 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
12395 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12396 (rs6000_hard_regno_mode_ok): Rename to...
12397 (rs6000_hard_regno_mode_ok_uncached): ...this.
12398 (rs6000_init_hard_regno_mode_ok): Update accordingly.
12399 (rs6000_hard_regno_mode_ok): New function.
12400 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
12401 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
12402 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
12403 (riscv_hard_regno_mode_ok): ...this and make static.
12404 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12405 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
12406 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
12407 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12408 (rl78_hard_regno_mode_ok): Make static and return bool.
12409 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
12410 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
12412 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
12413 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12414 (rs6000_hard_regno_mode_ok): Rename to...
12415 (rs6000_hard_regno_mode_ok_uncached): ...this.
12416 (rs6000_init_hard_regno_mode_ok): Update accordingly.
12417 (rs6000_hard_regno_mode_ok): New function.
12418 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
12419 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
12420 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12421 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
12422 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
12423 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
12424 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12425 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
12426 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
12427 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12428 (sh_hard_regno_mode_ok): Make static.
12429 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
12430 instead of HARD_REGNO_MODE_OK.
12431 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
12432 (sparc_mode_class): Delete.
12433 (HARD_REGNO_MODE_OK): Delete.
12434 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12435 (hard_regno_mode_classes): Make static.
12436 (sparc_mode_class): Likewise.
12437 (sparc_hard_regno_mode_ok): New function.
12438 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
12439 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
12440 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
12442 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12443 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
12444 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
12445 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
12446 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
12447 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12448 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
12449 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
12450 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12451 (visium_hard_regno_mode_ok): New function.
12452 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
12453 instead of HARD_REGNO_MODE_OK.
12454 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
12455 (HARD_REGNO_MODE_OK): Delete.
12456 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
12457 (xtensa_hard_regno_mode_ok_p): ...this and make static.
12458 (xtensa_option_override): Update accordingly.
12459 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12460 (xtensa_hard_regno_mode_ok): New function.
12461 * system.h (HARD_REGNO_MODE_OK): Poison.
12463 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12464 Alan Hayward <alan.hayward@arm.com>
12465 David Sherwood <david.sherwood@arm.com>
12467 * target.def (hard_regno_call_part_clobbered): New hook.
12468 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
12469 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
12470 * doc/tm.texi: Regenerate.
12471 * hooks.h (hook_bool_uint_mode_false): Declare.
12472 * hooks.c (hook_bool_uint_mode_false): New function.
12473 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12474 * cselib.c (cselib_process_insn): Use
12475 targetm.hard_regno_call_part_clobbered instead of
12476 HARD_REGNO_CALL_PART_CLOBBERED.
12477 * ira-conflicts.c (ira_build_conflicts): Likewise.
12478 * ira-costs.c (ira_tune_allocno_costs): Likewise.
12479 * lra-constraints.c (need_for_call_save_p): Likewise.
12480 * lra-lives.c: Include target.h.
12481 (check_pseudos_live_through_calls): Use
12482 targetm.hard_regno_call_part_clobbered instead of
12483 HARD_REGNO_CALL_PART_CLOBBERED.
12484 * regcprop.c: Include target.h.
12485 (copyprop_hardreg_forward_1): Use
12486 targetm.hard_regno_call_part_clobbered instead of
12487 HARD_REGNO_CALL_PART_CLOBBERED.
12488 * reginfo.c (choose_hard_reg_mode): Likewise.
12489 * regrename.c (check_new_reg_p): Likewise.
12490 * reload.c (find_equiv_reg): Likewise.
12491 * reload1.c (emit_reload_insns): Likewise.
12492 * sched-deps.c (deps_analyze_insn): Likewise.
12493 * sel-sched.c (init_regs_for_mode): Likewise.
12494 (mark_unavailable_hard_regs): Likewise.
12495 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
12496 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12497 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
12499 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12500 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12501 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
12503 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
12505 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12506 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12507 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
12509 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12510 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12511 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
12513 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12514 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
12516 * config/powerpcspe/powerpcspe.c
12517 (rs6000_hard_regno_call_part_clobbered): New function.
12518 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12519 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12520 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
12522 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12523 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12524 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
12526 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12527 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12528 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
12530 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12531 Alan Hayward <alan.hayward@arm.com>
12532 David Sherwood <david.sherwood@arm.com>
12534 * rtl.h (subreg_memory_offset): Declare.
12535 * emit-rtl.c (subreg_memory_offset): New function.
12536 * expmed.c (store_bit_field_1): Use it.
12537 * expr.c (undefined_operand_subword_p): Likewise.
12538 * simplify-rtx.c (simplify_subreg): Likewise.
12540 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
12542 PR rtl-optimization/57448
12545 * optabs.c (expand_atomic_load): Place compiler memory barriers if
12546 using atomic_load pattern.
12547 (expand_atomic_store): Likewise.
12549 2017-09-04 Jakub Jelinek <jakub@redhat.com>
12552 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
12553 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
12554 handling. Use replace_call_with_value with NULL instead of
12555 gsi_replace, unlink_stmt_vdef and release_defs.
12557 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
12560 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
12562 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12565 * rtl.h (emit_library_call_value_1): Declare.
12566 (emit_library_call): Replace declaration with a series of overloads.
12567 Remove the parameter count argument.
12568 (emit_library_call_value): Likewise.
12569 * calls.c (emit_library_call_value_1): Make global. Replace varargs
12570 with an "rtx_mode_t *".
12571 (emit_library_call_value): Delete.
12572 (emit_library_call): Likewise.
12573 * asan.c (asan_emit_stack_protection): Update calls accordingly.
12574 (asan_emit_allocas_unpoison): Likewise.
12575 * builtins.c (expand_builtin_powi): Likewise.
12576 (expand_asan_emit_allocas_unpoison): Likewise.
12577 * cfgexpand.c (expand_main_function): Likewise.
12578 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
12579 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
12580 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
12581 * config/arm/arm.c (arm_trampoline_init): Likewise.
12582 (arm_call_tls_get_addr): Likewise.
12583 (arm_expand_divmod_libfunc): Likewise.
12584 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
12585 (smulsi3_highpart): Likewise.
12586 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
12587 (c6x_expand_compare): Likewise.
12588 (c6x_expand_movmem): Likewise.
12589 * config/frv/frv.c (frv_trampoline_init): Likewise.
12590 * config/i386/i386.c (ix86_trampoline_init): Likewise.
12591 (ix86_expand_divmod_libfunc): Likewise.
12592 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
12593 (ia64_expand_compare): Likewise.
12594 (ia64_profile_hook): Likewise.
12595 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
12596 (nonlocal_goto): Likewise.
12597 (restore_stack_nonlocal): Likewise.
12598 * config/m32r/m32r.c (block_move_call): Likewise.
12599 (m32r_trampoline_init): Likewise.
12600 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
12601 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
12602 (m68k_call_m68k_read_tp): Likewise.
12603 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
12604 (microblaze_expand_divide): Likewise.
12605 * config/mips/mips.h (mips_args): Likewise.
12606 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
12607 (MIPS_ICACHE_SYNC): Likewise.
12608 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
12609 (nios2_trampoline_init): Likewise.
12610 * config/pa/pa.c (hppa_tls_call): Likewise.
12611 (pa_trampoline_init): Likewise.
12612 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
12613 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
12614 (expand_strn_compare): Likewise.
12615 (rs6000_generate_compare): Likewise.
12616 (rs6000_expand_float128_convert): Likewise.
12617 (output_profile_hook): Likewise.
12618 (rs6000_trampoline_init): Likewise.
12619 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
12620 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
12621 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
12622 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
12623 (rs6000_generate_compare): Likewise.
12624 (rs6000_expand_float128_convert): Likewise.
12625 (output_profile_hook): Likewise.
12626 (rs6000_trampoline_init): Likewise.
12627 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
12628 * config/sh/sh.c (sh_trampoline_init): Likewise.
12629 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
12630 (sparc_emit_float_lib_cmp): Likewise.
12631 (sparc32_initialize_trampoline): Likewise.
12632 (sparc64_initialize_trampoline): Likewise.
12633 (sparc_profile_hook): Likewise.
12634 * config/spu/spu.c (ea_load_store): Likewise.
12635 * config/spu/spu.md (floatunssidf2): Likewise.
12636 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
12637 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
12638 * config/visium/visium.c (expand_block_move_4): Likewise.
12639 (expand_block_move_2): Likewise.
12640 (expand_block_move_1): Likewise.
12641 (expand_block_set_4): Likewise.
12642 (expand_block_set_2): Likewise.
12643 (expand_block_set_1): Likewise.
12644 (visium_trampoline_init): Likewise.
12645 (visium_profile_hook): Likewise.
12646 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
12647 (xtensa_setup_frame_addresses): Likewise.
12648 (xtensa_trampoline_init): Likewise.
12649 * except.c (sjlj_emit_function_enter): Likewise.
12650 (sjlj_emit_function_exit): Likewise.
12651 * explow.c (allocate_dynamic_stack_space): Likewise.
12652 (probe_stack_range): Likewise.
12653 * expr.c (convert_mode_scalar): Likewise.
12654 * optabs.c (expand_binop): Likewise.
12655 (expand_twoval_binop_libfunc): Likewise.
12656 (expand_unop): Likewise.
12657 (prepare_cmp_insn): Likewise.
12658 (prepare_float_lib_cmp): Likewise.
12659 (expand_float): Likewise.
12660 (expand_fix): Likewise.
12661 (expand_fixed_convert): Likewise.
12662 (maybe_emit_sync_lock_test_and_set): Likewise.
12663 (expand_atomic_compare_and_swap): Likewise.
12664 (expand_mem_thread_fence): Likewise.
12665 (expand_atomic_fetch_op): Likewise.
12667 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
12669 * doc/generic.texi (OpenACC): Adjust URL.
12670 * doc/invoke.texi (C Dialect Options): Ditto.
12672 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
12674 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
12675 predicate for operand 1. Add (m,<S>) constraint.
12676 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
12677 Prevent memory operand 1 with register operand 2.
12679 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
12681 PR rtl-optimization/82024
12682 * combine.c (try_combine): If the combination result is a PARALLEL,
12683 and we only need to retain the SET in there that would be placed
12684 at I2, check that we can place that at I3 instead, before doing so.
12686 2017-09-01 Jakub Jelinek <jakub@redhat.com>
12689 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
12691 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
12692 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
12695 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
12696 Jeff Law <law@redhat.com>
12698 * varasm.c (bss_initializer_p): Do not put constants into .bss
12699 (categorize_decl_for_section): Handle bss_initializer_p returning
12700 false when DECL_INITIAL is NULL.
12702 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12705 * config/s390/s390.c (s390_can_inline_p): New function.
12707 2017-09-01 Jeff Law <law@redhat.com>
12709 PR tree-optimization/82052
12710 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
12711 Always initialize the returned slot after a hash table miss
12712 when INSERT is true.
12714 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
12716 * config/s390/s390.md (mem_signal_fence): Remove.
12717 * doc/md.texi (mem_signal_fence): Remove.
12718 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
12720 * target-insns.def (mem_signal_fence): Remove.
12722 2017-09-01 Jakub Jelinek <jakub@redhat.com>
12725 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
12728 * asan.c (create_odr_indicator): Strip name encoding from assembler
12729 name before appending it after __odr_asan_.
12731 2017-09-01 Martin Liska <mliska@suse.cz>
12733 PR tree-optimization/82059
12734 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
12735 frequency only when an edge is redirected.
12737 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12739 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
12740 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
12741 (arc_conditional_register_usage): Remove ARC600 lp_count
12743 (arc_file_start): Emit Tag_ARC_CPU_variation.
12744 (arc_can_use_doloop_p): New conditions to use ZOLs.
12745 (hwloop_fail): New function.
12746 (hwloop_optimize): Likewise.
12747 (hwloop_pattern_reg): Likewise.
12748 (arc_doloop_hooks): New struct, to be used with reorg_loops.
12749 (arc_reorg_loops): New function, calls reorg_loops.
12750 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
12751 (arc600_corereg_hazard): Remove ZOL checking, case handled by
12753 (arc_loop_hazard): Remove function, functionality moved into
12755 (arc_hazard): Remove arc_loop_hazard call.
12756 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
12757 into hwloop_optimize.
12758 (arc_label_align): Remove ZOL handling.
12759 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
12760 * config/arc/arc.md (doloop_begin): Remove pattern.
12761 (doloop_begin_i): Likewise.
12762 (doloop_end_i): Likewise.
12763 (doloop_fallback): Likewise.
12764 (doloop_fallback_m): Likewise.
12765 (doloop_end): Reimplement expand.
12766 (arc_lp): New pattern for LP instruction.
12767 (loop_end): New pattern.
12768 (loop_fail): Likewise.
12769 (decrement_and_branch_until_zero): Likewise.
12770 * config/arc/arc.opt (mlpc-width): New option.
12771 * doc/invoke.texi (mlpc-width): Document option.
12773 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12775 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
12776 (arc_ccfsm_advance): Fix checking for delay slots.
12777 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
12779 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12781 * config/arc/arc.md (movqi_insn): Add stores to save constant long
12783 (movhi_insn): Update store instruction constraint which are saving
12784 6-bit short immediates.
12785 (movsi_insn): Consider also short scaled load operations.
12786 (zero_extendhisi2_i): Use Usd constraint instead of T.
12787 (extendhisi2_i): Add q constraint.
12788 (arc_clzsi2): Add type and length attributes.
12789 (arc_ctzsi2): Likewise.
12790 * config/arc/constraints.md (Usc): Update constraint, the
12791 assembler can parse two relocations for a single instruction.
12793 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12795 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
12796 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
12798 2017-08-31 Olivier Hainque <hainque@adacore.com>
12800 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
12801 match as powerpc-wrs-vxworks*.
12803 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
12805 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
12806 register constraint for by-element operand.
12807 (aarch64_mls_elt_merge<mode>): Likewise.
12809 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12811 * config/arc/arc.c (arc_can_follow_jump): Check for short
12814 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12816 * config.gcc: Use g.opt for arc.
12817 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
12818 functionality moved to ...
12819 (legitimate_scaled_address_p): New function, ...here.
12820 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
12821 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
12822 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
12824 (arc_override_options): Handle G option.
12825 (arc_output_pic_addr_const): Correct function definition.
12826 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
12827 (arc_decl_anon_ns_mem_p): Delete.
12828 (arc_in_small_data_p): Overhaul this function to take into
12829 consideration the value given via G option.
12830 (arc_rewrite_small_data_1): Renamed and corrected old
12831 arc_rewrite_small_data function.
12832 (arc_rewrite_small_data): New function.
12833 (small_data_pattern): Don't use pic_offset_table_rtx.
12834 * config/arc/arc.h (CC1_SPEC): Recognize G option.
12835 * config/arc/simdext.md (movmisalignv2hi): Use
12836 prepare_move_operands function.
12838 (movmisalign*): Likewise.
12839 * doc/invoke.texi (ARC options): Document -G option.
12841 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12843 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
12845 * config/arc/arc.c (arc_print_operand): Output scalled address for
12846 sdata whenever is possible.
12847 (arc_in_small_data_p): Allow sdata for 64bit datum when double
12848 load/stores are available.
12849 (compact_sda_memory_operand): Check for the alignment required by
12850 code density instructions.
12851 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
12853 * config/arc/constraints.md (Usd): Update constraint.
12854 (Us0): New constraint.
12855 (Usc): Update constraint.
12857 2017-08-31 Richard Biener <rguenther@suse.de>
12859 PR middle-end/82054
12860 * dwarf2out.c (dwarf2out_early_global_decl): Process each
12861 function only once.
12863 2017-08-31 Tamar Christina <tamar.christina@arm.com>
12865 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
12866 Resize type_signature.
12868 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12869 Alan Hayward <alan.hayward@arm.com>
12870 David Sherwood <david.sherwood@arm.com>
12872 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
12873 subregs whose inner modes can be stored in GPRs.
12874 (aarch64_classify_index): Likewise.
12876 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12877 Alan Hayward <alan.hayward@arm.com>
12878 David Sherwood <david.sherwood@arm.com>
12880 * config/aarch64/iterators.md (V_cmp_result): Rename to...
12881 (V_INT_EQUIV): ...this.
12882 (v_cmp_result): Rename to...
12883 (v_int_equiv): ...this.
12884 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
12885 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
12886 (copysign<mode>3): Likewise.
12887 (aarch64_simd_bsl<mode>_internal): Likewise.
12888 (aarch64_simd_bsl<mode>): Likewise.
12889 (vec_cmp<mode><mode>): Likewise.
12890 (vcond<mode><mode>): Likewise.
12891 (vcond<v_cmp_mixed><mode>): Likewise.
12892 (vcondu<mode><v_cmp_mixed>): Likewise.
12893 (aarch64_cm<optab><mode>): Likewise.
12894 (aarch64_cmtst<mode>): Likewise.
12895 (aarch64_fac<optab><mode>): Likewise.
12896 (vec_perm_const<mode>): Likewise.
12897 (vcond_mask_<mode><v_cmp_result>): Rename to...
12898 (vcond_mask_<mode><v_int_equiv>): ...this.
12899 (vec_cmp<mode><v_cmp_result>): Rename to...
12900 (vec_cmp<mode><v_int_equiv>): ...this.
12902 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12903 Alan Hayward <alan.hayward@arm.com>
12904 David Sherwood <david.sherwood@arm.com>
12906 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
12908 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
12909 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
12910 (UNSPEC_LD4_DREG): New unspecs.
12911 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
12912 (aarch64_ld2<mode>_dreg_be): Replace with...
12913 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
12915 (aarch64_ld3<mode>_dreg_le)
12916 (aarch64_ld3<mode>_dreg_be): Replace with...
12917 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
12919 (aarch64_ld4<mode>_dreg_le)
12920 (aarch64_ld4<mode>_dreg_be): Replace with...
12921 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
12924 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12926 PR tree-optimization/81987
12927 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
12928 insert an initializer in a location not dominated by the stride
12931 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
12933 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
12934 on the entire header of the finally block in the fallthru case.
12936 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
12938 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
12940 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
12942 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
12943 rs6000_emit_move_from_cr and call renamed function.
12944 (rs6000_emit_prologue): Call renamed functions.
12945 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
12946 movesi_from_cr, remove volatile CRs.
12948 2017-08-30 Jon Beniston <jon@beniston.com>
12949 Richard Biener <rguenther@suse.de>
12951 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
12952 of VECTOR_MODE_P check.
12953 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
12954 element vector types.
12956 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12958 * df.h (df_read_modify_subreg_p): Remove in favor of...
12959 * rtl.h (read_modify_subreg_p): ...this new function. Take a
12960 const_rtx instead of an rtx.
12961 * cprop.c (local_cprop_find_used_regs): Update accordingly.
12962 * df-problems.c (df_word_lr_mark_ref): Likewise.
12963 * ira-lives.c (mark_pseudo_reg_live): Likewise.
12964 (mark_pseudo_reg_dead): Likewise.
12965 (mark_ref_dead): Likewise.
12966 * reginfo.c (init_subregs_of_mode): Likewise.
12967 * sched-deps.c (sched_analyze_1): Likewise.
12968 * df-scan.c (df_def_record_1): Likewise.
12969 (df_uses_record): Likewise.
12970 (df_read_modify_subreg_p): Remove in favor of...
12971 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
12972 const_rtx instead of an rtx.
12974 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12975 Alan Hayward <alan.hayward@arm.com>
12976 David Sherwood <david.sherwood@arm.com>
12978 * rtl.h (partial_subreg_p): New function.
12979 * caller-save.c (save_call_clobbered_regs): Use it.
12980 * calls.c (expand_call): Likewise.
12981 * combine.c (combinable_i3pat): Likewise.
12982 (simplify_set): Likewise.
12983 (make_extraction): Likewise.
12984 (make_compound_operation_int): Likewise.
12985 (gen_lowpart_or_truncate): Likewise.
12986 (force_to_mode): Likewise.
12987 (make_field_assignment): Likewise.
12988 (reg_truncated_to_mode): Likewise.
12989 (record_truncated_value): Likewise.
12990 (move_deaths): Likewise.
12991 * cse.c (record_jump_cond): Likewise.
12992 (cse_insn): Likewise.
12993 * cselib.c (cselib_lookup_1): Likewise.
12994 * expmed.c (extract_bit_field_using_extv): Likewise.
12995 * function.c (assign_parm_setup_reg): Likewise.
12996 * ifcvt.c (noce_convert_multiple_sets): Likewise.
12997 * ira-build.c (create_insn_allocnos): Likewise.
12998 * lra-coalesce.c (merge_pseudos): Likewise.
12999 * lra-constraints.c (match_reload): Likewise.
13000 (simplify_operand_subreg): Likewise.
13001 (curr_insn_transform): Likewise.
13002 * lra-lives.c (process_bb_lives): Likewise.
13003 * lra.c (new_insn_reg): Likewise.
13004 (lra_substitute_pseudo): Likewise.
13005 * regcprop.c (mode_change_ok): Likewise.
13006 (maybe_mode_change): Likewise.
13007 (copyprop_hardreg_forward_1): Likewise.
13008 * reload.c (push_reload): Likewise.
13009 (find_reloads): Likewise.
13010 (find_reloads_subreg_address): Likewise.
13011 * reload1.c (alter_reg): Likewise.
13012 (eliminate_regs_1): Likewise.
13013 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13015 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
13017 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
13018 back to if statements, including unpack.
13020 2017-08-30 Martin Liska <mliska@suse.cz>
13022 PR inline-asm/82001
13023 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
13025 (func_checker::compare_asm_inputs_outputs): ... this function.
13026 (func_checker::compare_gimple_asm): Use the function to compare
13027 also ASM constrains.
13028 * ipa-icf-gimple.h: Rename the function.
13030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13031 Alan Hayward <alan.hayward@arm.com>
13032 David Sherwood <david.sherwood@arm.com>
13034 * coretypes.h (complex_mode): New type.
13035 * gdbhooks.py (build_pretty_printer): Handle it.
13036 * machmode.h (complex_mode): New class.
13037 (complex_mode::includes_p): New function.
13038 (is_complex_int_mode): Likewise.
13039 (is_complex_float_mode): Likewise.
13040 * genmodes.c (get_mode_class): Handle complex mode classes.
13041 * function.c (expand_function_end): Use is_complex_int_mode.
13043 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13044 Alan Hayward <alan.hayward@arm.com>
13045 David Sherwood <david.sherwood@arm.com>
13047 * coretypes.h (scalar_mode_pod): New typedef.
13048 * gdbhooks.py (build_pretty_printer): Handle it.
13049 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
13050 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
13051 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
13052 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
13053 as_a <scalar_mode>.
13055 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13056 Alan Hayward <alan.hayward@arm.com>
13057 David Sherwood <david.sherwood@arm.com>
13059 * machmode.h (mode_for_vector): Take a scalar_mode instead
13061 * stor-layout.c (mode_for_vector): Likewise.
13062 * explow.c (promote_mode): Use as_a <scalar_mode>.
13063 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
13065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13066 Alan Hayward <alan.hayward@arm.com>
13067 David Sherwood <david.sherwood@arm.com>
13069 * target.def (preferred_simd_mode): Take a scalar_mode
13070 instead of a machine_mode.
13071 * targhooks.h (default_preferred_simd_mode): Likewise.
13072 * targhooks.c (default_preferred_simd_mode): Likewise.
13073 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
13074 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
13075 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
13076 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
13077 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
13078 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
13079 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
13080 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
13082 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
13083 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
13084 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
13085 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
13086 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
13087 * doc/tm.texi: Regenerate.
13088 * optabs-query.c (can_vec_mask_load_store_p): Return false for
13091 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13092 Alan Hayward <alan.hayward@arm.com>
13093 David Sherwood <david.sherwood@arm.com>
13095 * target.def (scalar_mode_supported_p): Take a scalar_mode
13096 instead of a machine_mode.
13097 * targhooks.h (default_scalar_mode_supported_p): Likewise.
13098 * targhooks.c (default_scalar_mode_supported_p): Likewise.
13099 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
13100 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13101 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
13102 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
13103 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
13104 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
13105 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
13106 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
13107 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
13108 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
13109 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
13110 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
13112 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
13113 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
13114 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13115 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13116 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
13118 * doc/tm.texi: Regenerate.
13120 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13121 Alan Hayward <alan.hayward@arm.com>
13122 David Sherwood <david.sherwood@arm.com>
13124 * coretypes.h (opt_scalar_mode): New typedef.
13125 * gdbhooks.py (build_pretty_printers): Handle it.
13126 * machmode.h (mode_iterator::get_2xwider): Add overload for
13128 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
13130 * expr.c (convert_mode_scalar): Likewise.
13131 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13132 * optabs.c (expand_float): Likewise.
13133 (expand_fix): Likewise.
13134 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13137 Alan Hayward <alan.hayward@arm.com>
13138 David Sherwood <david.sherwood@arm.com>
13140 * optabs.c (expand_float): Explicitly check for scalars before
13141 using a branching expansion.
13142 (expand_fix): Likewise.
13144 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13145 Alan Hayward <alan.hayward@arm.com>
13146 David Sherwood <david.sherwood@arm.com>
13148 * expr.c (convert_mode): Split scalar handling out into...
13149 (convert_mode_scalar): ...this new function. Treat the modes
13152 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13153 Alan Hayward <alan.hayward@arm.com>
13154 David Sherwood <david.sherwood@arm.com>
13156 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
13158 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
13160 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13161 Alan Hayward <alan.hayward@arm.com>
13162 David Sherwood <david.sherwood@arm.com>
13164 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
13165 rather than a machine_mode.
13166 (fixed_from_string): Likewise.
13167 (fixed_convert): Likewise.
13168 (fixed_convert_from_int): Likewise.
13169 (fixed_convert_from_real): Likewise.
13170 (real_convert_from_fixed): Likewise.
13171 * fixed-value.c (fixed_from_double_int): Likewise.
13172 (fixed_from_string): Likewise.
13173 (fixed_convert): Likewise.
13174 (fixed_convert_from_int): Likewise.
13175 (fixed_convert_from_real): Likewise.
13176 (real_convert_from_fixed): Likewise.
13177 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
13179 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13180 Alan Hayward <alan.hayward@arm.com>
13181 David Sherwood <david.sherwood@arm.com>
13183 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
13184 of separate mode class checks. Do not allow vector modes here.
13185 (immed_wide_int_const): Use as_a <scalar_mode>.
13186 * explow.c (trunc_int_for_mode): Likewise.
13187 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
13188 (wi::shwi): Likewise.
13189 (wi::min_value): Likewise.
13190 (wi::max_value): Likewise.
13191 * dwarf2out.c (loc_descriptor): Likewise.
13192 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
13193 for CONST_WIDE_INT.
13195 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13196 Alan Hayward <alan.hayward@arm.com>
13197 David Sherwood <david.sherwood@arm.com>
13199 * tree.h (SCALAR_TYPE_MODE): New macro.
13200 * expr.c (expand_expr_addr_expr_1): Use it.
13201 (expand_expr_real_2): Likewise.
13202 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
13203 (fold_convert_const_fixed_from_int): Likewise.
13204 (fold_convert_const_fixed_from_real): Likewise.
13205 (native_encode_fixed): Likewise
13206 (native_encode_complex): Likewise
13207 (native_encode_vector): Likewise.
13208 (native_interpret_fixed): Likewise.
13209 (native_interpret_real): Likewise.
13210 (native_interpret_complex): Likewise.
13211 (native_interpret_vector): Likewise.
13212 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
13213 (simd_clone_adjust_argument_types): Likewise.
13214 (simd_clone_init_simd_arrays): Likewise.
13215 (simd_clone_adjust): Likewise.
13216 * stor-layout.c (layout_type): Likewise.
13217 * tree.c (build_minus_one_cst): Likewise.
13218 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
13219 * tree-inline.c (estimate_move_cost): Likewise.
13220 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
13221 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
13222 (vectorizable_reduction): Likewise.
13223 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
13224 (vect_recog_mixed_size_cond_pattern): Likewise.
13225 (check_bool_pattern): Likewise.
13226 (adjust_bool_pattern): Likewise.
13227 (search_type_for_mask_1): Likewise.
13228 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
13229 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13230 (vectorizable_load): Likewise.
13231 (vectorizable_store): Likewise.
13232 * ubsan.c (ubsan_encode_value): Likewise.
13233 * varasm.c (output_constant): Likewise.
13235 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13236 Alan Hayward <alan.hayward@arm.com>
13237 David Sherwood <david.sherwood@arm.com>
13239 * coretypes.h (scalar_mode): New class.
13240 * machmode.h (scalar_mode): Likewise.
13241 (scalar_mode::includes_p): New function.
13242 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
13243 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
13244 * genmodes.c (get_mode_class): Handle remaining scalar modes.
13245 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
13246 * expmed.c (store_bit_field_1): Likewise.
13247 (extract_bit_field_1): Likewise.
13248 * expr.c (write_complex_part): Likewise.
13249 (read_complex_part): Likewise.
13250 (emit_move_complex_push): Likewise.
13251 (expand_expr_real_2): Likewise.
13252 * function.c (assign_parm_setup_reg): Likewise.
13253 (assign_parms_unsplit_complex): Likewise.
13254 * optabs.c (expand_binop): Likewise.
13255 * rtlanal.c (subreg_get_info): Likewise.
13256 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13257 * varasm.c (output_constant_pool_2): Likewise.
13259 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13260 Alan Hayward <alan.hayward@arm.com>
13261 David Sherwood <david.sherwood@arm.com>
13263 * expmed.c (extract_high_half): Use scalar_int_mode and remove
13265 (expmed_mult_highpart_optab): Likewise.
13266 (expmed_mult_highpart): Likewise.
13268 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13269 Alan Hayward <alan.hayward@arm.com>
13270 David Sherwood <david.sherwood@arm.com>
13272 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
13273 instead of a machine_mode.
13274 (builtin_memset_read_str): Likewise.
13275 * builtins.c (c_readstr): Likewise.
13276 (builtin_memcpy_read_str): Likewise.
13277 (builtin_strncpy_read_str): Likewise.
13278 (builtin_memset_read_str): Likewise.
13279 (builtin_memset_gen_str): Likewise.
13280 (expand_builtin_signbit): Use scalar_int_mode for local variables.
13281 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
13282 instead of a machine_mode.
13283 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
13285 (make_extraction): Likewise.
13286 (try_widen_shift_mode): Take and return scalar_int_modes instead
13288 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
13289 a scalar_int_mode instead of a machine_mode.
13290 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
13291 (avr_addr_space_pointer_mode): Likewise.
13292 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
13293 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
13294 (msp430_unwind_word_mode): Likewise.
13295 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
13296 (spu_addr_space_pointer_mode): Likewise.
13297 (spu_addr_space_address_mode): Likewise.
13298 (spu_libgcc_cmp_return_mode): Likewise.
13299 (spu_libgcc_shift_count_mode): Likewise.
13300 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
13301 (rl78_addr_space_pointer_mode): Likewise.
13302 (fl78_unwind_word_mode): Likewise.
13303 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
13305 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
13306 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
13307 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
13308 (mips_valid_pointer_mode): Likewise.
13309 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
13310 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
13311 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
13313 (ft32_addr_space_address_mode): Likewise.
13314 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
13315 scalar_int_mode instead of a machine_mode.
13316 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
13318 (m32c_addr_space_address_mode): Likewise.
13319 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
13320 (rs6000_eh_return_filter_mode): Likewise.
13321 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
13322 (rs6000_eh_return_filter_mode): Likewise.
13323 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
13324 (s390_libgcc_shift_count_mode): Likewise.
13325 (s390_unwind_word_mode): Likewise.
13326 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
13328 * target.def (mode_rep_extended): Likewise.
13329 (valid_pointer_mode): Likewise.
13330 (addr_space.valid_pointer_mode): Likewise.
13331 (eh_return_filter_mode): Return a scalar_int_mode rather than
13333 (libgcc_cmp_return_mode): Likewise.
13334 (libgcc_shift_count_mode): Likewise.
13335 (unwind_word_mode): Likewise.
13336 (addr_space.pointer_mode): Likewise.
13337 (addr_space.address_mode): Likewise.
13338 * doc/tm.texi: Regenerate.
13339 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
13341 (do_jump): Use scalar_int_mode for local variables.
13342 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
13343 rather than a machine_mode.
13344 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
13345 (scompare_loc_descriptor_wide): Likewise.
13346 (scompare_loc_descriptor_narrow): Likewise.
13347 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
13349 * except.c (sjlj_emit_dispatch_table): Likewise.
13350 (expand_builtin_eh_copy_values): Likewise.
13351 * explow.c (convert_memory_address_addr_space_1): Likewise.
13352 Take a scalar_int_mode rather than a machine_mode.
13353 (convert_memory_address_addr_space): Take a scalar_int_mode rather
13354 than a machine_mode.
13355 (memory_address_addr_space): Use scalar_int_mode for local variables.
13356 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
13357 rather than a machine_mode.
13358 * expmed.c (mask_rtx): Likewise.
13359 (init_expmed_one_conv): Likewise.
13360 (expand_mult_highpart_adjust): Likewise.
13361 (extract_high_half): Likewise.
13362 (expmed_mult_highpart_optab): Likewise.
13363 (expmed_mult_highpart): Likewise.
13364 (expand_smod_pow2): Likewise.
13365 (expand_sdiv_pow2): Likewise.
13366 (emit_store_flag_int): Likewise.
13367 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
13369 (extract_low_bits): Likewise.
13370 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
13372 * expr.c (pieces_addr::adjust): Likewise.
13373 (can_store_by_pieces): Likewise.
13374 (store_by_pieces): Likewise.
13375 (clear_by_pieces_1): Likewise.
13376 (expand_expr_addr_expr_1): Likewise.
13377 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
13378 (expand_expr_real_1): Likewise.
13379 (try_casesi): Likewise.
13380 * final.c (shorten_branches): Likewise.
13381 * fold-const.c (fold_convert_const_int_from_fixed): Change the
13382 type of "mode" to machine_mode.
13383 * internal-fn.c (expand_arith_overflow_result_store): Take a
13384 scalar_int_mode rather than a machine_mode.
13385 (expand_mul_overflow): Use scalar_int_mode for local variables.
13386 * loop-doloop.c (doloop_modify): Likewise.
13387 (doloop_optimize): Likewise.
13388 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
13389 than a machine_mode.
13390 (expand_doubleword_shift_condmove): Likewise.
13391 (expand_doubleword_shift): Likewise.
13392 (expand_doubleword_clz): Likewise.
13393 (expand_doubleword_popcount): Likewise.
13394 (expand_doubleword_parity): Likewise.
13395 (expand_absneg_bit): Use scalar_int_mode for local variables.
13396 (prepare_float_lib_cmp): Likewise.
13397 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
13398 rather than a machine_mode.
13399 (convert_memory_address_addr_space): Likewise.
13400 (get_mode_bounds): Likewise.
13401 (get_address_mode): Return a scalar_int_mode rather than a
13403 * rtlanal.c (get_address_mode): Likewise.
13404 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
13405 than a machine_mode.
13406 * targhooks.c (default_mode_rep_extended): Likewise.
13407 (default_valid_pointer_mode): Likewise.
13408 (default_addr_space_valid_pointer_mode): Likewise.
13409 (default_eh_return_filter_mode): Return a scalar_int_mode rather
13410 than a machine_mode.
13411 (default_libgcc_cmp_return_mode): Likewise.
13412 (default_libgcc_shift_count_mode): Likewise.
13413 (default_unwind_word_mode): Likewise.
13414 (default_addr_space_pointer_mode): Likewise.
13415 (default_addr_space_address_mode): Likewise.
13416 * targhooks.h (default_eh_return_filter_mode): Likewise.
13417 (default_libgcc_cmp_return_mode): Likewise.
13418 (default_libgcc_shift_count_mode): Likewise.
13419 (default_unwind_word_mode): Likewise.
13420 (default_addr_space_pointer_mode): Likewise.
13421 (default_addr_space_address_mode): Likewise.
13422 (default_mode_rep_extended): Take a scalar_int_mode rather than
13424 (default_valid_pointer_mode): Likewise.
13425 (default_addr_space_valid_pointer_mode): Likewise.
13426 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
13428 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
13429 rather than a machine_mode.
13430 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
13431 for local variables.
13432 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13433 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
13434 than a machine_mode.
13436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13437 Alan Hayward <alan.hayward@arm.com>
13438 David Sherwood <david.sherwood@arm.com>
13440 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
13441 the mode argument to scalar_int_mode.
13442 (do_jump_by_parts_zero_rtx): Likewise.
13443 (do_jump_by_parts_equality_rtx): Likewise.
13444 (do_jump_by_parts_greater): Take a mode argument.
13445 (do_jump_by_parts_equality): Likewise.
13446 (do_jump_1): Update calls accordingly.
13448 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13449 Alan Hayward <alan.hayward@arm.com>
13450 David Sherwood <david.sherwood@arm.com>
13452 * is-a.h (safe_dyn_cast): New function.
13453 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
13454 (jump_table_for_label): Likewise.
13455 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
13456 instead of an rtx_insn *.
13457 (shorten_branches): Use dyn_cast instead of LABEL_P and
13458 JUMP_TABLE_DATA_P. Use jump_table_for_label and
13459 rtx_jump_table_data::get_data_mode.
13460 (final_scan_insn): Likewise.
13462 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13463 Alan Hayward <alan.hayward@arm.com>
13464 David Sherwood <david.sherwood@arm.com>
13466 * combine.c (try_combine): Use is_a <scalar_int_mode> when
13467 trying to combine a full-register integer set with a subreg
13470 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13471 Alan Hayward <alan.hayward@arm.com>
13472 David Sherwood <david.sherwood@arm.com>
13474 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
13475 that is always either address_mode or pointer_mode.
13477 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13478 Alan Hayward <alan.hayward@arm.com>
13479 David Sherwood <david.sherwood@arm.com>
13481 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
13482 when the two are known to be equal.
13484 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13485 Alan Hayward <alan.hayward@arm.com>
13486 David Sherwood <david.sherwood@arm.com>
13488 * simplify-rtx.c (simplify_const_unary_operation): Use
13489 is_a <scalar_int_mode> instead of checking for a nonzero
13490 precision. Forcibly convert op_mode to a scalar_int_mode
13491 in that case. More clearly differentiate the operand and
13492 result modes and use the former when deciding what the value
13493 of a count-bits operation should be. Use is_int_mode instead
13494 of checking for a MODE_INT. Remove redundant check for whether
13495 this mode has a zero precision.
13497 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13498 Alan Hayward <alan.hayward@arm.com>
13499 David Sherwood <david.sherwood@arm.com>
13501 * optabs.c (widen_leading): Change the type of the mode argument
13502 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
13503 (widen_bswap): Likewise.
13504 (expand_parity): Likewise.
13505 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
13506 (expand_ffs): Likewise.
13507 (epand_unop): Check for scalar integer modes before calling the
13510 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13511 Alan Hayward <alan.hayward@arm.com>
13512 David Sherwood <david.sherwood@arm.com>
13514 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
13516 (expand_expr_real_1): Update call accordingly.
13518 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13519 Alan Hayward <alan.hayward@arm.com>
13520 David Sherwood <david.sherwood@arm.com>
13522 * expmed.c (store_bit_field_using_insv): Add op0_mode and
13523 value_mode arguments. Use scalar_int_mode internally.
13524 (store_bit_field_1): Rename the new integer mode from imode
13525 to op0_mode and use it instead of GET_MODE (op0). Update calls
13526 to store_split_bit_field, store_bit_field_using_insv and
13527 store_fixed_bit_field.
13528 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
13529 Use scalar_int_mode internally. Use a bit count rather than a mode
13530 when calculating the largest bit size for get_best_mode.
13531 Update calls to store_split_bit_field and store_fixed_bit_field_1.
13532 (store_fixed_bit_field_1): Add mode and value_mode arguments.
13533 Remove assertion that OP0 has a scalar integer mode.
13534 (store_split_bit_field): Add op0_mode and value_mode arguments.
13535 Update calls to extract_fixed_bit_field.
13536 (extract_bit_field_using_extv): Add an op0_mode argument.
13537 Use scalar_int_mode internally.
13538 (extract_bit_field_1): Rename the new integer mode from imode to
13539 op0_mode and use it instead of GET_MODE (op0). Update calls to
13540 extract_split_bit_field, extract_bit_field_using_extv and
13541 extract_fixed_bit_field.
13542 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
13543 to extract_split_bit_field and extract_fixed_bit_field_1.
13544 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
13545 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
13546 on the target mode.
13547 (extract_split_bit_field): Add an op0_mode argument. Update call
13548 to extract_fixed_bit_field.
13550 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13551 Alan Hayward <alan.hayward@arm.com>
13552 David Sherwood <david.sherwood@arm.com>
13554 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
13555 * explow.c (hard_function_value): Likewise.
13556 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
13557 convert_to_mode call outside the loop.
13558 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
13559 for the mode iterator. Require the mode specified by max_pieces
13561 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
13563 (copy_blkmode_to_reg): Likewise.
13564 (set_storage_via_setmem): Likewise.
13565 * optabs.c (prepare_cmp_insn): Likewise.
13566 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13567 * stor-layout.c (finish_bitfield_representative): Likewise.
13569 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13570 Alan Hayward <alan.hayward@arm.com>
13571 David Sherwood <david.sherwood@arm.com>
13573 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
13574 * expr.c (convert_move): Use them.
13575 (convert_modes): Likewise.
13576 (store_expr_with_bounds): Likewise.
13578 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13579 Alan Hayward <alan.hayward@arm.com>
13580 David Sherwood <david.sherwood@arm.com>
13582 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
13583 parameter for the mode of "x". Remove the "known_x", "known_mode"
13584 and "known_ret" arguments. Change the type of the mode argument
13585 to scalar_int_mode.
13586 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
13587 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
13588 (reg_num_sign_bit_copies_for_combine): Likewise.
13589 * rtlanal.c (nonzero_bits1): Likewise.
13590 (num_sign_bit_copies1): Likewise.
13591 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
13592 (reg_num_sign_bit_copies_general): Likewise.
13593 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
13594 (reg_nonzero_bits_general): Likewise.
13596 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13597 Alan Hayward <alan.hayward@arm.com>
13598 David Sherwood <david.sherwood@arm.com>
13600 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
13601 than in subroutines. Return 1 for non-integer modes.
13602 (cached_num_sign_bit_copies): Change the type of the mode parameter
13603 to scalar_int_mode.
13604 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
13605 classes. Handle CONST_INT_P first and then check whether X also
13606 has a scalar integer mode. Check the same thing for inner registers
13607 of a SUBREG and for values that are being extended or truncated.
13609 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13610 Alan Hayward <alan.hayward@arm.com>
13611 David Sherwood <david.sherwood@arm.com>
13613 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
13614 in subroutines. Return the mode mask for non-integer modes.
13615 (cached_nonzero_bits): Change the type of the mode parameter
13616 to scalar_int_mode.
13617 (nonzero_bits1): Likewise. Remove early exit for other mode
13618 classes. Handle CONST_INT_P first and then check whether X
13619 also has a scalar integer mode.
13621 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13622 Alan Hayward <alan.hayward@arm.com>
13623 David Sherwood <david.sherwood@arm.com>
13625 * expr.c (widest_int_mode_for_size): Make the comment match the code.
13626 Return a scalar_int_mode and assert that the size is greater than
13628 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
13629 (op_by_pieces_d::op_by_pieces_d): Likewise.
13630 (op_by_pieces_d::run): Likewise.
13631 (can_store_by_pieces): Likewise.
13633 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13634 Alan Hayward <alan.hayward@arm.com>
13635 David Sherwood <david.sherwood@arm.com>
13637 * combine.c (extract_left_shift): Add a mode argument and update
13639 (make_compound_operation_int): Change the type of the mode parameter
13640 to scalar_int_mode and update the call to extract_left_shift.
13642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13643 Alan Hayward <alan.hayward@arm.com>
13644 David Sherwood <david.sherwood@arm.com>
13646 * combine.c (simplify_and_const_int): Change the type of the mode
13647 parameter to scalar_int_mode.
13648 (simplify_and_const_int_1): Likewise. Update recursive call.
13650 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13651 Alan Hayward <alan.hayward@arm.com>
13652 David Sherwood <david.sherwood@arm.com>
13654 * combine.c (simplify_compare_const): Check that the mode is a
13655 scalar_int_mode (rather than VOIDmode) before testing its
13657 (simplify_comparison): Move COMPARISON_P handling out of the
13658 loop and restrict the latter part of the loop to scalar_int_modes.
13659 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
13660 and when considering SUBREG_REGs. Use is_int_mode instead of
13661 checking GET_MODE_CLASS against MODE_INT.
13663 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13664 Alan Hayward <alan.hayward@arm.com>
13665 David Sherwood <david.sherwood@arm.com>
13667 * combine.c (try_widen_shift_mode): Move check for equal modes to...
13668 (simplify_shift_const_1): ...here. Use scalar_int_mode for
13669 shift_unit_mode and for modes involved in scalar shifts.
13671 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13672 Alan Hayward <alan.hayward@arm.com>
13673 David Sherwood <david.sherwood@arm.com>
13675 * combine.c (force_int_to_mode): New function, split out from...
13676 (force_to_mode): ...here. Keep xmode up-to-date and use it
13677 instead of GET_MODE (x).
13679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13680 Alan Hayward <alan.hayward@arm.com>
13681 David Sherwood <david.sherwood@arm.com>
13683 * optabs-query.h (extraction_insn::struct_mode): Change type to
13684 opt_scalar_int_mode and update comment.
13685 (extraction_insn::field_mode): Change type to scalar_int_mode.
13686 (extraction_insn::pos_mode): Likewise.
13687 * combine.c (make_extraction): Update accordingly.
13688 * optabs-query.c (get_traditional_extraction_insn): Likewise.
13689 (get_optab_extraction_insn): Likewise.
13690 * recog.c (simplify_while_replacing): Likewise.
13691 * expmed.c (narrow_bit_field_mem): Change the type of the mode
13692 parameter to opt_scalar_int_mode.
13694 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13695 Alan Hayward <alan.hayward@arm.com>
13696 David Sherwood <david.sherwood@arm.com>
13698 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
13699 to a scalar_int_mode instead of a machine_mode.
13700 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
13701 (get_best_mode): Return a boolean and use a pointer argument to store
13702 the selected mode. Replace the limit mode parameter with a bit limit.
13703 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
13704 for the values returned by bit_field_mode_iterator::next_mode.
13705 (store_bit_field): Update call to get_best_mode.
13706 (store_fixed_bit_field): Likewise.
13707 (extract_fixed_bit_field): Likewise.
13708 * expr.c (optimize_bitfield_assignment_op): Likewise.
13709 * fold-const.c (optimize_bit_field_compare): Likewise.
13710 (fold_truth_andor_1): Likewise.
13711 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
13712 Update for new type of m_mode.
13713 (get_best_mode): As above.
13715 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13716 Alan Hayward <alan.hayward@arm.com>
13717 David Sherwood <david.sherwood@arm.com>
13719 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
13720 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
13721 (store_bit_field): Check is_a <scalar_int_mode> before calling
13722 strict_volatile_bitfield_p.
13723 (extract_bit_field): Likewise.
13725 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13726 Alan Hayward <alan.hayward@arm.com>
13727 David Sherwood <david.sherwood@arm.com>
13729 * target.def (cstore_mode): Return a scalar_int_mode.
13730 * doc/tm.texi: Regenerate.
13731 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
13732 * targhooks.h (default_cstore_mode): Likewise.
13733 * targhooks.c (default_cstore_mode): Likewise, using a forced
13735 * expmed.c (emit_cstore): Expect the target of the cstore to be
13738 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13739 Alan Hayward <alan.hayward@arm.com>
13740 David Sherwood <david.sherwood@arm.com>
13742 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
13744 (niter_desc): Likewise mode.
13745 (iv_analyze): Add a mode parameter.
13747 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
13748 and change its type to scalar_int_mode.
13749 * loop-iv.c: Update commentary at head of file.
13750 (iv_constant): Pass the mode paraeter before the rtx it describes
13751 and change its type to scalar_int_mode. Remove VOIDmode handling.
13752 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
13753 (iv_extend): Likewise.
13754 (shorten_into_mode): Likewise.
13755 (iv_add): Use scalar_int_mode.
13756 (iv_mult): Likewise.
13757 (iv_shift): Likewise.
13758 (canonicalize_iv_subregs): Likewise.
13759 (get_biv_step_1): Pass the outer_mode parameter before the rtx
13760 it describes and change its mode to scalar_int_mode. Also change
13761 the type of the returned inner_mode to scalar_int_mode.
13762 (get_biv_step): Likewise, turning outer_mode from a pointer
13763 into a direct parameter. Update call to get_biv_step_1.
13764 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
13765 iv_constant and get_biv_step.
13766 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
13767 and change its type to scalar_int_mode. Don't initialise iv->mode
13768 to VOIDmode and remove later checks for its still being VOIDmode.
13769 Update calls to iv_analyze_op and iv_analyze_expr. Check
13770 is_a <scalar_int_mode> when changing the mode under consideration.
13771 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
13772 Update call to iv_analyze_expr.
13773 (iv_analyze_op): Add a mode parameter. Reject subregs whose
13774 inner register is not also a scalar_int_mode. Update call to
13776 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
13777 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
13778 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
13779 separate mode class checks. Update calls to iv_analyze. Remove
13780 fix-up of VOIDmodes after iv_analyze_biv.
13781 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
13782 don't have a scalar_int_mode. Update call to biv_p.
13784 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13785 Alan Hayward <alan.hayward@arm.com>
13786 David Sherwood <david.sherwood@arm.com>
13788 * cfgexpand.c (convert_debug_memory_address): Use
13789 as_a <scalar_int_mode>.
13790 * combine.c (expand_compound_operation): Likewise.
13791 (make_extraction): Likewise.
13792 (change_zero_ext): Likewise.
13793 (simplify_comparison): Likewise.
13794 * cse.c (cse_insn): Likewise.
13795 * dwarf2out.c (minmax_loc_descriptor): Likewise.
13796 (mem_loc_descriptor): Likewise.
13797 (loc_descriptor): Likewise.
13798 * expmed.c (init_expmed_one_mode): Likewise.
13799 (synth_mult): Likewise.
13800 (emit_store_flag_1): Likewise.
13801 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
13802 of a comparison with size.
13803 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
13804 (reduce_to_bit_field_precision): Likewise.
13805 * function.c (expand_function_end): Likewise.
13806 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
13807 * loop-doloop.c (doloop_modify): Likewise.
13808 * optabs.c (expand_binop): Likewise.
13809 (expand_unop): Likewise.
13810 (expand_copysign_absneg): Likewise.
13811 (prepare_cmp_insn): Likewise.
13812 (maybe_legitimize_operand): Likewise.
13813 * recog.c (const_scalar_int_operand): Likewise.
13814 * rtlanal.c (get_address_mode): Likewise.
13815 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13816 (simplify_cond_clz_ctz): Likewise.
13817 * tree-nested.c (get_nl_goto_field): Likewise.
13818 * tree.c (build_vector_type_for_mode): Likewise.
13819 * var-tracking.c (use_narrower_mode): Likewise.
13821 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13822 Alan Hayward <alan.hayward@arm.com>
13823 David Sherwood <david.sherwood@arm.com>
13825 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
13826 * builtins.c (expand_builtin_signbit): Use it.
13827 * cfgexpand.c (expand_debug_expr): Likewise.
13828 * dojump.c (do_jump): Likewise.
13829 (do_compare_and_jump): Likewise.
13830 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
13831 * expmed.c (make_tree): Likewise.
13832 * expr.c (expand_expr_real_2): Likewise.
13833 (expand_expr_real_1): Likewise.
13834 (try_casesi): Likewise.
13835 * fold-const-call.c (fold_const_call_ss): Likewise.
13836 * fold-const.c (unextend): Likewise.
13837 (extract_muldiv_1): Likewise.
13838 (fold_single_bit_test): Likewise.
13839 (native_encode_int): Likewise.
13840 (native_encode_string): Likewise.
13841 (native_interpret_int): Likewise.
13842 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
13843 * internal-fn.c (expand_addsub_overflow): Likewise.
13844 (expand_neg_overflow): Likewise.
13845 (expand_mul_overflow): Likewise.
13846 (expand_arith_overflow): Likewise.
13847 * match.pd: Likewise.
13848 * stor-layout.c (layout_type): Likewise.
13849 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
13850 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
13851 * tree-ssanames.c (get_range_info): Likewise.
13852 * tree-switch-conversion.c (array_value_type) Likewise.
13853 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
13854 (vect_recog_divmod_pattern): Likewise.
13855 (vect_recog_mixed_size_cond_pattern): Likewise.
13856 * tree-vrp.c (extract_range_basic): Likewise.
13857 (simplify_float_conversion_using_ranges): Likewise.
13858 * tree.c (int_fits_type_p): Likewise.
13859 * ubsan.c (instrument_bool_enum_load): Likewise.
13860 * varasm.c (mergeable_string_section): Likewise.
13861 (narrowing_initializer_constant_valid_p): Likewise.
13862 (output_constant): Likewise.
13864 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13865 Alan Hayward <alan.hayward@arm.com>
13866 David Sherwood <david.sherwood@arm.com>
13868 * machmode.h (NARROWEST_INT_MODE): New macro.
13869 * expr.c (alignment_for_piecewise_move): Use it instead of
13870 GET_CLASS_NARROWEST_MODE (MODE_INT).
13871 (push_block): Likewise.
13872 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
13874 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13876 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13877 Alan Hayward <alan.hayward@arm.com>
13878 David Sherwood <david.sherwood@arm.com>
13880 * postreload.c (move2add_valid_value_p): Change the type of the
13881 mode parameter to scalar_int_mode.
13882 (move2add_use_add2_insn): Add a mode parameter and use it instead
13884 (move2add_use_add3_insn): Likewise.
13885 (reload_cse_move2add): Update accordingly.
13887 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13888 Alan Hayward <alan.hayward@arm.com>
13889 David Sherwood <david.sherwood@arm.com>
13891 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
13893 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
13894 * optabs.c (expand_unop): Likewise.
13896 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13897 Alan Hayward <alan.hayward@arm.com>
13898 David Sherwood <david.sherwood@arm.com>
13900 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
13901 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
13902 (popcount_loc_descriptor): Likewise.
13903 (bswap_loc_descriptor): Likewise.
13904 (rotate_loc_descriptor): Likewise.
13905 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
13906 calling the functions above.
13908 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13909 Alan Hayward <alan.hayward@arm.com>
13910 David Sherwood <david.sherwood@arm.com>
13912 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
13914 (try_combine): Likewise.
13915 (simplify_if_then_else): Likewise.
13916 * cse.c (cse_insn): Likewise.
13917 * dwarf2out.c (mem_loc_descriptor): Likewise.
13918 * emit-rtl.c (gen_lowpart_common): Likewise.
13919 * simplify-rtx.c (simplify_truncation): Likewise.
13920 (simplify_binary_operation_1): Likewise.
13921 (simplify_const_relational_operation): Likewise.
13922 (simplify_ternary_operation): Likewise.
13923 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
13925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13926 Alan Hayward <alan.hayward@arm.com>
13927 David Sherwood <david.sherwood@arm.com>
13929 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
13930 * reload.c (push_reload): Likewise.
13931 (find_reloads): Likewise.
13933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13934 Alan Hayward <alan.hayward@arm.com>
13935 David Sherwood <david.sherwood@arm.com>
13937 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
13938 (make_compound_operation_int): Likewise.
13939 (change_zero_ext): Likewise.
13940 * expr.c (convert_move): Likewise.
13941 (convert_modes): Likewise.
13942 * fwprop.c (forward_propagate_subreg): Likewise.
13943 * loop-iv.c (get_biv_step_1): Likewise.
13944 * optabs.c (widen_operand): Likewise.
13945 * postreload.c (move2add_valid_value_p): Likewise.
13946 * recog.c (simplify_while_replacing): Likewise.
13947 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13948 (simplify_binary_operation_1): Likewise. Remove redundant
13949 mode equality check.
13951 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13952 Alan Hayward <alan.hayward@arm.com>
13953 David Sherwood <david.sherwood@arm.com>
13955 * combine.c (combine_simplify_rtx): Add checks for
13956 is_a <scalar_int_mode>.
13957 (simplify_if_then_else): Likewise.
13958 (make_field_assignment): Likewise.
13959 (simplify_comparison): Likewise.
13960 * ifcvt.c (noce_try_bitop): Likewise.
13961 * loop-invariant.c (canonicalize_address_mult): Likewise.
13962 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13964 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13965 Alan Hayward <alan.hayward@arm.com>
13966 David Sherwood <david.sherwood@arm.com>
13968 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
13969 is_a <scalar_int_mode> instead of != BLKmode.
13971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13972 Alan Hayward <alan.hayward@arm.com>
13973 David Sherwood <david.sherwood@arm.com>
13975 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
13976 instead of != VOIDmode.
13977 * combine.c (if_then_else_cond): Likewise.
13978 (change_zero_ext): Likewise.
13979 * dwarf2out.c (mem_loc_descriptor): Likewise.
13980 (loc_descriptor): Likewise.
13981 * rtlanal.c (canonicalize_condition): Likewise.
13982 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
13984 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13985 Alan Hayward <alan.hayward@arm.com>
13986 David Sherwood <david.sherwood@arm.com>
13988 * simplify-rtx.c (simplify_binary_operation_1): Use
13989 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
13991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13992 Alan Hayward <alan.hayward@arm.com>
13993 David Sherwood <david.sherwood@arm.com>
13995 * wide-int.h (int_traits<unsigned char>) New class.
13996 (int_traits<unsigned short>) Likewise.
13997 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
13998 Use GET_MODE_UNIT_PRECISION and remove redundant test for
14000 * combine.c (set_nonzero_bits_and_sign_copies): Use
14001 is_a <scalar_int_mode>.
14002 (find_split_point): Likewise.
14003 (combine_simplify_rtx): Likewise.
14004 (simplify_logical): Likewise.
14005 (expand_compound_operation): Likewise.
14006 (expand_field_assignment): Likewise.
14007 (make_compound_operation): Likewise.
14008 (extended_count): Likewise.
14009 (change_zero_ext): Likewise.
14010 (simplify_comparison): Likewise.
14011 * dwarf2out.c (scompare_loc_descriptor): Likewise.
14012 (ucompare_loc_descriptor): Likewise.
14013 (minmax_loc_descriptor): Likewise.
14014 (mem_loc_descriptor): Likewise.
14015 (loc_descriptor): Likewise.
14016 * expmed.c (init_expmed_one_mode): Likewise.
14017 * lra-constraints.c (lra_constraint_offset): Likewise.
14018 * optabs.c (prepare_libcall_arg): Likewise.
14019 * postreload.c (move2add_note_store): Likewise.
14020 * reload.c (operands_match_p): Likewise.
14021 * rtl.h (load_extend_op): Likewise.
14022 * rtlhooks.c (gen_lowpart_general): Likewise.
14023 * simplify-rtx.c (simplify_truncation): Likewise.
14024 (simplify_unary_operation_1): Likewise.
14025 (simplify_binary_operation_1): Likewise.
14026 (simplify_const_binary_operation): Likewise.
14027 (simplify_const_relational_operation): Likewise.
14028 (simplify_subreg): Likewise.
14029 * stor-layout.c (bitwise_mode_for_mode): Likewise.
14030 * var-tracking.c (adjust_mems): Likewise.
14031 (prepare_call_arguments): Likewise.
14033 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14034 Alan Hayward <alan.hayward@arm.com>
14035 David Sherwood <david.sherwood@arm.com>
14037 * machmode.h (is_int_mode): New fuction.
14038 * combine.c (find_split_point): Use it.
14039 (combine_simplify_rtx): Likewise.
14040 (simplify_if_then_else): Likewise.
14041 (simplify_set): Likewise.
14042 (simplify_shift_const_1): Likewise.
14043 (simplify_comparison): Likewise.
14044 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
14045 * cse.c (notreg_cost): Likewise.
14046 (cse_insn): Likewise.
14047 * cselib.c (cselib_lookup_1): Likewise.
14048 * dojump.c (do_jump_1): Likewise.
14049 (do_compare_rtx_and_jump): Likewise.
14050 * dse.c (get_call_args): Likewise.
14051 * dwarf2out.c (rtl_for_decl_init): Likewise.
14052 (native_encode_initializer): Likewise.
14053 * expmed.c (emit_store_flag_1): Likewise.
14054 (emit_store_flag): Likewise.
14055 * expr.c (convert_modes): Likewise.
14056 (store_field): Likewise.
14057 (expand_expr_real_1): Likewise.
14058 * fold-const.c (fold_read_from_constant_string): Likewise.
14059 * gimple-ssa-sprintf.c (get_format_string): Likewise.
14060 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
14061 * optabs.c (expand_binop): Likewise.
14062 (expand_unop): Likewise.
14063 (expand_abs_nojump): Likewise.
14064 (expand_one_cmpl_abs_nojump): Likewise.
14065 * simplify-rtx.c (mode_signbit_p): Likewise.
14066 (val_signbit_p): Likewise.
14067 (val_signbit_known_set_p): Likewise.
14068 (val_signbit_known_clear_p): Likewise.
14069 (simplify_relational_operation_1): Likewise.
14070 * tree.c (vector_type_mode): Likewise.
14072 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14073 Alan Hayward <alan.hayward@arm.com>
14074 David Sherwood <david.sherwood@arm.com>
14076 * machmode.h (smallest_mode_for_size): Fix formatting.
14077 (smallest_int_mode_for_size): New function.
14078 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
14079 instead of smallest_mode_for_size.
14080 * combine.c (make_extraction): Likewise.
14081 * config/arc/arc.c (arc_expand_movmem): Likewise.
14082 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
14083 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
14084 * config/s390/s390.c (s390_expand_insv): Likewise.
14085 * config/sparc/sparc.c (assign_int_registers): Likewise.
14086 * config/spu/spu.c (spu_function_value): Likewise.
14087 (spu_function_arg): Likewise.
14088 * coverage.c (get_gcov_type): Likewise.
14089 (get_gcov_unsigned_t): Likewise.
14090 * dse.c (find_shift_sequence): Likewise.
14091 * expmed.c (store_bit_field_1): Likewise.
14092 * expr.c (convert_move): Likewise.
14093 (store_field): Likewise.
14094 * internal-fn.c (expand_arith_overflow): Likewise.
14095 * optabs-query.c (get_best_extraction_insn): Likewise.
14096 * optabs.c (expand_twoval_binop_libfunc): Likewise.
14097 * stor-layout.c (layout_type): Likewise.
14098 (initialize_sizetypes): Likewise.
14099 * targhooks.c (default_get_mask_mode): Likewise.
14100 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
14102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14103 Alan Hayward <alan.hayward@arm.com>
14104 David Sherwood <david.sherwood@arm.com>
14106 * machmode.h (opt_mode::else_blk): New function.
14107 (int_mode_for_mode): Declare.
14108 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
14109 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
14111 * cfgexpand.c (expand_debug_expr): Likewise.
14112 * combine.c (gen_lowpart_or_truncate): Likewise.
14113 (gen_lowpart_for_combine): Likewise.
14114 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
14115 * config/avr/avr.c (avr_to_int_mode): Likewise.
14116 (avr_out_plus_1): Likewise.
14117 (avr_out_plus): Likewise.
14118 (avr_out_round): Likewise.
14119 * config/i386/i386.c (ix86_split_to_parts): Likewise.
14120 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
14121 (s390_expand_vcond): Likewise.
14122 * config/spu/spu.c (spu_split_immediate): Likewise.
14123 (spu_expand_mov): Likewise.
14124 * dse.c (get_stored_val): Likewise.
14125 * expmed.c (store_bit_field_1): Likewise.
14126 (convert_extracted_bit_field): Use int_mode_for_mode instead of
14128 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
14129 (extract_low_bits): Likewise.
14130 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
14131 handling rather than repeating the check.
14132 (emit_group_store): Likewise.
14133 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
14134 * optabs.c (expand_absneg_bit): Likewise.
14135 (expand_copysign_absneg): Likewise.
14136 (expand_copysign_bit): Likewise.
14137 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
14138 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
14139 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
14140 * var-tracking.c (prepare_call_arguments): Likewise.
14141 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
14142 int_mode_for_mode instead of mode_for_size.
14143 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
14145 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14146 Alan Hayward <alan.hayward@arm.com>
14147 David Sherwood <david.sherwood@arm.com>
14149 * machmode.h (int_mode_for_size): New function.
14150 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
14151 instead of mode_for_size.
14152 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
14154 * combine.c (expand_field_assignment): Use int_mode_for_size
14155 instead of mode_for_size.
14156 (make_extraction): Likewise.
14157 (simplify_shift_const_1): Likewise.
14158 (simplify_comparison): Likewise.
14159 * dojump.c (do_jump): Likewise.
14160 * dwarf2out.c (mem_loc_descriptor): Likewise.
14161 * emit-rtl.c (init_derived_machine_modes): Likewise.
14162 * expmed.c (flip_storage_order): Likewise.
14163 (convert_extracted_bit_field): Likewise.
14164 * expr.c (copy_blkmode_from_reg): Likewise.
14165 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
14166 * internal-fn.c (expand_mul_overflow): Likewise.
14167 * lower-subreg.c (simple_move): Likewise.
14168 * optabs-libfuncs.c (init_optabs): Likewise.
14169 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14170 * tree.c (vector_type_mode): Likewise.
14171 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
14172 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
14173 * tree-vect-generic.c (expand_vector_parallel): Likewise.
14174 * tree-vect-stmts.c (vectorizable_load): Likewise.
14175 (vectorizable_store): Likewise.
14177 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14178 Alan Hayward <alan.hayward@arm.com>
14179 David Sherwood <david.sherwood@arm.com>
14181 * coretypes.h (pod_mode): New type.
14182 (scalar_int_mode_pod): New typedef.
14183 * machmode.h (pod_mode): New class.
14184 (int_n_data_t::m): Change type to scalar_int_mode_pod.
14185 * genmodes.c (emit_mode_int_n): Update accordingly.
14186 * lower-subreg.h (target_lower_subreg): Change type to
14187 scalar_int_mode_pod.
14188 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
14189 scalar_int_mode_pod.
14191 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14192 Alan Hayward <alan.hayward@arm.com>
14193 David Sherwood <david.sherwood@arm.com>
14195 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
14196 machine_mode to scalar_int_mode.
14197 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
14198 (rs6000_option_override_internal): Remove cast to int.
14199 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
14200 machine_mode to scalar_int_mode.
14201 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
14202 (rs6000_option_override_internal): Remove cast to int.
14203 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
14204 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
14206 * config/s390/s390.c (s390_expand_builtin): Likewise.
14207 * coretypes.h (scalar_int_mode): New type.
14208 (opt_scalar_int_mode): New typedef.
14209 * machmode.h (scalar_int_mode): New class.
14210 (scalar_int_mode::includes_p): New function.
14211 (byte_mode): Change type to scalar_int_mode.
14212 (word_mode): Likewise.
14213 (ptr_mode): Likewise.
14214 * emit-rtl.c (byte_mode): Likewise.
14215 (word_mode): Likewise.
14216 (ptr_mode): Likewise.
14217 (init_derived_machine_modes): Update accordingly.
14218 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
14219 and MODE_PARTIAL_INT.
14220 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
14221 opt_scalar_int_mode.
14223 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14224 Alan Hayward <alan.hayward@arm.com>
14225 David Sherwood <david.sherwood@arm.com>
14227 * target.def (libgcc_floating_mode_supported_p): Take a
14229 * doc/tm.texi: Regenerate.
14230 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
14232 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
14233 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
14236 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14237 Alan Hayward <alan.hayward@arm.com>
14238 David Sherwood <david.sherwood@arm.com>
14240 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
14241 * doc/tm.texi: Regenerate.
14242 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
14243 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
14244 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
14245 * targhooks.h (default_floatn_mode): Likewise.
14246 * targhooks.c (default_floatn_mode): Likewise.
14247 * tree.c (build_common_tree_nodes): Update accordingly.
14249 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14250 Alan Hayward <alan.hayward@arm.com>
14251 David Sherwood <david.sherwood@arm.com>
14253 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
14254 (mode_iterator::iterate_p): Likewise.
14255 (mode_iterator::get_wider): Likewise.
14256 * expr.c (init_expr_target): Use opt_scalar_float_mode.
14258 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14259 Alan Hayward <alan.hayward@arm.com>
14260 David Sherwood <david.sherwood@arm.com>
14262 * coretypes.h (opt_scalar_float_mode): New typedef.
14263 * machmode.h (float_mode_for_size): New function.
14264 * emit-rtl.c (double_mode): Delete.
14265 (init_emit_once): Use float_mode_for_size.
14266 * stor-layout.c (layout_type): Likewise.
14267 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
14269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14270 Alan Hayward <alan.hayward@arm.com>
14271 David Sherwood <david.sherwood@arm.com>
14273 * output.h (assemble_real): Take a scalar_float_mode.
14274 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
14275 * config/arm/arm.md (consttable_4): Likewise.
14276 (consttable_8): Likewise.
14277 (consttable_16): Likewise.
14278 * config/mips/mips.md (consttable_float): Likewise.
14279 * config/s390/s390.c (s390_output_pool_entry): Likewise.
14280 * varasm.c (assemble_real): Take a scalar_float_mode.
14281 (output_constant_pool_2): Update accordingly.
14282 (output_constant): Likewise.
14284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14285 Alan Hayward <alan.hayward@arm.com>
14286 David Sherwood <david.sherwood@arm.com>
14288 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
14289 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
14290 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
14291 (native_encode_real): Likewise.
14292 (native_interpret_real): Likewise.
14293 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
14294 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14297 Alan Hayward <alan.hayward@arm.com>
14298 David Sherwood <david.sherwood@arm.com>
14300 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
14301 <scalar_float_mode>. Simplify.
14302 (gen_extend_conv_libfunc): Likewise.
14304 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14305 Alan Hayward <alan.hayward@arm.com>
14306 David Sherwood <david.sherwood@arm.com>
14308 * coretypes.h (scalar_float_mode): New type.
14309 * machmode.h (mode_traits::from_int): Use machine_mode if
14310 USE_ENUM_MODES is defined.
14311 (is_a): New function.
14313 (dyn_cast): Likewise.
14314 (scalar_float_mode): New class.
14315 (scalar_float_mode::includes_p): New function.
14316 (is_float_mode): Likewise.
14317 * gdbhooks.py (MachineModePrinter): New class.
14318 (build_pretty_printer): Use it for scalar_float_mode.
14319 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
14320 (format_helper::format_helper): Turn into a template.
14321 * genmodes.c (get_mode_class): New function.
14322 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
14323 or machine_mode if none.
14324 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
14325 as_a <scalar_float_mode>.
14326 * dwarf2out.c (mem_loc_descriptor): Likewise.
14327 (insert_float): Likewise.
14328 (add_const_value_attribute): Likewise.
14329 * simplify-rtx.c (simplify_immed_subreg): Likewise.
14330 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
14331 (expand_unop): Update accordingly.
14332 (expand_abs_nojump): Likewise.
14333 (expand_copysign_absneg): Take a scalar_float_mode.
14334 (expand_copysign_bit): Likewise.
14335 (expand_copysign): Update accordingly.
14337 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14338 Alan Hayward <alan.hayward@arm.com>
14339 David Sherwood <david.sherwood@arm.com>
14341 * coretypes.h (opt_mode): New class.
14342 * machmode.h (opt_mode): Likewise.
14343 (opt_mode::else_void): New function.
14344 (opt_mode::require): Likewise.
14345 (opt_mode::exists): Likewise.
14346 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
14347 (GET_MODE_2XWIDER_MODE): Likewise.
14348 (mode_iterator::get_wider): Update accordingly.
14349 (mode_iterator::get_2xwider): Likewise.
14350 (mode_iterator::get_known_wider): Likewise, turning into a template.
14351 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
14352 forcing a wider mode to exist.
14353 * config/cr16/cr16.h (LONG_REG_P): Likewise.
14354 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14355 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
14356 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
14357 * lower-subreg.c (init_lower_subreg): Likewise.
14358 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
14359 on the final iteration.
14360 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
14361 a wider mode exists before asking for a move pattern.
14362 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
14363 forcing a wider mode to exist.
14364 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
14365 returning false if no such mode exists.
14366 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
14367 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
14368 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
14369 Avoid checking for a MODE_INT if we already know the mode is not a
14371 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
14372 forcing a wider mode to exist.
14373 (expmed_mult_highpart_optab): Likewise.
14374 (expmed_mult_highpart): Likewise.
14375 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
14377 * lto-streamer-in.c (lto_input_mode_table): Likewise.
14378 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
14379 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
14380 * internal-fn.c (expand_mul_overflow): Update use of
14381 GET_MODE_2XWIDER_MODE.
14382 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14383 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
14384 GET_MODE_WIDER_MODE.
14385 (convert_plusminus_to_widen): Likewise.
14386 * tree-switch-conversion.c (array_value_type): Likewise.
14387 * var-tracking.c (emit_note_insn_var_location): Likewise.
14388 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14389 Return false inside rather than outside the loop if no wider mode
14391 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
14392 and GET_MODE_2XWIDER_MODE
14393 (can_compare_p): Use else_void.
14394 * gdbhooks.py (OptMachineModePrinter): New class.
14395 (build_pretty_printer): Use it for opt_mode.
14397 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14398 Alan Hayward <alan.hayward@arm.com>
14399 David Sherwood <david.sherwood@arm.com>
14401 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
14402 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
14404 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14405 Alan Hayward <alan.hayward@arm.com>
14406 David Sherwood <david.sherwood@arm.com>
14408 * machmode.h (mode_traits): New structure.
14409 (get_narrowest_mode): New function.
14410 (mode_iterator::start): Likewise.
14411 (mode_iterator::iterate_p): Likewise.
14412 (mode_iterator::get_wider): Likewise.
14413 (mode_iterator::get_known_wider): Likewise.
14414 (mode_iterator::get_2xwider): Likewise.
14415 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
14416 (FOR_EACH_MODE): Likewise.
14417 (FOR_EACH_MODE_FROM): Likewise.
14418 (FOR_EACH_MODE_UNTIL): Likewise.
14419 (FOR_EACH_WIDER_MODE): Likewise.
14420 (FOR_EACH_2XWIDER_MODE): Likewise.
14421 * builtins.c (expand_builtin_strlen): Use new mode iterators.
14422 * combine.c (simplify_comparison): Likewise
14423 * config/i386/i386.c (type_natural_mode): Likewise.
14424 * cse.c (cse_insn): Likewise.
14425 * dse.c (find_shift_sequence): Likewise.
14426 * emit-rtl.c (init_derived_machine_modes): Likewise.
14427 (init_emit_once): Likewise.
14428 * explow.c (hard_function_value): Likewise.
14429 * expmed.c (extract_fixed_bit_field_1): Likewise.
14430 (extract_bit_field_1): Likewise.
14431 (expand_divmod): Likewise.
14432 (emit_store_flag_1): Likewise.
14433 * expr.c (init_expr_target): Likewise.
14434 (convert_move): Likewise.
14435 (alignment_for_piecewise_move): Likewise.
14436 (widest_int_mode_for_size): Likewise.
14437 (emit_block_move_via_movmem): Likewise.
14438 (copy_blkmode_to_reg): Likewise.
14439 (set_storage_via_setmem): Likewise.
14440 (compress_float_constant): Likewise.
14441 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14442 * optabs-query.c (get_best_extraction_insn): Likewise.
14443 * optabs.c (expand_binop): Likewise.
14444 (expand_twoval_unop): Likewise.
14445 (expand_twoval_binop): Likewise.
14446 (widen_leading): Likewise.
14447 (widen_bswap): Likewise.
14448 (expand_parity): Likewise.
14449 (expand_unop): Likewise.
14450 (prepare_cmp_insn): Likewise.
14451 (prepare_float_lib_cmp): Likewise.
14452 (expand_float): Likewise.
14453 (expand_fix): Likewise.
14454 (expand_sfix_optab): Likewise.
14455 * postreload.c (move2add_use_add2_insn): Likewise.
14456 * reg-stack.c (reg_to_stack): Likewise.
14457 * reginfo.c (choose_hard_reg_mode): Likewise.
14458 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14459 * stor-layout.c (mode_for_size): Likewise.
14460 (smallest_mode_for_size): Likewise.
14461 (mode_for_vector): Likewise.
14462 (finish_bitfield_representative): Likewise.
14463 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
14464 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
14465 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14466 * var-tracking.c (prepare_call_arguments): Likewise.
14468 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14469 Alan Hayward <alan.hayward@arm.com>
14470 David Sherwood <david.sherwood@arm.com>
14472 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
14473 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
14474 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
14475 * machmode.h (mode_size): Move earlier in file.
14476 (mode_precision): Likewise.
14477 (mode_inner): Likewise.
14478 (mode_nunits): Likewise.
14479 (mode_unit_size): Likewise.
14480 (unit_unit_precision): Likewise.
14481 (mode_wider): Likewise.
14482 (mode_2xwider): Likewise.
14483 (machine_mode): New class.
14484 (mode_to_bytes): New function.
14485 (mode_to_bits): Likewise.
14486 (mode_to_precision): Likewise.
14487 (mode_to_inner): Likewise.
14488 (mode_to_unit_size): Likewise.
14489 (mode_to_unit_precision): Likewise.
14490 (mode_to_nunits): Likewise.
14491 (GET_MODE_SIZE): Use mode_to_bytes.
14492 (GET_MODE_BITSIZE): Use mode_to_bits.
14493 (GET_MODE_PRECISION): Use mode_to_precision.
14494 (GET_MODE_INNER): Use mode_to_inner.
14495 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
14496 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
14497 (GET_MODE_NUNITS): Use mode_to_nunits.
14498 * system.h (ALWAYS_INLINE): New macro.
14499 * config/powerpcspe/powerpcspe-c.c
14500 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
14501 int for arg1_mode and arg2_mode.
14503 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14504 Alan Hayward <alan.hayward@arm.com>
14505 David Sherwood <david.sherwood@arm.com>
14507 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
14508 Prefix mode names with E_ in case statements.
14509 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14510 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
14511 (aarch64_split_simd_move): Likewise.
14512 (aarch64_gen_storewb_pair): Likewise.
14513 (aarch64_gen_loadwb_pair): Likewise.
14514 (aarch64_gen_store_pair): Likewise.
14515 (aarch64_gen_load_pair): Likewise.
14516 (aarch64_get_condition_code_1): Likewise.
14517 (aarch64_constant_pool_reload_icode): Likewise.
14518 (get_rsqrte_type): Likewise.
14519 (get_rsqrts_type): Likewise.
14520 (get_recpe_type): Likewise.
14521 (get_recps_type): Likewise.
14522 (aarch64_gimplify_va_arg_expr): Likewise.
14523 (aarch64_simd_container_mode): Likewise.
14524 (aarch64_emit_load_exclusive): Likewise.
14525 (aarch64_emit_store_exclusive): Likewise.
14526 (aarch64_expand_compare_and_swap): Likewise.
14527 (aarch64_gen_atomic_cas): Likewise.
14528 (aarch64_emit_bic): Likewise.
14529 (aarch64_emit_atomic_swap): Likewise.
14530 (aarch64_emit_atomic_load_op): Likewise.
14531 (aarch64_evpc_trn): Likewise.
14532 (aarch64_evpc_uzp): Likewise.
14533 (aarch64_evpc_zip): Likewise.
14534 (aarch64_evpc_ext): Likewise.
14535 (aarch64_evpc_rev): Likewise.
14536 (aarch64_evpc_dup): Likewise.
14537 (aarch64_gen_ccmp_first): Likewise.
14538 (aarch64_gen_ccmp_next): Likewise.
14539 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
14540 (alpha_emit_xfloating_libcall): Likewise.
14541 (emit_insxl): Likewise.
14542 (alpha_arg_type): Likewise.
14543 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
14544 (arc_preferred_simd_mode): Likewise.
14545 (arc_secondary_reload): Likewise.
14546 (get_arc_condition_code): Likewise.
14547 (arc_print_operand): Likewise.
14548 (arc_legitimate_constant_p): Likewise.
14549 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14550 * config/arc/arc.md (casesi_load): Likewise.
14551 (casesi_compact_jump): Likewise.
14552 * config/arc/predicates.md (proper_comparison_operator): Likewise.
14553 (cc_use_register): Likewise.
14554 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14555 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
14556 (arm_init_iwmmxt_builtins): Likewise.
14557 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
14558 (neon_expand_vector_init): Likewise.
14559 (arm_attr_length_move_neon): Likewise.
14560 (maybe_get_arm_condition_code): Likewise.
14561 (arm_emit_vector_const): Likewise.
14562 (arm_preferred_simd_mode): Likewise.
14563 (arm_output_iwmmxt_tinsr): Likewise.
14564 (thumb1_output_casesi): Likewise.
14565 (thumb2_output_casesi): Likewise.
14566 (arm_emit_load_exclusive): Likewise.
14567 (arm_emit_store_exclusive): Likewise.
14568 (arm_expand_compare_and_swap): Likewise.
14569 (arm_evpc_neon_vuzp): Likewise.
14570 (arm_evpc_neon_vzip): Likewise.
14571 (arm_evpc_neon_vrev): Likewise.
14572 (arm_evpc_neon_vtrn): Likewise.
14573 (arm_evpc_neon_vext): Likewise.
14574 (arm_validize_comparison): Likewise.
14575 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
14576 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
14577 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
14578 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
14579 (c6x_preferred_simd_mode): Likewise.
14580 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
14581 (epiphany_rtx_costs): Likewise.
14582 * config/epiphany/predicates.md (proper_comparison_operator):
14584 * config/frv/frv.c (condexec_memory_operand): Likewise.
14585 (frv_emit_move): Likewise.
14586 (output_move_single): Likewise.
14587 (output_condmove_single): Likewise.
14588 (frv_hard_regno_mode_ok): Likewise.
14589 (frv_matching_accg_mode): Likewise.
14590 * config/h8300/h8300.c (split_adds_subs): Likewise.
14591 (h8300_rtx_costs): Likewise.
14592 (h8300_print_operand): Likewise.
14593 (compute_mov_length): Likewise.
14594 (output_logical_op): Likewise.
14595 (compute_logical_op_length): Likewise.
14596 (compute_logical_op_cc): Likewise.
14597 (h8300_shift_needs_scratch_p): Likewise.
14598 (output_a_shift): Likewise.
14599 (compute_a_shift_length): Likewise.
14600 (compute_a_shift_cc): Likewise.
14601 (expand_a_rotate): Likewise.
14602 (output_a_rotate): Likewise.
14603 * config/i386/i386.c (classify_argument): Likewise.
14604 (function_arg_advance_32): Likewise.
14605 (function_arg_32): Likewise.
14606 (function_arg_64): Likewise.
14607 (function_value_64): Likewise.
14608 (ix86_gimplify_va_arg): Likewise.
14609 (ix86_legitimate_constant_p): Likewise.
14610 (put_condition_code): Likewise.
14611 (split_double_mode): Likewise.
14612 (ix86_avx256_split_vector_move_misalign): Likewise.
14613 (ix86_expand_vector_logical_operator): Likewise.
14614 (ix86_split_idivmod): Likewise.
14615 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
14616 (ix86_build_const_vector): Likewise.
14617 (ix86_build_signbit_mask): Likewise.
14618 (ix86_match_ccmode): Likewise.
14619 (ix86_cc_modes_compatible): Likewise.
14620 (ix86_expand_branch): Likewise.
14621 (ix86_expand_sse_cmp): Likewise.
14622 (ix86_expand_sse_movcc): Likewise.
14623 (ix86_expand_int_sse_cmp): Likewise.
14624 (ix86_expand_vec_perm_vpermi2): Likewise.
14625 (ix86_expand_vec_perm): Likewise.
14626 (ix86_expand_sse_unpack): Likewise.
14627 (ix86_expand_int_addcc): Likewise.
14628 (ix86_split_to_parts): Likewise.
14629 (ix86_vectorize_builtin_gather): Likewise.
14630 (ix86_vectorize_builtin_scatter): Likewise.
14631 (avx_vpermilp_parallel): Likewise.
14632 (inline_memory_move_cost): Likewise.
14633 (ix86_tieable_integer_mode_p): Likewise.
14634 (x86_maybe_negate_const_int): Likewise.
14635 (ix86_expand_vector_init_duplicate): Likewise.
14636 (ix86_expand_vector_init_one_nonzero): Likewise.
14637 (ix86_expand_vector_init_one_var): Likewise.
14638 (ix86_expand_vector_init_concat): Likewise.
14639 (ix86_expand_vector_init_interleave): Likewise.
14640 (ix86_expand_vector_init_general): Likewise.
14641 (ix86_expand_vector_set): Likewise.
14642 (ix86_expand_vector_extract): Likewise.
14643 (emit_reduc_half): Likewise.
14644 (ix86_emit_i387_round): Likewise.
14645 (ix86_mangle_type): Likewise.
14646 (ix86_expand_round_sse4): Likewise.
14647 (expand_vec_perm_blend): Likewise.
14648 (canonicalize_vector_int_perm): Likewise.
14649 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
14650 (expand_vec_perm_1): Likewise.
14651 (expand_vec_perm_interleave3): Likewise.
14652 (expand_vec_perm_even_odd_pack): Likewise.
14653 (expand_vec_perm_even_odd_1): Likewise.
14654 (expand_vec_perm_broadcast_1): Likewise.
14655 (ix86_vectorize_vec_perm_const_ok): Likewise.
14656 (ix86_expand_vecop_qihi): Likewise.
14657 (ix86_expand_mul_widen_hilo): Likewise.
14658 (ix86_expand_sse2_abs): Likewise.
14659 (ix86_expand_pextr): Likewise.
14660 (ix86_expand_pinsr): Likewise.
14661 (ix86_preferred_simd_mode): Likewise.
14662 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
14663 * config/i386/sse.md (*andnot<mode>3): Likewise.
14664 (<mask_codefor><code><mode>3<mask_name>): Likewise.
14665 (*<code><mode>3): Likewise.
14666 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
14667 (ia64_expand_atomic_op): Likewise.
14668 (ia64_arg_type): Likewise.
14669 (ia64_mode_to_int): Likewise.
14670 (ia64_scalar_mode_supported_p): Likewise.
14671 (ia64_vector_mode_supported_p): Likewise.
14672 (expand_vec_perm_broadcast): Likewise.
14673 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
14674 (iq2000_function_arg_advance): Likewise.
14675 (iq2000_function_arg): Likewise.
14676 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
14677 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
14678 (m68k_libcall_value): Likewise.
14679 (m68k_function_value): Likewise.
14680 (sched_attr_op_type): Likewise.
14681 * config/mcore/mcore.c (mcore_output_move): Likewise.
14682 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
14684 (microblaze_function_arg): Likewise.
14685 * config/mips/mips.c (mips16_build_call_stub): Likewise.
14686 (mips_print_operand): Likewise.
14687 (mips_mode_ok_for_mov_fmt_p): Likewise.
14688 (mips_vector_mode_supported_p): Likewise.
14689 (mips_preferred_simd_mode): Likewise.
14690 (mips_expand_vpc_loongson_even_odd): Likewise.
14691 (mips_expand_vec_unpack): Likewise.
14692 (mips_expand_vi_broadcast): Likewise.
14693 (mips_expand_vector_init): Likewise.
14694 (mips_expand_vec_reduc): Likewise.
14695 (mips_expand_msa_cmp): Likewise.
14696 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
14697 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
14698 (cc_flags_for_mode): Likewise.
14699 * config/msp430/msp430.c (msp430_print_operand): Likewise.
14700 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
14701 (nds32_output_casesi_pc_relative): Likewise.
14702 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14703 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
14704 (nvptx_gen_unpack): Likewise.
14705 (nvptx_gen_pack): Likewise.
14706 (nvptx_gen_shuffle): Likewise.
14707 (nvptx_gen_wcast): Likewise.
14708 (nvptx_preferred_simd_mode): Likewise.
14709 * config/pa/pa.c (pa_secondary_reload): Likewise.
14710 * config/pa/predicates.md (base14_operand): Likewise.
14711 * config/powerpcspe/powerpcspe-c.c
14712 (altivec_resolve_overloaded_builtin): Likewise.
14713 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
14715 (rs6000_preferred_simd_mode): Likewise.
14716 (output_vec_const_move): Likewise.
14717 (rs6000_expand_vector_extract): Likewise.
14718 (rs6000_split_vec_extract_var): Likewise.
14719 (reg_offset_addressing_ok_p): Likewise.
14720 (rs6000_legitimate_offset_address_p): Likewise.
14721 (rs6000_legitimize_address): Likewise.
14722 (rs6000_emit_set_const): Likewise.
14723 (rs6000_const_vec): Likewise.
14724 (rs6000_emit_move): Likewise.
14725 (spe_build_register_parallel): Likewise.
14726 (rs6000_darwin64_record_arg_recurse): Likewise.
14727 (swap_selector_for_mode): Likewise.
14728 (spe_init_builtins): Likewise.
14729 (paired_init_builtins): Likewise.
14730 (altivec_init_builtins): Likewise.
14731 (do_load_for_compare): Likewise.
14732 (rs6000_generate_compare): Likewise.
14733 (rs6000_expand_float128_convert): Likewise.
14734 (emit_load_locked): Likewise.
14735 (emit_store_conditional): Likewise.
14736 (rs6000_output_function_epilogue): Likewise.
14737 (rs6000_handle_altivec_attribute): Likewise.
14738 (rs6000_function_value): Likewise.
14739 (emit_fusion_gpr_load): Likewise.
14740 (emit_fusion_p9_load): Likewise.
14741 (emit_fusion_p9_store): Likewise.
14742 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
14743 (fusion_gpr_mem_load): Likewise.
14744 (fusion_addis_mem_combo_load): Likewise.
14745 (fusion_addis_mem_combo_store): Likewise.
14746 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
14747 (fusion_gpr_mem_load): Likewise.
14748 (fusion_addis_mem_combo_load): Likewise.
14749 (fusion_addis_mem_combo_store): Likewise.
14750 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14752 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
14753 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
14754 (rs6000_preferred_simd_mode): Likewise.
14755 (output_vec_const_move): Likewise.
14756 (rs6000_expand_vector_extract): Likewise.
14757 (rs6000_split_vec_extract_var): Likewise.
14758 (reg_offset_addressing_ok_p): Likewise.
14759 (rs6000_legitimate_offset_address_p): Likewise.
14760 (rs6000_legitimize_address): Likewise.
14761 (rs6000_emit_set_const): Likewise.
14762 (rs6000_const_vec): Likewise.
14763 (rs6000_emit_move): Likewise.
14764 (rs6000_darwin64_record_arg_recurse): Likewise.
14765 (swap_selector_for_mode): Likewise.
14766 (paired_init_builtins): Likewise.
14767 (altivec_init_builtins): Likewise.
14768 (rs6000_expand_float128_convert): Likewise.
14769 (emit_load_locked): Likewise.
14770 (emit_store_conditional): Likewise.
14771 (rs6000_output_function_epilogue): Likewise.
14772 (rs6000_handle_altivec_attribute): Likewise.
14773 (rs6000_function_value): Likewise.
14774 (emit_fusion_gpr_load): Likewise.
14775 (emit_fusion_p9_load): Likewise.
14776 (emit_fusion_p9_store): Likewise.
14777 * config/rx/rx.c (rx_gen_move_template): Likewise.
14778 (flags_from_mode): Likewise.
14779 * config/s390/predicates.md (s390_alc_comparison): Likewise.
14780 (s390_slb_comparison): Likewise.
14781 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
14782 (s390_vector_mode_supported_p): Likewise.
14783 (s390_cc_modes_compatible): Likewise.
14784 (s390_match_ccmode_set): Likewise.
14785 (s390_canonicalize_comparison): Likewise.
14786 (s390_emit_compare_and_swap): Likewise.
14787 (s390_branch_condition_mask): Likewise.
14788 (s390_rtx_costs): Likewise.
14789 (s390_secondary_reload): Likewise.
14790 (__SECONDARY_RELOAD_CASE): Likewise.
14791 (s390_expand_cs): Likewise.
14792 (s390_preferred_simd_mode): Likewise.
14793 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
14794 * config/sh/sh.c (sh_print_operand): Likewise.
14795 (dump_table): Likewise.
14796 (sh_secondary_reload): Likewise.
14797 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14798 * config/sh/sh.md (casesi_worker_1): Likewise.
14799 (casesi_worker_2): Likewise.
14800 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
14801 (fcc_comparison_operator): Likewise.
14802 * config/sparc/sparc.c (sparc_expand_move): Likewise.
14803 (emit_soft_tfmode_cvt): Likewise.
14804 (sparc_preferred_simd_mode): Likewise.
14805 (output_cbranch): Likewise.
14806 (sparc_print_operand): Likewise.
14807 (sparc_expand_vec_perm_bmask): Likewise.
14808 (vector_init_bshuffle): Likewise.
14809 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14810 (spu_vector_mode_supported_p): Likewise.
14811 (spu_expand_insv): Likewise.
14812 (spu_emit_branch_or_set): Likewise.
14813 (spu_handle_vector_attribute): Likewise.
14814 (spu_builtin_splats): Likewise.
14815 (spu_builtin_extract): Likewise.
14816 (spu_builtin_promote): Likewise.
14817 (spu_expand_sign_extend): Likewise.
14818 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14819 (tilegx_simd_int): Likewise.
14820 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
14821 (tilepro_simd_int): Likewise.
14822 * config/v850/v850.c (const_double_split): Likewise.
14823 (v850_print_operand): Likewise.
14824 (ep_memory_offset): Likewise.
14825 * config/vax/vax.c (vax_rtx_costs): Likewise.
14826 (vax_output_int_move): Likewise.
14827 (vax_output_int_add): Likewise.
14828 (vax_output_int_subtract): Likewise.
14829 * config/visium/predicates.md (visium_branch_operator): Likewise.
14830 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
14831 (visium_print_operand_address): Likewise.
14832 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14833 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
14834 (xtensa_expand_conditional_branch): Likewise.
14835 (xtensa_copy_incoming_a7): Likewise.
14836 (xtensa_output_literal): Likewise.
14837 * dfp.c (decimal_real_maxval): Likewise.
14838 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
14840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14841 Alan Hayward <alan.hayward@arm.com>
14842 David Sherwood <david.sherwood@arm.com>
14844 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
14845 (mode_nunits_inline): Likewise.
14846 (mode_inner_inline): Likewise.
14847 (mode_unit_size_inline): Likewise.
14848 (mode_unit_precision_inline): Likewise.
14849 (emit_insn_modes_h): Likewise. Also emit a #define of the
14851 (emit_mode_wider): Add an E_ prefix to mode names.
14852 (emit_mode_complex): Likewise.
14853 (emit_mode_inner): Likewise.
14854 (emit_mode_adjustments): Likewise.
14855 (emit_mode_int_n): Likewise.
14856 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
14857 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
14858 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
14859 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
14860 (CRC32_BUILTIN, ENTRY): Likewise.
14861 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
14862 (aarch64_pop_regs): Likewise.
14863 (aarch64_process_components): Likewise.
14864 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
14865 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
14866 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
14867 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
14868 * config/arm/arm.c (arm_init_libfuncs): Likewise.
14869 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
14871 * config/i386/i386-builtin.def (pcmpestr): Likewise.
14872 (pcmpistr): Likewise.
14873 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
14874 * config/mmix/mmix.c (mmix_output_condition): Likewise.
14875 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
14877 * config/rl78/rl78.c (mduc_regs): Likewise.
14878 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14879 (htm_expand_builtin): Likewise.
14880 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
14881 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
14882 * config/xtensa/xtensa.c (print_operand): Likewise.
14883 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
14884 (NUM_MODE_VECTOR_INT): Likewise.
14885 * genoutput.c (null_operand): Likewise.
14886 (output_operand_data): Likewise.
14887 * genrecog.c (print_parameter_value): Likewise.
14888 * lra.c (debug_operand_data): Likewise.
14890 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14892 * dwarf2out.c (scompare_loc_descriptor_wide)
14893 (scompare_loc_descriptor_narrow): New functions, split out from...
14894 (scompare_loc_descriptor): ...here.
14895 * expmed.c (emit_store_flag_int): New function, split out from...
14896 (emit_store_flag): ...here.
14898 2017-08-30 Richard Biener <rguenther@suse.de>
14900 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
14901 (dwarf2out_early_finish): Move setting of AT_pubnames from
14902 early debug output to early finish.
14904 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14906 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
14907 and -mdata-region to the assembler.
14909 2017-08-30 Richard Biener <rguenther@suse.de>
14911 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
14913 (gen_subprogram_die): Add DW_AT_object_pointer only early.
14914 (dwarf2out_early_global_decl): Only generate a DIE for the
14915 abstract origin if it doesn't already exist or is a declaration DIE.
14916 (resolve_addr): Do not add the linkage name twice when
14917 generating a stub DIE for the DW_TAG_GNU_call_site target.
14919 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14921 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14922 Use machine_mode rather than int for arg1_mode.
14924 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
14927 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
14928 that the second argument of the built-in functions to unpack
14929 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
14930 switch statement instead a lot of if statements.
14931 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
14932 Allow 64-bit values to be in Altivec registers as well as
14933 traditional floating point registers.
14934 (pack<mode>, FMOVE128_VSX iterator): Likewise.
14936 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
14938 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
14939 MAX_REGS_PER_ADDRESS == 1.
14941 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
14943 * config/i386/i386.opt (flag_fentry): Do not init to -1.
14944 * config/i386/i386.c (ix86_option_override_internal): Simplify
14945 setting of opts->x_flag_entry.
14947 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14948 Jakub Jelinek <jakub@redhat.com>
14949 Richard Biener <rguenther@suse.de>
14951 PR tree-optimization/81503
14952 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
14953 folded constant fits in the target type; reorder tests for clarity.
14955 2017-08-29 Martin Liska <mliska@suse.cz>
14957 * passes.def: Include pass_lower_switch.
14958 * stmt.c (dump_case_nodes): Remove and move to
14959 tree-switch-conversion.
14960 (case_values_threshold): Likewise.
14961 (expand_switch_as_decision_tree_p): Likewise.
14962 (emit_case_decision_tree): Likewise.
14963 (expand_case): Likewise.
14964 (balance_case_nodes): Likewise.
14965 (node_has_low_bound): Likewise.
14966 (node_has_high_bound): Likewise.
14967 (node_is_bounded): Likewise.
14968 (emit_case_nodes): Likewise.
14969 (struct simple_case_node): New struct.
14970 (add_case_node): Remove.
14971 (emit_case_dispatch_table): Use vector instead of case_list.
14972 (reset_out_edges_aux): Remove.
14973 (compute_cases_per_edge): Likewise.
14974 (expand_case): Build list of simple_case_node.
14975 (expand_sjlj_dispatch_table): Use it.
14976 * tree-switch-conversion.c (struct case_node): Moved from
14977 stmt.c and adjusted.
14978 (emit_case_nodes): Likewise.
14979 (node_has_low_bound): Likewise.
14980 (node_has_high_bound): Likewise.
14981 (node_is_bounded): Likewise.
14982 (case_values_threshold): Likewise.
14983 (reset_out_edges_aux): Likewise.
14984 (compute_cases_per_edge): Likewise.
14985 (add_case_node): Likewise.
14986 (dump_case_nodes): Likewise.
14987 (balance_case_nodes): Likewise.
14988 (expand_switch_as_decision_tree_p): Likewise.
14989 (emit_jump): Likewise.
14990 (emit_case_decision_tree): Likewise.
14991 (try_switch_expansion): Likewise.
14992 (do_jump_if_equal): Likewise.
14993 (emit_cmp_and_jump_insns): Likewise.
14994 (fix_phi_operands_for_edge): New function.
14995 (record_phi_operand_mapping): Likewise.
14996 (class pass_lower_switch): New pass.
14997 (pass_lower_switch::execute): New function.
14998 (make_pass_lower_switch): Likewise.
14999 (conditional_probability):
15000 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
15001 * tree-pass.h: Add make_pass_lower_switch.
15003 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15006 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
15009 2017-08-29 Richard Biener <rguenther@suse.de>
15011 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
15012 we do not add a DW_AT_inline attribute twice.
15013 (gen_subprogram_die): Remove code setting DW_AT_inline on
15014 DECL_ABSTRACT_P nodes.
15016 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
15018 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
15019 calls to internal functions.
15020 (gimplify_modify_expr): Likewise.
15021 * tree-call-cdce.c (use_internal_fn): Likewise.
15022 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
15023 (convert_to_divmod): Set the nothrow flag.
15024 * tree-if-conv.c (predicate_mem_writes): Likewise.
15025 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15026 (vectorizable_call): Likewise.
15027 (vectorizable_store): Likewise.
15028 (vectorizable_load): Likewise.
15029 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
15030 (vect_recog_mask_conversion_pattern): Likewise.
15032 2017-08-29 Martin Liska <mliska@suse.cz>
15035 * gcc.c (driver_handle_option): Add new argument.
15036 * opts-common.c (handle_option): Pass
15037 target_option_override_hook.
15038 * opts-global.c (lang_handle_option): Add new option.
15039 (set_default_handlers): Add new argument.
15040 (decode_options): Likewise.
15041 * opts.c (target_handle_option): Likewise.
15042 (common_handle_option): Call target_option_override_hook.
15043 * opts.h (struct cl_option_handler_func): Add hook for
15044 target option override.
15045 (struct cl_option_handlers): Likewise.
15046 (set_default_handlers): Add new argument.
15047 (decode_options): Likewise.
15048 (common_handle_option): Likewise.
15049 (target_handle_option): Likewise.
15050 * toplev.c (toplev::main): Pass targetm.target_option.override
15053 2017-08-29 Richard Biener <rguenther@suse.de>
15054 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
15056 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
15057 life to the active subtree.
15059 2017-08-28 Jeff Law <law@redhat.com>
15061 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
15062 derive_equivalences.
15063 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
15064 Code moved into....
15065 (edge_info::derive_equivalences): New private member function
15067 * tree-ssa-dom.c (class edge_info): Changed from a struct
15068 to a class. Add ctor/dtor, methods and data members.
15069 (edge_info::edge_info): Renamed from allocate_edge_info.
15070 Initialize additional members.
15071 (edge_info::~edge_info): New.
15072 (free_dom_edge_info): Delete the edge info.
15073 (record_edge_info): Use new class & associated member functions.
15074 Tighten forms for testing for edge equivalences.
15075 (record_temporary_equivalences): Iterate over the simple
15076 equivalences rather than assuming there's only one per edge.
15077 (cprop_into_successor_phis): Iterate over the simple
15078 equivalences rather than assuming there's only one per edge.
15079 (optimize_stmt): Use operand_equal_p rather than pointer
15080 equality for mini-DSE code.
15082 2017-08-28 Nathan Sidwell <nathan@acm.org>
15084 * gcc.c (execute): Fold SIGPIPE handling into switch
15085 statement. Adjust internal error message.
15087 2017-08-28 Richard Biener <rguenther@suse.de>
15090 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
15091 Do nothing for removed DIEs.
15093 2017-08-28 Richard Biener <rguenther@suse.de>
15095 PR tree-optimization/81977
15096 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
15099 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
15102 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
15103 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
15104 using targetm.gen_mem_thread_fence.
15106 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
15109 * config/i386/i386.md (*<btsc><mode>): Change operand 2
15110 predicate to register_operand. Reorder operands.
15111 (*btr<mode>): Ditto.
15112 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
15113 (*btr<mode>_mask): Ditto.
15115 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
15117 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
15118 * config/rs6000/xmmintrin.h: New file.
15119 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
15121 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15124 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
15125 parameter and_insn and return it.
15126 (recombine_lvx_pattern): Insert a copy to ensure availability of
15127 the base register of the copied masking operation at the point of
15128 the instruction replacement.
15129 (recombine_stvx_pattern): Likewise.
15131 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
15133 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
15134 undocumented switches.
15135 (-mpower9-dform-vector): Likewise.
15136 (-mpower9-dform): Likewise.
15137 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
15138 comments to delete references to -mpower9-dform* switches.
15139 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
15140 Delete reference to -mpower9-dform* switches, test for
15141 -mpower9-vector instead.
15142 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
15143 (OTHER_P9_VECTOR_MASKS): Likewise.
15144 (POWERPC_MASKS): Likewise.
15145 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
15146 tests against -mpower9-dform* to -mpower9-vector. Delete code
15147 that checked for -mpower9-dform* consistancy with other options.
15148 Add test for -mpower9-misc to enable other power9 switches.
15149 (rs6000_init_hard_regno_mode_ok): Likewise.
15150 (rs6000_option_override_internal): Likewise.
15151 (rs6000_emit_prologue): Likewise.
15152 (rs6000_emit_epilogue): Likewise.
15153 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
15154 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
15155 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
15157 (emit_fusion_p9_store): Likewise.
15158 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
15159 resetting these macros if the assembler does not support ISA 3.0
15161 (TARGET_P9_DFORM_VECTOR): Likewise.
15162 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
15163 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
15165 2017-08-25 Alan Modra <amodra@gmail.com>
15167 PR rtl-optimization/81747
15168 * cse.c (cse_extended_basic_block): Don't attempt to record
15169 equivalences for degenerate conditional branches that branch
15170 to their fall-through.
15172 2017-08-24 Martin Sebor <msebor@redhat.com>
15174 PR middle-end/81908
15175 * gimple-fold.c (size_must_be_zero_p): New function.
15176 (gimple_fold_builtin_memory_op): Call it.
15178 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
15180 * config/rs6000/mm_malloc.h: New file.
15182 2017-08-24 Bin Cheng <bin.cheng@arm.com>
15184 PR tree-optimization/81913
15185 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
15186 analysis when either IVs in condition can wrap.
15188 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
15190 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
15191 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
15193 2017-08-24 Richard Biener <rguenther@suse.de>
15196 * targhooks.c (default_target_can_inline_p): Properly
15197 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
15198 is present and always compare.
15199 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
15200 infer -mfpmath=sse from TARGET_SSE_P.
15201 (ix86_can_inline_p): Properly use target_option_default_node when
15202 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
15204 2017-08-24 Richard Biener <rguenther@suse.de>
15207 * dwarf2out.c (output_die): Handle flag_generate_offload like
15209 (output_comp_unit): Likewise.
15210 (gen_array_type_die): Likewise.
15211 (dwarf2out_early_finish): Likewise.
15212 (note_variable_value_in_expr): Likewise.
15213 (dwarf2out_finish): Likewise. Adjust assert.
15214 * cgraphunit.c (symbol_table::compile): Move setting of
15215 flag_generate_offload earlier ...
15216 (symbol_table::finalize_compilation_unit): ... here, before
15217 early debug finalization.
15219 2017-08-24 Richard Biener <rguenther@suse.de>
15221 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
15222 and ipa-fnsummary.h.
15223 (ix86_can_inline_p): When ix86_fpmath flags do not match
15224 check whether the callee uses FP math at all.
15226 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
15228 PR middle-end/81931
15229 * tree-ssanames.c (get_nonzero_bits): Use element_precision
15230 instead of TYPE_PRECISION.
15232 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
15233 Alan Hayward <alan.hayward@arm.com>
15234 David Sherwood <david.sherwood@arm.com>
15236 * combine.c (make_extraction): Use subreg_offset_from_lsb.
15238 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
15240 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
15242 (ix86_frame::stack_realign_allocate): New field.
15243 (struct machine_frame_state): Modify comments.
15244 (machine_frame_state::sp_realigned_fp_end): New field.
15245 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
15246 layout calculation.
15247 (sp_valid_at): Add assertion to assure no attempt to access invalid
15248 offset of a realigned stack.
15249 (fp_valid_at): Likewise.
15250 (choose_baseaddr): Modify comments.
15251 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
15252 ix86_expand_prologue.
15253 (ix86_expand_prologue): Modify stack realignment and allocation.
15254 (ix86_expand_epilogue): Modify comments.
15255 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
15256 avx2_runtime, avx512f, and avx512f_runtime.
15258 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
15260 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
15261 (mstackrealign): Do not init to -1.
15262 * config/i386/i386.c (ix86_option_override_internal):
15263 Check opts_set, not opts when setting default value of
15264 opts->x_ix86_force_align_arg_pointer.
15266 2017-08-23 Richard Biener <rguenther@suse.de>
15268 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
15269 lang_hooks.decl_printable_name.
15270 * print-rtl-function.c (print_rtx_function): Likewise.
15271 * tree-pretty-print.c (dump_function_header): Likewise.
15273 2017-08-23 Richard Biener <rguenther@suse.de>
15276 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
15277 -g0 at compile-time.
15279 2017-08-23 Tamar Christina <tamar.christina@arm.com>
15281 PR middle-end/19706
15282 * doc/sourcebuild.texi (Other hardware attributes):
15285 2017-08-23 Tamar Christina <tamar.christina@arm.com>
15287 PR middle-end/19706
15288 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
15289 Fix single-use check.
15291 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
15293 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
15295 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
15297 * doc/install.texi: Modify to add more details on running selected
15300 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
15302 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
15303 is combined with -mabi=ms.
15304 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
15307 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15309 PR tree-optimization/81488
15310 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
15311 and cached_basis fields.
15312 (MAX_SPREAD): New constant.
15313 (alloc_cand_and_find_basis): Initialize new fields.
15314 (clear_visited): New function.
15315 (create_phi_basis_1): Rename from create_phi_basis, set visited
15316 and cached_basis fields.
15317 (create_phi_basis): New wrapper function.
15318 (phi_add_costs_1): Rename from phi_add_costs, add spread
15319 parameter, set visited field, short-circuit when limits reached.
15320 (phi_add_costs): New wrapper function.
15321 (record_phi_increments_1): Rename from record_phi_increments, set
15323 (record_phi_increments): New wrapper function.
15324 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
15325 (phi_incr_cost): New wrapper function.
15326 (all_phi_incrs_profitable_1): Rename from
15327 all_phi_incrs_profitable, set visited field.
15328 (all_phi_incrs_profitable): New wrapper function.
15330 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
15331 Alan Hayward <alan.hayward@arm.com>
15332 David Sherwood <david.sherwood@arm.com>
15334 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
15335 that takes the outer and inner modes.
15336 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
15337 comparison as the canonical test for a paradoxical subreg.
15338 * combine.c (simplify_set): Use paradoxical_subreg_p.
15339 (make_extraction): Likewise.
15340 (force_to_mode): Likewise.
15341 (rtx_equal_for_field_assignment_p): Likewise.
15342 (gen_lowpart_for_combine): Likewise.
15343 (simplify_comparison): Likewise.
15344 * cse.c (equiv_constant): Likewise.
15345 * expmed.c (store_bit_field_1): Likewise.
15346 * final.c (alter_subreg): Likewise.
15347 * fwprop.c (propagate_rtx): Likewise.
15348 (forward_propagate_subreg): Likewise.
15349 * ira-conflicts.c (ira_build_conflicts): Likewise.
15350 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
15351 * lra-constraints.c (curr_insn_transform): Likewise.
15352 (split_reg): Likewise.
15353 * lra-eliminations.c (move_plus_up): Likewise.
15354 (lra_eliminate_regs_1): Likewise.
15355 * recog.c (general_operand): Likewise.
15356 * ree.c (combine_reaching_defs): Likewise.
15357 * reload.c (push_reload): Likewise.
15358 (find_reloads): Likewise.
15359 * reload1.c (elimination_effects): Likewise.
15360 (compute_reload_subreg_offset): Likewise.
15361 (choose_reload_regs): Likewise.
15362 * rtlanal.c (subreg_lsb_1): Likewise.
15363 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15364 (simplify_subreg): Likewise.
15365 * var-tracking.c (track_loc_p): Likewise.
15366 * emit-rtl.c (byte_lowpart_offset): Likewise.
15367 (paradoxical_subreg_p): Delete out-of-line definition.
15369 2017-08-22 Jeff Law <law@redhat.com>
15371 PR tree-optimization/81741
15372 PR tree-optimization/71947
15373 * tree-ssa-dom.c: Include tree-inline.h.
15374 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
15375 equivalences if one is more expensive to compute than the other.
15376 * tree-ssa-scopedtables.h (class const_or_copies): Make
15377 record_const_or_copy_raw method private.
15378 (class avail_exprs_stack): New method simplify_binary_operation.
15379 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
15380 avail_exprs_stack::simplify_binary_operation as needed.
15381 (avail_exprs_stack::simplify_binary_operation): New function.
15383 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
15385 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
15386 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
15387 (DOT_SYMBOLS): Likewise.
15388 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
15389 (RELOCATABLE_NEEDS_FIXUP): Likewise.
15390 (RS6000_ABI_NAME): Likewise.
15391 (TARGET_CMODEL): Likewise.
15392 (TOC_SECTION_ASM_OP): Likewise.
15393 (SET_CMODEL): New macro.
15394 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
15396 2017-08-22 Richard Biener <rguenther@suse.de>
15398 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
15399 to allow for free-lang-data replacements similar to verify_type_variant.
15401 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
15402 Alan Hayward <alan.hayward@arm.com>
15403 David Sherwood <david.sherwood@arm.com>
15405 * config/aarch64/aarch64.md (casesi): Use DImode rather than
15406 VOIDmode for the LABEL_REF.
15408 2017-08-22 Richard Biener <rguenther@suse.de>
15410 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
15412 2017-08-22 Richard Biener <rguenther@suse.de>
15414 * common.opt (feliminate-dwarf2-dups): Ignore.
15415 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
15416 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
15417 same_die_p_wrap, compute_section_prefix,
15418 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
15419 (comdat_symbol_id, comdat_symbol_number): Likewise.
15420 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
15422 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
15423 (output_die): Mark unreachable path unreachable.
15424 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
15425 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
15426 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
15427 (dwarf2out_early_finish): Likewise.
15429 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
15431 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
15433 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
15436 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
15437 not VAR_P. Filter attribute warnings with OPT_Wattributes.
15438 (avr_attribute_table) <io, io_low, address>: Initialize
15439 .decl_required with true.
15441 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
15443 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
15444 undocumented debugging options.
15445 (-mvsx-scalar-double): Likewise.
15446 (-mallow-df-permute): Likewise.
15447 (-mvectorize-builtins): Likewise.
15448 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15449 (rs6000_builtin_vectorized_function): Likewise.
15450 (rs6000_builtin_md_vectorized_function): Likewise.
15451 (rs6000_opt_vars): Likewise.
15453 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
15456 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
15457 (*btrq_imm): Rename from *btrq.
15458 (*btcq_imm): Rename from *btcq.
15459 (btsc): New code attribute.
15460 (*<btsc><mode>): New insn pattern.
15461 (*btr<mode>): Ditto.
15462 (*<btsc><mode>_mask): New insn_and_split pattern.
15463 (*btr<mode>_mask): Ditto.
15465 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15466 Alan Hayward <alan.hayward@arm.com>
15467 David Sherwood <david.sherwood@arm.com>
15469 * function.c (pad_below): Simplify padding calculation.
15471 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15472 Alan Hayward <alan.hayward@arm.com>
15473 David Sherwood <david.sherwood@arm.com>
15475 * target.def (function_prologue): Remove frame size argument.
15476 (function_epilogue): Likewise.
15477 * doc/tm.texi: Regenerate.
15478 * final.c (final_start_function): Update call to function_prologue.
15479 (final_end_function): Update call to function_epilogue.
15480 (default_function_pro_epilogue): Remove frame size argument.
15481 * output.h (default_function_pro_epilogue): Likewise.
15482 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
15483 (arm_output_function_prologue): Likewise.
15484 * config/frv/frv.c (frv_function_prologue): Likewise.
15485 (frv_function_epilogue): Likewise.
15486 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
15487 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
15488 (ia64_output_function_epilogue): Likewise.
15489 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
15490 (m32r_output_function_epilogue): Likewise.
15491 * config/microblaze/microblaze.c (microblaze_function_prologue)
15492 (microblaze_function_epilogue): Likewise.
15493 * config/mips/mips.c (mips_output_function_prologue): Likewise.
15494 (mips_output_function_epilogue): Likewise.
15495 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
15496 (mmix_target_asm_function_epilogue): Likewise.
15497 * config/msp430/msp430.c (msp430_start_function): Likewise.
15498 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
15499 (nds32_asm_function_epilogue): Likewise.
15500 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
15501 * config/pa/pa.c (pa_output_function_prologue): Likewise.
15502 (pa_output_function_epilogue): Likewise.
15503 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
15504 (rs6000_output_function_epilogue): Likewise.
15505 * config/rl78/rl78.c (rl78_start_function): Likewise.
15506 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
15507 (rs6000_output_function_epilogue): Likewise.
15508 * config/rx/rx.c (rx_output_function_prologue): Likewise.
15509 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
15510 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
15511 (sparc_asm_function_epilogue): Likewise.
15513 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15515 * tree.h (type_has_mode_precision_p): New function.
15516 * convert.c (convert_to_integer_1): Use it.
15517 * expr.c (expand_expr_real_2): Likewise.
15518 (expand_expr_real_1): Likewise.
15519 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
15520 * match.pd: Likewise.
15521 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
15522 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
15523 * tree-tailcall.c (process_assignment): Likewise.
15524 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15525 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
15526 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
15527 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15528 (vectorizable_assignment): Likewise.
15529 (vectorizable_shift): Likewise.
15530 (vectorizable_operation): Likewise.
15531 * tree-vrp.c (register_edge_assert_for_2): Likewise.
15533 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
15535 * match.pd: Add pow (C, x) simplification.
15537 2017-08-21 Richard Biener <rguenther@suse.de>
15539 PR tree-optimization/81900
15540 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
15541 for blocks with abnormal predecessors.
15542 (compute_antic): Do not set visited flag prematurely.
15544 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
15547 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
15549 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15551 * stor-layout.h (vector_type_mode): Move to...
15552 * tree.h (vector_type_mode): ...here.
15553 * stor-layout.c (vector_type_mode): Move to...
15554 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
15556 2017-08-21 Richard Biener <rguenther@suse.de>
15558 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
15559 register_external_die hooks.
15560 (debug_false_tree_charstarstar_uhwistar): Declare.
15561 (debug_nothing_tree_charstar_uhwi): Likewise.
15562 * debug.c (do_nothing_debug_hooks): Adjust.
15563 (debug_false_tree_charstarstar_uhwistar): New do nothing.
15564 (debug_nothing_tree_charstar_uhwi): Likewise.
15565 * dbxout.c (dbx_debug_hooks): Adjust.
15566 (xcoff_debug_hooks): Likewise.
15567 * sdbout.c (sdb_debug_hooks): Likewise.
15568 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15569 * dwarf2out.c (macinfo_label_base): New global.
15570 (dwarf2out_register_external_die): New function for the
15571 register_external_die hook.
15572 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
15573 (dwarf2_debug_hooks): Use them.
15574 (dwarf2_lineno_debug_hooks): Adjust.
15575 (struct die_struct): Add with_offset flag.
15576 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
15577 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
15578 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
15579 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
15580 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
15581 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
15582 defining section names for the early LTO debug variants.
15583 (reset_indirect_string): New helper.
15584 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
15585 (print_dw_val): Add support for offsetted symbol references.
15586 (get_ultimate_context): Split out from is_cxx.
15587 (is_cxx): Use get_ultimate_context.
15588 (is_fortran): Add decl overload.
15589 (compute_comp_unit_symbol): Split out worker from
15590 compute_section_prefix.
15591 (compute_section_prefix): Call compute_comp_unit_symbol and
15592 set comdat_type_p here.
15593 (output_die): Skip DIE symbol output for the LTO added one.
15594 Handle DIE symbol references with offset.
15595 (output_comp_unit): Guard section name mangling properly.
15596 For LTO debug sections emit a symbol at the section beginning
15597 which we use to refer to its DIEs.
15598 (add_abstract_origin_attribute): For DIEs registered via
15599 dwarf2out_register_external_die directly refer to the early
15600 DIE rather than indirectly through the shadow one we created.
15601 Remove obsolete call to dwarf2out_abstract_function for
15602 non-function/block origins.
15603 (gen_array_type_die): When generating early LTO debug do
15604 not emit DW_AT_string_length.
15605 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
15606 late when in LTO. As suggested place a gcc_unreachable for
15607 the DECL_ABSTRACT_P case.
15608 (gen_subprogram_die): Avoid another specification DIE
15609 for early built declarations/definitions for the late LTO case.
15610 (gen_variable_die): Add type references for late duplicated VLA dies
15612 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
15613 we have the abstract instance already.
15614 (process_scope_var): Adjust decl DIE contexts in LTO which
15615 first puts them in limbo.
15616 (gen_decl_die): Do not generate type DIEs late apart from
15617 types for VLAs or for decls we do not yet have a DIE. Do not
15618 call dwarf2out_abstract_function late.
15619 (dwarf2out_early_global_decl): Make sure to create DIEs
15620 for abstract instances of a decl first.
15621 (dwarf2out_late_global_decl): Adjust comment.
15622 (output_macinfo_op): With multiple macro sections use
15623 macinfo_label_base to distinguish labels.
15624 (output_macinfo): Likewise. Update macinfo_label_base.
15625 Pass in the line info label.
15626 (note_variable_value_in_expr): When generating LTO resolve
15627 all variable values here by generating DIEs as needed.
15628 (init_sections_and_labels): Add early LTO debug flag parameter
15629 and generate different sections and names if set. Add generation
15630 counter for the labels so we can have multiple of them.
15631 (reset_dies): Helper to allow DIEs to be output multiple times.
15632 (dwarf2out_finish): When outputting DIEs to the fat part of an
15633 LTO object first reset DIEs.
15634 (dwarf2out_early_finish): Output early DIEs when generating LTO.
15635 (modified_type_die): Check for decl_ultimate_origin being self
15637 (gen_type_die_with_usage): Likewise.
15638 (gen_typedef_die): Allow decl_ultimate_origin being self.
15639 (set_decl_abstract_flags): Remove.
15640 (set_block_abstract_flags): Likewise.
15641 (dwarf2out_abstract_function): Treat the early generated DIEs
15642 as the abstract copy and only add DW_AT_inline and
15643 DW_AT_artificial here and call set_decl_origin_self.
15644 If the DIE has an abstract origin don't do anything.
15645 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
15646 if we have none yet (Go fails to build one, PR78628).
15647 (variably_modified_type_p): Prevent endless recursion for Ada
15648 cyclic pointer types.
15649 * lto-streamer-in.c: Include debug.h.
15650 (dref_queue): New global.
15651 (lto_read_tree_1): Stream in DIE references.
15652 (lto_input_tree): Register DIE references.
15653 (input_function): Stream DECL_DEBUG_ARGS.
15654 * lto-streamer-out.c: Include debug.h.
15655 (lto_write_tree_1): Output DIE references.
15656 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
15657 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
15658 (output_function): Stream DECL_DEBUG_ARGS.
15659 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
15660 Stream DECL_ABSTRACT_ORIGIN.
15661 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
15662 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
15663 DECL_CONTEXT for file-scope decls.
15664 * lto-streamer.h (struct dref_entry): Declare.
15665 (dref_queue): Likewise.
15666 * cfgexpand.c (pass_expand::execute): Do not call the
15667 outlining_inline_function hook here.
15668 * lto-wrapper.c (debug_obj): New global.
15669 (tool_cleanup): Unlink it if required.
15670 (debug_objcopy): New function.
15671 (run_gcc): Handle early debug sections in the IL files by
15672 extracting them to separate files, partially linkin them and
15673 feeding the result back as result to the linker.
15674 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
15675 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
15676 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
15677 sections into a separate segment.
15678 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
15680 (darwin_asm_dwarf_section): Likewise.
15681 (darwin_asm_output_dwarf_offset): Likewise.
15682 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
15684 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15685 Alan Hayward <alan.hayward@arm.com>
15686 David Sherwood <david.sherwood@arm.com>
15688 * read-md.h (md_reader::record_potential_iterator_use): Replace
15689 pointer argument with an rtx and an index.
15690 * read-rtl.c (iterator_group::apply_iterator): Likewise.
15691 (apply_mode_iterator): Likewise.
15692 (apply_code_iterator): Likewise.
15693 (apply_int_iterator): Likewise.
15694 (apply_subst_iterator): Likewise.
15695 (record_iterator_use): Likewise.
15696 (record_attribute_use): Likewise.
15697 (md_reader::record_potential_iterator_use): Likewise. Update calls
15698 to record_iterator_use and apply_iterator.
15699 (iterator_use): Replace ptr with x and index.
15700 (attribute_use): Likewise.
15701 (apply_attribute_uses): Update calls to apply_iterator.
15702 (apply_iterators): Likewise. Update initialization of iterator_use.
15703 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
15704 and record_potential_iterator_use.
15705 (rtx_reader::read_rtx_operand): Likewise.
15707 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15708 Alan Hayward <alan.hayward@arm.com>
15709 David Sherwood <david.sherwood@arm.com>
15711 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
15714 2017-08-21 Richard Biener <rguenther@suse.de>
15716 PR middle-end/81884
15717 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
15718 at struct end conservatively when comparing common bases.
15720 2017-08-21 Richard Biener <rguenther@suse.de>
15722 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
15723 (mem_ref_in_stmt): Remove.
15724 (determine_max_movement): Use ref index to get at the reference.
15725 (invariantness_dom_walker::before_dom_children): Deal with
15726 lim data already initialized.
15727 (gather_mem_refs_stmt): Initialize lim data and record ref index.
15729 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
15731 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
15732 (TARGET_ISA_ROUND): Ditto.
15733 (TARGET_ROUND): Ditto.
15734 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
15735 * config/i386/i386.md: Ditto.
15736 * config/i386/sse.md: Ditto.
15737 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
15738 with OPTION_MASK_ISA_SSE4_1.
15740 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
15743 * doc/extend.texi (x86 Built-in Functions): Correct the name of
15744 __builtin_ia32_lzcnt_u16.
15746 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
15749 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
15750 (rs6000_set_current_function): Rewrite function to use it.
15752 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
15755 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
15756 and TYPE_WARN_IF_NOT_ALIGN.
15757 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
15758 (handle_warn_if_not_align): New.
15759 (place_union_field): Call handle_warn_if_not_align.
15760 (place_field): Call handle_warn_if_not_align.
15761 Copy TYPE_WARN_IF_NOT_ALIGN.
15762 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
15763 (layout_type): Likewise.
15764 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
15766 (tree_decl_common): Add warn_if_not_align.
15767 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
15768 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
15769 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
15770 (DECL_WARN_IF_NOT_ALIGN): Likewise.
15771 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
15772 * doc/extend.texi: Document warn_if_not_aligned attribute.
15773 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
15775 2017-08-17 Martin Liska <mliska@suse.cz>
15778 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
15779 (get_data_dependence): Use it as pointer type.
15780 (distribute_loop): Likewise.
15782 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15784 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
15785 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
15786 (p8_vmrgow_<mode>_direct): New define_insn.
15787 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
15788 handle endianness for vmrgew and vmrgow permute patterns.
15790 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
15792 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
15793 * config/rs6000/rs6000-cpus.def: Remove comment.
15794 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
15795 (POWERPC_MASKS): Likewise.
15796 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
15797 use of TARGET_VSX_TIMODE.
15798 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
15799 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
15800 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
15801 (rs6000_option_override_internal): Remove dead code.
15802 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
15803 (rs6000_legitimize_reload_address): Likewise.
15804 (rs6000_legitimate_address_p): Likewise.
15805 (rs6000_opt_masks): Delete "vsx-timode".
15806 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
15807 from function comment.
15808 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
15809 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
15810 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
15812 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
15813 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
15814 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
15815 (VSX_TI): Likewise.
15817 (define_peephole2): Likewise.
15819 2017-08-17 Martin Sebor <msebor@redhat.com>
15822 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
15823 past the end of an array.
15824 (test_pp_format): Add test cases.
15826 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
15828 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
15829 missing ECF_NOTHROW flags.
15831 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
15834 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
15835 operands residing in integer registers.
15836 (*vsx_le_perm_load_<mode>): Likewise.
15837 (*vsx_le_perm_store_<mode>): Likewise.
15838 (define_peephole2): Add peepholes to optimize the above.
15840 2017-08-17 Marek Polacek <polacek@redhat.com>
15842 PR middle-end/81814
15843 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
15844 to mimic what shorten_compare did. Change the return type to bool.
15845 (fold_cond_expr_with_comparison): Update call to
15846 operand_equal_for_comparison_p.
15847 (fold_ternary_loc): Likewise.
15849 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
15851 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
15853 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
15855 2017-08-17 Richard Biener <rguenther@suse.de>
15857 * tree-ssa-structalias.c (solve_graph): When propagating
15858 to successors update the graphs succ edges and avoid duplicate work.
15860 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
15863 * config/i386/i386.c (ix86_option_override_internal): Save target
15864 specific options after ix86_stack_protector_guard_reg was changed.
15866 2017-08-17 Richard Biener <rguenther@suse.de>
15868 PR tree-optimization/81827
15869 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
15870 (new_var_info): Initialize it conservatively.
15871 (get_call_vi): Mark register vars.
15872 (new_scalar_tmp_constraint_exp): Likewise.
15873 (handle_rhs_call): Likewise.
15874 (handle_const_call): Likewise.
15875 (create_function_info_for): Likewise.
15876 (solve_constraints): Sort varinfos to separate register from
15877 non-register vars to pack points-to solution bitmaps during
15880 2017-08-17 Marek Polacek <polacek@redhat.com>
15882 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
15884 2017-08-17 Richard Biener <rguenther@suse.de>
15886 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
15887 to true when overflow is undefined and we saturated the result.
15889 2017-08-17 Alan Modra <amodra@gmail.com>
15892 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
15893 Don't use store multiple if only one reg needs saving.
15894 (interesting_frame_related_regno): New function.
15895 (rs6000_frame_related): Don't emit frame info for regs that
15897 (rs6000_emit_epilogue): Likewise.
15899 2017-08-16 Nathan Sidwell <nathan@acm.org>
15901 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
15902 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
15903 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
15904 * tree.c (free_lang_data_in_type): Use else-if chain. Always
15905 clear TYPE_LANG_1. Remove obsolete member-function stripping.
15906 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
15907 (verify_type): Adjust for TYPE_BINFO move.
15908 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
15909 process TYPE_BINFO directly.
15910 (hash_tree): Likewise.
15911 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15913 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15916 2017-08-16 David Malcolm <dmalcolm@redhat.com>
15918 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
15920 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
15923 * config/i386/i386.md (*anddi_1_btr): Change predicates of
15924 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
15925 Add ix86_binary_operator_ok to insn constraint.
15926 (*iordi_1_bts): Ditto.
15927 (*xordi_1_btc): Ditto.
15928 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
15929 Update corresponding peephole2 pattern.
15933 2017-08-16 Bin Cheng <bin.cheng@arm.com>
15935 PR tree-optimization/81832
15936 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
15937 copy loop header which has IFN_LOOP_DIST_ALIAS call.
15939 2017-08-16 Marek Polacek <polacek@redhat.com>
15942 * fold-const.c (fold_indirect_ref_1): Restore original behavior
15943 regarding size_zero_node.
15945 2017-08-16 Martin Liska <mliska@suse.cz>
15948 * config.gcc: Respect previously set extra_objs in case
15951 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
15953 PR tree-optimization/81835
15954 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
15955 the phi SSA_NAME. Check that the condition in a COND_EXPR does
15956 not depend on the phi.
15958 2017-08-16 Alan Modra <amodra@gmail.com>
15960 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
15963 2017-08-16 Alan Modra <amodra@gmail.com>
15965 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
15966 (save_reg_p): ..into this. Update all callers.
15967 (first_reg_to_save): Simplify.
15969 2017-08-16 Alan Modra <amodra@gmail.com>
15971 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
15974 2017-08-15 Joseph Myers <joseph@codesourcery.com>
15978 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
15979 constant count if that count is less than 32.
15981 2017-08-15 Nathan Sidwell <nathan@acm.org>
15983 * gcc.c (execute): Emit friendlier message if inferior is killed
15984 by an external cause.
15986 2017-08-15 Richard Biener <rguenther@suse.de>
15988 PR tree-optimization/81790
15989 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
15990 CONSTRUCTORs from simplifying and VN.
15992 2017-08-14 Martin Sebor <msebor@redhat.com>
15994 * builtin-attrs.def: Add comments.
15996 2017-08-14 Martin Sebor <msebor@redhat.com>
15999 * doc/extend.texi (attribute nonstring): Document new attribute.
16001 2017-08-14 Martin Sebor <msebor@redhat.com>
16004 * tree-diagnostic.c (default_tree_printer): Handle %G.
16005 * gimple-pretty-print.h (percent_G_format): Declare new function.
16006 * gimple-pretty-print.c (percent_G_format): Define.
16007 * tree-pretty-print.c (percent_K_format): Add argument.
16009 2017-08-14 Martin Sebor <msebor@redhat.com>
16011 PR translation/79998
16012 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
16013 Remove a stray space.
16015 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
16018 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
16019 (*iordi_1_bts): Ditto.
16020 (*xordi_1_btc): Ditto.
16022 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16025 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
16026 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16028 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
16029 quoted strings, and make more translator-friendly.
16030 (darwin_rs6000_override_options): Likewise.
16031 (rs6000_option_override_internal): Likewise.
16032 (rs6000_return_in_memory): Fix overlong line.
16033 (init_cmulative_args): Use quoted strings, and make more
16034 translator-friendly.
16035 (rs6000_pass_by_reference): Fix overlong line.
16036 (def_builtin): Use quoted strings.
16037 (altivec_expand_predicate_builtin): Use quoted strings, and make
16038 more translator-friendly.
16039 (htm_expand_builtin): Use quoted strings.
16040 (cpu_expand_builtin): Use quoted strings, and make more
16041 translator-friendly.
16042 (altivec_expand_builtin): Likewise.
16043 (paired_expand_predicate_builtin): Likewise.
16044 (rs6000_invalid_builtin): Likewise.
16045 (builtin_function_type): Use quoted strings.
16046 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
16047 more translator-friendly.
16048 (rs6000_trampoline_init): Likewise.
16049 (rs6000_handle_altivec_attribute): Likewise.
16050 (rs6000_inner_target_options): Use quoted strings.
16051 (rs6000_disable_incompatible_switches): Likewise.
16052 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
16053 strings, and make more translator-friendly.
16054 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
16056 2017-08-14 Bin Cheng <bin.cheng@arm.com>
16058 PR tree-optimization/81799
16059 * tree-loop-distribution.c (version_loop_by_alias_check): Force
16060 cond_expr to simple gimple operand.
16062 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
16064 PR middle-end/46932
16065 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
16067 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
16071 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
16072 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
16073 TARGET_GASISR_PROLOGUES.
16074 * config/avr/avr.c (avr_option_override): Same.
16075 (avr_pass_pre_proep::execute): Same.
16077 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
16080 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
16081 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
16083 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
16085 * config/i386/i386.md (*load_tp_<mode>): Redefine as
16086 define_insn_and_split. Split to a memory load from 0 in
16087 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
16088 using PTR mode iterator.
16089 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
16090 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
16091 (*add_tp_<mode>): Redefine as define_insn_and_split.
16092 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
16093 address space. Merge with *add_tp_x32 using PTR mode iterator.
16094 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
16095 Split to an add with a memory load from 0 in
16096 DEFAULT_TLS_SEG_REG address space.
16098 2017-08-12 Andrew Pinski <apinski@cavium.com>
16100 * config/aarch64/aarch64-option-extensions.def (rdma):
16101 Fix feature string to what Linux prints out in /proc/cpuinfo.
16103 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
16106 * dwarf2out.c (modified_type_die): For C typedef types that have
16107 an ultimate origin, process the ultimate origin instead of the
16109 (gen_typedef_die): Assert that input DECLs have no ultimate
16111 (gen_type_die_with_usage): For typedef variants that have an
16112 ultimate origin, just call gen_decl_die on the original DECL.
16113 (process_scope_var): Avoid creating DIEs for local typedefs and
16114 concrete static variables.
16116 2017-08-12 Alan Modra <amodra@gmail.com>
16120 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
16121 match gnu-user.h startfile.
16122 (ENDFILE_LINUX_SPEC): Similarly.
16124 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
16127 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
16129 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
16131 2017-08-11 Tamar Christina <tamar.christina@arm.com>
16132 * config/aarch64/aarch64.md (mov<mode>): Change.
16133 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16134 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
16135 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
16137 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
16139 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
16140 for storage order barriers.
16142 2017-08-11 Martin Liska <mliska@suse.cz>
16145 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
16146 variables of void type.
16148 2017-08-11 Martin Liska <mliska@suse.cz>
16150 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
16151 TARGET_SUPPORTS_ALIASES.
16152 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
16153 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
16154 (optimize_weakref): Likewise.
16155 * symtab.c (symtab_node::noninterposable_alias): Likewise.
16156 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
16157 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
16159 2017-08-11 Martin Liska <mliska@suse.cz>
16162 * config/i386/i386.c (make_resolver_func): Do complete
16163 refactoring of the function.
16165 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
16168 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
16169 * config/i386/i386.c (ix86_stack_protect_guard): Use
16170 ix86_stack_protect_guard_symbol_str to generate varible declaration.
16171 * doc/invoke.texi (x86 Options): Document
16172 -mstack-protector-guard-symbol= option.
16174 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
16176 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
16177 * config/i386/i386.c (ix86_split_stack_guard): New function.
16178 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
16179 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
16180 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
16181 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
16182 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
16183 (split_stack_space_check): Call ix86_split_stack_guard.
16185 2017-08-10 Martin Sebor <msebor@redhat.com>
16187 * print-tree.c (print_node): Print location using the established
16189 Replace spaces with colons.
16190 (debug_raw, debug): Ditto.
16192 2017-08-10 Martin Sebor <msebor@redhat.com>
16195 * pretty-print.c (pp_format): Correct the handling of %s precision.
16197 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
16200 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
16202 (ix86_finalize_stack_frame_flags): This. Also clear
16203 frame_pointer_needed if -fno-omit-frame-pointer is used without
16205 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
16206 with ix86_finalize_stack_frame_flags.
16207 (ix86_expand_epilogue): Likewise.
16208 (ix86_expand_split_stack_prologue): Likewise.
16209 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
16211 2017-08-10 Martin Liska <mliska@suse.cz>
16214 * c-attribs.c (handle_target_attribute):
16215 Report warning for an empty string argument of target attribute.
16217 2017-08-09 Jakub Jelinek <jakub@redhat.com>
16220 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
16222 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
16223 or DECL_NONLOCAL labels.
16224 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
16225 or DECL_NONLOCAL labels here.
16227 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
16229 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
16230 to indicate when early gimple folding has been disabled.
16231 (rs6000_gimple_fold_builtin): Add debug content.
16232 (rs6000_invalid_builtin): Fix whitespace.
16233 (rs6000_expand_builtin): Fix whitespace.
16234 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
16236 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
16239 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
16240 SAVE_MULTIPLE if not all the registers that saves, should be saved.
16242 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
16244 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
16245 (qdf24xx): Likewise.
16246 * config/aarch64/aarch64.md: Include falkor.md.
16247 * config/aarch64/falkor.md: New.
16249 2017-08-09 Marek Polacek <polacek@redhat.com>
16252 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
16253 * diagnostic.c (emit_diagnostic): Add a comment.
16254 (emit_diagnostic_valist): New function.
16256 2017-08-09 Marek Polacek <polacek@redhat.com>
16259 * input.c (make_location): New overload.
16260 * input.h (make_location): Declare.
16262 2017-08-08 Alan Modra <amodra@gmail.com>
16263 H.J. Lu <hongjiu.lu@intel.com>
16266 * gcc.c (NO_PIE_SPEC): Delete.
16267 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
16269 (LINK_PIE_SPEC): ..to here.
16270 (LINK_COMMAND_SPEC): Support -no-pie.
16271 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
16272 chain of crtbegin*.o selection, update for PIE_SPEC changes and
16274 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
16275 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
16276 (ENDFILE_CRTEND_SPEC): Similarly.
16278 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
16281 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
16282 (mstack-protector-guard-offset=): Ditto.
16283 * config/i386/i386.c (ix86_option_override): Handle
16284 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
16286 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
16287 ix86_stack_protect_guard_offset variables.
16288 (TARGET_STACK_PROTECT_GUARD): Always define.
16289 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
16290 and -mstack-protector-guard-offset= options.
16292 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16294 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
16295 boundary case for the last candidate.
16297 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16299 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
16300 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
16302 2017-08-08 Tamar Christina <tamar.christina@arm.com>
16304 PR middle-end/19706
16305 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
16306 * config/aarch64/aarch64-builtins.c
16307 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
16308 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
16309 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
16311 2017-08-08 Tamar Christina <tamar.christina@arm.com>
16312 Andrew Pinski <pinskia@gmail.com>
16314 PR middle-end/19706
16315 * internal-fn.def (XORSIGN): New.
16316 * optabs.def (xorsign_optab): New.
16317 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
16318 (convert_expand_mult_copysign): New.
16319 (pass_optimize_widening_mul::execute): Call
16320 convert_expand_mult_copysign.
16322 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16324 PR tree-optimization/81354
16325 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
16326 Insert on edges rather than explicitly creating landing pads.
16327 (analyze_candidates_and_replace): Commit edge inserts.
16329 2017-08-08 Richard Biener <rguenther@suse.de>
16331 PR middle-end/81719
16332 * tree-ssa-loop-niter.c: Include tree-dfa.h.
16333 (expand_simple_operations): Also look through ADDR_EXPRs with
16334 MEM_REF bases treating them as POINTER_PLUS_EXPR.
16336 2017-08-08 Richard Biener <rguenther@suse.de>
16338 PR tree-optimization/81723
16339 * tree-vect-slp.c (struct bst_traits): New hash traits.
16340 (bst_fail): New global.
16341 (vect_build_slp_tree_2): New worker, split out from ...
16342 (vect_build_slp_tree): ... this now wrapping it with using
16343 bst_fail set to cache SLP tree build fails. Properly handle
16345 (vect_analyze_slp_instance): Allocate and free bst_fail.
16347 2017-08-08 Martin Liska <mliska@suse.cz>
16350 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
16351 LABEL_DECLs that can be from a different function.
16353 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16355 PR tree-optimization/81744
16356 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
16357 loop's number of iterations.
16359 2017-08-08 Martin Liska <mliska@suse.cz>
16361 * asan.c: Include header files.
16362 * attribs.c (build_decl_attribute_variant): New function moved
16364 (build_type_attribute_qual_variant): Likewise.
16365 (cmp_attrib_identifiers): Likewise.
16366 (simple_cst_list_equal): Likewise.
16367 (omp_declare_simd_clauses_equal): Likewise.
16368 (attribute_value_equal): Likewise.
16369 (comp_type_attributes): Likewise.
16370 (build_type_attribute_variant): Likewise.
16371 (lookup_ident_attribute): Likewise.
16372 (remove_attribute): Likewise.
16373 (merge_attributes): Likewise.
16374 (merge_type_attributes): Likewise.
16375 (merge_decl_attributes): Likewise.
16376 (merge_dllimport_decl_attributes): Likewise.
16377 (handle_dll_attribute): Likewise.
16378 (attribute_list_equal): Likewise.
16379 (attribute_list_contained): Likewise.
16380 * attribs.h (lookup_attribute): New function moved from tree.[ch].
16381 (lookup_attribute_by_prefix): Likewise.
16382 * bb-reorder.c: Include header files.
16383 * builtins.c: Likewise.
16384 * calls.c: Likewise.
16385 * cfgexpand.c: Likewise.
16386 * cgraph.c: Likewise.
16387 * cgraphunit.c: Likewise.
16388 * convert.c: Likewise.
16389 * dwarf2out.c: Likewise.
16390 * final.c: Likewise.
16391 * fold-const.c: Likewise.
16392 * function.c: Likewise.
16393 * gimple-expr.c: Likewise.
16394 * gimple-fold.c: Likewise.
16395 * gimple-pretty-print.c: Likewise.
16396 * gimple.c: Likewise.
16397 * gimplify.c: Likewise.
16398 * hsa-common.c: Likewise.
16399 * hsa-gen.c: Likewise.
16400 * internal-fn.c: Likewise.
16401 * ipa-chkp.c: Likewise.
16402 * ipa-cp.c: Likewise.
16403 * ipa-devirt.c: Likewise.
16404 * ipa-fnsummary.c: Likewise.
16405 * ipa-inline.c: Likewise.
16406 * ipa-visibility.c: Likewise.
16408 * lto-cgraph.c: Likewise.
16409 * omp-expand.c: Likewise.
16410 * omp-general.c: Likewise.
16411 * omp-low.c: Likewise.
16412 * omp-offload.c: Likewise.
16413 * omp-simd-clone.c: Likewise.
16414 * opts-global.c: Likewise.
16415 * passes.c: Likewise.
16416 * predict.c: Likewise.
16417 * sancov.c: Likewise.
16418 * sanopt.c: Likewise.
16419 * symtab.c: Likewise.
16420 * toplev.c: Likewise.
16421 * trans-mem.c: Likewise.
16422 * tree-chkp.c: Likewise.
16423 * tree-eh.c: Likewise.
16424 * tree-into-ssa.c: Likewise.
16425 * tree-object-size.c: Likewise.
16426 * tree-parloops.c: Likewise.
16427 * tree-profile.c: Likewise.
16428 * tree-ssa-ccp.c: Likewise.
16429 * tree-ssa-live.c: Likewise.
16430 * tree-ssa-loop.c: Likewise.
16431 * tree-ssa-sccvn.c: Likewise.
16432 * tree-ssa-structalias.c: Likewise.
16433 * tree-ssa.c: Likewise.
16434 * tree-streamer-in.c: Likewise.
16435 * tree-vectorizer.c: Likewise.
16436 * tree-vrp.c: Likewise.
16437 * tsan.c: Likewise.
16438 * ubsan.c: Likewise.
16439 * varasm.c: Likewise.
16440 * varpool.c: Likewise.
16441 * tree.c: Remove functions moved to attribs.[ch].
16442 * tree.h: Likewise.
16443 * config/aarch64/aarch64.c: Add attrs.h header file.
16444 * config/alpha/alpha.c: Likewise.
16445 * config/arc/arc.c: Likewise.
16446 * config/arm/arm.c: Likewise.
16447 * config/avr/avr.c: Likewise.
16448 * config/bfin/bfin.c: Likewise.
16449 * config/c6x/c6x.c: Likewise.
16450 * config/cr16/cr16.c: Likewise.
16451 * config/cris/cris.c: Likewise.
16452 * config/darwin.c: Likewise.
16453 * config/epiphany/epiphany.c: Likewise.
16454 * config/fr30/fr30.c: Likewise.
16455 * config/frv/frv.c: Likewise.
16456 * config/ft32/ft32.c: Likewise.
16457 * config/h8300/h8300.c: Likewise.
16458 * config/i386/winnt.c: Likewise.
16459 * config/ia64/ia64.c: Likewise.
16460 * config/iq2000/iq2000.c: Likewise.
16461 * config/lm32/lm32.c: Likewise.
16462 * config/m32c/m32c.c: Likewise.
16463 * config/m32r/m32r.c: Likewise.
16464 * config/m68k/m68k.c: Likewise.
16465 * config/mcore/mcore.c: Likewise.
16466 * config/microblaze/microblaze.c: Likewise.
16467 * config/mips/mips.c: Likewise.
16468 * config/mmix/mmix.c: Likewise.
16469 * config/mn10300/mn10300.c: Likewise.
16470 * config/moxie/moxie.c: Likewise.
16471 * config/msp430/msp430.c: Likewise.
16472 * config/nds32/nds32-isr.c: Likewise.
16473 * config/nds32/nds32.c: Likewise.
16474 * config/nios2/nios2.c: Likewise.
16475 * config/nvptx/nvptx.c: Likewise.
16476 * config/pa/pa.c: Likewise.
16477 * config/pdp11/pdp11.c: Likewise.
16478 * config/powerpcspe/powerpcspe.c: Likewise.
16479 * config/riscv/riscv.c: Likewise.
16480 * config/rl78/rl78.c: Likewise.
16481 * config/rx/rx.c: Likewise.
16482 * config/s390/s390.c: Likewise.
16483 * config/sh/sh.c: Likewise.
16484 * config/sol2.c: Likewise.
16485 * config/sparc/sparc.c: Likewise.
16486 * config/spu/spu.c: Likewise.
16487 * config/stormy16/stormy16.c: Likewise.
16488 * config/tilegx/tilegx.c: Likewise.
16489 * config/tilepro/tilepro.c: Likewise.
16490 * config/v850/v850.c: Likewise.
16491 * config/vax/vax.c: Likewise.
16492 * config/visium/visium.c: Likewise.
16493 * config/xtensa/xtensa.c: Likewise.
16495 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16498 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
16499 constraints since the -mupper-regs-* switches have been
16501 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
16502 into a vector from a double word element that was extracted from
16503 another vector, and eliminate extra XXPERMDI instructions.
16504 (vsx_concat_<mode>_2): Likewise.
16505 (vsx_concat_<mode>_3): Likewise.
16506 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
16507 concat to allow optimizing inserts from previous extracts.
16509 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
16511 * config/i386/i386.c (ix86_stack_protect_guard): Generate
16512 memory reference to a SSP offset in TLS address space.
16513 (ix86_print_operand) <case '@'>: Remove.
16514 (ix86_print_operand_punct_valid_p): Remove '@' code.
16515 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
16516 UNSPEC_SP_TLS_TEST.
16517 (stack_tls_protect_set_<mode>): Remove.
16518 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
16519 (stack_tls_protect_test_<mode>): Remove.
16520 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
16522 2017-08-07 Olivier Hainque <hainque@adacore.com>
16525 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
16527 2017-08-07 Douglas Rupp <rupp@adacore.com>
16529 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
16530 variable was referenced as multidir in command.
16532 2017-08-07 Jakub Jelinek <jakub@redhat.com>
16535 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
16538 2017-08-07 Martin Liska <mliska@suse.cz>
16540 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
16541 * config/rl78/rl78.c: Add include of attribs.h.
16542 * config/sh/sh.c: Likewise.
16543 * config/v850/v850.c: Likewise.
16545 2017-08-07 Tom de Vries <tom@codesourcery.com>
16547 PR middle-end/78266
16548 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
16550 2017-08-07 Martin Liska <mliska@suse.cz>
16552 * config/mips/mips.c: Include attribs.h.
16554 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
16557 * doc/invoke.texi: Document change in behvaior for -Ofast for
16560 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
16562 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
16564 (aarch64_pop_regs): Likewise.
16565 (aarch64_gen_load_pair): Likewise.
16566 (aarch64_save_callee_saves): Likewise.
16567 (aarch64_restore_callee_saves): Likewise.
16569 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16571 * config/i386/i386.c: Revert the last change.
16573 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16576 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
16578 (ix86_finalize_stack_frame_flags): This. Also clear
16579 frame_pointer_needed if -fno-omit-frame-pointer is used without
16581 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
16582 with ix86_finalize_stack_frame_flags.
16583 (ix86_expand_epilogue): Likewise.
16584 (ix86_expand_split_stack_prologue): Likewise.
16586 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16589 * config/i386/i386.c (get_builtin_code_for_version): Set priority
16590 to P_AES for Westmere.
16592 2017-08-07 Jonathan Yong <10walls@gmail.com>
16594 * config/i386/mingw.opt (fset-stack-executable): Removed.
16595 * config/i386/cygming.opt (fset-stack-executable): Moved
16597 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
16599 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
16601 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
16603 2017-08-07 Marek Polacek <polacek@redhat.com>
16605 PR middle-end/81737
16606 * fold-const.c (fold_indirect_ref_1): Check type_domain.
16608 2017-08-07 Martin Liska <mliska@suse.cz>
16610 * attribs.h (canonicalize_attr_name): New function.
16611 (cmp_attribs): Move from c-format.c and adjusted.
16612 (is_attribute_p): Moved from tree.h.
16613 * tree-inline.c: Add new includes.
16614 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
16615 (private_is_attribute_p): Remove.
16616 (private_lookup_attribute): Likewise.
16617 (private_lookup_attribute_by_prefix): Simplify.
16618 (remove_attribute): Use is_attribute_p.
16619 * tree.h: Remove removed declarations.
16621 2017-08-07 Jakub Jelinek <jakub@redhat.com>
16623 PR middle-end/81698
16624 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
16625 instead of computing it in the function. Formatting fix.
16626 (expand_case): Don't rely on default_edge being the first edge,
16627 clear it if removing it, pass default_edge to
16628 emit_case_dispatch_table.
16629 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
16632 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
16634 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
16635 insn in the function, emit NOP after the insn.
16637 2017-08-06 Tom de Vries <tom@codesourcery.com>
16639 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
16642 2017-08-06 Tom de Vries <tom@codesourcery.com>
16644 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
16647 2017-08-04 Yury Gribov <tetra2005@gmail.com>
16649 PR tree-optimization/57371
16650 * match.pd: New pattern.
16652 2017-08-04 Marek Polacek <polacek@redhat.com>
16654 PR middle-end/81695
16655 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
16656 perform the computation in offset_int.
16658 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16660 PR tree-optimization/81136
16661 * tree-vectorizer.h: Include tree-hash-traits.h.
16662 (vec_base_alignments): New typedef.
16663 (vec_info): Add a base_alignments field.
16664 (vect_record_base_alignments): Declare.
16665 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
16667 (DR_IS_CONDITIONAL_IN_STMT): New macro.
16668 (create_data_ref): Add an is_conditional_in_stmt argument.
16669 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
16670 the is_conditional_in_stmt field.
16671 (data_ref_loc): Add an is_conditional_in_stmt field.
16672 (get_references_in_stmt): Set the is_conditional_in_stmt field.
16673 (find_data_references_in_stmt): Update call to create_data_ref.
16674 (graphite_find_data_references_in_stmt): Likewise.
16675 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
16676 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
16677 (vect_record_base_alignment): New function.
16678 (vect_record_base_alignments): Likewise.
16679 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
16680 for nested statements even if we fail to compute a misalignment.
16681 Use pooled base alignments for unconditional references.
16682 (vect_find_same_alignment_drs): Compare base addresses instead
16684 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
16685 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
16687 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16689 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
16690 Add an explicit name for the enum. Use auto_vec for slp_instances
16691 and grouped_stores.
16692 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
16694 (_bb_vec_info): Add a constructor and destructor.
16695 (vinfo_for_stmt): Return NULL for uids of -1 as well.
16696 (destroy_loop_vec_info): Delete.
16697 (vect_destroy_datarefs): Likewise.
16698 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
16699 (vec_info::vec_info): New function.
16700 (vec_info::~vec_info): Likewise.
16701 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
16702 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
16703 destroy_loop_vec_info.
16704 * tree-vect-loop.c (new_loop_vec_info): Replace with...
16705 (_loop_vec_info::_loop_vec_info): ...this.
16706 (destroy_loop_vec_info): Replace with...
16707 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
16708 the stmt_vec_infos. Leave handling of vec_info information to its
16709 destructor. Remove explicit vector releases.
16710 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
16711 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
16712 * tree-vect-slp.c (new_bb_vec_info): Replace with...
16713 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
16714 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
16715 (destroy_bb_vec_info): Replace with...
16716 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
16717 information to its destructor.
16718 (vect_slp_analyze_bb_1): Use new and delete instead of
16719 new_bb_vec_info and destroy_bb_vec_info.
16720 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
16723 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16725 * tree-data-ref.h (subscript): Add access_fn field.
16726 (data_dependence_relation): Add could_be_independent_p.
16727 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
16728 (same_access_functions): Move to tree-data-ref.c.
16729 * tree-data-ref.c (ref_contains_union_access_p): New function.
16730 (access_fn_component_p): Likewise.
16731 (access_fn_components_comparable_p): Likewise.
16732 (dr_analyze_indices): Add a reference to access_fn_component_p.
16733 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
16735 (constant_access_functions): Likewise.
16736 (add_other_self_distances): Likewise.
16737 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
16738 (initialize_data_dependence_relation): Use XCNEW and remove
16739 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
16740 of access functions that have the same type. Allow the
16741 subsequence to end with different bases in some circumstances.
16742 Record the chosen access functions in SUB_ACCESS_FN.
16743 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
16744 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
16745 (subscript_dependence_tester_1): Likewise dra and drb.
16746 (build_classic_dist_vector): Update calls accordingly.
16747 (subscript_dependence_tester): Likewise.
16748 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
16749 DDR_COULD_BE_INDEPENDENT_P.
16750 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
16751 comp_alias_ddrs instead of may_alias_ddrs.
16752 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
16754 (vect_analyze_data_ref_dependence): Use it if
16755 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
16756 distance vectors if that fails.
16757 (dependence_distance_ge_vf): New function.
16758 (vect_prune_runtime_alias_test_list): Use it. Don't clear
16759 LOOP_VINFO_MAY_ALIAS_DDRS.
16761 2017-08-04 Richard Biener <rguenther@suse.de>
16763 PR middle-end/81705
16764 * fold-const.c (fold_binary_loc): Properly restrict
16765 minus_var0 && minus_var1 case when associating undefined overflow
16768 2017-08-04 Tom de Vries <tom@codesourcery.com>
16770 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
16772 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16774 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16775 Don't start diagnostic messages with a capital letter.
16776 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16778 (rs6000_invalid_builtin): Likewise.
16779 (rs6000_trampoline_init): Likewise.
16781 2017-08-03 Jakub Jelinek <jakub@redhat.com>
16784 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
16785 after setting changeable df flags.
16787 2017-08-03 Richard Biener <rguenther@suse.de>
16789 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
16790 up if the use is in USE - X.
16792 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
16794 * toplev.c (dumpfile.h): New include.
16795 (internal_error_reentered): New static function. Use it...
16796 (internal_error_function): ...here to handle reentered internal_error.
16798 2017-08-03 Richard Biener <rguenther@suse.de>
16800 PR middle-end/81148
16801 * fold-const.c (split_tree): Add minus_var and minus_con
16802 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
16803 here but always use minus_*.
16804 (associate_trees): Assert we never associate with MINUS_EXPR
16805 and NULL first operand. Do not recurse for PLUS_EXPR operands
16806 when associating as MINUS_EXPR either.
16807 (fold_binary_loc): Track minus_var and minus_con.
16809 2017-08-03 Tom de Vries <tom@codesourcery.com>
16812 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
16813 ACCEL_COMPILER, apply finish_options on
16814 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
16816 2017-08-03 Tom de Vries <tom@codesourcery.com>
16819 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
16820 function_entry_patch_area_size > 0.
16822 2017-08-03 Jakub Jelinek <jakub@redhat.com>
16825 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
16826 instead of 10??LU, perform unit multiplication in wide_int,
16827 don't change alloc_object_size_limit if the limit is larger
16830 PR tree-optimization/81655
16831 PR tree-optimization/81588
16832 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
16833 the case when ranges[i].low and high are 1 for unsigned type with
16836 PR middle-end/81052
16837 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
16838 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
16840 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16842 * tree-vrp.h: Add include guard.
16844 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
16847 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
16848 (ud2): New insn pattern.
16849 * config/i386/i386.c (ix86_expand_epilogue):
16850 For naked functions, generate ud2 instead of trap insn.
16852 2017-08-02 Marek Polacek <polacek@redhat.com>
16855 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
16857 2017-08-02 Tom de Vries <tom@codesourcery.com>
16858 Cesar Philippidis <cesar@codesourcery.com>
16860 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
16861 Add missing edge probabilities.
16863 2017-08-02 Tamar Christina <tamar.christina@arm.com>
16865 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
16866 Correct endianness.
16868 2017-08-02 Jakub Jelinek <jakub@redhat.com>
16870 PR middle-end/79499
16871 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
16872 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
16873 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
16874 prologue_seq sequences - if any.
16876 2017-08-02 Richard Biener <rguenther@suse.de>
16878 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
16879 via vectors if supported, integer extracts via punning if supported
16880 or otherwise vector extracts.
16882 2017-08-02 Richard Biener <rguenther@suse.de>
16884 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
16886 (bitmap_insert_into_set): ... this.
16888 2017-08-02 Richard Biener <rguenther@suse.de>
16890 PR tree-optimization/81633
16892 2015-08-17 Alan Hayward <alan.hayward@arm.com>
16894 PR tree-optimization/71752
16895 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
16897 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
16899 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
16900 (machine_function::call_ms2sysv_pad_out): Remove field.
16901 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
16902 (ix86_compute_frame_layout): Likewise.
16904 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
16907 * config/i386/i386.c (ix86_set_func_type): Disallow naked
16908 attribute with interrupt attribute.
16910 2017-08-01 Andrew Pinski <apinski@cavium.com>
16912 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
16913 BIT_INSERT_EXPR's operand 1
16914 to see if the types precision matches.
16916 2017-08-01 Martin Liska <mliska@suse.cz>
16918 PR middle-end/70140
16919 * builtins.c (expand_builtin_memcpy_args): Remove.
16920 (expand_builtin_memcpy): Call newly added function
16921 expand_builtin_memory_copy_args.
16922 (expand_builtin_memcpy_with_bounds): Likewise.
16923 (expand_builtin_mempcpy): Remove last argument.
16924 (expand_builtin_mempcpy_with_bounds): Likewise.
16925 (expand_builtin_memory_copy_args): New function created from
16926 expand_builtin_mempcpy_args with small modifications.
16927 (expand_builtin_mempcpy_args): Remove.
16928 (expand_builtin_stpcpy): Remove unused argument.
16929 (expand_builtin): Likewise.
16930 (expand_builtin_with_bounds): Likewise.
16932 2017-08-01 Martin Liska <mliska@suse.cz>
16935 Make mempcpy more optimal (PR middle-end/70140).
16937 2017-08-01 Jakub Jelinek <jakub@redhat.com>
16940 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16941 __builtin_vec_cmpne verify both arguments are compatible vectors
16942 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
16943 verify arg1_type is a pointer or array type. For __builtin_vec_st,
16944 move computation of aligned to after checking the argument types.
16948 * config/rs6000/vsx.md (vextract_fp_from_shorth,
16949 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
16952 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
16953 Doug Rupp <rupp@adacore.com>
16954 Olivier Hainque <hainque@adacore.com>
16956 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
16957 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
16959 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
16960 for TARGET_OS_CPP_BUILTIN.
16961 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
16962 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
16964 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
16965 passing required abi options to the assembler for EABI configurations.
16966 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
16967 of .text.hot and .text.unlikely sections for kernel modules when
16968 using ARM style exceptions.
16969 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
16970 options. Add EXTRA_CC1_SPEC.
16971 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
16973 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
16975 (ARM_TARGET2_DWARF_FORMAT): Define.
16976 * config/arm/t-vxworks: Adjust multilib control to removal of the
16977 Diab command line options.
16979 2017-08-01 Martin Liska <mliska@suse.cz>
16981 PR gcov-profile/81561
16982 * gcov.c (unblock): Make unblocking safe as we need to preserve
16983 index correspondence of blocks and block_lists.
16985 2017-08-01 Richard Biener <rguenther@suse.de>
16987 PR tree-optimization/81181
16988 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
16989 (compute_antic): ... end of iteration here.
16991 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
16993 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
16994 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
16995 (ftree-slp-vectorize): Likewise.
16996 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
16997 can no longer be set independent of flag_tree_loop_vectorize.
16998 * omp-general.c (emp_max_vf): Likewise.
16999 * opts.c (enable_fdo_optimizations): Remove references to
17000 flag_tree_vectorize, these are now implicit.
17001 (common_handle_option): Remove handling for OPT_ftree_vectorize,
17002 and leave it for the options machinery.
17004 2017-08-01 Martin Liska <mliska@suse.cz>
17006 PR middle-end/70140
17007 * builtins.c (expand_builtin_memcpy_args): Remove.
17008 (expand_builtin_memcpy): Call newly added function
17009 expand_builtin_memory_copy_args.
17010 (expand_builtin_memcpy_with_bounds): Likewise.
17011 (expand_builtin_mempcpy): Remove last argument.
17012 (expand_builtin_mempcpy_with_bounds): Likewise.
17013 (expand_builtin_memory_copy_args): New function created from
17014 expand_builtin_mempcpy_args with small modifications.
17015 (expand_builtin_mempcpy_args): Remove.
17016 (expand_builtin_stpcpy): Remove unused argument.
17017 (expand_builtin): Likewise.
17018 (expand_builtin_with_bounds): Likewise.
17020 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
17023 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
17024 print "ds:" only for immediates in generic address space.
17026 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
17029 * config/i386/i386.c (ix86_funciton_naked): New prototype.
17030 (ix86_function_ok_for_sibcall): Return false for naked functions.
17032 2017-08-01 Richard Biener <rguenther@suse.de>
17034 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
17035 (compute_antic): Seed worklist with exit block predecessors.
17036 * cfganal.c (dfs_find_deadend): For a cycle return the source
17037 of the edge closing it.
17039 2017-08-01 Tamar Christina <tamar.christina@arm.com>
17041 * config/aarch64/aarch64.c
17042 (aarch64_can_const_movi_rtx_p): Move 0 check.
17044 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17046 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
17047 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
17049 * match.pd: Ditto in address comparison pattern.
17051 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17053 PR tree-optimization/81627
17054 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
17055 closed ssa form for store-store chain.
17057 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17059 PR tree-optimization/81620
17060 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
17061 for store-store chain.
17063 2017-08-01 Jakub Jelinek <jakub@redhat.com>
17065 PR tree-optimization/81588
17066 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
17067 ranges[i].in_p, invert comparison code ccode. For >/>=,
17068 swap rhs1 and rhs2 and comparison code unconditionally,
17069 for </<= don't do that. Don't swap rhs1/rhs2 again if
17070 ranges[i].in_p, instead invert comparison code ccode if
17071 opcode or oe->rank is BIT_IOR_EXPR.
17074 * optabs.def (vec_extract_optab, vec_init_optab): Change from
17075 a direct optab to conversion optab.
17076 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
17077 with GET_MODE_INNER as last argument instead of optab_handler.
17078 * expmed.c (extract_bit_field_1): Likewise. Use vector from
17079 vector extraction if possible and optab is available.
17080 * expr.c (store_constructor): Use convert_optab_handler instead
17081 of optab_handler. Use vector initialization from smaller
17082 vectors if possible and optab is available.
17083 * tree-vect-stmts.c (vectorizable_load): Likewise.
17084 * doc/md.texi (vec_extract, vec_init): Document that the optabs
17085 now have two modes.
17086 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
17087 of vec_init from half-sized vectors with the same element mode.
17088 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
17089 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
17090 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
17091 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
17092 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
17093 after mode in gen_vec_extract* calls.
17094 (vec_extract<mode>): Renamed to ...
17095 (vec_extract<mode><ssescalarmodelower>): ... this.
17096 (vec_extract<mode><ssehalfvecmodelower>): New expander.
17097 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
17098 element mode after mode in gen_vec_init* calls.
17099 (VEC_INIT_HALF_MODE): New mode iterator.
17100 (vec_init<mode>): Renamed to ...
17101 (vec_init<mode><ssescalarmodelower>): ... this.
17102 (vec_init<mode><ssehalfvecmodelower>): New expander.
17103 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
17104 (vec_extractv2sfsf): ... this.
17105 (vec_initv2sf): Renamed to ...
17106 (vec_initv2sfsf): ... this.
17107 (vec_extractv2si): Renamed to ...
17108 (vec_extractv2sisi): ... this.
17109 (vec_initv2si): Renamed to ...
17110 (vec_initv2sisi): ... this.
17111 (vec_extractv4hi): Renamed to ...
17112 (vec_extractv4hihi): ... this.
17113 (vec_initv4hi): Renamed to ...
17114 (vec_initv4hihi): ... this.
17115 (vec_extractv8qi): Renamed to ...
17116 (vec_extractv8qiqi): ... this.
17117 (vec_initv8qi): Renamed to ...
17118 (vec_initv8qiqi): ... this.
17119 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
17120 (vec_init<mode>): Renamed to ...
17121 (vec_init<mode><VEC_base_l>): ... this.
17122 (vec_extract<mode>): Renamed to ...
17123 (vec_extract<mode><VEC_base_l>): ... this.
17124 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
17125 (vec_initv2sfsf): ... this.
17126 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
17127 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
17128 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
17129 element mode after mode in gen_vec_init* calls.
17130 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
17131 (vec_init<mode><Vel>): ... this.
17132 (vec_extract<mode>): Renamed to ...
17133 (vec_extract<mode><Vel>): ... this.
17134 * config/aarch64/iterators.md (Vel): New mode attribute.
17135 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
17136 Add element mode after mode in gen_vec_extract* calls.
17137 * config/s390/vector.md (non_vec_l): New mode attribute.
17138 (vec_extract<mode>): Renamed to ...
17139 (vec_extract<mode><non_vec_l>): ... this.
17140 (vec_init<mode>): Renamed to ...
17141 (vec_init<mode><non_vec_l>): ... this.
17142 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
17143 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
17145 * config/arm/iterators.md (V_elem_l): New mode attribute.
17146 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
17147 (vec_extract<mode><V_elem_l>): ... this.
17148 (vec_extractv2di): Renamed to ...
17149 (vec_extractv2didi): ... this.
17150 (vec_init<mode>): Renamed to ...
17151 (vec_init<mode><V_elem_l>): ... this.
17152 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
17153 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
17154 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
17155 Add element mode after gen_vec_extract* calls.
17156 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
17157 (vec_init<mode><unitmode>): ... this.
17158 (vec_extract<mode>): Renamed to ...
17159 (vec_extract<mode><unitmode>): ... this.
17160 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
17161 (vec_init<mode><unitmode>): ... this.
17162 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
17163 (vec_initv2sfsf): ... this.
17164 (vec_extractv2sf): Renamed to ...
17165 (vec_extractv2sfsf): ... this.
17166 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
17167 Add element mode after gen_vec_extract* calls.
17168 * config/mips/mips.md (unitmode): New mode iterator.
17169 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
17170 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
17171 * config/spu/spu.md (inner_l): New mode attribute.
17172 (vec_init<mode>): Renamed to ...
17173 (vec_init<mode><inner_l>): ... this.
17174 (vec_extract<mode>): Renamed to ...
17175 (vec_extract<mode><inner_l>): ... this.
17176 * config/sparc/sparc.md (veltmode): New mode iterator.
17177 (vec_init<VMALL:mode>): Renamed to ...
17178 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
17179 * config/ia64/vect.md (vec_initv2si): Renamed to ...
17180 (vec_initv2sisi): ... this.
17181 (vec_initv2sf): Renamed to ...
17182 (vec_initv2sfsf): ... this.
17183 (vec_extractv2sf): Renamed to ...
17184 (vec_extractv2sfsf): ... this.
17185 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
17186 (vec_init<mode>): Renamed to ...
17187 (vec_init<mode><VEC_base_l>): ... this.
17188 (vec_extract<mode>): Renamed to ...
17189 (vec_extract<mode><VEC_base_l>): ... this.
17190 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
17191 (vec_initv2sfsf): ... this.
17192 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
17193 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
17194 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
17195 gen_vec_init* calls.
17197 2017-08-01 Richard Biener <rguenther@suse.de>
17199 PR tree-optimization/81297
17200 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
17201 TREE_OVERFLOW from INTEGER_CSTs.
17203 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
17205 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
17207 2017-07-31 Carl Love <cel@us.ibm.com>
17209 * config/rs6000/rs6000-c: Add support for built-in functions
17210 vector signed char vec_xl_be (signed long long, signed char *);
17211 vector unsigned char vec_xl_be (signed long long, unsigned char *);
17212 vector signed int vec_xl_be (signed long long, signed int *);
17213 vector unsigned int vec_xl_be (signed long long, unsigned int *);
17214 vector signed long long vec_xl_be (signed long long, signed long long *);
17215 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
17216 vector signed short vec_xl_be (signed long long, signed short *);
17217 vector unsigned short vec_xl_be (signed long long, unsigned short *);
17218 vector double vec_xl_be (signed long long, double *);
17219 vector float vec_xl_be (signed long long, float *);
17220 * config/rs6000/altivec.h (vec_xl_be): Add #define.
17221 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
17222 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
17224 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
17225 (altivec_expand_builtin): Add switch statement to call
17226 altivec_expand_xl_be for each builtin.
17227 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
17228 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
17229 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
17230 __builtin_vsx_le_be_v16qi.
17231 * doc/extend.texi: Update the built-in documentation file for the
17232 new built-in functions.
17234 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
17237 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
17239 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17241 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17243 * config.gcc: Add z14.
17244 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
17245 CPU model numbers for z13s and z14.
17246 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
17248 * config/s390/s390-opts.h (enum processor_type): Rename
17249 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
17250 * config/s390/s390.c (processor_table): Add field for CPU name to
17251 be passed to Binutils.
17252 (s390_asm_output_machine_for_arch): Use the new field in
17253 processor_table for Binutils.
17254 (s390_expand_builtin): Replace arch12 with z14.
17255 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
17256 (s390_get_sched_attrmask): Likewise.
17257 (s390_get_unit_mask): Likewise.
17258 * config/s390/s390.opt: Add z14 to processor_type enum.
17260 2017-07-31 Martin Jambor <mjambor@suse.cz>
17263 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
17264 regardless of optimization level.
17266 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
17267 Martin Liska <mliska@suse.cz>
17269 * predict.def: Remove old comment and adjust probability.
17270 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
17271 PREDICT statements.
17273 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
17276 * config/i386/i386.c (ix86_function_naked): New function.
17277 (ix86_can_use_return_insn_p): Return false for naked functions.
17278 (ix86_expand_prologue): Skip prologue for naked functions.
17279 (ix86_expand_epilogue): Skip epilogue for naked functions
17280 and emit trap instruction.
17281 (ix86_warn_func_return): New function.
17282 (ix86_attribute_table): Add "naked" attribute specification.
17283 (TARGET_WARN_FUNC_RETURN): Define.
17284 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
17286 2017-07-31 Martin Liska <mliska@suse.cz>
17288 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
17289 (dump_gimple_bb_header): Always dump BB info.
17290 (pp_cfg_jump): Do not append info about BB when dumping a jump.
17292 2017-07-31 Martin Liska <mliska@suse.cz>
17295 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
17296 also with current_function_decl non-null equality.
17298 2017-07-31 Jakub Jelinek <jakub@redhat.com>
17301 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
17302 change type to the element type, instead add eltype variable and
17303 use it where we are interested in the element type.
17305 PR tree-optimization/81603
17306 * ipa-polymorphic-call.c
17307 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
17308 offset arithmetic in offset_int, bail out if the resulting bit offset
17309 doesn't fit into shwi.
17311 2017-07-31 Martin Liska <mliska@suse.cz>
17313 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
17314 (gimplify_save_expr): Fix comment.
17316 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
17319 * config/i386/i386.c (ix86_function_arg): Update arguments for
17321 (ix86_compute_frame_layout): Set the initial stack offset to
17322 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
17323 INCOMING_FRAME_SP_OFFSET.
17324 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
17325 stack before exception handler returns.
17326 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
17327 the 'ERROR_CODE' for exception handler.
17329 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
17331 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
17332 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
17333 (ASM_OUTPUT_REG_POP): Ditto.
17334 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
17335 instead of asm_fprintf to output pure string.
17337 2017-07-29 Jakub Jelinek <jakub@redhat.com>
17339 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
17340 to imported_module_or_decl hook.
17341 (debug_nothing_tree_tree_tree_bool): Remove.
17342 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
17343 * debug.c (do_nothing_debug_hooks): Use
17344 debug_nothing_tree_tree_tree_bool_bool instead of
17345 debug_nothing_tree_tree_tree_bool.
17346 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17347 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
17348 * sdbout.c (sdb_debug_hooks): Likewise.
17349 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
17350 (gen_namespace_die): Add DW_AT_export_symbols attribute if
17352 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
17353 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
17354 attribute, don't add anything.
17356 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17358 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
17359 (fold_build2_stat_loc): Likewise.
17360 (fold_build3_stat_loc): Likewise.
17361 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
17362 (fold_build1_loc): Remove macro.
17363 (fold_build2_loc): Likewise.
17364 (fold_build3_loc): Likewise.
17366 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17368 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
17369 (gimple_build_debug_bind_source_stat): Likewise.
17370 * gimple.h (gimple_build_debug_bind): Remove macro.
17371 (gimple_build_debug_bind_source): Likewise.
17373 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17375 * bitmap.c (bitmap_alloc): Adjust.
17376 (bitmap_gc_alloc): Likewise.
17377 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
17379 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17381 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
17382 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
17383 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
17384 (bitmap_gc_alloc_stat): Likewise.
17385 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
17387 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17389 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
17390 * rtl.h (shallow_copy_rtx): Remove macro.
17392 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17394 * emit-rtl.c (gen_raw_REG): Adjust.
17395 * gengenrtl.c (gendef): Likewise.
17396 * rtl.c (rtx_alloc_stat): Remove _stat from name.
17397 * rtl.h (rtx_alloc): Remove macro.
17399 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17401 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
17402 (build_tree_list_stat): Likewise.
17403 * tree.h (build_tree_list): Remove macro.
17404 (build_tree_list_vec): Likewise.
17406 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17408 * tree.c (make_vector_stat): Remove _stat from name.
17409 (build_vector_stat): Likewise.
17410 * tree.h (make_vector_stat): Remove macro.
17411 (build_vector_stat): Likewise.
17413 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17415 * tree.h (build_var_debug_value): Remove prototype.
17417 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17419 * tree.c (tree_cons_stat): Remove _stat from name.
17420 * tree.h (tree_cons): Remove macro.
17422 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17424 * tree.c (build_vl_exp_stat): Remove _stat from name.
17425 * tree.h (build_vl_exp): Remove macro.
17427 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17429 * tree.c (build_decl_stat): Remove _stat from name.
17430 * tree.h (build_decl): Remove macro.
17432 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17434 * gimple.c (gimple_build_with_ops_stat): Adjust.
17435 (gimple_alloc_stat): Remove _stat from name.
17436 * gimple.h (gimple_alloc): Remove macro.
17438 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17440 * tree.c (make_tree_vec_stat): Remove _stat from name.
17441 (grow_tree_vec_stat): Likewise.
17442 * tree.h (make_tree_vec_stat): Adjust prototype.
17443 (grow_tree_vec_stat): Likewise.
17444 (make_tree_vec): Remove macro.
17445 (grow_tree_vec): Likewise.
17447 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17449 * fold-const.c (fold_build1_stat_loc): Adjust.
17450 (fold_build2_stat_loc): Likewise.
17451 (fold_build3_stat_loc): Likewise.
17452 * tree.c (build0_stat): Remove _stat from name.
17453 (build1_stat): Likewise.
17454 (build2_stat): Likewise.
17455 (build3_stat): Likewise.
17456 (build4_stat): Likewise.
17457 (build5_stat): Likewise.
17458 * tree.h (build1_loc): Remove macro, and rename _stat function
17460 (build2_loc): Likewise.
17461 (build3_loc): Likewise.
17462 (build4_loc): Likewise.
17463 (build5_loc): Likewise.
17465 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17467 * tree.c (make_int_cst_stat): Remove _stat from name.
17468 * tree.h (make_int_cst_stat): Adjust prototype.
17469 (make_int_cst): Remove macro.
17471 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17473 * tree.c (make_tre_binfo_stat): Remove _stat from name.
17474 * tree.h (make_tree_binfo_stat): Adjust prototype.
17475 (make_tree_binfo): Remove.
17477 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17479 * tree.c (copy_node_stat): Rename to copy_node.
17480 (build_distinct_type_copy): Adjust.
17481 * tree.h (copy_node_stat): Adjust prototype.
17482 (copy_node): Remove macro.
17484 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17486 * tree.c (make_node_stat): rename to make_node.
17487 (build_tree_list_stat): Adjust.
17488 (build0_stat): Likewise.
17489 (build2_stat): Likewise.
17490 (build3_stat): Likewise.
17491 (build4_stat): Likewise.
17492 (build5_stat): Likewise.
17493 (build_decl_stat): Likewise.
17494 * tree.h (make_node_stat): Adjust prototype.
17495 (make_node): remove macro.
17497 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
17499 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
17500 (PPC_FEATURE2_SCV): Likewise.
17501 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
17503 2017-07-28 Tamar Christina <tamar.christina@arm.com>
17505 * config/aarch64/aarch64.c
17506 (aarch64_internal_mov_immediate): Add new special pattern.
17507 * config/aarch64/aarch64.md (*movdi_aarch64):
17508 Add reg/32bit const mov case.
17510 2017-07-28 Tamar Christina <tamar.christina@arm.com>
17511 Richard Sandiford <richard.sandiford@linaro.org>
17513 * config/aarch64/aarch64.md (mov<mode>): Generalize.
17514 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
17515 Add integer and movi cases.
17516 (movi-split-hf-df-sf split, fp16): New.
17517 (enabled): Added TARGET_FP_F16INST.
17518 * config/aarch64/iterators.md (GPF_HF): New.
17519 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
17521 2017-07-28 Tamar Christina <tamar.christina@arm.com>
17523 * config/aarch64/aarch64.c
17524 (aarch64_simd_container_mode): Add prototype.
17525 (aarch64_expand_mov_immediate): Add HI support.
17526 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
17527 (aarch64_can_const_movi_rtx_p): New.
17528 (aarch64_preferred_reload_class):
17529 Remove restrictions of using FP registers for certain SIMD operations.
17530 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
17531 (aarch64_valid_floating_const): Add integer move validation.
17532 (aarch64_simd_imm_scalar_p): Remove.
17533 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
17534 (aarch64_legitimate_constant_p): Expand list of supported cases.
17535 * config/aarch64/aarch64-protos.h
17536 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
17537 (aarch64_reinterpret_float_as_int): New.
17538 (aarch64_simd_imm_scalar_p): Remove.
17539 * config/aarch64/constraints.md (Uvi): New.
17540 (Dd): Split into Ds and new Dd.
17541 * config/aarch64/aarch64.md (*movsi_aarch64):
17543 (*movdi_aarch64): Add SIMD mov case.
17545 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17547 * tree-predcom.c: (struct chain): Handle store-store chain in which
17548 stores for elimination only store loop invariant values.
17549 (execute_pred_commoning_chain): Ditto.
17550 (prepare_initializers_chain_store_elim): Ditto.
17551 (prepare_finalizers): Ditto.
17552 (is_inv_store_elimination_chain): New function.
17553 (initialize_root_vars_store_elim_1): New function.
17555 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17557 * tree-predcom.c: Revise general description of the pass.
17558 (enum chain_type): New enum type for store elimination.
17559 (struct chain): New field supporting store elimination.
17560 (struct component): Ditto.
17561 (dump_chain): Dump store-stores chain.
17562 (release_chain): Release resources.
17563 (split_data_refs_to_components): Compute and create component
17564 contains only stores for elimination.
17565 (get_chain_last_ref_at): New function.
17566 (make_invariant_chain): Initialization.
17567 (make_rooted_chain): Specify chain type in parameter and record it.
17568 (add_looparound_copies): Skip for store-stores chain.
17569 (determine_roots_comp): Compute type of chain and pass it to
17571 (initialize_root_vars_store_elim_2): New function.
17572 (finalize_eliminated_stores): New function.
17573 (remove_stmt): Handle store for elimination.
17574 (execute_pred_commoning_chain): Execute predictive commoning on
17575 store-store chains.
17576 (determine_unroll_factor): Skip unroll for store-stores chain.
17577 (prepare_initializers_chain_store_elim): New function.
17578 (prepare_initializers_chain): Hanlde store-store chain.
17579 (prepare_finalizers_chain, prepare_finalizers): New function.
17580 (tree_predictive_commoning_loop): Return integer value indicating
17581 if loop is unrolled or lcssa form is corrupted.
17582 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
17584 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17586 * tree-predcom.c (initialize_root): Delete.
17587 (execute_pred_commoning_chain): Initialize root vars and replace
17588 reference of non-combined chain directly, rather than call above
17591 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17593 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
17594 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
17596 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17598 * tree-predcom.c (struct chain): New field init_seq.
17599 (release_chain): Release init_seq.
17600 (prepare_initializers_chain): Record intialization stmts in above
17602 (insert_init_seqs): New function.
17603 (tree_predictive_commoning_loop): Call insert_init_seqs.
17605 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17607 * tree-predcom.c (determine_roots_comp): Skip trivial components.
17609 2017-07-28 Richard Biener <rguenther@suse.de>
17611 * match.pd: Remove superfluous :c.
17612 * genmatch.c (simplify::id): Add member.
17613 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
17615 (current_id): New global.
17616 (dt_node::parent): Move from ...
17617 (dt_operand::parent): ... here. Add for_id member.
17618 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
17619 (decision_tree::find_node): Relax order requirement when
17620 merging DT_TRUE nodes to ones inbetween the current simplify
17621 and the one we try to merge with. Add diagnostic whenever
17622 we need to enforce pattern order by not merging.
17623 (decision_tree::insert): Set current_id.
17624 (decision_tree::print_node): Dump parent node and for_id.
17625 (parser::last_id): Add member.
17626 (parser::push_simplify): Assign unique id.
17627 (parser::parser): Initialize last_id.
17629 2017-07-28 Martin Liska <mliska@suse.cz>
17632 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
17633 gimple_build_debug_bind.
17635 2017-07-28 Richard Biener <rguenther@suse.de>
17637 PR tree-optimization/81502
17638 * match.pd: Add pattern combining BIT_INSERT_EXPR with
17640 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
17642 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
17643 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
17644 for BIT_FIELD_REF args.
17645 * fold-const.c (make_bit_field_ref): Likewise.
17646 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
17648 2017-07-28 Jakub Jelinek <jakub@redhat.com>
17651 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
17652 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
17653 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
17654 Or it into SANITIZER_UNDEFINED.
17655 * ubsan.c: Include gimple-fold.h and varasm.h.
17656 (ubsan_expand_ptr_ifn): New function.
17657 (instrument_pointer_overflow): New function.
17658 (maybe_instrument_pointer_overflow): New function.
17659 (instrument_object_size): Formatting fix.
17660 (pass_ubsan::execute): Call instrument_pointer_overflow
17661 and maybe_instrument_pointer_overflow.
17662 * internal-fn.c (expand_UBSAN_PTR): New function.
17663 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
17664 * sanitizer.def (__ubsan_handle_pointer_overflow,
17665 __ubsan_handle_pointer_overflow_abort): New builtins.
17666 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
17667 * internal-fn.def (UBSAN_PTR): New internal function.
17668 * opts.c (sanitizer_opts): Add pointer-overflow.
17669 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
17670 * fold-const.c (build_range_check): Compute pointer range check in
17671 integral type if pointer arithmetics would be needed. Formatting
17674 2017-07-28 Martin Liska <mliska@suse.cz>
17677 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
17678 parameters that are of a variable-length.
17680 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
17682 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
17684 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
17685 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
17686 (CRT_CALL_STATIC_FUNCTION): Likewise.
17687 (ASM_DEFAULT_SPEC): New define.
17688 (ASM_SPEC32): Likewise.
17689 (ASM_SPEC64): Likewise.
17690 (ASM_SPEC_COMMON): Likewise.
17691 (ASM_SPEC): Likewise.
17692 (INVALID_64BIT): Likewise.
17693 (LINK_OS_DEFAULT_SPEC): Likewise.
17694 (LINK_OS_SPEC32): Likewise.
17695 (LINK_OS_SPEC64): Likewise.
17696 (POWERPC_LINUX): Likewise.
17697 (PTRDIFF_TYPE): Likewise.
17698 (RESTORE_FP_PREFIX): Likewise.
17699 (RESTORE_FP_SUFFIX): Likewise.
17700 (SAVE_FP_PREFIX): Likewise.
17701 (SAVE_FP_SUFFIX): Likewise.
17702 (SIZE_TYPE): Likewise.
17703 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
17704 (TARGET_64BIT): Likewise.
17705 (TARGET_64BIT): Likewise.
17706 (TARGET_AIX): Likewise.
17707 (WCHAR_TYPE_SIZE): Likewise.
17708 (WCHAR_TYPE): Undefine.
17709 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
17710 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
17711 (CPP_OS_RTEMS_SPEC): Delete.
17712 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
17713 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
17715 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
17717 2017-07-28 Jakub Jelinek <jakub@redhat.com>
17719 PR tree-optimization/81578
17720 * tree-parloops.c (build_new_reduction): Bail out if
17721 reduction_code isn't one of the standard OpenMP reductions.
17722 Move the details printing after that decision.
17724 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
17726 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
17727 related to reload_in_progress.
17728 (splat_input_operand): Likewise.
17729 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
17731 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
17733 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
17734 (TARGET_INSTANTIATE_DECLS): Likewise.
17735 (legitimate_indexed_address_p): Delete reload_in_progress code.
17736 (rs6000_debug_legitimate_address_p): Likewise.
17737 (rs6000_eliminate_indexed_memrefs): Likewise.
17738 (rs6000_emit_le_vsx_store): Likewise.
17739 (rs6000_emit_move_si_sf_subreg): Likewise.
17740 (rs6000_emit_move): Likewise.
17741 (register_to_reg_type): Likewise.
17742 (rs6000_pre_atomic_barrier): Likewise.
17743 (rs6000_machopic_legitimize_pic_address): Likewise.
17744 (rs6000_allocate_stack_temp): Likewise.
17745 (rs6000_address_for_fpconvert): Likewise.
17746 (rs6000_address_for_altivec): Likewise.
17747 (rs6000_secondary_memory_needed_rtx): Delete function.
17748 (rs6000_check_sdmode): Likewise.
17749 (rs6000_alloc_sdmode_stack_slot): Likewise.
17750 (rs6000_instantiate_decls): Likewise.
17751 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
17752 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
17753 Delete reload_in_progress.
17754 (*vec_reload_and_plus_<mptrsize>): Likewise.
17755 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
17756 (vsx_div_v2di): Likewise.
17757 (vsx_udiv_v2di): Likewise.
17759 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
17761 * config/rs6000/rs6000.opt (mlra): Replace with stub.
17762 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
17763 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
17764 (rs6000_debug_reg_global): Delete print of LRA status.
17765 (rs6000_option_override_internal): Delete dead LRA related code.
17766 (rs6000_lra_p): Delete function.
17767 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
17769 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
17771 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
17772 * config/riscv/rtems.h: New file.
17774 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17775 Sudakshina Das <sudi.das@arm.com>
17777 * config/aarch64/aarch64.md
17778 (define_split for and<mode>3nr_compare): Move
17779 non aarch64_logical_operand to a register.
17780 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
17781 register immediate operand to a register.
17782 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
17784 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
17786 PR middle-end/81564
17787 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
17789 2017-07-27 Richard Biener <rguenther@suse.de>
17791 PR tree-optimization/81573
17792 PR tree-optimization/81494
17793 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
17794 multi defuse cycle case.
17796 2017-07-27 Richard Biener <rguenther@suse.de>
17798 PR tree-optimization/81571
17799 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
17802 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
17804 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
17805 earlier and only if MASK_FPU is set. Adjust formatting.
17807 2017-07-27 Martin Liska <mliska@suse.cz>
17809 * opt-functions.awk: Add validation of value of Init.
17810 * optc-gen.awk: Pass new argument.
17812 2017-07-27 Martin Liska <mliska@suse.cz>
17814 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
17815 Fix wrong condition.
17817 2017-07-27 Martin Liska <mliska@suse.cz>
17819 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
17820 BBs and edges seen by autoFDO.
17822 2017-07-27 Richard Biener <rguenther@suse.de>
17824 PR tree-optimization/81502
17825 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
17826 with incompatible but same sized type.
17827 (execute_update_addresses_taken): Likewise.
17829 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
17831 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
17832 flag_tree_loop_vectorize rather than flag_tree_vectorize.
17834 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17837 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
17838 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
17839 Change s_operand to memory_operand.
17841 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
17843 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
17844 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
17845 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
17846 Emit instructions rather than returning an expression. Handle TFmode
17847 and KFmode by casting to TImode.
17848 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
17849 (rs6000_emit_le_vsx_store): Likewise.
17850 * config/rs6000/vsx.md (VSX_TI): New iterator.
17851 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
17852 (*vsx_le_undo_permute_<mode>): Likewise.
17853 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
17854 emit the split sequence.
17855 (*vsx_le_perm_store_<mode>): Likewise.
17857 2017-07-27 Jakub Jelinek <jakub@redhat.com>
17859 PR tree-optimization/81555
17860 PR tree-optimization/81556
17861 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
17862 if true, force CHANGED for the recursive invocation.
17863 (reassociate_bb): Remember original length of ops array, pass
17864 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
17866 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
17867 attributes for noipa attribute. For naked attribute use
17868 lookup_attribute first before lookup_attribute_spec.
17869 * final.c (rest_of_handle_final): Disable IPA RA for functions with
17871 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
17872 for functions with noipa attribute.
17873 (cgraph_externally_visible_p): Return true for functions with noipa
17875 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
17876 for functions with noipa attribute.
17877 * doc/extend.texi: Document noipa function attribute.
17878 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
17879 also for functions with noipa attribute.
17880 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
17882 2017-07-26 Andrew Pinski <apinski@cavium.com>
17884 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
17885 vec_unalign_load_cost and vec_unalign_store_cost.
17887 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
17889 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
17890 -mvsx-small-integer option.
17891 (ISA_3_0_MASKS_IEEE): Likewise.
17892 (OTHER_VSX_VECTOR_MASKS): Likewise.
17893 (POWERPC_MASKS): Likewise.
17894 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
17895 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
17896 code, only testing for DImode being allowed in non-VSX floating
17898 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
17899 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
17901 (rs6000_option_override_internal): Delete -mvsx-small-integer.
17902 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
17903 TARGET_P8_VECTOR test.
17904 (rs6000_secondary_reload_simple_move): Likewise.
17905 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
17906 since TARGET_P9_VECTOR was already tested.
17907 (rs6000_opt_masks): Remove -mvsx-small-integer.
17908 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
17909 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17911 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
17912 test for TARGET_VEXTRACTUB was used, and that uses
17914 (p9 extract splitter): Likewise.
17915 (vsx_extract_<mode>_di_p9): Likewise.
17916 (vsx_extract_<mode>_store_p9): Likewise.
17917 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
17918 for TARGET_P9_VECTOR was used. Delete code that is now dead with
17919 the elimination of TARGET_VSX_SMALL_INTEGER.
17920 (vsx_extract_<mode>_p8): Likewise.
17921 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
17922 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
17923 (vsx_set_<mode>_p9): Likewise.
17924 (vsx_set_v4sf_p9): Likewise.
17925 (vsx_set_v4sf_p9_zero): Likewise.
17926 (vsx_insert_extract_v4sf_p9): Likewise.
17927 (vsx_insert_extract_v4sf_p9_2): Likewise.
17928 * config/rs6000/rs6000.md (sign extend splitter): Change
17929 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
17930 (floatsi<mode>2_lfiwax_mem): Likewise.
17931 (floatunssi<mode>2_lfiwzx_mem): Likewise.
17932 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
17933 since a test for TARGET_P9_VECTOR was used.
17934 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17935 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
17936 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17937 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
17938 TARGET_P8_VECTOR test.
17939 (fix_trunc<mode>si2_stfiwx): Likewise.
17940 (fix_trunc<mode>si2_internal): Likewise.
17941 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
17942 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17944 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
17945 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
17946 TARGET_P8_VECTOR test.
17947 (fixuns_trunc<mode>si2_stfiwx): Likewise.
17948 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
17949 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17951 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
17952 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
17953 since a test for TARGET_P9_VECTOR was used.
17954 (splitter for loading small constants): Likewise.
17956 2017-07-26 Andrew Pinski <apinski@cavium.com>
17958 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
17961 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
17964 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
17965 (fp_valid_at): Likewise.
17967 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
17969 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
17970 (qdf24xx_addrcost_table): Likewise.
17971 (cortexa57_tunings): Update to use generic_branch_cost.
17972 (cortexa72_tunings): Likewise.
17973 (cortexa73_tunings): Likewise.
17974 (qdf24xx_tunings): Likewise.
17976 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
17978 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
17979 (thunderx2t99_branch_cost): Likewise.
17980 (cortexa35_tunings): Update to use generic_branch_cost.
17981 (cortexa53_tunings): Likewise.
17982 (cortexa57_tunings): Likewise.
17983 (cortexa72_tunings): Likewise.
17984 (cortexa73_tunings): Likewise.
17985 (thunderx2t99_tunings): Likewise.
17987 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
17989 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
17990 (sparc_option_override): Honour MASK_FSMULD.
17991 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
17992 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
17993 * config/sparc/sparc.opt (mfsmuld): New option.
17994 * doc/invoke.texi (mfsmuld): Document option.
17996 2017-07-26 Marek Polacek <polacek@redhat.com>
17998 PR middle-end/70992
17999 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
18001 2017-07-26 Richard Biener <rguenther@suse.de>
18003 * gimple-match-head.c (do_valueize): Return OP if valueize
18005 (get_def): New helper to get at the def stmt of a SSA name
18006 if valueize allows.
18007 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
18008 do_valueize to get at the def stmt.
18009 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
18011 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
18013 PR middle-end/46932
18014 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
18016 2017-07-26 Martin Liska <mliska@suse.cz>
18019 * function.c (expand_function_start): Make expansion of
18020 nonlocal_goto_save_area after parm_birth_insn.
18022 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
18024 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
18025 from all CPU target flags enable members.
18027 2017-07-26 Richard Biener <rguenther@suse.de>
18029 * genmatch.c (dt_simplify::gen): Make iterator vars const.
18030 (decision_tree::gen): Make 'type' const.
18031 (write_predicate): Likewise.
18033 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
18035 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
18036 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
18037 (rs6000_option_override_internal): Likewise.
18038 (rs6000_expand_vector_set): Likewise.
18039 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
18040 (TARGET_UPPER_REGS_SF): Likewise.
18041 (TARGET_UPPER_REGS_DI): Likewise.
18042 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
18043 (TARGET_DIRECT_MOVE_64BIT): Likewise.
18044 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
18045 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18046 (Splitters for DI constants in Altivec registers): Likewise.
18047 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
18048 (vsx_set_v4sf_p9): Likewise.
18049 (vsx_set_v4sf_p9_zero): Likewise.
18050 (vsx_insert_extract_v4sf_p9): Likewise.
18051 (vsx_insert_extract_v4sf_p9_2): Likewise.
18053 2017-07-25 Carl Love <cel@us.ibm.com>
18055 * doc/extend.texi: Update the built-in documentation file for the
18056 existing built-in functions
18057 vector signed char vec_cnttz (vector signed char);
18058 vector unsigned char vec_cnttz (vector unsigned char);
18059 vector signed short vec_cnttz (vector signed short);
18060 vector unsigned short vec_cnttz (vector unsigned short);
18061 vector signed int vec_cnttz (vector signed int);
18062 vector unsigned int vec_cnttz (vector unsigned int);
18063 vector signed long long vec_cnttz (vector signed long long);
18064 vector unsigned long long vec_cnttz (vector unsigned long long);
18066 2017-07-25 Andrew Pinski <apinski@cavium.com>
18068 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
18069 accesses where the use is for the first operand of a BIT_INSERT.
18071 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
18074 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
18075 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
18077 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
18079 * config/i386/gstabs.h: Delete.
18080 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
18082 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
18084 * config/i386/i386.c (ix86_decompose_address): Do not check for
18085 register RTX when looking at index_reg or base_reg.
18086 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
18088 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
18090 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
18091 to update EH info here.
18093 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18095 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
18097 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18099 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
18101 2017-07-25 Torsten Duwe <duwe@suse.de>
18103 * common.opt: Introduce -fpatchable-function-entry
18104 command line option, and its variables function_entry_patch_area_size
18105 and function_entry_patch_area_start.
18106 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
18107 including a two-value parser.
18108 * target.def (print_patchable_function_entry): New target hook.
18109 * targhooks.h (default_print_patchable_function_entry): New function.
18110 * targhooks.c (default_print_patchable_function_entry): Likewise.
18111 * toplev.c (process_options): Switch off IPA-RA if
18112 patchable function entries are being generated.
18113 * varasm.c (assemble_start_function): Look at the
18114 patchable-function-entry command line switch and current
18115 function attributes and maybe generate NOP instructions by
18116 calling the print_patchable_function_entry hook.
18117 * doc/extend.texi: Document patchable_function_entry attribute.
18118 * doc/invoke.texi: Document -fpatchable_function_entry
18119 command line option.
18120 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
18122 * doc/tm.texi: Re-generate.
18124 2017-07-25 Jakub Jelinek <jakub@redhat.com>
18127 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
18128 TARGET_AVX512DQ rather than TARGET_AVX512BW.
18130 2017-07-25 Tamar Christina <tamar.christina@arm.com>
18132 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
18134 2017-07-25 Richard Biener <rguenther@suse.de>
18136 PR tree-optimization/81455
18137 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
18138 not walk in cycles when looking for guards.
18140 2017-07-25 Richard Biener <rguenther@suse.de>
18142 PR tree-optimization/81529
18143 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
18144 when optimizing backedge uses.
18146 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
18148 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
18150 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
18151 to dl_section_ref. On AIX, append an expression to subtract
18152 the size of the section length to dl_section_ref.
18154 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
18156 * configure.ac: If any of the config.* scripts fail, exit 1.
18157 * configure: Regenerate.
18159 2017-07-25 Richard Biener <rguenther@suse.de>
18161 PR middle-end/81546
18162 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
18163 of immediate uses, be more verbose on errors.
18165 2017-07-25 Richard Biener <rguenther@suse.de>
18167 PR tree-optimization/81510
18168 * tree-vect-loop.c (vect_is_simple_reduction): When the
18169 reduction stmt is not inside the loop bail out.
18171 2017-07-25 Richard Biener <rguenther@suse.de>
18173 PR tree-optimization/81303
18174 * tree-vect-loop-manip.c (vect_loop_versioning): Build
18175 profitability check against LOOP_VINFO_NITERSM1.
18177 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18179 * domwalk.c (cmp_bb_postorder): Simplify.
18180 (sort_bbs_postorder): New function. Use it...
18181 (dom_walker::walk): ...here to optimize common cases.
18183 2017-07-25 Martin Liska <mliska@suse.cz>
18186 * ipa-visibility.c (function_and_variable_visibility): Make the
18187 redirection just on target that supports aliasing.
18188 Fix GNU coding style.
18190 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
18193 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
18195 * config/arm/rtems.h: Likewise.
18196 * config/bfin/rtems.h: Likewise.
18197 * config/i386/rtemself.h: Likewise.
18198 * config/lm32/rtems.h: Likewise.
18199 * config/m32c/rtems.h: Likewise.
18200 * config/m68k/rtemself.h: Likewise.
18201 * config/microblaze/rtems.h: Likewise.
18202 * config/mips/rtems.h: Likewise.
18203 * config/moxie/rtems.h: Likewise.
18204 * config/nios2/rtems.h: Likewise.
18205 * config/powerpcspe/rtems.h: Likewise.
18206 * config/rs6000/rtems.h: Likewise.
18207 * config/rtems.h: Likewise.
18208 * config/sh/rtems.h: Likewise.
18209 * config/sh/rtemself.h: Likewise.
18210 * config/sparc/rtemself.h: Likewise.
18212 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
18215 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
18216 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
18217 * tree-ssa-structalias.c (alias_get_name): Same.
18219 2017-07-25 Bin Cheng <bin.cheng@arm.com>
18222 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
18223 instructions if no du chain is found.
18225 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
18227 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
18229 2017-07-25 Richard Biener <rguenther@suse.de>
18231 PR middle-end/81505
18232 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
18235 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
18237 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
18238 upper-regs options.
18239 (ISA_2_7_MASKS_SERVER): Likewise.
18240 (ISA_3_0_MASKS_IEEE): Likewise.
18241 (OTHER_P8_VECTOR_MASKS): Likewise.
18242 (OTHER_VSX_VECTOR_MASKS): Likewise.
18243 (POWERPC_MASKS): Likewise.
18244 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
18245 duplicate list of options.
18246 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
18247 explicit -mupper-regs options.
18248 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
18249 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
18250 alias for -mupper-regs-df.
18251 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
18252 (rs6000_init_hard_regno_mode_ok): Likewise.
18253 (rs6000_option_override_internal): Likewise.
18254 (rs6000_opt_masks): Likewise.
18255 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
18256 options in terms of whether -mvsx or -mpower8-vector was used.
18257 (TARGET_UPPER_REGS_DI): Likewise.
18258 (TARGET_UPPER_REGS_SF): Likewise.
18259 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
18260 -mupper-regs-* options.
18262 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
18264 * passes.c (emergency_dump_function): Print some empty lines and a
18265 header before the RTL dump.
18267 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
18269 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
18271 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
18274 * config/aarch64/aarch64.c (aarch64_classify_symbol):
18275 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
18277 2017-07-24 Carl Love <cel@us.ibm.com>
18279 * config/rs6000/rs6000-c.c: Add support for built-in functions
18280 vector float vec_extract_fp32_from_shorth (vector unsigned short);
18281 vector float vec_extract_fp32_from_shortl (vector unsigned short);
18282 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
18283 vec_extract_fp_from_shortl): Add defines for the two builtins.
18284 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
18285 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
18287 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
18288 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
18289 * doc/extend.texi: Update the built-in documentation file for the
18290 new built-in function.
18292 2017-07-24 Jakub Jelinek <jakub@redhat.com>
18295 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
18297 * doc/generic.texi: Likewise.
18298 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
18299 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
18301 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
18303 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
18304 (aarch64_mls_elt_merge<mode>): Likewise.
18306 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
18308 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18309 having __cxa_atexit.
18311 2017-07-23 Michael Collison <michael.collison@arm.com>
18313 * config/arm/arm.c (arm_option_override): Deprecate
18314 use of -mstructure-size-boundary.
18315 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
18316 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
18318 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18321 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
18322 Reduce cost estimate for direct moves.
18324 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
18327 * config/i386/i386.c (ix86_option_override_internal): Disable
18328 BMI, BMI2 and TBM instructions for -m16.
18330 2017-07-21 Carl Love <cel@us.ibm.com>
18332 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18333 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18334 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18335 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18336 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18337 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18338 VMULOSW): New enum "unspec" values.
18339 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18340 altivec_vmulosw): New patterns.
18341 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18342 VMULOSW): Add definitions.
18344 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
18346 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
18347 (qdf24xx): Likewise.
18348 * config/aarch64/aarch64-options-extensions.def (rdma); New.
18349 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
18350 (AARCH64_FL_V8_1): Renumber.
18351 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
18352 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
18353 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
18354 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
18355 rdma to feature modifiers list.
18357 2017-07-21 Yury Gribov <tetra2005@gmail.com>
18359 PR middle-end/56727
18360 * ipa-visibility (function_and_variable_visibility): Convert
18361 recursive PLT call to direct call if appropriate.
18363 2017-07-21 Andrew Pinski <apinski@cavium.com>
18365 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
18366 operand 1 to see if the types precision matches.
18367 * fold-const.c (operand_equal_p): Likewise.
18369 2017-07-21 Richard Biener <rguenther@suse.de>
18371 PR tree-optimization/81303
18372 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
18373 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
18374 (vect_peeling_hash_get_lowest_cost): Adjust.
18375 (vect_enhance_data_refs_alignment): Likewise. Use
18376 vect_get_peeling_costs_all_drs to compute the penalty for no
18377 peeling to match up costs.
18379 2017-07-21 Richard Biener <rguenther@suse.de>
18381 PR tree-optimization/81500
18382 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
18383 we didn't identify a reduction path.
18385 2017-07-21 Tom de Vries <tom@codesourcery.com>
18386 Cesar Philippidis <cesar@codesourcery.com>
18388 PR gcov-profile/81442
18389 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
18392 2017-07-21 Tom de Vries <tom@codesourcery.com>
18395 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
18397 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
18398 nvptx_override_options_after_change.
18400 2017-07-21 Ulrich Drepper <drepper@redhat.com>
18402 * dwarf2out.c (output_file_names): Avoid double testing for
18403 dwarf_version >= 5.
18405 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
18407 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
18409 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
18411 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
18413 (try_crossjump_to_edge): Do not punt on partitioned functions.
18415 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
18417 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
18418 Put all BBs reachable only via paths crossing cold region to cold
18420 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
18422 2016-07-21 Richard Biener <rguenther@suse.de>
18424 PR tree-optimization/81303
18425 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
18426 into account prologue and epilogue iterations when raising
18427 min_profitable_iters to sth at least covering one vector iteration.
18429 2017-07-21 Tamar Christina <tamar.christina@arm.com>
18431 * config/arm/arm.c (arm_test_cpu_arch_dat):
18434 2017-07-20 Nathan Sidwell <nathan@acm.org>
18436 Remove TYPE_METHODS.
18437 * tree.h (TYPE_METHODS): Delete.
18438 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
18439 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
18440 (dbxout_type_methods): Scan TYPE_FIELDS.
18441 (dbxout_type): Don't check TYPE_METHODS here.
18442 * function.c (use_register_for_decl): Always ignore register for
18443 class types when not optimizing.
18444 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
18445 * tree.c (free_lang_data_in_type): Stitch out member functions and
18446 templates from TYPE_FIELDS.
18447 (build_distinct_type_copy, verify_type_variant,
18448 verify_type): Member fns are on TYPE_FIELDS.
18449 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
18450 * tree-pretty-print.c (dump_generic_node): Likewise.
18452 2017-07-20 Jakub Jelinek <jakub@redhat.com>
18455 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
18456 V2TImode and V4TImode.
18457 (ix86_expand_vector_extract): Likewise.
18458 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
18459 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
18460 (ssescalarmode): Handle V4TImode and V2TImode.
18461 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
18462 (*vec_extractv2ti, *vec_extractv4ti): New insns.
18463 (VEXTRACTI128_MODE): New mode iterator.
18464 (splitter for *vec_extractv?ti first element): New.
18465 (VEC_INIT_MODE): New mode iterator.
18466 (vec_init<mode>): Consolidate 3 expanders into one using
18467 VEC_INIT_MODE mode iterator.
18469 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
18471 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
18472 non_spilled_static_chain_regno_p.
18474 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
18476 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
18478 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
18480 * bb-reorder.c (connect_traces): Allow copying of blocks within
18483 2017-07-20 Richard Biener <rguenther@suse.de>
18485 * gimple.h (gimple_phi_result): Add gphi * overload.
18486 (gimple_phi_result_ptr): Likewise.
18487 (gimple_phi_arg): Likewise. Adjust index assert to only
18488 allow actual argument accesses rather than all slots available
18490 (gimple_phi_arg_def): Add gphi * overload.
18491 * tree-phinodes.c (make_phi_node): Initialize only actual
18493 (resize_phi_node): Clear memory not covered by old node,
18494 do not initialize excess argument slots.
18495 (reserve_phi_args_for_new_edge): Initialize new argument slot
18498 2017-07-20 Bin Cheng <bin.cheng@arm.com>
18500 PR tree-optimization/81388
18502 2016-07-21 Bin Cheng <bin.cheng@arm.com>
18504 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
18505 by removing computation of may_be_zero.
18507 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18508 Tom de Vries <tom@codesourcery.com>
18510 PR middle-end/81030
18511 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
18512 when gimple level profile disagrees with what RTL expander did.
18514 2017-07-20 Richard Biener <rguenther@suse.de>
18516 PR tree-optimization/61171
18517 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
18518 (vect_analyze_stmt): Add slp instance parameter.
18519 (vectorizable_reduction): Likewise.
18520 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
18521 (vect_is_simple_reduction): Deal with chains not detected
18522 as SLP reduction chain, specifically not properly associated
18523 chains containing a mix of plus/minus.
18524 (get_reduction_op): Remove.
18525 (get_initial_defs_for_reduction): Simplify, pass in whether
18526 this is a reduction chain, pass in the SLP node for the PHIs.
18527 (vect_create_epilog_for_reduction): Get the SLP instance as
18529 (vectorizable_reduction): Get the SLP instance as arg.
18530 During analysis remember the SLP node with the PHIs in the
18531 instance. Simplify getting at the vectorized reduction PHIs.
18532 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
18533 through SLP instance.
18534 (vect_slp_analyze_operations): Likewise.
18535 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
18536 (vect_transform_stmt): Likewise.
18538 2017-07-20 Tom de Vries <tom@codesourcery.com>
18540 PR tree-optimization/81489
18541 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
18542 read of phi arg location to before loop that modifies phi.
18544 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18546 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
18549 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18551 PR middle-end/81331
18552 * except.c (execute): Fix ordering issue.
18554 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
18556 PR rtl-optimization/81423
18557 * combine.c (make_compound_operation_int): Don't try to optimize
18558 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
18560 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
18562 PR rtl-optimization/81423
18563 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
18564 with a constant that is -1 in the truncated to mode.
18566 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18568 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
18569 (determine_unlikely_bbs): ... here.
18570 * predict.h (propagate_unlikely_bbs_forward): Declare.
18571 * cfgexpand.c (pass_expand::execute): Use it.
18572 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
18574 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
18575 propagate_unlikely_bbs_forward.
18577 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18579 PR middle-end/81331
18580 * except.c (maybe_add_nop_after_section_switch): New function.
18583 2017-07-19 Tom de Vries <tom@codesourcery.com>
18585 * gimple.h (gimple_phi_set_arg): Make assert more strict.
18587 2017-07-19 Tom de Vries <tom@codesourcery.com>
18589 * gimple.h (gimple_phi_arg): Make assert more strict.
18591 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
18593 * config.gcc (powerpc*-*-*): Add mmintrin.h.
18594 * config/rs6000/mmintrin.h: New file.
18595 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
18597 2017-07-19 Jakub Jelinek <jakub@redhat.com>
18599 PR tree-optimization/81346
18600 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
18602 2017-07-19 Tom de Vries <tom@codesourcery.com>
18604 * config/nvptx/nvptx.md (VECIM): Add V2DI.
18606 2017-07-19 Tom de Vries <tom@codesourcery.com>
18608 * config/nvptx/nvptx-modes.def: Add V2DImode.
18609 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
18610 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
18611 (nvptx_output_mov_insn): Handle lack of mov.b128.
18612 (nvptx_print_operand): Handle 'H' and 'L' codes.
18613 (nvptx_vector_mode_supported): Allow V2DImode.
18614 (nvptx_preferred_simd_mode): New function.
18615 (nvptx_data_alignment): New function.
18616 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
18617 nvptx_preferred_simd_mode.
18618 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
18620 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
18622 2017-07-19 Tom de Vries <tom@codesourcery.com>
18624 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
18625 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
18626 (nvptx_vector_mode_supported): New function. Allow V2SImode.
18627 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
18628 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
18629 (mov<VECIM>_insn): New define_insn.
18630 (define_expand "mov<VECIM>): New define_expand.
18632 2017-07-19 Tom de Vries <tom@codesourcery.com>
18634 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
18636 2017-07-19 Jakub Jelinek <jakub@redhat.com>
18638 PR tree-optimization/81346
18639 * fold-const.h (fold_div_compare, range_check_type): Declare.
18640 * fold-const.c (range_check_type): New function.
18641 (build_range_check): Use range_check_type.
18642 (fold_div_compare): No longer static, rewritten into
18643 a match.pd helper function.
18644 (fold_comparison): Don't call fold_div_compare here.
18645 * match.pd (X / C1 op C2): New optimization using fold_div_compare
18646 as helper function.
18648 2017-07-19 Nathan Sidwell <nathan@acm.org>
18650 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
18651 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
18652 * tree.c (find_decls_types_r, verify_type): Use
18653 TYPE_{MIN,MAX}_VALUE_RAW.
18654 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
18655 (hash_tree): Likewise.
18656 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
18658 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
18661 2017-07-18 Tom de Vries <tom@codesourcery.com>
18663 PR middle-end/81464
18664 * omp-expand.c (expand_omp_for_static_chunk): Handle
18665 equal-argument loop exit phi.
18667 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
18670 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
18671 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
18672 operand 2 predicate.
18673 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
18674 operand 2 predicate.
18675 (ror,rol -> rorx splitters): Use const_int_operand as
18676 operand 2 predicate.
18678 2017-06-18 Richard Biener <rguenther@suse.de>
18680 PR tree-optimization/81410
18681 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
18682 the gap in the ! slp_perm SLP case after each group.
18684 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18686 PR middle-end/81463
18687 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
18690 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18692 PR middle-end/81462
18693 * predict.c (set_even_probabilities): Cleanup; do not affect
18694 probabilities that are already known.
18695 (combine_predictions_for_bb): Call even when count is set.
18697 2017-07-18 Nathan Sidwell <nathan@acm.org>
18699 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
18702 2017-07-18 Bin Cheng <bin.cheng@arm.com>
18705 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
18706 optimization for loop niter analysis.
18708 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
18711 * config/avr/avr.c (avr_optimize_casesi): Don't use
18712 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
18714 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
18716 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
18717 body_cost_vec from _vect_peel_extended_info.
18718 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
18719 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
18722 2017-07-18 Bin Cheng <bin.cheng@arm.com>
18724 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
18726 2017-07-18 Richard Biener <rguenther@suse.de>
18728 PR tree-optimization/80620
18729 PR tree-optimization/81403
18730 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
18731 info when re-using a VN table entry.
18733 2017-07-18 Richard Biener <rguenther@suse.de>
18735 PR tree-optimization/81418
18736 * tree-vect-loop.c (vectorizable_reduction): Properly compute
18737 vectype_in. Verify that with lane-reducing reduction operations
18738 we have a single def-use cycle.
18740 2017-07-17 Carl Love <cel@us.ibm.com>
18742 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
18744 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18745 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18746 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18747 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18748 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18749 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18750 VMULOSW): New enum "unspec" values.
18751 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
18752 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
18753 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18754 altivec_vmulosw): New patterns.
18755 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18756 VMULOSW): Add definitions.
18757 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
18759 * config/alpha/alpha.c: Include predict.h.
18761 2017-07-17 Yury Gribov <tetra2005@gmail.com>
18763 * tree-vrp.c (compare_assert_loc): Fix comparison function
18764 to return predictable results.
18766 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18768 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
18770 (subdi3): Likewise.
18771 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
18772 * doc/invoke.texi (mexpand-adddi): Update text.
18774 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18776 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
18777 that also clobbers the CC register. The old expand code is moved
18779 (*arc_clzsi2): ... here.
18780 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
18781 the CC register. The old expand code is moved to ...
18782 (arc_ctzsi2): ... here.
18784 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18786 * config/arc/arc.opt (mindexed-loads): Use initial value
18787 TARGET_INDEXED_LOADS_DEFAULT.
18788 (mauto-modify-reg): Use initial value
18789 TARGET_AUTO_MODIFY_REG_DEFAULT.
18790 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
18791 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
18792 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
18793 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
18795 2017-07-17 Martin Liska <mliska@suse.cz>
18798 * opts.c (finish_options): Do not allow -fgnu-tm
18799 w/ -fsanitize={kernel-,}address. Say sorry.
18801 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18804 * tree-loop-distribution.c (classify_partition): Only assert on
18805 numer of iterations.
18806 (merge_dep_scc_partitions): Delete prameter. Update function call.
18807 (distribute_loop): Remove code handling loop with unknown niters.
18808 (pass_loop_distribution::execute): Skip loop with unknown niters.
18810 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18813 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
18814 function sort_partitions_by_post_order.
18816 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18818 PR tree-optimization/81374
18819 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
18820 the max index of basic blocks, rather than number of basic blocks.
18822 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18824 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
18826 (arc_legitimate_pic_operand_p): Likewise.
18827 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
18829 (arc_needs_pcl_p): Likewise.
18830 (arc_legitimate_pc_offset_p): Likewise.
18831 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
18832 function is also used in constrains.md.
18833 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
18834 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
18835 PLUS. Only return true/false in known cases, otherwise assert.
18836 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
18837 is already called in arc_legitimate_constant_p.
18838 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
18840 (LEGITIMATE_PIC_OPERAND_P): Use
18841 arc_raw_symbolic_reference_mentioned_p function.
18842 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
18847 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18848 Andrew Burgess <andrew.burgess@embecosm.com>
18850 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
18851 (arc_return_address_register): New function.
18852 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
18853 (arc_handle_fndecl_attribute): Add naked attribute.
18854 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
18855 (TARGET_WARN_FUNC_RETURN): Likewise.
18856 (arc_allocate_stack_slots_for_args): New function.
18857 (arc_warn_func_return): Likewise.
18858 (machine_function): Change type fn_type.
18859 (arc_compute_function_type): Consider new naked function type,
18860 change function return type.
18861 (arc_must_save_register): Adapt to handle new
18862 arc_compute_function_type's return type.
18863 (arc_expand_prologue): Likewise.
18864 (arc_expand_epilogue): Likewise.
18865 (arc_return_address_regs): Delete.
18866 (arc_return_address_register): New function.
18867 (arc_epilogue_uses): Use above function.
18868 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
18869 (arc_function_type): Change encoding, add naked type.
18870 (ARC_INTERRUPT_P): Change to handle the new encoding.
18871 (ARC_FAST_INTERRUPT_P): Likewise.
18872 (ARC_NORMAL_P): Define.
18873 (ARC_NAKED_P): Likewise.
18874 (arc_compute_function_type): Delete prototype.
18875 * config/arc/arc.md (in_ret_delay_slot): Use
18876 arc_return_address_register function.
18877 (simple_return): Likewise.
18878 (p_return_i): Likewise.
18880 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18882 PR tree-optimization/81428
18883 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
18884 can't be built for those types.
18886 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
18888 Remove stuff dead since r239246.
18890 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
18891 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
18892 (avr_inform_devices): Remove dead stuff.
18894 2017-07-17 Tamar Christina <tamar.christina@arm.com>
18896 * config/arm/arm_neon.h: Fix softp typo.
18898 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18900 PR tree-optimization/81365
18901 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
18902 aggregate moves onto bb predecessor edges, make sure there are no
18903 loads that could alias the lhs in between the start of bb and the
18906 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
18909 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
18910 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
18911 [LSHIFTRT, outer_code = TRUNCATE]: Same.
18913 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18915 PR tree-optimization/81396
18916 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
18917 (init_symbolic_number): Initialize it to 1.
18918 (perform_symbolic_merge): Add n_ops from both operands into the new
18920 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
18921 without base_addr as useless if they need more than one operation.
18922 (bswap_replace): Handle !bswap case for NULL base_addr.
18924 2017-07-17 Tom de Vries <tom@codesourcery.com>
18927 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
18930 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
18932 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
18933 conditional builtin define __FIX_LEON3FT_B2BST.
18935 2017-07-17 Daniel Cederman <cederman@gaisler.com>
18937 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
18938 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
18941 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
18943 PR rtl-optimization/81424
18944 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
18945 to remove potential trapping from operands if -fnon-call-exceptions.
18947 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18949 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
18950 profile_proability for scalling.
18951 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
18953 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18955 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
18957 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18959 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
18960 fixpoint arithmetics.
18962 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18964 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
18965 fixpoint arithmetics.
18967 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18969 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
18970 fixpoint arithmetics.
18972 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18974 * profile-count.h (profile_probability::from_reg_br_prob_note,
18975 profile_probability::to_reg_br_prob_note): New functions.
18976 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
18977 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
18978 * predict.c (probability_reliable_p): Update.
18979 (edge_probability_reliable_p): Update.
18980 (br_prob_note_reliable_p): Update.
18981 (invert_br_probabilities): Update.
18982 (add_reg_br_prob_note): New function.
18983 (combine_predictions_for_insn): Update.
18984 * asan.c (asan_clear_shadow): Update.
18985 * cfgbuild.c (compute_outgoing_frequencies): Update.
18986 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
18987 (update_br_prob_note): Update.
18988 (rtl_verify_edges): Update.
18989 (purge_dead_edges): Update.
18990 (fixup_reorder_chain): Update.
18991 * emit-rtl.c (try_split): Update.
18992 * ifcvt.c (cond_exec_process_insns): Update.
18993 (cond_exec_process_if_block): Update.
18994 (dead_or_predicable): Update.
18995 * internal-fn.c (expand_addsub_overflow): Update.
18996 (expand_neg_overflow): Update.
18997 (expand_mul_overflow): Update.
18998 * loop-doloop.c (doloop_modify): Update.
18999 * loop-unroll.c (compare_and_jump_seq): Update.
19000 * optabs.c (emit_cmp_and_jump_insn_1): Update.
19001 * predict.h: Update.
19002 * reorg.c (mostly_true_jump): Update.
19004 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
19005 * config/alpha/alpha.c (emit_unlikely_jump): Update.
19006 * config/arc/arc.c: (emit_unlikely_jump): Update.
19007 * config/arm/arm.c: (emit_unlikely_jump): Update.
19008 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
19009 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
19010 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
19011 (ix86_print_operand): Update.
19012 (ix86_split_fp_branch): Update.
19013 (predict_jump): Update.
19014 * config/ia64/ia64.c (ia64_print_operand): Update.
19015 * config/mmix/mmix.c (mmix_print_operand): Update.
19016 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
19017 (rs6000_expand_split_stack_prologue): Update.
19018 * config/rs6000/rs6000.c: Update.
19019 * config/s390/s390.c (s390_expand_vec_strlen): Update.
19020 (s390_expand_vec_movstr): Update.
19021 (s390_expand_cs_tdsi): Update.
19022 (s390_expand_split_stack_prologue): Update.
19023 * config/sh/sh.c (sh_print_operand): Update.
19024 (expand_cbranchsi4): Update.
19025 (expand_cbranchdi4): Update.
19026 * config/sparc/sparc.c (output_v9branch): Update.
19027 * config/spu/spu.c (get_branch_target): Update.
19028 (ea_load_store_inline): Update.
19029 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
19030 * config/tilepro/tilepro.c: Update.
19032 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
19034 * gimplify.c (mostly_copy_tree_r): Revert latest change.
19035 (gimplify_save_expr): Likewise.
19037 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19039 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
19041 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19043 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
19045 * timevar.def (TV_IPA_FNSUMMARY): Define.
19047 2017-07-16 Daniel Cederman <cederman@gaisler.com>
19049 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
19050 to back store errata sensitive sequence from being generated.
19051 (sqrtdf2_fix): Likewise.
19053 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19055 * tree-ssa-threadupdate.c (compute_path_counts,
19056 update_joiner_offpath_counts): Use profile_probability.
19058 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
19061 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19063 * config/arm/arm-c.c (arm_cpu_builtins): Define
19064 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
19066 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
19068 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19069 array entries to represent __ieee128 versions of the
19070 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
19071 scalar_extract_sig, and scalar_insert_exp built-in functions.
19072 (altivec_resolve_overloaded_builtin): Add special case handling
19073 for the __builtin_scalar_insert_exp function, as represented by
19074 the P9V_BUILTIN_VEC_VSIEDP constant.
19075 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
19076 exponent support for __ieee128 argument.
19077 (VSESQP): Add scalar extract signature support for __ieee128
19079 (VSTDCNQP): Add scalar test negative support for __ieee128
19081 (VSIEQP): Add scalar insert exponent support for __int128 argument
19082 with __ieee128 result.
19083 (VSIEQPF): Add scalar insert exponent support for __ieee128
19084 argument with __ieee128 result.
19085 (VSTDCQP): Add scalar test data class support for __ieee128
19087 (VSTDCNQP): Add overload support for scalar test negative with
19088 __ieee128 argument.
19089 (VSTDCQP): Add overload support for scalar test data class
19090 __ieee128 argument.
19091 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
19092 UNSPEC_VSX_SXSIGDP.
19093 (UNSPEC_VSX_SIEXPQP): New constant.
19094 (xsxexpqp): New insn for VSX scalar extract exponent quad
19096 (xsxsigqp): New insn for VSX scalar extract significand quad
19098 (xsiexpqpf): New insn for VSX scalar insert exponent quad
19099 precision with floating point argument.
19100 (xststdcqp): New expand for VSX scalar test data class quad
19102 (xststdcnegqp): New expand for VSX scalar test negative quad
19104 (xststdcqp): New insn to match expansions for VSX scalar test data
19105 class quad precision and VSX scalar test negative quad precision.
19106 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
19107 special case operand checking to enforce that second operand of
19108 VSX scalar test data class with quad precision argument is a 7-bit
19110 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19111 prototypes and descriptions of __ieee128 versions of
19112 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
19113 scalar_test_data_class, and scalar_test_neg built-in functions.
19115 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19117 PR tree-optimization/81162
19118 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
19119 replace a negate with an add.
19121 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
19123 * doc/invoke.texi (arm/-mcpu): Document +crypto.
19125 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19127 * config/arm/arm-c.c (arm_cpu_builtins): Define
19128 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
19130 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19132 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
19133 (armv8-r): Set ARM Cortex-R52 as default CPU.
19134 * config/arm/arm-tables.opt: Regenerate.
19135 * config/arm/arm-tune.md: Regenerate.
19136 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19138 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
19139 extension for -mcpu=cortex-r52.
19141 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19143 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
19144 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
19145 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
19146 (fp-armv8): Define it as FP_ARMv8 only.
19147 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
19148 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
19150 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
19151 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
19152 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
19153 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
19154 than TARGET_FPU_ARMV8.
19155 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
19156 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
19157 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
19159 * config/arm/neon.md (neon_vrint): Likewise.
19160 (neon_vcvt): Likewise.
19161 (neon_<fmaxmin_op><mode>): Likewise.
19162 (<fmaxmin><mode>3): Likewise.
19163 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
19164 * config/arm/predicates.md (arm_cond_move_operator): Check against
19165 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
19167 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
19169 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
19170 to top of function.
19172 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19174 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
19175 loop in comment with memset.
19177 2017-07-14 Martin Liska <mliska@suse.cz>
19179 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
19180 * dwarf2out.c (is_java): Remove the function.
19181 (output_pubname): Remove usage of the function.
19182 (lower_bound_default): Remove usage of DW_LANG_Java.
19183 (gen_compile_unit_die): Likewise.
19184 * gcc.c: Remove compiler defaults for .java and .zip files.
19185 * gimple-expr.c (remove_suffix): Change as there's no longer
19186 extension than 4-letter one.
19187 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
19188 (gimplify_save_expr): Likewise.
19189 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
19190 as it's possible even for other languages than Java.
19191 * langhooks.h (struct lang_hooks): Remove Java from a comment.
19192 * lto-opts.c (lto_write_options): Remove reference to Java.
19193 * opts.c (strip_off_ending): Update file extension handling.
19194 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
19195 * tree-eh.c (lower_resx): Likewise.
19196 * tree.c (free_lang_data_in_type): Remove dead code.
19197 (find_decls_types_r): Likewise.
19198 (build_common_builtin_nodes): Remove Java from a comment.
19199 (verify_type): Remove dead code.
19200 * varasm.c (assemble_external): Remove Java from a comment.
19202 2017-07-14 Martin Liska <mliska@suse.cz>
19204 * opts.c (finish_options): Add quotes.
19205 (common_handle_option): Likewise.
19207 2017-07-14 Martin Liska <mliska@suse.cz>
19209 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
19210 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
19211 Remove N_SO_PASCAL.
19212 * dwarf2out.c (lower_bound_default): Do not handle
19214 (gen_compile_unit_die): Likewise.
19215 * gcc.c: Remove default extension binding for GNU Pascal.
19216 * stmt.c: Remove Pascal language from a comment.
19217 * xcoffout.c: Likewise.
19219 2017-07-13 David Malcolm <dmalcolm@redhat.com>
19222 * diagnostic-show-locus.c (fixit_cmp): New function.
19223 (layout::layout): Sort m_fixit_hints.
19224 (column_range::column_range): Assert that the values are valid.
19225 (struct char_span): New struct.
19226 (correction::overwrite): New method.
19227 (struct source_line): New struct.
19228 (line_corrections::add_hint): Add assertions. Reimplement memcpy
19229 calls in terms of classes source_line and char_span, and
19230 correction::overwrite.
19231 (selftest::test_overlapped_fixit_printing_2): New function.
19232 (selftest::diagnostic_show_locus_c_tests): Call it.
19234 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
19236 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
19237 early if there is no lhs.
19239 2017-07-13 Martin Liska <mliska@suse.cz>
19241 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
19242 (gen_reference_type_die): Likewise.
19243 * stor-layout.c: Remove Pascal-related comment.
19245 2017-07-13 Martin Liska <mliska@suse.cz>
19247 * opts.c (finish_options): Add quotes to error messages.
19248 (parse_sanitizer_options): Likewise.
19250 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
19252 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
19254 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
19256 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
19258 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
19260 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
19262 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
19264 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
19267 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
19268 provides the hardware capability bits, define the macro
19269 __BUILTIN_CPU_SUPPORTS__.
19270 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
19271 if GLIBC does not provide the hardware capability bits. Add a
19272 gcc_unreachable call if the built-in cpu function is neither
19273 __builtin_cpu_is nor __builtin_cpu_supports.
19274 (rs6000_get_function_versions_dispatcher): Change the warning
19275 that an old GLIBC is used which does not export the capability
19276 bits to be an error.
19277 * doc/extend.texi (target_clones attribute): Document the
19278 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
19279 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
19280 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
19281 the macros defined by GCC if the newer GLIBC is available.
19283 2017-07-12 Jeff Law <law@redhat.com>
19285 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
19286 remaining includes slightly.
19287 * config/riscv/riscv-builtins.c: Include profile-count.h.
19289 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
19292 * config/avr/avr.c (avr_set_current_function): In diagnostic
19293 messages: Quote keywords and (parts of) identifiers.
19294 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
19297 2017-07-12 Carl Love <cel@us.ibm.com>
19299 * config/rs6000/rs6000-c.c: Add support for built-in functions
19300 vector bool char vec_revb (vector bool char);
19301 vector bool short vec_revb (vector short char);
19302 vector bool int vec_revb (vector bool int);
19303 vector bool long long vec_revb (vector bool long long);
19304 * doc/extend.texi: Update the built-in documentation file for the
19305 new built-in functions.
19307 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19309 * config/s390/s390.md: Remove movcc splitter.
19311 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19313 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
19314 load/store on condition.
19316 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
19319 * config/avr/avr.c (avr_encode_section_info)
19320 [progmem && !TREE_READONLY]: Error if progmem object needs
19323 2017-07-11 Michael Collison <michael.collison@arm.com>
19325 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
19328 2017-07-11 Carl Love <cel@us.ibm.com>
19330 * config/rs6000/rs6000-c.c: Add support for builtins
19331 vector unsigned int vec_parity_lsbb (vector signed int);
19332 vector unsigned int vec_parity_lsbb (vector unsigned int);
19333 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
19334 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
19335 vector unsigned long long vec_parity_lsbb (vector signed long long);
19336 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
19337 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
19338 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
19339 * doc/extend.texi: Update the built-in documentation file for the
19340 new built-in functions.
19342 2017-07-11 David Malcolm <dmalcolm@redhat.com>
19344 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
19345 (layout::m_primary_loc): New field.
19346 (layout::layout): Initialize new field. Move location filtering
19347 logic from here to...
19348 (layout::maybe_add_location_range): ...this new method. Add
19349 support for filtering to just the lines already specified by other
19351 (layout::will_show_line_p): New method.
19352 (gcc_rich_location::add_location_if_nearby): New method.
19353 (selftest::test_add_location_if_nearby): New test function.
19354 (selftest::diagnostic_show_locus_c_tests): Call it.
19355 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
19358 2017-07-11 Tom de Vries <tom@codesourcery.com>
19360 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
19361 (bb_first_real_insn): New function.
19362 (nvptx_single): Add extra initialization of broadcasted condition
19365 2017-07-11 Nathan Sidwell <nathan@acm.org>
19367 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
19369 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
19371 * doc/extend.texi (AVR Function Attributes): Remove weblink to
19372 Binutils doc as TEXI will mess them up.
19373 * doc/invoke.texi (AVR Options): Same here.
19375 2017-07-11 Daniel Cederman <cederman@gaisler.com>
19377 * config/sparc/sparc.opt (mfix-ut700): New option.
19378 (mfix-gr712rc): Likewise.
19379 (sparc_fix_b2bst): New variable.
19380 * doc/invoke.texi (SPARC options): Document them.
19381 (ARM options): Fix warnings.
19382 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
19383 instructions to prevent sequences that can trigger the store-store
19384 errata for certain LEON3FT processors.
19385 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
19386 (sparc_option_override): Set sparc_fix_b2bst appropriately.
19387 * config/sparc/sparc.md (fix_b2bst): New attribute.
19388 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
19390 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
19393 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
19395 (*rsqrtsf2_sse): Ditto.
19397 (div<mode>3): Macroize insn from divdf3 and divsf3
19398 using MODEF mode iterator.
19400 2017-07-10 Martin Sebor <msebor@redhat.com>
19402 PR tree-optimization/80397
19403 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
19404 instead of testing for equality to INTEGER_TYPE.
19406 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
19408 * config.gcc: Remove uclibc from arc target spec.
19410 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
19412 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
19414 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19417 * lto-wrapper.c (remove_option): New function.
19418 (merge_and_complain): Merge PIC/PIE options more realistically.
19420 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
19422 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
19426 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
19427 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
19428 * config.in: Regenerate.
19429 * configure: Regenerate.
19430 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
19431 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
19432 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
19433 (TARGET_GASISR_PROLOGUES): ...target mask.
19434 * common/config/avr/avr-common.c
19435 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
19436 Set -mgas-isr-prologues.
19437 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
19438 INSERT_PASS_BEFORE for it.
19439 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
19440 * config/avr/avr.c (avr_option_override)
19441 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
19442 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
19443 (avr_attribute_table) <no_gccisr>: Add new function attribute.
19444 (avr_set_current_function) <is_no_gccisr>: Init machine field.
19445 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
19447 (make_avr_pass_pre_proep): New function.
19448 (emit_push_sfr) <treg>: Add argument to function and use it
19449 instead of TMP_REG.
19450 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
19451 and set machine->gasisr.yes.
19452 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
19453 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
19454 __gcc_isr.n_pushed to .L__stack_usage.
19455 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
19456 (avr_asm_final_postscan_insn): ...this new static function.
19457 * config/avr/avr.h (machine_function)
19458 <is_no_gccisr, use_L__stack_usage>: New fields.
19459 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
19460 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
19461 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
19462 (gasisr, *gasisr): New expander and insn.
19463 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
19464 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
19465 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
19467 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
19469 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
19472 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
19474 Move jump-tables out of .text again.
19477 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
19478 (ASM_OUTPUT_ADDR_VEC): New function.
19479 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
19480 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
19481 INSN_ADDRESSes as asm comment.
19482 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
19483 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
19484 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
19485 * config/avr/avr.md (*tablejump): Adjust comment.
19486 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
19487 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
19489 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
19490 (avr_output_addr_vec): New proto.
19491 (avr_log_t) <insn_addresses>: New field.
19493 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
19496 * config/i386/i386.c (ix86_function_arg_advance): Set
19497 outgoing_args_on_stack to true if there are outgoing arguments
19499 (ix86_function_arg): Likewise.
19500 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
19501 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
19502 * config/i386/i386.h (machine_function): Add
19503 outgoing_args_on_stack.
19505 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
19507 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
19508 supporting pthreds.
19509 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
19511 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
19513 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
19514 (REAL_H): Remove $(MACHMODE_H).
19515 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
19517 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
19518 $(MACHMODE_H) and double-int.h.
19519 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
19521 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
19524 2017-07-07 Andrew Pinski <apinski@cavium.com>
19526 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
19527 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
19529 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
19531 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
19532 Add warning if GCC was not configured to link against a GLIBC that
19533 exports the hardware capability bits.
19534 (make_resolver_func): Make resolver function private and not a
19535 COMDAT function. Create the name with clone_function_name instead
19536 of make_unique_name.
19539 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
19540 correct operand in doing the split.
19542 2017-07-07 Carl Love <cel@us.ibm.com>
19544 * config/rs6000/rs6000-c: Add support for built-in function
19545 vector unsigned short vec_pack_to_short_fp32 (vector float,
19547 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
19548 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
19549 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
19550 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
19551 (convert_4f32_8i16): Add define_expand.
19552 * doc/extend.texi: Update the built-in documentation file for the
19553 new built-in function.
19555 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19557 * config/sparc/m8.md: New file.
19558 * config/sparc/sparc.md: Include m8.md.
19560 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19562 * config/sparc/sparc.opt: New option -mvis4b.
19563 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
19564 (sparc_option_override): Handle VIS4B.
19565 (enum sparc_builtins): Define
19566 SPARC_BUILTIN_DICTUNPACK{8,16,32},
19567 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
19568 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
19569 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
19570 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
19571 (check_constant_argument): New function.
19572 (sparc_vis_init_builtins): Define builtins
19573 __builtin_vis_dictunpack{8,16,32},
19574 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
19575 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
19576 __builtin_vis_fpcmpde{8,16,32}shl and
19577 __builtin_vis_fpcmpur{8,16,32}shl.
19578 (sparc_expand_builtin): Check that the constant operands to
19579 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
19580 constant and in range.
19581 * config/sparc/sparc-c.c (sparc_target_macros): Handle
19583 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
19584 (SPARC_IMM5_P): Likewise.
19585 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
19586 (enabled): Handle vis4b.
19587 (UNSPEC_DICTUNPACK): New unspec.
19588 (UNSPEC_FPCMPSHL): Likewise.
19589 (UNSPEC_FPUCMPSHL): Likewise.
19590 (UNSPEC_FPCMPDESHL): Likewise.
19591 (UNSPEC_FPCMPURSHL): Likewise.
19592 (cpu_feature): New CPU feature `vis4b'.
19593 (dictunpack{8,16,32}): New insns.
19594 (FPCSMODE): New mode iterator.
19595 (fpcscond): New code iterator.
19596 (fpcsucond): Likewise.
19597 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
19598 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
19599 (fpcmpde{8,16,32}{si,di}shl): Likewise.
19600 (fpcmpur{8,16,32}{si,di}shl): Likewise.
19601 * config/sparc/constraints.md: Define constraints `q' for unsigned
19602 2-bit integer constants and `t' for unsigned 5-bit integer
19604 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
19606 (imm5_operand_dictunpack16): Likewise.
19607 (imm5_operand_dictunpack32): Likewise.
19608 (imm2_operand): Likewise.
19609 * doc/invoke.texi (SPARC Options): Document -mvis4b.
19610 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
19611 ditunpack* and fpcmp*shl builtins.
19613 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19615 * config.gcc: Handle m8 in --with-{cpu,tune} options.
19616 * config.in: Add HAVE_AS_SPARC6 define.
19617 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
19619 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
19621 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
19622 (CPP_CPU_SPEC): Handle m8.
19623 (ASM_CPU_SPEC): Likewise.
19624 * config/sparc/sparc-opts.h (enum processor_type): Add
19626 * config/sparc/sparc.c (m8_costs): New struct.
19627 (sparc_option_override): Handle TARGET_CPU_m8.
19628 (sparc32_initialize_trampoline): Likewise.
19629 (sparc64_initialize_trampoline): Likewise.
19630 (sparc_issue_rate): Likewise.
19631 (sparc_register_move_cost): Likewise.
19632 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
19633 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
19634 (ASM_CPU64_DEFAULT_SPEC): Likewise.
19635 (CPP_CPU_SPEC): Handle M8.
19636 (ASM_CPU_SPEC): Likewise.
19637 (AS_M8_FLAG): Define.
19638 * config/sparc/sparc.md: Add m8 to the cpu attribute.
19639 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
19640 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
19642 * configure: Regenerate.
19643 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
19646 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19648 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
19650 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
19651 ("*movdi_insn_sp32"): Do not set v3pipe.
19652 ("*movsi_insn"): Likewise.
19653 ("*movdi_insn_sp64"): Likewise.
19654 ("*movsf_insn"): Likewise.
19655 ("*movdf_insn_sp32"): Likewise.
19656 ("*movdf_insn_sp64"): Likewise.
19657 ("*zero_extendsidi2_insn_sp64"): Likewise.
19658 ("*sign_extendsidi2_insn"): Likewise.
19659 ("*mov<VM32:mode>_insn"): Likewise.
19660 ("*mov<VM64:mode>_insn_sp64"): Likewise.
19661 ("*mov<VM64:mode>_insn_sp32"): Likewise.
19662 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
19663 ("<vlop:code><VL:mode>3"): Likewise.
19664 ("*not_<vlop:code><VL:mode>3"): Likewise.
19665 ("*nand<VL:mode>_vis"): Likewise.
19666 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
19667 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
19668 ("one_cmpl<VL:mode>2"): Likewise.
19669 ("faligndata<VM64:mode>_vis"): Likewise.
19670 ("alignaddrsi_vis"): Likewise.
19671 ("alignaddrdi_vis"): Likweise.
19672 ("alignaddrlsi_vis"): Likewise.
19673 ("alignaddrldi_vis"): Likewise.
19674 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
19675 ("bmaskdi_vis"): Likewise.
19676 ("bmasksi_vis"): Likewise.
19677 ("bshuffle<VM64:mode>_vis"): Likewise.
19678 ("cmask8<P:mode>_vis"): Likewise.
19679 ("cmask16<P:mode>_vis"): Likewise.
19680 ("cmask32<P:mode>_vis"): Likewise.
19681 ("pdistn<P:mode>_vis"): Likewise.
19682 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
19684 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19686 * config/sparc/sparc.md ("subtype"): New insn attribute.
19687 ("*wrgsr_sp64"): Set insn subtype.
19688 ("*rdgsr_sp64"): Likewise.
19689 ("alignaddrsi_vis"): Likewise.
19690 ("alignaddrdi_vis"): Likewise.
19691 ("alignaddrlsi_vis"): Likewise.
19692 ("alignaddrldi_vis"): Likewise.
19693 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
19694 ("fexpand_vis"): Likewise.
19695 ("fpmerge_vis"): Likewise.
19696 ("faligndata<VM64:mode>_vis"): Likewise.
19697 ("bshuffle<VM64:mode>_vis"): Likewise.
19698 ("cmask8<P:mode>_vis"): Likewise.
19699 ("cmask16<P:mode>_vis"): Likewise.
19700 ("cmask32<P:mode>_vis"): Likewise.
19701 ("fchksm16_vis"): Likewise.
19702 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
19703 ("fmean16_vis"): Likewise.
19704 ("fp<plusminus_insn>64_vis"): Likewise.
19705 ("<plusminus_insn>v8qi3"): Likewise.
19706 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
19707 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
19708 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
19709 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
19710 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
19711 ("*movqi_insn"): Likewise.
19712 ("*movhi_insn"): Likewise.
19713 ("*movsi_insn"): Likewise.
19714 ("movsi_pic_gotdata_op"): Likewise.
19715 ("*movdi_insn_sp32"): Likewise.
19716 ("*movdi_insn_sp64"): Likewise.
19717 ("movdi_pic_gotdata_op"): Likewise.
19718 ("*movsf_insn"): Likewise.
19719 ("*movdf_insn_sp32"): Likewise.
19720 ("*movdf_insn_sp64"): Likewise.
19721 ("*zero_extendhisi2_insn"): Likewise.
19722 ("*zero_extendqihi2_insn"): Likewise.
19723 ("*zero_extendqisi2_insn"): Likewise.
19724 ("*zero_extendqidi2_insn"): Likewise.
19725 ("*zero_extendhidi2_insn"): Likewise.
19726 ("*zero_extendsidi2_insn_sp64"): Likewise.
19727 ("ldfsr"): Likewise.
19728 ("prefetch_64"): Likewise.
19729 ("prefetch_32"): Likewise.
19730 ("tie_ld32"): Likewise.
19731 ("tie_ld64"): Likewise.
19732 ("*tldo_ldub_sp32"): Likewise.
19733 ("*tldo_ldub1_sp32"): Likewise.
19734 ("*tldo_ldub2_sp32"): Likewise.
19735 ("*tldo_ldub_sp64"): Likewise.
19736 ("*tldo_ldub1_sp64"): Likewise.
19737 ("*tldo_ldub2_sp64"): Likewise.
19738 ("*tldo_ldub3_sp64"): Likewise.
19739 ("*tldo_lduh_sp32"): Likewise.
19740 ("*tldo_lduh1_sp32"): Likewise.
19741 ("*tldo_lduh_sp64"): Likewise.
19742 ("*tldo_lduh1_sp64"): Likewise.
19743 ("*tldo_lduh2_sp64"): Likewise.
19744 ("*tldo_lduw_sp32"): Likewise.
19745 ("*tldo_lduw_sp64"): Likewise.
19746 ("*tldo_lduw1_sp64"): Likewise.
19747 ("*tldo_ldx_sp64"): Likewise.
19748 ("*mov<VM32:mode>_insn"): Likewise.
19749 ("*mov<VM64:mode>_insn_sp64"): Likewise.
19750 ("*mov<VM64:mode>_insn_sp32"): Likewise.
19752 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19754 * config/sparc/sparc.md ("type"): New insn type viscmp.
19755 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
19757 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
19758 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
19759 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
19760 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
19762 ("n7_vis_logical_11cycle"): Likewise.
19763 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
19764 * config/sparc/niagara2.md ("niag3_vis": Likewise.
19765 * config/sparc/niagara.md ("niag_vis"): Likewise.
19766 * config/sparc/ultra3.md ("us3_fga"): Likewise.
19767 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
19769 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19771 * config/sparc/sparc.md: New instruction type `bmask'.
19772 (bmaskdi_vis): Use the `bmask' type.
19773 (bmasksi_vis): Likewise.
19774 * config/sparc/ultra3.md (us3_array): Likewise.
19775 * config/sparc/niagara7.md (n7_array): Likewise.
19776 * config/sparc/niagara4.md (n4_array): Likewise.
19777 * config/sparc/niagara2.md (niag2_vis): Likewise.
19778 (niag3_vis): Likewise.
19779 * config/sparc/niagara.md (niag_vis): Likewise.
19781 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19783 * ipa-comdats.c: Remove optimize check from gate.
19784 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
19785 for functions not optimized.
19786 (ipa_fn_summary_read): Skip optimize check.
19787 (ipa_fn_summary_write): Likewise.
19788 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
19790 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
19792 (can_inline_edge_p): Check flag_pcc_struct_return for match.
19793 (check_callers): Give up on caller which is not optimized.
19794 (inline_small_functions): Likewise.
19795 (ipa_inline): Do not give up when not optimizing.
19796 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
19797 away unoptimizes cdtors.
19798 (whole_program_function_and_variable_visibility): Do
19799 ipa_discover_readonly_nonaddressable_vars in LTO mode.
19800 * ipa.c (process_references): Do not check optimize.
19801 (symbol_table::remove_unreachable_nodes): Update optimize check.
19802 (set_writeonly_bit): Update optimize check.
19803 (pass_ipa_cdtor_merge::gate): Do not check optimize.
19804 (pass_ipa_single_use::gate): Remove.
19806 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19808 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
19809 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
19810 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
19811 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
19812 permute_load, permute_store, adjust_extract, adjust_splat,
19813 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
19814 replace_swap_with_copy, dump_swap_insn_table,
19815 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
19816 recombine_lvx_pattern, recombine_stvx_pattern,
19817 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
19818 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
19819 to file rs6000-p8swap.c.
19820 * config/rs6000/rs6000-p8swap.c: New file.
19821 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
19822 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
19823 and rs6000*-*-* targets.
19825 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19827 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
19829 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19831 * lto-wrapper.c (merge_and_complain): Do not merge
19832 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
19833 fsigned_zeros, ftrapping_math, fwrapv.
19834 (append_compiler_options): Do not track these options.
19835 (append_linker_options): Likewie
19837 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19839 * cgraphunit.c (cgraph_node::finalize_function): When
19840 !flag_toplevel_reorde set no_reorder flag.
19841 (varpool_node::finalize_decl): Likewise.
19842 (symbol_table::compile): Drop no toplevel reorder path.
19844 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19846 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
19847 edges; zero probability is not better than uninitialized.
19849 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
19851 * asan.h (asan_sanitize_allocas_p): Declare.
19852 * asan.c (asan_sanitize_allocas_p): New function.
19853 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
19854 (handle_builtin_alloca): Likewise.
19855 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
19856 if !asan_sanitize_allocas_p.
19857 * params.def (asan-instrument-allocas): Add new option.
19858 * params.h (ASAN_PROTECT_ALLOCAS): Define.
19859 * opts.c (common_handle_option): Disable allocas sanitization for
19862 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
19864 * asan.c: Include gimple-fold.h.
19865 (get_last_alloca_addr): New function.
19866 (handle_builtin_stackrestore): Likewise.
19867 (handle_builtin_alloca): Likewise.
19868 (asan_emit_allocas_unpoison): Likewise.
19869 (get_mem_refs_of_builtin_call): Add new parameter, remove const
19870 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
19871 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
19872 (instrument_builtin_call): Pass gimple iterator to
19873 get_mem_refs_of_builtin_call.
19874 (last_alloca_addr): New global.
19875 * asan.h (asan_emit_allocas_unpoison): Declare.
19876 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
19877 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
19878 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
19879 if function calls alloca.
19880 * gimple-fold.c (replace_call_with_value): Remove static keyword.
19881 * gimple-fold.h (replace_call_with_value): Declare.
19882 * internal-fn.c: Include asan.h.
19883 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
19884 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
19886 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19888 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
19889 (C_SELFTEST_FLAGS): New.
19890 (CPP_SELFTEST_FLAGS): New.
19891 (SELFTEST_DEPS): New, from deps of s-selftest.
19892 (C_SELFTEST_DEPS): New, from deps of s-selftest.
19893 (CPP_SELFTEST_DEPS): New.
19894 (selftest): Add dependency on s-selftest-c++.
19895 (s-selftest): Rename to...
19896 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
19897 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
19898 than SELFTEST_FLAGS.
19899 (selftest-gdb): Rename to...
19900 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
19902 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
19903 (selftest-valgrind): Rename to...
19904 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
19906 (selftest-valgrind): Reintroduce as an alias for
19907 selftest-c-valgrind.
19908 (s-selftest-c++): New.
19909 (selftest-c++-gdb): New.
19910 (selftest-c++-valgrind): New.
19912 2017-07-06 Olivier Hainque <hainque@adacore.com>
19914 * gcc.c (process_command): When deciding if undefined variables
19915 should be ignored when processing specs, accept "gcc -v" as well.
19917 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19919 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
19920 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
19922 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
19924 * config/arm/arm-cpus.in (armv8-r): Add new entry.
19925 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
19926 * config/arm/arm-tables.opt: Regenerate.
19927 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
19929 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
19931 2017-07-06 Carl Love <cel@us.ibm.com>
19933 * ChangeLog: Clean up from mid air collision
19935 2017-07-06 Carl Love <cel@us.ibm.com>
19937 * config/rs6000/rs6000-c.c: Add support for built-in functions
19938 vector signed int vec_subc (vector signed int, vector signed int);
19939 vector signed __int128 vec_subc (vector signed __int128,
19940 vector signed __int128);
19941 vector unsigned __int128 vec_subc (vector unsigned __int128,
19942 vector unsigned __int128);
19943 vector signed int vec_sube (vector signed int, vector signed int,
19944 vector signed int);
19945 vector unsigned int vec_sube (vector unsigned int,
19946 vector unsigned int,
19947 vector unsigned int);
19948 vector signed __int128 vec_sube (vector signed __int128,
19949 vector signed __int128,
19950 vector signed__int128);
19951 vector unsigned __int128 vec_sube (vector unsigned __int128,
19952 vector unsigned __int128,
19953 vector unsigned __int128);
19954 vector signed int vec_subec (vector signed int, vector signed int,
19955 vector signed int);
19956 vector unsigned int vec_subec (vector unsigned int,
19957 vector unsigned int,
19958 vector unsigned int);
19959 vector signed __int128 vec_subec (vector signed __int128,
19960 vector signed __int128,
19961 vector signed__int128);
19962 vector unsigned __int128 vec_subec (vector unsigned __int128,
19963 vector unsigned __int128,
19964 vector unsigned __int128);
19965 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
19966 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
19967 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
19968 BU_ALTIVEC_OVERLOAD_X definitions.
19969 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
19970 * doc/extend.texi: Update the built-in documentation file for the new
19971 built-in functions.
19973 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19976 * diagnostic-show-locus.c (layout::layout): Use start and finish
19977 spelling location for the start and finish of each range.
19978 * genmatch.c (linemap_client_expand_location_to_spelling_point):
19979 Add unused aspect param.
19980 * input.c (expand_location_1): Add "aspect" param, and use it
19981 to access the correct part of the location.
19982 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
19984 (expand_location_to_spelling_point): Likewise.
19985 (linemap_client_expand_location_to_spelling_point): Add "aspect"
19986 param, and pass it to expand_location_1.
19988 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
19990 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
19991 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
19992 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
19993 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
19994 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
19995 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
19996 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
19997 _mm_maskz_getmant_ss): New intrinsics.
19998 (__builtin_ia32_getexpss128_mask): Changed to ...
19999 __builtin_ia32_getexpss128_round ... this.
20000 (__builtin_ia32_getexpsd128_mask): Changed to ...
20001 __builtin_ia32_getexpsd128_round ... this.
20002 * config/i386/i386-builtin-types.def
20003 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
20004 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
20005 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
20006 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
20007 __builtin_ia32_getmantss_mask_round): New builtins.
20008 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
20009 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
20010 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
20011 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
20012 * config/i386/sse.md
20013 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
20014 avx512f_sgetexp<mode><mask_scalar_name>
20015 <round_saeonly_scalar_name> ... this.
20016 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
20017 %0, %1, %2<round_saeonly_op3>}): Changed to ...
20018 vgetexp<ssescalarmodesuffix>
20019 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20020 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
20021 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
20022 avx512f_vgetmant<mode><mask_scalar_name>
20023 <round_saeonly_scalar_name> ... this.
20024 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
20025 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
20026 vgetmant<ssescalarmodesuffix>
20027 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
20028 %0<mask_scalar_operand4>, %1, %2
20029 <round_saeonly_scalar_mask_op4>, %3} ... this.
20030 * config/i386/subst.md (mask_scalar_operand4,
20031 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
20032 round_saeonly_scalar_nimm_predicate): New subst attributes.
20034 2017-07-06 Julia Koval <julia.koval@intel.com>
20036 * config/i386/i386.c (ix86_erase_embedded_rounding):
20037 Remove code for old rounding pattern.
20039 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
20041 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
20043 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
20045 * doc/sourcebuild.texi (Test Directives, Variants of
20046 dg-require-support): Add documentation for dg-require-stack-check.
20048 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
20050 * config/i386/subst.md (mask_scalar, round_scalar,
20051 round_saeonly_scalar): New meta-templates.
20052 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
20053 round_scalar_mask_operand3, round_scalar_mask_op3,
20054 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
20055 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
20056 round_saeonly_scalar_constraint,
20057 round_saeonly_scalar_prefix): New subst attribute.
20058 * config/i386/sse.md
20059 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
20060 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
20061 <round_scalar_name> ... this.
20062 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
20063 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
20064 <round_scalar_name> ... this.
20065 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
20066 <sse>_vm<code><mode>3<mask_scalar_name>
20067 <round_saeonly_scalar_name> ... this.
20068 (v<plusminus_mnemonic><ssescalarmodesuffix>
20069 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
20070 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
20071 v<plusminus_mnemonic><ssescalarmodesuffix>
20072 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20073 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
20074 (v<multdiv_mnemonic><ssescalarmodesuffix>
20075 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
20076 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
20077 v<multdiv_mnemonic><ssescalarmodesuffix>
20078 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20079 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
20080 (v<maxmin_float><ssescalarmodesuffix>
20081 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
20082 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
20083 v<maxmin_float><ssescalarmodesuffix>
20084 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20085 %0<mask_scalar_operand3>, %1, %<iptr>2
20086 <round_saeonly_scalar_mask_op3>} ... this.
20088 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
20090 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
20091 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
20093 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
20094 Alan Hayward <alan.hayward@arm.com>
20095 David Sherwood <david.sherwood@arm.com>
20097 * combine.c (simplify_if_then_else): Remove "enum" before
20099 * compare-elim.c (can_eliminate_compare): Likewise.
20100 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
20102 (aarch64_lookup_simd_builtin_type): Likewise.
20103 (aarch64_simd_builtin_type): Likewise.
20104 (aarch64_init_simd_builtin_types): Likewise.
20105 (aarch64_simd_expand_args): Likewise.
20106 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
20108 (aarch64_reverse_mask): Likewise.
20109 (aarch64_simd_emit_reg_reg_move): Likewise.
20110 (aarch64_gen_adjusted_ldpstp): Likewise.
20111 (aarch64_ccmp_mode_to_code): Likewise.
20112 (aarch64_operands_ok_for_ldpstp): Likewise.
20113 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
20114 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
20116 (aarch64_min_divisions_for_recip_mul): Likewise.
20117 (aarch64_reassociation_width): Likewise.
20118 (aarch64_get_condition_code_1): Likewise.
20119 (aarch64_simd_emit_reg_reg_move): Likewise.
20120 (aarch64_simd_attr_length_rglist): Likewise.
20121 (aarch64_reverse_mask): Likewise.
20122 (aarch64_operands_ok_for_ldpstp): Likewise.
20123 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
20124 (aarch64_gen_adjusted_ldpstp): Likewise.
20125 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
20127 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
20128 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
20129 (arm_lookup_simd_builtin_type): Likewise.
20130 (arm_simd_builtin_type): Likewise.
20131 (arm_init_simd_builtin_types): Likewise.
20132 (arm_expand_builtin_args): Likewise.
20133 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
20134 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
20135 (ft32_setup_incoming_varargs): Likewise.
20136 (ft32_function_arg): Likewise.
20137 (ft32_function_arg_advance): Likewise.
20138 (ft32_pass_by_reference): Likewise.
20139 (ft32_arg_partial_bytes): Likewise.
20140 (ft32_valid_pointer_mode): Likewise.
20141 (ft32_addr_space_pointer_mode): Likewise.
20142 (ft32_addr_space_legitimate_address_p): Likewise.
20143 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
20145 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
20146 (ix86_emit_outlined_ms2sysv_restore): Likewise.
20147 (iamcu_alignment): Likewise.
20148 (canonicalize_vector_int_perm): Likewise.
20149 (ix86_noce_conversion_profitable_p): Likewise.
20150 (ix86_mpx_bound_mode): Likewise.
20151 (ix86_operands_ok_for_move_multiple): Likewise.
20152 * config/microblaze/microblaze-protos.h
20153 (microblaze_expand_conditional_branch_reg): Likewise.
20154 * config/microblaze/microblaze.c
20155 (microblaze_expand_conditional_branch_reg): Likewise.
20156 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20158 (rs6000_reassociation_width): Likewise.
20159 (rs6000_invalid_binary_op): Likewise.
20160 (fusion_p9_p): Likewise.
20161 (emit_fusion_p9_load): Likewise.
20162 (emit_fusion_p9_store): Likewise.
20163 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
20165 (riscv_hard_regno_mode_ok_p): Likewise.
20166 (riscv_address_insns): Likewise.
20167 (riscv_split_symbol): Likewise.
20168 (riscv_legitimize_move): Likewise.
20169 (riscv_function_value): Likewise.
20170 (riscv_hard_regno_nregs): Likewise.
20171 (riscv_expand_builtin): Likewise.
20172 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
20173 (riscv_build_integer): Likewise.
20174 (riscv_split_integer): Likewise.
20175 (riscv_legitimate_constant_p): Likewise.
20176 (riscv_cannot_force_const_mem): Likewise.
20177 (riscv_regno_mode_ok_for_base_p): Likewise.
20178 (riscv_valid_base_register_p): Likewise.
20179 (riscv_valid_offset_p): Likewise.
20180 (riscv_valid_lo_sum_p): Likewise.
20181 (riscv_classify_address): Likewise.
20182 (riscv_legitimate_address_p): Likewise.
20183 (riscv_address_insns): Likewise.
20184 (riscv_load_store_insns): Likewise.
20185 (riscv_force_binary): Likewise.
20186 (riscv_split_symbol): Likewise.
20187 (riscv_force_address): Likewise.
20188 (riscv_legitimize_address): Likewise.
20189 (riscv_move_integer): Likewise.
20190 (riscv_legitimize_const_move): Likewise.
20191 (riscv_legitimize_move): Likewise.
20192 (riscv_address_cost): Likewise.
20193 (riscv_subword): Likewise.
20194 (riscv_output_move): Likewise.
20195 (riscv_canonicalize_int_order_test): Likewise.
20196 (riscv_emit_int_order_test): Likewise.
20197 (riscv_function_arg_boundary): Likewise.
20198 (riscv_pass_mode_in_fpr_p): Likewise.
20199 (riscv_pass_fpr_single): Likewise.
20200 (riscv_pass_fpr_pair): Likewise.
20201 (riscv_get_arg_info): Likewise.
20202 (riscv_function_arg): Likewise.
20203 (riscv_function_arg_advance): Likewise.
20204 (riscv_arg_partial_bytes): Likewise.
20205 (riscv_function_value): Likewise.
20206 (riscv_pass_by_reference): Likewise.
20207 (riscv_setup_incoming_varargs): Likewise.
20208 (riscv_print_operand): Likewise.
20209 (riscv_elf_select_rtx_section): Likewise.
20210 (riscv_save_restore_reg): Likewise.
20211 (riscv_for_each_saved_reg): Likewise.
20212 (riscv_register_move_cost): Likewise.
20213 (riscv_hard_regno_mode_ok_p): Likewise.
20214 (riscv_hard_regno_nregs): Likewise.
20215 (riscv_class_max_nregs): Likewise.
20216 (riscv_memory_move_cost): Likewise.
20217 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
20218 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
20219 (rl78_addr_space_address_mode): Likewise.
20220 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20222 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20223 (rs6000_reassociation_width): Likewise.
20224 (rs6000_invalid_binary_op): Likewise.
20225 (fusion_p9_p): Likewise.
20226 (emit_fusion_p9_load): Likewise.
20227 (emit_fusion_p9_store): Likewise.
20228 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
20229 (ok_for_simple_move_operands): Likewise.
20230 (ok_for_simple_move_strict_operands): Likewise.
20231 (ok_for_simple_arith_logic_operands): Likewise.
20232 (visium_legitimize_reload_address): Likewise.
20233 (visium_select_cc_mode): Likewise.
20234 (output_cbranch): Likewise.
20235 (visium_split_double_move): Likewise.
20236 (visium_expand_copysign): Likewise.
20237 (visium_expand_int_cstore): Likewise.
20238 (visium_expand_fp_cstore): Likewise.
20239 * config/visium/visium.c (visium_pass_by_reference): Likewise.
20240 (visium_function_arg): Likewise.
20241 (visium_function_arg_advance): Likewise.
20242 (visium_libcall_value): Likewise.
20243 (visium_setup_incoming_varargs): Likewise.
20244 (visium_legitimate_constant_p): Likewise.
20245 (visium_legitimate_address_p): Likewise.
20246 (visium_legitimize_address): Likewise.
20247 (visium_secondary_reload): Likewise.
20248 (visium_register_move_cost): Likewise.
20249 (visium_memory_move_cost): Likewise.
20250 (prepare_move_operands): Likewise.
20251 (ok_for_simple_move_operands): Likewise.
20252 (ok_for_simple_move_strict_operands): Likewise.
20253 (ok_for_simple_arith_logic_operands): Likewise.
20254 (visium_function_value_1): Likewise.
20255 (rtx_ok_for_offset_p): Likewise.
20256 (visium_legitimize_reload_address): Likewise.
20257 (visium_split_double_move): Likewise.
20258 (visium_expand_copysign): Likewise.
20259 (visium_expand_int_cstore): Likewise.
20260 (visium_expand_fp_cstore): Likewise.
20261 (visium_split_cstore): Likewise.
20262 (visium_select_cc_mode): Likewise.
20263 (visium_split_cbranch): Likewise.
20264 (output_cbranch): Likewise.
20265 (visium_print_operand_address): Likewise.
20266 * expmed.c (flip_storage_order): Likewise.
20267 * expmed.h (emit_cstore): Likewise.
20268 (flip_storage_order): Likewise.
20269 * genrecog.c (validate_pattern): Likewise.
20270 * hsa-gen.c (gen_hsa_addr): Likewise.
20271 * internal-fn.c (expand_arith_overflow): Likewise.
20272 * ira-color.c (allocno_copy_cost_saving): Likewise.
20273 * lra-assigns.c (find_hard_regno_for_1): Likewise.
20274 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
20275 (process_invariant_for_inheritance): Likewise.
20276 * lra-eliminations.c (move_plus_up): Likewise.
20277 * omp-low.c (lower_oacc_reductions): Likewise.
20278 * simplify-rtx.c (simplify_subreg): Likewise.
20279 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
20280 (TARGET_CHKP_BOUND_MODE): Likewise..
20281 * targhooks.c (default_chkp_bound_mode): Likewise.
20282 (default_setup_incoming_vararg_bounds): Likewise.
20283 * targhooks.h (default_chkp_bound_mode): Likewise.
20284 (default_setup_incoming_vararg_bounds): Likewise.
20285 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
20286 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
20287 (have_whole_vector_shift): Likewise.
20288 * tree-vect-stmts.c (vectorizable_load): Likewise.
20289 * doc/tm.texi: Regenerate.
20291 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
20293 Graceful degrade if Binutils PR21472 is not available.
20296 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
20297 .rodata in flash test fails.
20298 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
20299 * confgure: Regenerate.
20300 * config.in: Regenerate.
20301 * config/avr/avr.c (avr_asm_named_section)
20302 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
20303 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
20304 (avr_asm_init_sections): Same.
20306 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20308 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
20309 (fma<VH:mode>4_intrinsic): Likewise.
20310 (*fmsub<VCVTF:mode>4): Likewise.
20311 (*fmsub<VH:mode>4_intrinsic): Likewise.
20313 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
20316 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
20317 Don't depend on "optimize > 0".
20318 (out_movhi_r_mr, out_movqi_mr_r): Same.
20319 (out_movhi_mr_r, out_movqi_r_mr): Same.
20320 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
20321 io_address_operand on "optimize > 0".
20323 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20325 * tree-loop-distribution.c: Add general explanantion on the pass.
20326 (generate_loops_for_partition): Mark distributed loop.
20327 (pg_add_dependence_edges): New parameter. Handle alias data
20328 dependence specially and record it in the parameter if asked.
20329 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
20330 (init_partition_graph_vertices, add_partition_graph_edge): New.
20331 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
20332 (free_partition_graph_vdata, build_partition_graph): New.
20333 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
20334 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
20335 (data_ref_segment_size, latch_dominated_by_data_ref): New.
20336 (compute_alias_check_pairs, version_loop_by_alias_check): New.
20337 (version_for_distribution_p, finalize_partitions): New.
20338 (distribute_loop): Handle alias data dependence specially. Factor
20339 out loop fusion code as functions and call these functions.
20341 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20343 * tree-loop-distribution.c (classify_partition): New parameter and
20344 better handle reduction statement.
20345 (rdg_build_partitions): Revise comment.
20346 (distribute_loop): Compute statements in all partitions and pass it
20347 to classify_partition.
20349 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20351 * tree-loop-distribution.c (enum partition_type): New.
20352 (struct partition): New field type.
20353 (partition_merge_into): Add parameter. Update partition type.
20354 (data_dep_in_cycle_p, update_type_for_merge): New functions.
20355 (build_rdg_partition_for_vertex): Compute partition type.
20356 (rdg_build_partitions): Dump partition type.
20357 (distribute_loop): Update calls to partition_merge_into.
20359 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20361 * tree-loop-distribution.c (struct ddr_hasher): New.
20362 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
20364 (classify_partition): Call get_data_dependence.
20365 (pg_add_dependence_edges): Ditto.
20366 (distribute_loop): Release data dependence hash table.
20368 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20370 * tree-loop-distribution.c (ref_base_address): Delete.
20371 (similar_memory_accesses): Rename ...
20372 (share_memory_accesses): ... to this. Check if partitions access
20373 the same memory reference.
20374 (distribute_loop): Call share_memory_accesses.
20376 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20378 * tree-loop-distribution.c (struct partition): New field recording
20379 its data reference.
20380 (partition_alloc, partition_free): Init and release data refs.
20381 (partition_merge_into): Merge data refs.
20382 (build_rdg_partition_for_vertex): Collect data refs for partition.
20383 (pg_add_dependence_edges): Change parameters from vector to bitmap.
20385 (distribute_loop): Remve data refs from vertice data of partition
20388 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20390 * tree-loop-distribution.c (params.h): Include header file.
20391 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
20392 (datarefs_vec): New global var.
20393 (create_rdg_vertices): Use datarefs_vec directly.
20394 (free_rdg): Don't free data references.
20395 (build_rdg): Update use. Don't free data references.
20396 (distribute_loop): Compute global variable for data references.
20397 Bail out if there are too many data references.
20399 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20401 * tree-loop-distribution.c (loop_nest): New global var.
20402 (build_rdg): Use loop directly, rather than loop nest.
20403 (pg_add_dependence_edges): Remove loop nest parameter. Use global
20405 (distribute_loop): Compute global variable loop nest. Update use.
20407 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20409 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
20410 (partition_merge_into): New parameter. Dump reason for fusion.
20411 (distribute_loop): Update use of partition_merge_into.
20413 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20415 * tree-loop-distribution.c (bb_top_order_index): New.
20416 (bb_top_order_index_size, bb_top_order_cmp): New.
20417 (stmts_from_loop): Use topological order.
20418 (pass_loop_distribution::execute): Compute and release topological
20419 order for basic blocks.
20421 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20423 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
20426 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20428 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
20429 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
20430 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
20431 * internal-fn.def (LOOP_DIST_ALIAS): New.
20432 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
20433 (fold_loop_internal_call): ... this.
20434 (vect_loop_dist_alias_call): New function.
20435 (set_uid_loop_bbs): Call fold_loop_internal_call.
20436 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
20439 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
20442 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
20443 Require dead FLAGS_REG at the beginning of a peephole.
20445 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
20448 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
20449 arguments in the call to __builtin_ia32_sbb_u32.
20450 (_subborrow_u64): Swap _X and _Y arguments in the call to
20451 __builtin_ia32_sbb_u64.
20453 2017-07-04 Jakub Jelinek <jakub@redhat.com>
20456 * tree-vrp.c (compare_assert_loc): Turn into a function template
20457 with stable template parameter. Only test if a->e is NULL,
20458 !a->e == !b->e has been verified already. Use e == NULL or
20459 e != NULL instead of e or ! e tests. If stable is true, don't use
20460 iterative_hash_expr, on the other side allow a or b or both NULL
20461 and sort the NULLs last.
20462 (process_assert_insertions): Sort using compare_assert_loc<false>
20463 instead of compare_assert_loc, later sort using
20464 compare_assert_loc<true> before calling process_assert_insertions_for
20465 in a loop. Use break instead of continue once seen NULL pointer.
20467 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20469 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
20470 Cortex-R7 and Cortex-R8 processors.
20472 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
20474 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
20475 uninitialized while src is not.
20477 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
20479 * common/config/arm/arm-common.c: Adjust include path for
20481 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
20482 (arm-cpu.h): Create in build directory. Adjust dependency rules.
20483 (arm-cpu-data.h): Likewise.
20484 (arm-cpu-cdata.h): Likewise.
20485 * config/arm/arm-cpu.h: Delete.
20486 * config/arm/arm-cpu-cdata.h: Delete.
20487 * config/arm/arm-cpu-data.h: Delete.
20489 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
20491 * config/arm/arm-cpus.in (cortex-a55): New.
20492 (cortex-a75): Likewise.
20493 (cortex-a75.cortex-a55): Likewise.
20494 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
20496 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
20497 * config/arm/arm-cpu-cdata.h: Regenerate.
20498 * config/arm/arm-cpu-data.h: Regenerate.
20499 * config/arm/arm-cpu.h: Regenerate.
20500 * config/arm/arm-tables.opt: Regenerate.
20501 * config/arm/arm-tune.md: Regenerate.
20503 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
20505 * haifa-sched.c (sched_create_recovery_edges): Update profile.
20507 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
20509 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
20512 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
20514 PR tree-optimization/81292
20515 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
20516 full_string_p, also call adjust_related_strinfos if the adjustment
20517 is simple, otherwise invalidate related strinfos.
20519 2017-07-04 Martin Liska <mliska@suse.cz>
20522 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
20523 newly created variable as DECL_IGNORED_P.
20525 2017-07-04 Martin Liska <mliska@suse.cz>
20528 * ipa-inline.c (inline_small_functions):
20529 Use xstrdup_for_dump.
20531 2017-07-04 Tom de Vries <tom@codesourcery.com>
20533 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
20535 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
20538 * config/darwin.c (darwin_function_switched_text_sections):
20541 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
20543 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
20545 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
20547 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
20549 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20551 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
20552 min_profitable_iters, and th as inclusive lower bounds.
20553 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
20554 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
20555 for min_profitable_iters and min_profitable_estimate.
20556 (vect_transform_loop): Treat th as an inclusive lower bound.
20557 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
20559 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
20562 * config/darwin.c (darwin_function_switched_text_sections):
20563 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
20564 in two pieces, and suppress the use of buf.
20566 2017-07-03 Nathan Sidwell <nathan@acm.org>
20568 * hash-table.h (hash_table_mod1): Fix indentation.
20570 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20572 PR middle-end/81290
20573 * predict.c (force_edge_cold): Be more careful about propagation
20575 * profile-count.h (profile_probability::guessed,
20576 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
20578 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
20580 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
20582 * doc/invoke.texi (rcpc architecture extension): Document it.
20584 2017-07-03 Richard Biener <rguenther@suse.de>
20586 PR tree-optimization/60510
20587 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
20588 the scalar reduction PHI and use it.
20589 (vectorizable_reduction): Properly guard the single_defuse_cycle
20590 path for non-SLP reduction chains where we cannot use it.
20591 Rework reduc_def/index and vector type deduction. Rework
20592 vector operand gathering during reduction op code-gen.
20593 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
20594 chains dissolve the chain and leave it to non-SLP reduction
20597 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20599 * tree-data-ref.h (dr_alignment): Declare.
20600 * tree-data-ref.c (dr_alignment): New function.
20601 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
20602 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
20604 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
20606 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20608 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
20609 and base_misalignment fields.
20610 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
20611 * tree-data-ref.c: Include builtins.h.
20612 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
20613 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
20614 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
20615 * tree-vect-data-refs.c: Include tree-cfg.h.
20616 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
20617 fields instead of calculating an alignment here.
20618 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
20619 innermost_loop_behavior fields.
20621 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20623 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
20625 (DR_STEP_ALIGNMENT): New macro.
20626 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
20627 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
20628 (create_data_ref): Print it.
20629 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
20630 to tell whether the step preserves vector (mis)alignment.
20631 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
20632 Move the check for an integer step and generalise to all INTEGER_CST.
20633 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
20634 Print the outer step alignment.
20636 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20638 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
20639 with offset_alignment.
20640 (DR_ALIGNED_TO): Delete.
20641 (DR_OFFSET_ALIGNMENT): New macro.
20642 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
20643 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
20644 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
20645 (create_data_ref): Likewise.
20646 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
20647 (vect_analyze_data_refs): Likewise.
20648 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
20649 creating dummy innermost behavior.
20651 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20653 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
20654 with a "innermost_loop_behavior *" and refeence tree.
20655 * tree-data-ref.c (dr_analyze_innermost): Likewise.
20656 (create_data_ref): Update call accordingly.
20657 * tree-predcom.c (find_looparound_phi): Likewise.
20659 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20661 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
20662 fields with dr_wrt_vec_loop.
20663 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
20664 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
20665 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
20666 (vect_dr_behavior): New function.
20667 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
20668 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
20669 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
20670 track whether the step preserves the misalignment.
20671 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
20672 Use vect_dr_behavior.
20673 (vect_setup_realignment): Update call accordingly.
20674 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
20675 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
20676 call to vect_create_addr_base_for_vector_ref.
20677 (vect_create_cond_for_align_checks): Likewise.
20678 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
20679 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
20680 (vect_recog_mask_conversion_pattern): Likewise.
20681 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
20682 (new_stmt_vec_info): Remove redundant zeroing.
20684 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
20686 * common/config/arm/arm-common.c (arm_be8_option): New function.
20687 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
20688 (ISA_ARMv6): Add isa_bit_be8.
20689 * config/arm/arm.h (arm_be8_option): Add prototype.
20690 (BE8_SPEC_FUNCTION): New define.
20691 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
20692 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
20693 (mlittle-endian): Similarly.
20694 (mbe8, mbe32): New options.
20695 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
20696 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
20698 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20700 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
20702 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20704 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
20705 (cleanup_tree_cfg_bb): Use it.
20706 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
20708 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
20710 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20713 * loop-doloop.c (add_test): Update profile.
20715 2017-07-03 Martin Liska <mliska@suse.cz>
20718 * sanopt.c (rewrite_usage_of_param): New function.
20719 (sanitize_rewrite_addressable_params): Likewise.
20720 (pass_sanopt::execute): Call rewrite_usage_of_param.
20722 2017-07-03 Richard Biener <rguenther@suse.de>
20724 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
20725 back to using VIEW_CONVERT_EXPR.
20727 2017-07-03 Martin Liska <mliska@suse.cz>
20730 * doc/extend.texi: Document when a resolver function is
20731 generated for target_clones.
20733 2017-07-03 Martin Liska <mliska@suse.cz>
20735 * asan.c (asan_emit_stack_protection): Unpoison just red zones
20736 and shadow memory of auto variables which are subject of
20737 use-after-scope sanitization.
20738 (asan_expand_mark_ifn): Add do set only when is_poison.
20740 2016-07-03 Richard Biener <rguenther@suse.de>
20742 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
20744 (vect_force_simple_reduction): Record reduction def -> phi mapping.
20745 (vectorizable_reduction): Perform reduction PHI creation when
20746 visiting a reduction PHI and adjust and simplify code generation
20747 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
20748 (vect_transform_loop): Visit reduction PHIs.
20749 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
20750 defs into the SLP tree.
20751 (vect_build_slp_tree): Reduction defs terminate the recursion.
20752 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
20754 (vect_get_vec_defs_for_stmt_copy): Export.
20755 (vect_get_vec_defs): Likewise.
20756 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
20758 (vect_get_vec_defs_for_stmt_copy): Declare.
20759 (vect_get_vec_defs): Likewise.
20761 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20763 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
20764 parameter with a "loop" parameter and use it instead of the
20765 loop containing DR_STMT. Don't check simple_iv when doing
20766 BB analysis. Describe the two analysis modes in the comment.
20768 2017-07-03 Tom de Vries <tom@codesourcery.com>
20770 PR tree-optimization/69468
20771 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
20772 (find_same_succ_bb): Handle ignore_edge_flags.
20774 2017-07-03 Tom de Vries <tom@codesourcery.com>
20776 PR tree-optimization/81192
20777 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
20779 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
20781 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
20783 2017-07-03 Tom de Vries <tom@codesourcery.com>
20785 PR tree-optimization/81192
20786 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
20787 BB_SAME_SUCC (bb) == NULL.
20789 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20791 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
20794 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20796 * dumpfile.c: Include profile-count.h
20797 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
20799 (insert_cond_bb): Update profile.
20800 * tree-cfg.h (insert_cond_bb): Update prototype.
20801 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
20802 * tree-dump.c: Do not include tree-cfg.
20804 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20806 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
20808 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20810 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
20813 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20815 * tree-complex.c (expand_complex_div_wide): update profile.
20817 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20818 Alan Hayward <alan.hayward@arm.com>
20819 David Sherwood <david.sherwood@arm.com>
20821 * Makefile.in (MACHMODE_H): Remove insn-modes.h
20822 (CORETYPES_H): New define.
20823 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
20824 (insn-modes-inline.h, s-modes-inline-h): New rules.
20825 (generated_files): Add insn-modes-inline.h.
20826 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
20827 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
20828 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
20829 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
20830 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
20831 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
20832 (build/gencodes.o, build/genconditions.o): Likewise.
20833 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
20834 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
20835 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
20836 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
20837 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
20838 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
20839 * coretypes.h: Include everything up to real.h for generators.
20840 Include insn-modes.h first. Include wide-int-print.h after
20841 wide-int.h. Include insn-modes-inline.h and then machmode.h.
20842 * machmode.h: Don't include insn-modes.h here.
20843 * function-tests.c: Remove includes of signop.h, machmode.h,
20844 double-int.h and wide-int.h.
20846 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
20848 * optc-save-gen.awk: Likewise.
20849 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
20850 * godump.c: Remove include of wide-int-print.h.
20851 * pretty-print.h: Likewise.
20852 * wide-int-print.cc: Likewise.
20853 * wide-int.cc: Likewise.
20854 * hash-map-tests.c: Remove include of signop.h.
20855 * hash-set-tests.c: Likewise.
20856 * rtl-tests.c: Likewise.
20857 * mkconfig.sh: Remove include of machmode.h.
20858 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
20860 (emit_insn_modes_inline_h): ...this new function. Emit the code
20861 into an insn-modes-inline.h header file, adding appropriate
20862 include guards and end comments.
20863 (emit_insn_modes_c_header): Remove include of machmode.h.
20864 (emit_min_insn_modes_c_header): Include coretypes.h rather than
20866 (main): Handle -i flag and call emit_insn_modes_inline_h when
20869 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20871 * tree-ssa-strlen.c (strinfo): Rename the length field to
20872 nonzero_chars. Add a full_string_p field.
20873 (compare_nonzero_chars, zero_length_string_p): New functions.
20874 (get_addr_stridx): Add an offset_out parameter.
20875 Use compare_nonzero_chars.
20876 (get_stridx): Update accordingly. Use compare_nonzero_chars.
20877 (new_strinfo): Update after above changes to strinfo.
20878 (set_endptr_and_length): Set full_string_p.
20879 (get_string_length): Update after above changes to strinfo.
20880 (unshare_strinfo): Update call to new_strinfo.
20881 (maybe_invalidate): Likewise.
20882 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
20883 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
20884 as a uhwi instead of an shwi. Update after above changes to
20885 strinfo and new_strinfo.
20886 (zero_length_string): Assert that chainsi contains full strings.
20887 Use zero_length_string_p. Update call to new_strinfo.
20888 (adjust_related_strinfos): Update after above changes to strinfo.
20889 Copy full_string_p from origsi.
20890 (adjust_last_stmt): Use zero_length_string_p.
20891 (handle_builtin_strlen): Update after above changes to strinfo and
20892 new_strinfo. Install the lhs as the string length if the previous
20893 entry didn't describe a full string.
20894 (handle_builtin_strchr): Update after above changes to strinfo
20896 (handle_builtin_strcpy): Likewise.
20897 (handle_builtin_strcat): Likewise.
20898 (handle_builtin_malloc): Likewise.
20899 (handle_pointer_plus): Likewise.
20900 (handle_builtin_memcpy): Likewise. Track nonzero characters
20901 that aren't necessarily followed by a nul terminator.
20902 (handle_char_store): Likewise.
20904 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20906 PR tree-optimization/80769
20907 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
20908 for malloc and calloc. Document the new invariant that all related
20909 strinfos have delayed lengths or none do.
20910 (verify_related_strinfos): Move earlier in file.
20911 (set_endptr_and_length): New function, split out from...
20912 (get_string_length): ...here. Also set the lengths of related
20914 (zero_length_string): Assert that chainsi has known (rather than
20916 (adjust_related_strinfos): Likewise.
20918 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20920 PR tree-optimization/81136
20921 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
20922 assert that two references with the same misalignment have the same
20923 compile-time misalignment if those compile-time misalignments
20926 2017-07-01 Andi Kleen <ak@linux.intel.com>
20928 * print-tree.c (print_node): Print all attributes.
20930 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20932 * cfg.c (scale_bbs_frequencies): New function.
20933 * cfg.h (scale_bbs_frequencies): Declare it.
20934 * cfgloopanal.c (single_likely_exit): Cleanup.
20935 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
20937 (scale_loop_profile): Likewise.
20938 (loop_version): Likewise.
20939 (create_empty_loop_on_edge): Update.
20940 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
20941 scale_loop_frequencies, scale_loop_profile, loopify,
20942 loop_version): Update prototypes.
20943 * modulo-sched.c (sms_schedule): Update.
20944 * predict.c (unlikely_executed_edge_p): Also check probability.
20945 (probably_never_executed_edge_p): Fix typo.
20946 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20947 * tree-parloops.c (gen_parallel_loop): Update.
20948 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
20949 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20950 * tree-ssa-loop-split.c (split_loop): Update.
20951 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20952 * tree-vect-loop-manip.c (vect_do_peeling): Update.
20953 (vect_loop_versioning): Update.
20954 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20956 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20958 * trans-mem.c (split_bb_make_tm_edge): Update profile.
20960 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20962 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
20963 to keep profile consistent.
20965 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20967 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
20968 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
20969 * profile-count.h (max_safe_multiplier): Make unsigned.
20970 (profile_count::guessed_zero): New.
20972 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20974 * bb-reorder.c (fix_up_crossing_landing_pad,
20975 fix_crossing_conditional_branches): Use make_single_succ_edge
20976 to keep profile consistent.
20978 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20980 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
20983 2017-07-01 Jakub Jelinek <jakub@redhat.com>
20986 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
20987 the right scopes, make sure cond_jump isn't preserved between multiple
20988 iterations. Search for fallthru edge whenever there are 3+ edges and
20989 use find_fallthru_edge for it.
20991 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20993 Patch by Alexander Monakov <amonakov@ispras.ru>
20994 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
20995 probabilities consistently.
20997 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20999 * pa.c (pa_expand_compare_and_swap_loop): Update call of
21000 emit_cmp_and_jump_insns.
21002 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21005 * tree-inline.c (expand_call_inline): Combine profile statuses.
21007 2017-06-30 Andrew Pinski <apinski@cavium.com>
21009 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
21010 fold_stmt returned true.
21012 2017-06-30 Nathan Sidwell <nathan@acm.org>
21014 * ggc.h (empty_string): Delete.
21015 * cfgexpand.c (expand_asm_stmt): Use plain "".
21016 * optabs.c (expand_asm_memory_barrier): Likewise.
21017 * stringpool.c (empty_string): Delete.
21018 (digit_vector, digit_string): Delete.
21019 (ggc_alloc_string): Use plain "", don't optimize single digit
21020 strings. Use ggc_alloc_atomic.
21022 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
21024 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
21025 comparison set and one other set, use the cost of the non-comparison
21028 2017-06-30 Nathan Sidwell <nathan@acm.org>
21030 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
21033 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
21035 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
21036 loops. Remove now unneeded calls to gimple_switch_set_label() that
21037 just set removed labels to NULL_TREE.
21039 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
21041 * tree-ssanames.c (set_range_info_raw): Abstract from ...
21042 (set_range_info): ...here. Only call set_range_info_raw if domain
21044 (set_nonzero_bits): Call set_range_info_raw.
21045 * tree-ssanames.h (set_range_info_raw): New.
21047 2017-06-30 Jakub Jelinek <jakub@redhat.com>
21050 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
21051 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
21052 of nonimmediate_operand and <store_mask_constraint> instead of m
21053 for the input operand. For V8FI iterator, always split if input
21054 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
21055 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
21056 <store_mask_predicate> instead of register_operand and
21057 <store_mask_constraint> instead of v for the input operand. Make
21058 sure both operands aren't MEMs for if not <mask_applied>.
21060 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
21062 * lto-wrapper.c (copy_file) Close both file descriptors before
21065 2017-06-30 Martin Liska <mliska@suse.cz>
21068 * multiple_target.c (create_dispatcher_calls): Make ifunc
21069 also for function that don't have calls or are not referenced.
21071 2017-06-30 Richard Biener <rguenther@suse.de>
21073 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
21074 analyze the first scalar stmt. Move vector type computation
21075 for the BB case here from ...
21076 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
21077 live operation processing in the SLP case properly.
21079 2017-06-30 Richard Biener <rguenther@suse.de>
21081 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
21083 2017-06-30 Martin Liska <mliska@suse.cz>
21086 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
21087 before BUILT_IN_UNWIND_RESUME when ASAN is used.
21089 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
21091 * doc/invoke.texi (AArch64): Add missing options and remove redundant
21094 2017-06-30 Richard Biener <rguenther@suse.de>
21096 PR tree-optimization/81249
21097 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
21098 condition reduction result to original scalar type.
21100 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21102 * profile-count.h (enum profile_quality): Fix typos and whitespace
21105 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21107 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
21108 type for branch probabilities.
21110 2017-06-29 Julian Brown <julian@codesourcery.com>
21111 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21113 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
21114 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
21115 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
21116 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
21118 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21120 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
21121 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
21122 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
21123 CC usage from generic code to here.
21124 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
21125 CC usage into the target macros.
21127 2017-06-29 Maya Rashish <coypu@sdf.org>
21129 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
21132 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21134 * arm/arm-builtins.c: Include profile-count.h
21135 * except.c (sjlj_emit_function_enter): Use
21136 profile_probability::unlikely.
21138 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21140 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
21141 and tocrel_offset be pointer args rather than implicitly using
21143 (legitimate_constant_pool_address_p, rs6000_emit_move,
21144 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
21145 tocrel_offset and use in toc_relative_expr_p call.
21146 (print_operand, print_operand_address): Use static tocrel_base_oac
21147 and tocrel_offset_oac.
21148 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
21151 2017-06-29 Maya Rashish <coypu@sdf.org>
21153 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
21155 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
21157 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
21158 objects, take into account only the alignment of 'op0' and 'mode1' if
21161 2017-06-29 Steve Ellcey <sellcey@cavium.com>
21163 * ccmp.c (ccmp_tree_comparison_p): New function.
21164 (ccmp_candidate_p): Update to use above function.
21165 (get_compare_parts): New function.
21166 (expand_ccmp_next): Update to use new functions.
21167 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
21169 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
21170 take mode as argument.
21171 * ccmp.h (expand_ccmp_expr): Add mode as argument.
21172 * expr.c (expand_expr_real_1): Pass mode as argument.
21174 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
21176 * combine.c (combine_instructions): Print insns to dump_file, together
21179 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21181 * asan.c (asan_emit_stack_protection): Update.
21182 (create_cond_insert_point): Update.
21183 * auto-profile.c (afdo_propagate_circuit): Update.
21184 * basic-block.h (struct edge_def): Turn probability to
21185 profile_probability.
21186 (EDGE_FREQUENCY): Update.
21187 * bb-reorder.c (find_traces_1_round): Update.
21188 (better_edge_p): Update.
21189 (sanitize_hot_paths): Update.
21190 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
21191 (make_single_succ_edge): Update.
21192 (check_bb_profile): Update.
21193 (dump_edge_info): Update.
21194 (update_bb_profile_for_threading): Update.
21195 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
21196 probabilitycount to 0.
21197 * cfgbuild.c (compute_outgoing_frequencies): Update.
21198 * cfgcleanup.c (try_forward_edges): Update.
21199 (outgoing_edges_match): Update.
21200 (try_crossjump_to_edge): Update.
21201 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
21202 (expand_gimple_tailcall): Update.
21203 (construct_init_block): Use make_single_succ_edge.
21204 (construct_exit_block): Use make_single_succ_edge.
21205 * cfghooks.c (verify_flow_info): Update.
21206 (redirect_edge_succ_nodup): Update.
21207 (split_edge): Update.
21208 (account_profile_record): Update.
21209 * cfgloopanal.c (single_likely_exit): Update.
21210 * cfgloopmanip.c (scale_loop_profile): Update.
21211 (set_zero_probability): Remove.
21212 (duplicate_loop_to_header_edge): Update.
21213 * cfgloopmanip.h (loop_version): Update prototype.
21214 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
21215 (force_nonfallthru_and_redirect): Update.
21216 (update_br_prob_note): Update.
21217 (rtl_verify_edges): Update.
21218 (purge_dead_edges): Update.
21219 (rtl_lv_add_condition_to_bb): Update.
21220 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
21221 * cgraphunit.c (init_lowered_empty_function): Update.
21222 (cgraph_node::expand_thunk): Update.
21223 * cilk-common.c: Include profile-count.h
21224 * dojump.c (inv): Remove.
21225 (jumpifnot): Update.
21226 (jumpifnot_1): Update.
21227 (do_jump_1): Update.
21229 (do_jump_by_parts_greater_rtx): Update.
21230 (do_compare_rtx_and_jump): Update.
21231 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
21232 do_jump_1. do_compare_rtx_and_jump): Update prototype.
21233 * dwarf2cfi.c: Include profile-count.h
21234 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
21235 (sjlj_emit_dispatch_table): Likewise.
21236 * explow.c: Include profile-count.h
21237 * expmed.c (emit_store_flag_force): Update.
21238 (do_cmp_and_jump): Update.
21239 * expr.c (compare_by_pieces_d::generate): Update.
21240 (compare_by_pieces_d::finish_mode): Update.
21241 (emit_block_move_via_loop): Update.
21242 (store_expr_with_bounds): Update.
21243 (store_constructor): Update.
21244 (expand_expr_real_2): Update.
21245 (expand_expr_real_1): Update.
21246 * expr.h (try_casesi, try_tablejump): Update prototypes.
21247 * gimple-pretty-print.c (dump_probability): Update.
21248 (dump_profile): New.
21249 (dump_gimple_label): Update.
21250 (dump_gimple_bb_header): Update.
21251 * graph.c (draw_cfg_node_succ_edges): Update.
21252 * hsa-gen.c (convert_switch_statements): Update.
21253 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
21254 (find_if_case_1): Update.
21255 (find_if_case_2): Update.
21256 * internal-fn.c (expand_arith_overflow_result_store): Update.
21257 (expand_addsub_overflow): Update.
21258 (expand_neg_overflow): Update.
21259 (expand_mul_overflow): Update.
21260 (expand_vector_ubsan_overflow): Update.
21261 * ipa-cp.c (good_cloning_opportunity_p): Update.
21262 * ipa-split.c (split_function): Use make_single_succ_edge.
21263 * ipa-utils.c (ipa_merge_profiles): Update.
21264 * loop-doloop.c (add_test): Update.
21265 (doloop_modify): Update.
21266 * loop-unroll.c (compare_and_jump_seq): Update.
21267 (unroll_loop_runtime_iterations): Update.
21268 * lra-constraints.c (lra_inheritance): Update.
21269 * lto-streamer-in.c (input_cfg): Update.
21270 * lto-streamer-out.c (output_cfg): Update.
21271 * mcf.c (adjust_cfg_counts): Update.
21272 * modulo-sched.c (sms_schedule): Update.
21273 * omp-expand.c (expand_omp_for_init_counts): Update.
21274 (extract_omp_for_update_vars): Update.
21275 (expand_omp_ordered_sink): Update.
21276 (expand_omp_for_ordered_loops): Update.
21277 (expand_omp_for_generic): Update.
21278 (expand_omp_for_static_nochunk): Update.
21279 (expand_omp_for_static_chunk): Update.
21280 (expand_cilk_for): Update.
21281 (expand_omp_simd): Update.
21282 (expand_omp_taskloop_for_outer): Update.
21283 (expand_omp_taskloop_for_inner): Update.
21284 * omp-simd-clone.c (simd_clone_adjust): Update.
21285 * optabs.c (expand_doubleword_shift): Update.
21286 (expand_abs): Update.
21287 (emit_cmp_and_jump_insn_1): Update.
21288 (expand_compare_and_swap_loop): Update.
21289 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
21290 * predict.c (predictable_edge_p): Update.
21291 (edge_probability_reliable_p): Update.
21292 (set_even_probabilities): Update.
21293 (combine_predictions_for_insn): Update.
21294 (combine_predictions_for_bb): Update.
21295 (propagate_freq): Update.
21296 (estimate_bb_frequencies): Update.
21297 (force_edge_cold): Update.
21298 * profile-count.c (profile_count::dump): Add missing space into dump.
21299 (profile_count::debug): Add newline.
21300 (profile_count::differs_from_p): Explicitly convert to unsigned.
21301 (profile_count::stream_in): Update.
21302 (profile_probability::dump): New member function.
21303 (profile_probability::debug): New member function.
21304 (profile_probability::differs_from_p): New member function.
21305 (profile_probability::differs_lot_from_p): New member function.
21306 (profile_probability::stream_in): New member function.
21307 (profile_probability::stream_out): New member function.
21308 * profile-count.h (profile_count_quality): Rename to ...
21309 (profile_quality): ... this one.
21310 (profile_probability): New.
21311 (profile_count): Update.
21312 * profile.c (compute_branch_probabilities): Update.
21313 * recog.c (peep2_attempt): Update.
21314 * sched-ebb.c (schedule_ebbs): Update.
21315 * sched-rgn.c (find_single_block_region): Update.
21316 (compute_dom_prob_ps): Update.
21317 (schedule_region): Update.
21318 * sel-sched-ir.c (compute_succs_info): Update.
21319 * stmt.c (struct case_node): Update.
21320 (do_jump_if_equal): Update.
21321 (get_outgoing_edge_probs): Update.
21322 (conditional_probability): Update.
21323 (emit_case_dispatch_table): Update.
21324 (expand_case): Update.
21325 (expand_sjlj_dispatch_table): Update.
21326 (emit_case_nodes): Update.
21327 * targhooks.c: Update.
21328 * tracer.c (better_p): Update.
21329 (find_best_successor): Update.
21330 * trans-mem.c (expand_transaction): Update.
21331 * tree-call-cdce.c: Update.
21332 * tree-cfg.c (gimple_split_edge): Upate.
21333 (move_sese_region_to_fn): Upate.
21334 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
21335 * tree-eh.c (lower_resx): Upate.
21336 (cleanup_empty_eh_move_lp): Upate.
21337 * tree-if-conv.c (version_loop_for_if_conversion): Update.
21338 * tree-inline.c (copy_edges_for_bb): Update.
21339 (copy_cfg_body): Update.
21340 * tree-parloops.c (gen_parallel_loop): Update.
21341 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
21342 (gimple_gen_time_profiler): Update.
21343 * tree-ssa-dce.c (remove_dead_stmt): Update.
21344 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21345 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
21346 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
21347 (unloop_loops): Update.
21348 (try_peel_loop): Update.
21349 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21350 * tree-ssa-loop-split.c (connect_loops): Update.
21351 (split_loop): Update.
21352 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
21353 (hoist_guard): Update.
21354 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
21355 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
21356 (value_replacement): Update.
21357 * tree-ssa-reassoc.c (branch_fixup): Update.
21358 * tree-ssa-tail-merge.c (replace_block_by): Update.
21359 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
21360 (create_edge_and_update_destination_phis): Update.
21361 (compute_path_counts): Update.
21362 (recompute_probabilities): Update.
21363 (update_joiner_offpath_counts): Update.
21364 (freqs_to_counts_path): Update.
21365 (duplicate_thread_path): Update.
21366 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
21367 (struct switch_conv_info): Update.
21368 (gen_inbound_check): Update.
21369 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21370 (vect_do_peeling): Update.
21371 (vect_loop_versioning): Update.
21372 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21373 (optimize_mask_stores): Update.
21374 * ubsan.c (ubsan_expand_null_ifn): Update.
21375 * value-prof.c (gimple_divmod_fixed_value): Update.
21376 (gimple_divmod_fixed_value_transform): Update.
21377 (gimple_mod_pow2): Update.
21378 (gimple_mod_pow2_value_transform): Update.
21379 (gimple_mod_subtract): Update.
21380 (gimple_mod_subtract_transform): Update.
21381 (gimple_ic): Update.
21382 (gimple_stringop_fixed_value): Update.
21383 (gimple_stringops_transform): Update.
21384 * value-prof.h: Update.
21386 2017-06-29 Carl Love <cel@us.ibm.com>
21388 * config/rs6000/rs6000-c.c: Add support for built-in functions
21389 vector signed int vec_signed (vector float);
21390 vector signed long long vec_signed (vector double);
21391 vector signed int vec_signed2 (vector double, vector double);
21392 vector signed int vec_signede (vector double);
21393 vector signed int vec_signedo (vector double);
21394 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
21395 instruction generator.
21396 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
21397 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
21398 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
21400 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
21401 vunsignede_v2df): Add define_expands.
21402 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
21403 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
21404 VEC_UNSIGNEDO): Add definitions.
21405 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
21406 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
21407 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
21408 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
21409 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
21410 * config/rs6000/altivec.h (vec_signed, vec_signed2,
21411 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
21412 vec_unsignede, vec_unsignedo): Add builtin defines.
21413 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
21415 * doc/extend.texi: Update the built-in documentation file for the
21416 new built-in functions.
21418 2017-06-29 Richard Biener <rguenther@suse.de>
21420 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
21421 reduction chains to LOOP_VINFO_REDUCTIONS.
21422 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
21423 SLP reductions after processing reduction chains.
21425 2017-06-29 Nathan Sidwell <nathan@acm.org>
21427 * builtins.c (fold_builtin_FUNCTION): Use
21428 lang_hooks.decl_printable_name.
21430 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
21432 PR middle-end/81194
21433 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
21434 with only one label.
21435 * stmt.c (expand_case): Assert NCASES is greater than one.
21437 2017-06-29 Richard Biener <rguenther@suse.de>
21439 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
21441 (group_case_labels): Likewise.
21442 (find_taken_edge): Push sanity checking on val to workers...
21443 (find_taken_edge_cond_expr): ... here
21444 (find_taken_edge_switch_expr): ... and here, handle cases
21445 with just a default label.
21446 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
21447 (group_case_labels): Likewise.
21448 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
21449 group_case_labels does anything cleanup the CFG again.
21451 2017-06-29 Bin Cheng <bin.cheng@arm.com>
21453 PR tree-optimization/81196
21454 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
21455 exit condition comparing two IVs.
21457 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
21459 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
21460 profile to the dummy entry at the end of the list of architectures.
21461 * config/arm/arm-cpu-cdata.h: Regenerated.
21463 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21464 Michael Collison <michael.collison@arm.com>
21467 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
21469 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
21470 (*aarch64_reg_<mode>3_minus_mask): New pattern.
21471 (*aarch64_<optab>_reg_di3_mask2): New pattern.
21472 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
21473 of shift when the shift amount is masked with constant equal to
21474 the size of the mode.
21475 * config/aarch64/predicates.md (subreg_lowpart_operator): New
21478 2017-06-29 Martin Liska <mliska@suse.cz>
21480 * config/i386/i386.opt: Change range from [1,5] to [0,5].
21482 2017-06-29 Yury Gribov <tetra2005@gmail.com>
21485 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
21488 (edge_growth_cache_entry::edge_growth_cache_entry): New
21490 (reset_edge_growth_cache): Update to use constructor.
21492 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21494 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
21495 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
21496 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
21498 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
21500 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
21501 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
21503 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
21505 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
21506 (*-linux-uclibc*): Add t-uclibc tmake_file.
21507 * config/t-musl: New.
21508 * config/t-uclibc: New.
21510 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
21512 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
21514 (gen_comm_data): Emit architectural setting of arch_prof.
21515 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
21517 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
21518 (armv8-m.base, armv8-m.main): Likewise.
21519 * arm-protos.h (arm_build_target): Add profile field.
21520 (arch_option): Likewise.
21521 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
21523 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
21524 arm_active_target.profile.
21526 2017-06-28 Richard Biener <rguenther@suse.de>
21528 PR middle-end/81227
21529 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
21530 TYPE_OVERFLOW_WRAPS.
21531 * match.pd (negate_expr_p): Likewise.
21532 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
21533 fold_build2, not fold_binary.
21535 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
21537 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
21538 Convert memory address to Pmode.
21539 (aarch64_print_operand): Assert MEM operands are always Pmode.
21541 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
21544 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
21545 Remove redundant if.
21546 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
21547 * config/arm/aarch-common-protos.h
21548 (aarch_forward_to_shift_is_not_shifted_re): Remove.
21549 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
21551 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
21554 * multiple_target.c (create_dispatcher_calls): Set the default
21555 clone to be static, not public.
21557 2017-06-28 Richard Biener <rguenther@suse.de>
21559 * tree-vect-loop.c (vectorizable_reduction): Move special
21560 cond reduction IV var creation ...
21561 (vect_create_epilog_for_reduction): ... here. Remove induction_index
21562 parameter. Use STMT_VINFO_VECTYPE.
21563 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
21566 2017-06-28 Martin Liska <mliska@suse.cz>
21569 * ipa-visibility.c (non_local_p): Handle visibility.
21571 2017-06-28 Martin Liska <mliska@suse.cz>
21574 * common.opt: Add IntegerRange to various options.
21575 * opt-functions.awk (integer_range_info): New function.
21576 * optc-gen.awk: Add integer_range_info to cl_options struct.
21577 * opts-common.c (decode_cmdline_option): Handle
21578 CL_ERR_INT_RANGE_ARG.
21579 (cmdline_handle_error): Likewise.
21580 * opts.c (print_filtered_help): Show valid interval in
21581 when --help is provided.
21582 * opts.h (struct cl_option): Add range_min and range_max fields.
21583 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
21585 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
21587 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
21588 (x * C EQ/NE y * C): New transformation.
21590 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
21592 * genmultilib (combination_space): Accept '+' in option names.
21594 2017-06-28 Martin Liska <mliska@suse.cz>
21597 * asan.c (instrument_derefs): Bail out inner references
21598 that are hard register variables.
21600 2017-06-28 Jakub Jelinek <jakub@redhat.com>
21603 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
21604 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
21606 2017-06-28 Richard Biener <rguenther@suse.de>
21608 * tree-vectorizer.h (vect_get_vec_defs): Remove.
21609 (vect_get_slp_defs): Adjust.
21610 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
21612 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
21614 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
21615 get_initial_defs_for_reduction instead of vect_get_vec_defs.
21616 (vectorizable_reduction): Adjust.
21617 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
21619 (vect_get_slp_defs): Likewise.
21620 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
21621 (vectorizable_bswap): Adjust.
21622 (vectorizable_call): Likewise.
21623 (vectorizable_conversion): Likewise.
21624 (vectorizable_assignment): Likewise.
21625 (vectorizable_shift): Likewise.
21626 (vectorizable_operation): Likewise.
21627 (vectorizable_store): Likewise.
21628 (vectorizable_condition): Likewise.
21629 (vectorizable_comparison): Likewise.
21631 2017-06-28 Michael Collison <michael.collison@arm.com>
21634 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
21636 (arm_gen_movmemqi): Removed unused variable 'i'.
21637 Convert 'for' loop into 'while' loop.
21638 (arm_expand_prologue): Remove last unnecessary set of insn.
21639 (thumb_pop): Remove unused variable 'pushed_words'.
21640 (thumb_exit): Remove last unnecessary set of regs_to_pop.
21642 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21644 * config/s390/predicates.md: Use s390_rel_address_ok_p.
21645 * config/s390/s390-protos.h: Add prototype of
21646 s390_rel_address_ok_p.
21647 * config/s390/s390.c (s390_got_symbol): New function.
21648 (s390_rel_address_ok_p): New function.
21649 (legitimize_pic_address): Use s390_rel_address_ok_p.
21650 (s390_load_got): Use s390_got_symbol.
21651 (s390_option_override): Issue error if
21652 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
21653 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
21655 * config/s390/s390.opt: New option mpic-data-is-text-relative.
21657 2017-06-27 Andrew Pinski <apinski@cavium.com>
21659 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
21660 (X * copysign (1.0, X)): New pattern.
21661 (X * copysign (1.0, -X)): New pattern.
21662 (copysign (-1.0, CST)): New pattern.
21664 2017-06-27 Joseph Myers <joseph@codesourcery.com>
21666 * genmultilib (combination_space): Remove variable.
21667 Validate reuse rules against regular expression for any sequence
21668 of multilib options in any order.
21670 2017-06-27 Michael Collison <michael.collison@arm.com>
21672 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
21673 call aarch64_split_simd_combine.
21674 * (aarch64_combine_internal<mode>): Delete pattern.
21675 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
21676 Allow register and subreg operands.
21678 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21680 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
21681 specific need, just fallback on defaults.
21682 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
21684 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21685 Olivier Hainque <hainque@adacore.com>
21687 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
21689 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
21690 targets. Pick a default if no particular attempt applied.
21691 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
21694 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21696 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
21697 (x86_64-wrs-vxworks7): Likewise.
21699 2017-06-27 Marek Polacek <polacek@redhat.com>
21702 * ubsan.c (instrument_null): Check get_base_address's result for null.
21704 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
21706 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
21708 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
21710 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
21711 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
21712 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
21713 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
21714 New function types.
21715 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
21716 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
21717 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
21718 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
21719 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
21720 BUILT_IN_FEUPDATEENV): New builtins.
21721 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
21722 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
21723 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
21724 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
21726 (builtin_structptr_types): Adjust size.
21727 * tree.c (builtin_structptr_types): Add four entries.
21729 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21730 Olivier Hainque <hainque@adacore.com>
21732 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
21733 (TLS_SYM): New local macro, forcing reference to __tls__ on
21734 link command lines for VxWorks 7 RTPs, triggering initialization
21736 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
21737 OS features TLS support, true for RTPs on VxWorks 7.
21738 * config/vxworks.c (vxworks_override_options): Setup emutls
21741 2017-06-27 Jakub Jelinek <jakub@redhat.com>
21743 * predict.c (test_prediction_value_range): Use -1U instead of -1
21744 to avoid narrowing conversion warning.
21745 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
21746 to avoid narrowing conversion warning.
21747 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
21749 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
21751 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21753 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
21754 64bit configurations.
21755 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
21756 (SIZE_TYPE): Likewise.
21757 * config/vxworks.c (vxworks_emutls_var_fields): Use
21758 long_unsigned_type_node instead of unsigned_type_node as the offset
21759 field type, which is "pointer" mode in emutls.c.
21761 2017-06-27 Jakub Jelinek <jakub@redhat.com>
21764 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
21766 PR middle-end/81207
21767 * gimple-fold.c (replace_call_with_call_and_fold): Handle
21768 gimple_vuse copying separately from gimple_vdef copying.
21770 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21772 * value-prof.c (free_hist): Remove call to memset and the enclosing if
21775 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
21776 Olivier Hainque <hainque@adacore.com>
21778 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
21779 for all vxworks7 targets.
21780 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
21781 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
21782 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
21783 variations for VX6/VX7 and 32/64bits later on in ...
21784 (VXWORKS_LIB_SPEC): Leverage new macros.
21785 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
21786 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
21788 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
21790 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
21791 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
21793 2017-06-26 Carl Love <cel@us.ibm.com>
21795 * config/rs6000/rs6000-c.c: Add support for built-in functions
21796 vector bool char vec_reve (vector bool char);
21797 vector signed char vec_reve (vector signed char);
21798 vector unsigned char vec_reve (vector unsigned char);
21799 vector bool int vec_reve (vector bool int);
21800 vector signed int vec_reve (vector signed int);
21801 vector unsigned int vec_reve (vector unsigned int);
21802 vector bool long long vec_reve (vector bool long long);
21803 vector signed long long vec_reve (vector signed long long);
21804 vector unsigned long long vec_reve (vector unsigned long long);
21805 vector bool short vec_reve (vector bool short);
21806 vector signed short vec_reve (vector signed short);
21807 vector double vec_reve (vector double);
21808 vector float vec_reve (vector float);
21809 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
21810 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
21811 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
21812 (altivec_vreve): New pattern.
21813 * config/rs6000/altivec.h (vec_reve): New define.
21814 * doc/extend.texi (vec_rev): Update the built-in documentation file
21815 for the new built-in functions.
21817 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21819 PR tree-optimization/71815
21820 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
21822 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
21824 (slsr_process_phi): Likewise.
21825 (replace_uncond_cands_and_profitable_phis): Don't replace a
21826 multiply candidate with a stride of 1 (copy or cast).
21827 (phi_incr_cost): Call uses_consumed_by_stmt rather than
21829 (lowest_cost_path): Likewise.
21830 (total_savings): Likewise.
21832 2017-06-26 Richard Biener <rguenther@suse.de>
21835 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
21836 Use def_builtin_pure for all gather builtins.
21838 2017-06-26 Richard Biener <rguenther@suse.de>
21840 PR tree-optimization/81203
21841 * tree-tailcall.c (find_tail_calls): Do not move stmts into
21842 non-dominating BBs.
21844 2017-06-26 Marek Polacek <polacek@redhat.com>
21847 * doc/invoke.texi: Document -Wmultistatement-macros.
21849 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
21851 * doc/sourcebuild.texi (ARM-specific attributes): Document new
21852 arm_neon_ok_no_float_abi effective target.
21854 2017-06-26 Richard Biener <rguenther@suse.de>
21856 PR tree-optimization/80928
21857 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
21858 (copy_bbs): Set BB_DUPLICATED flag early.
21859 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
21861 (execute_on_shrinking_pred): Likewise.
21862 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
21863 BB_DUPLICATED blocks.
21864 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
21865 iterate over all PHIs considering removal of *gsi.
21867 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
21869 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
21872 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21874 * config/rs6000/rs6000-string.c: (expand_block_clear,
21875 do_load_for_compare, select_block_compare_mode,
21876 compute_current_alignment, expand_block_compare,
21877 expand_strncmp_align_check, expand_strn_compare,
21878 expand_block_move, rs6000_output_load_multiple)
21879 Move functions related to string/block move/compare
21880 to a separate file.
21881 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
21882 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
21883 for this function which is now used in two files.
21884 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
21885 * config.gcc: Add rs6000-string.o to extra_objs for
21886 targets powerpc*-*-* and rs6000*-*-*.
21888 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
21891 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
21892 32-bit, since indexed is not valid for DImode.
21893 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
21894 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
21895 (define_peephole2 for Altivec d-form load): Add 32-bit support.
21896 (define_peephole2 for Altivec d-form store): Likewise.
21899 * multiple_target.c (create_dispatcher_calls): Only create the
21900 dispatcher call if the function is the default clone of a
21901 versioned function.
21903 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
21905 PR middle-end/80902
21906 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
21907 a call, force the call to not be a tail call.
21909 2017-06-23 Jeff Law <law@redhat.com>
21911 * doc/contrib.texi: Add entry for Steven Pemberton's work on
21914 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
21916 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
21917 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
21918 handling for early expansion of vector shifts (sl,sr,sra,rl).
21919 (builtin_function_type): Add vector shift right instructions
21920 to the unsigned argument list.
21922 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
21924 rtl-optimizatoin/79286
21925 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
21926 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
21927 trap. PIC register plus a const unspec without offset can never trap.
21929 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
21931 * tree.h (builtin_structptr_type): New type.
21932 (builtin_structptr_types): Declare new array.
21933 * tree.c (builtin_structptr_types): New array.
21934 (free_lang_data, build_common_tree_nodes): Use it.
21936 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
21939 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
21942 2017-06-22 Matt Turner <mattst88@gmail.com>
21944 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
21945 Lake models to skylake case. Assume skylake for unknown
21946 models with clflushopt.
21948 2017-06-22 Jeff Law <law@redhat.com>
21950 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
21951 frame sizes that do not satisfy aarch64_uimm12_shift.
21953 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
21955 * profile-count.h (apply_probability,
21956 apply_scale, probability_in): Fix checks for zero.
21958 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21960 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
21961 * doc/cppdiropts.texi (-I @var{dir}): Document it.
21963 2016-06-22 Richard Biener <rguenther@suse.de>
21965 * tree-vect-loop.c (vect_model_reduction_cost): Handle
21966 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
21967 REDUC_MAX_EXPR support.
21968 (vectorizable_reduction): Likewise.
21969 (vect_create_epilog_for_reduction): Likewise.
21971 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
21973 * match.pd (A / (1 << B) -> A >> B): New.
21974 * generic-match-head.c: Include optabs-tree.h.
21975 * gimple-match-head.c: Likewise.
21976 * optabs-tree.h (target_supports_op_p): New.
21977 * optabs-tree.c (target_supports_op_p): New.
21979 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21981 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
21982 $gcc_cv_ld --help output.
21983 (gcc_cv_ld_demangle): Likewise.
21984 (gcc_cv_ld_eh_frame_hdr): Likewise.
21985 (gcc_cv_ld_pie): Likewise.
21986 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
21987 (gcc_cv_ld_buildid): Likewise.
21988 (gcc_cv_ld_sysroot): Likewise.
21989 (ld_bndplt_support): Likewise.
21990 (ld_pushpopstate_support): Likewise.
21991 * configure: Regenerate.
21992 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
21994 2017-06-21 Jakub Jelinek <jakub@redhat.com>
21997 * config/i386/sse.md (round<mode>2): Renumber match_dup and
21998 operands indexes to avoid gap between operands and match_dups.
22000 2017-06-21 Andrew Pinski <apinski@cavium.com>
22002 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
22003 Increment Arith_shift and Arith_shift_reg by 1.
22004 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
22006 * config/aarch64/aarch64.c (thunderx_tunings): Enable
22007 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
22008 (aarch64_strip_extend): Add new argument and test for it.
22009 (aarch64_cheap_mult_shift_p): New function.
22010 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
22011 add a cost if it is true.
22012 Update calls to aarch64_strip_extend.
22013 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
22015 2017-06-21 Andrew Pinski <apinski@cavium.com>
22017 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
22019 (thunderxt88): Likewise.
22020 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
22021 (thunderx_prefetch_tune): New variable.
22022 (thunderx2t99_prefetch_tune): Update for the correct values.
22023 (thunderxt88_tunings): New variable.
22024 (thunderx_tunings): Use thunderx_prefetch_tune instead of
22025 generic_prefetch_tune.
22026 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
22028 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22030 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
22031 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
22032 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
22033 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
22034 (aarch64_atomic_cas<mode>, GPI): Likewise.
22036 2017-06-21 Martin Liska <mliska@suse.cz>
22038 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
22039 statements on cold and hot labels.
22040 * predict.c (tree_estimate_probability_bb): Remove the
22041 prediction from this place.
22043 2017-06-21 Martin Liska <mliska@suse.cz>
22045 PR tree-optimization/79489
22046 * gimplify.c (maybe_add_early_return_predict_stmt): New
22048 (gimplify_return_expr): Call the function.
22049 * predict.c (tree_estimate_probability_bb): Remove handling
22051 * predict.def: Update comment about early return predictor.
22052 * gimple-predict.h (is_gimple_predict): New function.
22053 * predict.def: Change default value of early return to 66.
22054 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
22056 * passes.def: Put pass_strip_predict_hints to the beginning of
22059 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
22061 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
22062 FUNCTION_DECL declarations.
22063 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
22065 (dwaf2out_decl): Likewise.
22066 * godump.c (go_early_global_decl): Skip call to the real debug hook
22067 for FUNCTION_DECL declarations.
22068 * passes.c (rest_of_decl_compilation): Skip call to the
22069 early_global_decl debug hook for FUNCTION_DECL declarations, unless
22070 -fdump-go-spec is passed.
22072 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
22074 * config/i386/i386.c (struct builtin_isa): New field pure_p.
22075 Reorder for compactness.
22076 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
22077 (def_builtin_pure, def_builtin_pure2): New functions.
22078 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
22080 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
22082 * match.pd (nop_convert): New predicate.
22083 ((A +- CST1) +- CST2): Allow some NOP conversions.
22085 2017-06-21 Jakub Jelinek <jakub@redhat.com>
22088 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
22089 with ctors/dtors if GOVD_SHARED is set.
22091 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22093 * config/aarch64/aarch64.md (movti_aarch64):
22094 Emit mov rather than orr.
22095 (movtf_aarch64): Likewise.
22096 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
22097 Emit mov rather than orr.
22099 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22101 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
22102 Swap alternatives, make integer dup more expensive.
22104 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22106 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
22107 Return true for non-tls symbols.
22109 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
22111 * config/aarch64/aarch64-cores.def (cortex-a55): New.
22112 (cortex-a75): Likewise.
22113 (cortex-a75.cortex-a55): Likewise.
22114 * config/aarch64/aarch64-tune.md: Regenerate.
22115 * doc/invoke.texi (-mtune): Document new values for -mtune.
22117 2017-06-21 Tom de Vries <tom@codesourcery.com>
22119 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
22120 stack_size feature.
22121 (Effective-Target Keywords, Other attributes): Suggest using
22122 dg-add-options stack_size feature to get stack limit in stack_size
22123 effective target documentation.
22125 2017-06-21 Julian Brown <julian@codesourcery.com>
22126 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22128 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
22129 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
22130 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
22132 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
22133 attribute type list for neon_multiply.
22134 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
22135 attribute type list for neon_multiply.
22136 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
22137 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
22138 attribute type list for neon_multiply.
22139 * config/arm/types.md (crypto_pmull): Add.
22140 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
22141 attribute type list.
22143 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
22145 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
22148 2017-06-20 Jakub Jelinek <jakub@redhat.com>
22150 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
22151 to make sure not to dereference a NULL cost_classes_ptr pointer.
22153 2017-06-20 Carl Love <cel@us.ibm.com>
22155 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22156 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22157 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22158 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22159 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22160 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22161 VMULOSW): New enum "unspec" values.
22162 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22163 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22164 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22165 altivec_vmulosw): New patterns.
22166 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22167 VMULOSW): Add definitions.
22169 2017-06-20 Julia Koval <julia.koval@intel.com>
22171 * config/i386/i386.c: Fix rounding expand for new pattern.
22172 * config/i386/subst.md: Fix pattern (parallel -> unspec).
22174 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22176 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
22177 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
22179 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22181 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
22184 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22186 * config/aarch64/aarch64-cores.def: Rearrange to sort by
22187 architecture, then by implementer ID.
22188 * config/aarch64/aarch64-tune.md: Regenerate.
22190 2017-06-20 Richard Biener <rguenther@suse.de>
22192 PR middle-end/81097
22193 * fold-const.c (split_tree): Fold to type before negating.
22195 2017-06-20 David Malcolm <dmalcolm@redhat.com>
22197 * diagnostic-show-locus.c
22198 (selftest::test_fixit_deletion_affecting_newline): New function.
22199 (selftest::diagnostic_show_locus_c_tests): Call it.
22201 2017-06-20 Andreas Schwab <schwab@suse.de>
22204 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
22207 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22209 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
22210 __ARM_FEATURE_COPROC according to support.
22212 2017-06-20 Jakub Jelinek <jakub@redhat.com>
22214 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22215 Rewritten to avoid overflow for > 32-bit pointers.
22218 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
22219 by removing enum keyword.
22220 (ubsan_type_descriptor): Likewise. Formatting fix.
22223 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
22224 splitter): Require TARGET_SSE2 in the condition.
22226 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
22229 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
22230 for doing vector set of SFmode on ISA 3.0.
22231 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
22232 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
22234 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
22235 SFmode value into a V4SF variable that was extracted from another
22236 V4SF variable without converting the element to double precision
22237 and back to single precision vector format.
22238 (vsx_insert_extract_v4sf_p9_2): Likewise.
22240 2017-06-19 Jakub Jelinek <jakub@redhat.com>
22242 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
22243 in UWHI to avoid undefined overflow.
22246 * ubsan.h (enum ubsan_encode_value_phase): New.
22247 (ubsan_encode_value): Change second argument to
22248 enum ubsan_encode_value_phase with default value of
22249 UBSAN_ENCODE_VALUE_GENERIC.
22250 * ubsan.c (ubsan_encode_value): Change second argument to
22251 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
22252 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
22253 create_tmp_var_raw instead of create_tmp_var and use a
22255 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
22256 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
22257 ubsan_encode_value callers.
22260 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
22261 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
22262 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
22264 2017-06-19 Richard Biener <rguenther@suse.de>
22266 PR middle-end/81118
22267 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
22268 estimates if we changed anything.
22270 2017-06-19 Richard Biener <rguenther@suse.de>
22272 PR tree-optimization/80887
22273 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
22274 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
22275 simplified lookups, then reset mprts_hook.
22276 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
22278 (try_to_simplify): Likewise.
22280 2017-06-19 Martin Liska <mliska@suse.cz>
22283 * gimplify.c (gimplify_switch_expr):
22284 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
22286 2017-06-19 Martin Liska <mliska@suse.cz>
22288 * doc/install.texi: Document that PGO runs in 4 stages.
22290 2017-06-19 Martin Liska <mliska@suse.cz>
22293 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
22294 to dispatcher function name.
22295 * multiple_target.c (replace_function_decl): New function.
22296 (create_dispatcher_calls): Redirect both edges and references.
22298 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
22300 * profile-count.c (profile_count::dump): Dump quality.
22301 (profile_count::differs_from_p): Update for unsigned val.
22302 * profile-count.h (profile_count_quality): New enum.
22303 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
22305 2017-06-19 Richard Biener <rguenther@suse.de>
22307 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
22308 struct function as arg.
22309 (estimate_numbers_of_iterations): Export overload with loop arg.
22310 (free_numbers_of_iterations_estimates_loop): Use an overload of
22311 free_numbers_of_iterations_estimates instead.
22312 * tree-cfg.c (remove_bb): Adjust.
22313 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
22314 * tree-parloops.c (gen_parallel_loop): Likewise.
22315 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
22317 (tree_unroll_loops_completely): Likewise.
22318 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
22319 Use an overload instead and export.
22320 (estimated_loop_iterations): Adjust.
22321 (max_loop_iterations): Likewise.
22322 (likely_max_loop_iterations): Likewise.
22323 (estimate_numbers_of_iterations): Take struct function as arg
22325 (loop_exits_before_overflow): Adjust.
22326 (free_numbers_of_iterations_estimates_loop): Use an overload.
22327 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
22328 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
22330 2017-06-19 Richard Biener <rguenther@suse.de>
22333 * ipa-prop.c (find_constructor_constant_at_offset): Handle
22334 RANGE_EXPR conservatively.
22336 2017-06-16 Carl Love <cel@us.ibm.com>
22338 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22339 definitions for vec_float, vec_float2, vec_floato,
22340 vec_floate built-ins.
22341 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
22342 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
22344 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
22345 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
22346 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
22347 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
22348 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
22349 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
22350 vec_floato): Add builtin defines.
22351 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
22352 Update the built-in documentation file for the new built-in
22355 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22357 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
22358 (mthumb): Mark as the negative of -marm.
22360 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22362 * doc/invoke.texi (ARM Options, -mcpu): Document supported
22364 (ARM Options, -mtune): Document that this accepts the same
22365 extension options as -mcpu.
22366 (ARM Options, -mfpu): Document addition of -mfpu=auto.
22368 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22370 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
22371 permitted extensions.
22373 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22375 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
22376 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
22377 (armv8-m.main): Add option +nodsp.
22378 * config/arm/arm-cpu-cdata.h: Regenerated.
22380 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22382 * config/arm/t-fuchsia: New file.
22383 * config.gcc (arm*-*-fuchsia*): Use it.
22385 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22387 * config/arm/t-symbian: Rewrite for new option infrastructure.
22389 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22391 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
22392 (MULTILIB_REQUIRED): Likewise.
22394 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22396 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
22397 (MULTILIB_RESUE): Likewise.
22398 (MULTILIB_MATCHES): Likewise.
22399 (MULTLIB_REQUIRED): Likewise.
22401 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22403 * config/arm/t-rtems: Rewrite for new option framework.
22405 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22407 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
22408 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
22409 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
22410 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
22411 * config/arm/t-multilib: ... here.
22412 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
22413 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
22414 armv7-a and armv8*-a when A-profile libraries have not been built.
22415 * config/arm/t-rmprofile: Rewrite.
22417 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22419 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
22420 with a backslash. Remove the backslash after substituting unescaped
22422 * doc/fragments.texi (MULTILIB_REUSE): Document it.
22424 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22426 * config.gcc: (arm*-*-*): When building a-profile libraries, force
22427 the driver to pass through the default setting of -mfloat-abi.
22428 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
22430 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
22431 (all_feat_combs): New rule.
22432 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
22434 * config/arm/t-aprofile: Rewrite.
22436 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22438 * config/arm/arm.h (FPUTYPE_AUTO): Define.
22439 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
22440 fpu is not specified by the user/command-line.
22441 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
22442 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
22443 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
22444 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
22445 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
22446 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
22448 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22450 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
22451 * config/arm/t-arm-elf: Rewritten.
22453 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22455 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
22456 have some floating-point instructions.
22457 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
22458 (TARGET_MAYBE_HARD_FLOAT): New macro.
22459 * config/arm/arm-builtins.c (arm_init_builtins): Use
22460 TARGET_MAYBE_HARD_FLOAT.
22461 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
22463 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22465 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
22466 (configargs.h): Include it.
22467 (arm_print_hint_for_fpu_option): New function.
22468 (arm_parse_fpu_option): New function.
22469 (candidate_extension): New class.
22470 (arm_canon_for_multilib): New function.
22471 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
22472 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
22473 (ARCH_CANONICAL_SPECS): New macro.
22474 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
22476 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22478 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
22479 are set after handling multilib fragments. Set target_cpu_default2
22482 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22484 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
22486 (arm*-*-*): Set target_cpu_default2 to a quoted string.
22487 * config/arm/parsecpu.awk (check_cpu): Validate any extension
22489 (check_arch): Likewise.
22490 * config/arm/arm.c (arm_configure_build_target): Handle
22491 TARGET_CPU_DEFAULT being a string constant. Scan any feature
22492 options in the default.
22494 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22496 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
22497 when an option is an alias of another.
22498 * config/arm/parsecpu.awk (optalias): New parser token.
22499 (gen_comm_data): Mark non-alias options as such. Emit entries
22500 for extension aliases.
22501 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
22502 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
22503 (armv6kz, armv6zk, armv6t2): Likewise.
22504 (armv7): Make vfpv3-d16 an alias.
22505 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
22507 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
22508 Sort in canonical order.
22509 (armv8-a): Sort in canonical order.
22510 (armv8.1-a, armv8.2-a): Likewise.
22511 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
22513 (cortex-a9): Sort in canonical order.
22514 * config/arm/arm.c (selftests.h): Include it.
22515 (arm_test_cpu_arch_data): New function.
22516 (arm_run_self_tests): New function.
22517 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
22518 (targetm): Move declaration to the end of the file.
22519 * arm-cpu-cdata.h: Regenerated.
22521 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22523 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
22524 call to target_mode_check describing the type of option passed.
22525 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
22526 (arm_target_thumb_only): Use arm_parse_arch_option_name or
22527 arm_parse_cpu_option_name to match parameters against list of
22529 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
22530 arm_arch_core_flags data structure.
22531 * config/arm/arm-cpu_cdata.h: Regenerated.
22533 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22535 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
22537 (arm_print_hint_for_cpu_option): Likewise.
22538 (arm_print_hint_for_arch_option): Likewise.
22539 (arm_parse_cpu_option_name): Likewise.
22540 (arm_parse_arch_option_name): Likewise.
22541 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
22542 of entries in the all_fpus list.
22543 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
22544 (arm_parse_cpu_option_name): Declare.
22545 (arm_parse_arch_option_name): Declare.
22546 (arm_parse_option_features): Declare.
22547 (arm_intialize_isa): Declare.
22548 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
22550 (gen_comm_data): ... here. Make definitions non-static.
22551 * config/arm/arm-cpu-data.h: Regenerated.
22552 * config/arm/arm-cpu-cdata.h: Regenerated.
22554 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22556 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
22557 (cpu_arch_extension): New structure.
22558 (cpu_arch_option, arch_option, cpu_option): New structures.
22559 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
22560 architecture types.
22561 (gen_data): Generate new format data tables.
22562 * config/arm/arm.c (cpu_tune): New structure.
22563 (cpu_option, processors): Delete.
22564 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
22565 (arm_print_hint_for_cpu_option): ... this and ...
22566 (arm_print_hint_for_arch_option): ... this.
22567 (arm_parse_arch_cpu_name): Delete. Replace with ...
22568 (arm_parse_cpu_option_name): ... this and ...
22569 (arm_parse_arch_option_name): ... this.
22570 (arm_unrecognized_feature): Change type of target parameter to
22572 (arm_parse_arch_cpu_features): Delete. Replace with ...
22573 (arm_parse_option_features): ... this.
22574 (arm_configure_build_target): Rework to use new configuration data
22576 (arm_print_tune_info): Rework for new configuration data tables.
22577 * config/arm/arm-cpu-data.h: Regenerated.
22578 * config/arm/arm-cpu.h: Regenerated.
22580 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22582 * Makefile.in (OBJS): Move sbitmap.o from here ...
22583 (OBJS-libcommon): ... to here.
22585 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22587 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
22588 (ISA_ALL_CRYPTO): New macro.
22589 (ISA_ALL_SIMD): New macro
22590 (ISA_ALL_FP): New macro.
22591 * config/arm/arm.c (fpu_bitlist): Update initializer.
22592 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
22594 (arm9e): Add fpu. Add option for nofp
22595 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
22596 (arm926ej-s, arm1026ej-s): Likewise.
22597 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
22598 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
22599 neon-fp16, neon-vfpv4, nofp and nosimd.
22600 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
22601 (cortex-a8): Add fpu. Add option for nofp.
22602 (cortex-a9): Add fpu. Add options for nosimd and nofp.
22603 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
22604 (cortex-r4f): Add fpu.
22605 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
22606 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
22608 (cortex-r8): Likewise.
22609 (cortex-m4): Add fpu. Add option for nofp.
22610 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
22611 (cortex-a17.cortex-a7): Likewise.
22612 (cortex-a32): Add fpu. Add options for crypto and nofp.
22613 (cortex-a35, cortex-a53): Likewise.
22614 (cortex-a57): Add fpu. Add option for crypto.
22615 (cortex-a72, cortex-a73): Likewise.
22616 (exynos-m1): Likewise.
22617 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
22618 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
22619 (cortex-m33): Add fpu. Add option for nofp.
22620 * config/arm/arm-cpu-cdata.h: Regenerated
22621 * config/arm/arm-cpu-data.h: Regenerated.
22623 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22625 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
22626 (armv5te, armv5tej): Likewise.
22627 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
22628 (armv7): Add options fp and vfpv3-d16.
22629 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
22630 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
22632 (armv7ve): Likewise.
22633 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
22634 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
22635 (armv8-a): Add nocrypto option.
22636 (armv8.1-a, armv8.2-a): Likewise.
22637 (armv8-m.main): add options fp, fp.dp and nofp.
22639 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22641 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
22643 (armv8-a+crc): Delete.
22644 (armv8.1-a): Add options simd, crypto and nofp.
22645 (armv8.2-a): Add options fp16, simd, crypto and nofp.
22646 (armv8.2-a+fp16): Delete.
22647 (armv8-m.main): Add option dsp.
22648 (armv8-m.main+dsp): Delete.
22649 (cortex-a8): Add fpu. Add nofp option.
22650 (cortex-a9): Add fpu. Add nofp and nosimd options.
22651 * config/arm/parsecpu.awk (gen_data): Generate option tables and
22652 link to main cpu and architecture data structures.
22653 (gen_comm_data): Only put isa attributes from the main architecture
22655 (option): New statement for architecture and CPU entries.
22656 * arm.c (struct cpu_option): New structure.
22657 (struct processors): Add entry for options.
22658 (arm_unrecognized_feature): New function.
22659 (arm_parse_arch_cpu_name): Ignore any characters after the first
22661 (arm_parse_arch_cpu_feature): New function.
22662 (arm_configure_build_target): Separate out any CPU and architecture
22663 features and parse separately. Don't error out if -mfpu=auto is
22664 used with only an architecture string.
22665 (arm_print_asm_arch_directives): New function.
22666 (arm_file_start): Call it.
22667 * config/arm/arm-cpu-cdata.h: Regenerated.
22668 * config/arm/arm-cpu-data.h: Likewise.
22669 * config/arm/arm-tables.opt: Likewise.
22671 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22673 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
22674 assembler when it is not -mfpu=auto.
22676 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22678 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
22679 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
22680 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
22681 (ASM_CPU_SPEC): Rewrite.
22682 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
22683 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
22684 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
22685 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
22686 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
22687 copied string is NUL-terminated. Also strip any characters prefixed
22689 (arm_rewrite_selected_arch): New function.
22690 (arm_rewrite_march): New function.
22692 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22694 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
22695 (x_arm_cpu_string, x_arm_tune_string): Likewise.
22696 (march, mcpu, mtune): Convert to string-based options.
22697 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
22698 (arm_parse_arch_cpu_name): New function.
22699 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
22700 identify selected architecture or CPU.
22701 (arm_option_save): New function.
22702 (TARGET_OPTION_SAVE): Redefine.
22703 (arm_option_restore): Restore string options.
22704 (arm_option_print): Print string options.
22706 2017-06-16 Martin Sebor <msebor@redhat.com>
22708 PR tree-optimization/80933
22709 PR tree-optimization/80934
22710 * builtins.c (fold_builtin_3): Do not handle bcmp here.
22711 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
22712 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
22713 (gimple_fold_builtin): Call them.
22715 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22717 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
22718 as unlikely; update profile.
22720 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22722 * predict.c (force_edge_cold): Handle declaring edges impossible
22725 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22727 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
22729 (try_unroll_loop_completely): Fix reporting.
22731 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22733 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
22735 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
22738 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
22739 if given a non-constant argument for an intrinsic which requires a
22742 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22744 * profile.c (compare_freqs): New function.
22745 (branch_prob): Sort edge list.
22746 (find_spanning_tree): Assume that the list is priority sorted.
22748 2017-06-16 Richard Biener <rguenther@suse.de>
22750 PR tree-optimization/81090
22751 * passes.def (pass_record_bounds): Remove.
22752 * tree-pass.h (make_pass_record_bounds): Likewise.
22753 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
22754 make_pass_record_bounds): Likewise.
22755 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
22756 not free niter estimates at the beginning but at the end.
22757 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
22759 2017-06-16 Richard Biener <rguenther@suse.de>
22761 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
22762 initializer to workaround ICE in host GCC 4.8.
22764 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22766 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
22768 (clone_inlined_nodes): Update.
22770 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22772 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
22773 prefetch settings, and enable prefetching by default at -O3.
22775 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22777 * config/aarch64/aarch64.c (aarch64_override_options_internal):
22778 Set flag_prefetch_loop_arrays according to tuning data.
22780 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22782 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
22783 New tune structure.
22784 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
22785 [Unrelated to main purpose of the patch] Place the pointer field last
22786 to enable type checking errors when tune structure are wrongly merged.
22787 * config/aarch64/aarch64.c (generic_prefetch_tune,)
22788 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
22789 (thunderx2t99_prefetch_tune): New tune constants.
22790 (tune_params *_tunings): Update all tunings (no functional change).
22791 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
22792 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
22793 from tunings structures.
22795 2017-06-16 Jakub Jelinek <jakub@redhat.com>
22798 * ubsan.c (instrument_null): Add T argument, use it instead
22799 of computing it based on IS_LHS.
22800 (instrument_object_size): Likewise.
22801 (pass_ubsan::execute): Adjust instrument_null and
22802 instrument_object_size callers to pass gimple_get_lhs or
22803 gimple_assign_rhs1 result to it. Use instrument_null instead of
22804 calling get_base_address and instrument_mem_ref. Handle
22805 aggregate call arguments for object-size sanitization.
22807 2017-06-16 Yury Gribov <tetra2005@gmail.com>
22809 PR tree-optimization/81089
22810 * tree-vrp.c (is_masked_range_test): Validate operands of
22813 2017-06-15 Martin Sebor <msebor@redhat.com>
22816 * dumpfile.c (dump_register): Avoid calling memset to initialize
22817 a class with a default ctor.
22818 * gcc.c (struct compiler): Remove const qualification.
22819 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
22820 * hash-table.h: Ditto.
22821 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
22823 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
22824 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
22826 * params.h (struct param_info): Make struct members non-const.
22827 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
22828 with default initialization.
22829 * vec.h (vec_copy_construct, vec_default_construct): New helper
22831 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
22832 with vec_copy_construct.
22833 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
22834 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
22835 * doc/invoke.texi (-Wclass-memaccess): Document.
22837 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22839 * emit-rtl.h (is_leaf): Update comment about local
22840 register allocator.
22842 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22845 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
22846 for a variable to have a section before checking if the section has a
22848 Set section to.persistent if persistent attribute is set.
22849 Warn if .persistent attribute is used on an automatic variable.
22851 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
22853 PR rtl-optimization/80474
22854 * reorg.c (update_block): Do not ignore instructions in a delay slot.
22856 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
22858 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
22861 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
22863 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
22864 (casesi): Emit bounds checking as RTL.
22865 (casesi_internal_mips16_<mode>): Remove bounds checking.
22867 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
22869 * config/xtensa/xtensa.c (xtensa_option_override): Append
22870 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
22871 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
22872 xtensa_doloop_hooks): Define unconditionally.
22873 (xtensa_reorg_loops): Only call reorg_loops in the presence of
22875 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
22876 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
22877 for it in xtensa_option_override.
22878 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
22879 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
22881 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
22883 * doc/cppopts.texi: Document '-' special value to -MF.
22885 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
22887 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
22888 (cortex_a53_fconst): Likewise.
22889 (cortex_a53_fpmul): Likewise.
22890 (cortex_a53_f_load_64): Likewise.
22891 (cortex_a53_f_load_many): Likewise.
22892 (cortex_a53_advsimd_alu): Likewise.
22893 (cortex_a53_advsimd_alu_q): Likewise.
22894 (cortex_a53_advsimd_mul): Likewise.
22895 (cortex_a53_advsimd_mul_q): Likewise.
22896 (fpmac bypass): Add new bypass for fpmac-fpmac case.
22897 Add missing fmul, r2f_cvt and fconst cases.
22899 2017-06-14 Richard Biener <rguenther@suse.de>
22901 PR middle-end/81088
22902 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
22904 (fold_binary_loc): When associating do not treat pre-existing
22905 TREE_OVERFLOW on literal constants as a reason to allow
22906 TREE_OVERFLOW on associated literal constants.
22908 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22910 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
22911 (MASK_FEATURES): New macro.
22912 * config/sparc/sparc.c (sparc_option_override): Remove the special
22913 handling of -mfpu and generalize it to all MASK_FEATURES switches.
22915 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22917 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
22918 a division of 0 if non-call exceptions are enabled.
22920 2017-06-14 Andrew Pinski <apinski@cavium.com>
22921 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22924 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
22925 Improve vector initialization code gen for only variable case.
22927 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22929 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
22931 2017-06-14 Richard Biener <rguenther@suse.de>
22933 PR tree-optimization/81083
22934 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
22937 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22939 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
22940 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
22941 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
22942 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
22943 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
22944 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
22946 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22948 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
22949 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
22951 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22953 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
22955 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22957 * config/rs6000/t-rtems: Don't handle SPE.
22959 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22961 * config/rs6000/t-linux: Don't handle SPE.
22963 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22965 * config/rs6000/eabispe.h: Delete file.
22967 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22969 * config/rs6000/t-spe: Delete file.
22971 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22973 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
22974 (rs6000_legitimate_offset_address_p): Return false for anything in
22975 V2SImode or V2SFmode.
22977 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22979 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
22980 except V2SF and V2SI. Rearrange the vector modes, and add comments.
22981 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
22983 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
22984 (rs6000_legitimate_offset_address_p): Ditto.
22985 (rs6000_emit_move): Ditto.
22986 (rs6000_init_builtins): Remove V4HI_type_node.
22988 2017-06-13 Martin Liska <mliska@suse.cz>
22991 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
22992 (gate_asan): Likewise.
22993 * asan.h (asan_no_sanitize_address_p): Remove the function.
22994 (sanitize_flags_p): New function.
22995 * builtins.def: Fix coding style.
22996 * common.opt: Use renamed enum value.
22997 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
22998 * doc/extend.texi: Document no_sanitize attribute.
22999 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
23000 to SANITIZE_UNDEFINED_NONDEFAULT.
23001 * gcc.c (sanitize_spec_function): Use the renamed enum value.
23002 * gimple-fold.c (optimize_atomic_compare_exchange_p):
23003 Use sanitize_flags_p.
23004 * gimplify.c (gimplify_function_tree): Likewise.
23005 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
23006 * opts.c (parse_no_sanitize_attribute): New function.
23007 (common_handle_option): Use renamed enum value.
23008 * opts.h (parse_no_sanitize_attribute): Declare.
23009 * tree.c (sanitize_flags_p): New function.
23010 * tree.h: Declared here.
23011 * tsan.c: Use sanitize_flags_p.
23012 * ubsan.c (ubsan_expand_null_ifn): Likewise.
23013 (instrument_mem_ref): Likewise.
23014 (instrument_bool_enum_load): Likewise.
23015 (do_ubsan_in_current_function): Remove the function.
23016 (pass_ubsan::execute): Use sanitize_flags_p.
23017 * ubsan.h: Remove do_ubsan_in_current_function
23018 * tree-cfg.c (print_no_sanitize_attr_value): New function.
23019 (dump_function_to_file): Use it here.
23021 2017-06-13 Martin Jambor <mjambor@suse.cz>
23023 PR tree-optimization/80803
23024 PR tree-optimization/81063
23025 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
23026 (propagate_subaccesses_across_link): Enqueue subtree whenever
23027 necessary instead of relying on the caller.
23029 2017-06-13 Martin Jambor <mjambor@suse.cz>
23031 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
23032 that have a first_link.
23033 (sort_and_splice_var_accesses): Do not check first_link before
23035 (subtree_mark_written_and_enqueue): Likewise.
23036 (propagate_all_subaccesses): Likewise and do not stop at first
23037 parent with a first_link.
23039 2017-06-13 Martin Jambor <mjambor@suse.cz>
23041 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
23044 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23046 * match.pd: New pattern.
23048 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23050 * tree-vrp.c (is_masked_range_test): New function.
23051 (register_edge_assert_for): Determine ranges for
23054 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23056 PR tree-optimization/67328
23057 * fold-const.c (maskable_range_p): New function.
23058 (build_range_check): Generate bittests if possible.
23060 2017-06-13 Martin Liska <mliska@suse.cz>
23062 * gimple-pretty-print.c (dump_probability): Add new argument.
23063 (dump_edge_probability): Dump both probability and count.
23064 (dump_gimple_label): Likewise.
23065 (dump_gimple_bb_header): Likewise.
23067 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
23070 * config/avr/avr-devices.c: Fix indentation.
23071 * config/avr/gen-avr-mmcu-specs.c: Dito.
23073 2017-06-13 Richard Biener <rguenther@suse.de>
23075 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
23076 instead get vector type from stmt_info.
23077 (vectorizable_reduction): Adjust. Remove dead code.
23079 2017-06-13 Richard Biener <rguenther@suse.de>
23081 PR middle-end/81065
23082 * fold-const.c (extract_muldiv_1): Remove bogus distribution
23083 case of C * (x * C2 + C3).
23084 (fold_addr_of_array_ref_difference): Properly fold index difference.
23086 2017-06-12 David S. Miller <davem@davemloft.net>
23089 * config/sparc/sparc.md (return expander): Emit frame blockage if
23090 function uses alloca.
23092 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
23094 * combine.c (make_field_assignment): Check len rather than the mode
23095 precision when calling force_to_mode.
23097 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
23099 Support multilibs and devices that see flash in RAM address range.
23102 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
23103 (avr_mcu_t) <flash_pm_offset>: New field.
23104 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
23105 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
23106 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
23107 (AVR_TINY_PM_OFFSET): Remove macro.
23108 * config/avr/avr.opt (-mshort-calls): New option.
23109 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23110 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
23111 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
23112 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
23113 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
23114 instead of avr_arch->have_jmp_call.
23115 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
23116 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
23117 avr_arch->flash_pm_offset to define.
23118 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
23119 new field flash_pm_offset. Add entry for avrxmega3.
23120 (avr_texinfo): Add entry for avrxmega3.
23121 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
23122 attiny212, attiny214,
23123 attiny412, attiny414, attiny416, attiny417,
23124 attiny814, attiny816, attiny817,
23125 attiny1614, attiny1616, attiny1617,
23126 attiny3214, attiny3216, attiny3217.
23127 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
23128 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
23129 (avr_print_operand_address) [AVR_TINY]: Same.
23130 (avr_asm_init_sections) <readonly_data_section>: Only patch
23131 callback if avr_arch->flash_pm_offset = 0.
23132 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
23133 for rodata if avr_arch->flash_pm_offset != 0.
23134 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
23135 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
23136 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
23137 (m_options): Append opt_rcall.
23138 (m_dirnames): Append dir_rcall.
23139 * config/avr/t-multilib: Regenerate.
23141 * configure.ac [target=avr]: Check whether avrxmega3 default
23142 linker description file works as needed.
23143 * configure: Regenerate.
23144 * doc/avr-mmcu.texi: Regenerate.
23145 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
23146 <__AVR_ARCH__>: Document avrxmega3 and 103.
23147 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
23148 <__AVR_SHORT_CALLS__>: Document it.
23149 <__AVR_PM_BASE_ADDRESS__>: Document it.
23150 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
23151 (AVR Variable Attributes) <progmem>: Document this is
23152 not needed for avrxmega3.
23153 (AVR Named Address Spaces) <__flash>: Dito.
23155 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
23157 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
23159 2017-06-12 Doug Rupp <rupp@adacore.com>
23161 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
23162 Append vxworks-stdint.h to the tm_file list.
23163 * config/vxworks-stdint.h: New file.
23165 2017-06-12 Martin Liska <mliska@suse.cz>
23167 PR tree-optimization/81041
23168 * tree-profile.c (gimple_gen_ic_func_profiler):
23169 Create an extra BB in profile-generate
23170 (gimple_gen_time_profiler): Likewise.
23172 2017-06-12 Jakub Jelinek <jakub@redhat.com>
23174 PR tree-optimization/81003
23175 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
23176 (update_range_test): Use it instead of force_gimple_operand_gsi.
23178 2017-06-12 Richard Biener <rguenther@suse.de>
23180 PR tree-optimization/81053
23181 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
23182 with backedge value not defined in loop. Simplify def stmt
23185 2017-06-11 Tom de Vries <tom@codesourcery.com>
23188 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
23190 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
23191 nvptx_cannot_force_const_mem.
23193 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23195 * opts.c (finish_options): Move test for flag_split_stack after
23196 it has been initialized.
23198 2017-06-11 Jason Merrill <jason@redhat.com>
23200 * tree.h (id_equal): New.
23201 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
23202 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
23203 instead of strcmp of IDENTIFIER_POINTER.
23205 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23207 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
23208 (mark_all_inlined_calls_cdtor): Fix formating.
23209 (inline_transform): Rescale profile before inlining.
23211 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23213 * cgraph.h (cgraph_edge::clone): Update prototype.
23214 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
23215 (cgraph_node::create_clone): Update.
23216 (cgraph_node::create_version_clone): Update.
23217 * tree-inline.c (copy_bb): Update.
23218 (expand_call_inline): Update.
23220 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
23222 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
23223 factored out from ...
23224 (rs6000_emit_prologue): ... here.
23226 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
23228 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
23229 factored out from ...
23230 (rs6000_emit_prologue): ... here.
23232 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23234 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
23236 (handle_missing_profiles): Fix computation of tp_first_run.
23237 (counts_to_freqs): Do not touch freqs when count is 0.
23239 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23241 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
23244 2017-06-10 Tom de Vries <tom@codesourcery.com>
23246 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
23247 attributes): Document signal effective target.
23249 2017-06-10 Tom de Vries <tom@codesourcery.com>
23251 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23252 Document effective target stack_size.
23254 2017-06-09 David Malcolm <dmalcolm@redhat.com>
23256 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
23257 to the edit_context if they can be auto-applied.
23259 2017-06-9 Ian Lance Taylor <iant@golang.org>
23261 * opts.c (finish_options): If -fsplit-stack, disable implicit
23262 -forder-blocks-and-partition.
23263 * doc/invoke.texi (Optimize Options): Document that when using
23264 -fsplit-stack -forder-blocks-and-partition is not implicitly
23267 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23269 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
23270 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
23271 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
23272 * builtins.def (abort, trap, unreachable): Declare cold.
23273 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
23274 * tree-core.h (ECF_COLD): New.
23275 * tree.c (set_call_expr_flags): Handle ECF_COLD.
23276 (build_common_builtin_nodes): Mark unreachable and abort as cold.
23278 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23280 * predict.c (unlikely_executed_stmt_p): Cleanup.
23282 2017-06-09 Richard Biener <rguenther@suse.de>
23284 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
23285 model if the ref is always written to.
23287 2017-06-09 Tamar Christina <tamar.christina@arm.com>
23289 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
23291 2017-06-09 Tamar Christina <tamar.christina@arm.com>
23293 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
23296 2017-06-09 Tom de Vries <tom@codesourcery.com>
23299 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
23300 "target cannot support label values" when encountering LABEL_REF.
23302 2017-06-09 Martin Liska <mliska@suse.cz>
23304 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
23305 (gimple_gen_ic_func_profiler): Emit direct comparison
23306 of __gcov_indirect_call_callee with NULL.
23307 (gimple_gen_time_profiler): Change probability from
23308 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
23310 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23312 * profile.c (edge_gcov_counts): Turn to pointer.
23313 (compute_branch_probabilities, compute_branch_probabilities): Update.
23314 (branch_prob): Do not clear edge_gcov_count.
23315 * profile.h (edge_gcov_counts): Turn to pointer.
23316 (edge_gcov_count): Update.
23318 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23320 * gimple.h (gimple_check_failed): Mark cold.
23322 2017-06-09 Richard Biener <rguenther@suse.de>
23324 PR tree-optimization/66623
23325 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
23326 refactor check_reduction into two parts, properly computing
23327 whether we have to check reduction validity for outer loop
23330 2017-06-09 Richard Biener <rguenther@suse.de>
23332 PR tree-optimization/79483
23333 * graphite-scop-detection.c (order): New global.
23334 (get_order): Compute bb to order mapping that satisfies code
23335 generation constraints.
23336 (cmp_pbbs): New helper.
23337 (build_scops): Start domwalk at entry block, sort generated
23340 2017-06-09 Richard Biener <rguenther@suse.de>
23342 PR middle-end/81007
23343 * ipa-polymorphic-call.c
23344 (ipa_polymorphic_call_context::restrict_to_inner_class):
23345 Skip FIELD_DECLs with error_mark_node type.
23346 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
23349 2017-06-09 Martin Liska <mliska@suse.cz>
23351 * predict.c (struct branch_predictor): New struct.
23352 (test_prediction_value_range): New test.
23353 (predict_c_tests): New function.
23354 * selftest-run-tests.c (selftest::run_tests): Run the function.
23355 * selftest.h: Declare new tests.
23357 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
23360 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
23361 gen_add3_insn did not fail.
23362 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
23363 r0, construct that number in a temporary reg and add that reg to r0.
23364 If asked to put the result in r0 as well, fail.
23366 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
23368 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23369 for early expansion of vec_eqv.
23371 2017-06-08 Jakub Jelinek <jakub@redhat.com>
23373 PR middle-end/81005
23374 * ubsan.c (instrument_null): Avoid pointless code temporary.
23375 (pass_ubsan::execute): Instrument aggregate arguments of calls.
23377 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
23381 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
23384 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
23387 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23389 * predict.c (unlikely_executed_edge_p): Move ahead.
23390 (probably_never_executed_edge_p): Use it.
23392 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
23394 PR middle-end/79988
23395 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
23396 gimple_call_builtin_p call.
23398 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23400 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
23401 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
23402 rtl_check_failed_type2, rtl_check_failed_code1,
23403 rtl_check_failed_code2, rtl_check_failed_code_mode,
23404 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
23405 rtvec_check_failed_bounds, rtl_check_failed_flag,
23406 _fatal_insn_not_found, _fatal_insn): Likewise.
23407 * tree.h (tree_contains_struct_check_failed,
23408 tree_check_failed, tree_not_check_failed,
23409 tree_class_check_failed, tree_range_check_failed,
23410 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
23411 tree_vec_elt_check_failed, phi_node_elt_check_failed,
23412 tree_operand_check_failed, omp_clause_check_failed,
23413 omp_clause_operand_check_failed, omp_clause_range_check_failed):
23416 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23418 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
23419 flag_branch_probabilities.
23420 * ipa-inline.c (edge_badness): Likewise.
23421 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
23422 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
23423 * predict.c (maybe_hot_frequency_p): Likewise.
23424 (probably_never_executed): Likewise.
23425 * sched-ebb.c (schedule_ebbs): Likewise.
23426 * sched-rgn.c (find_single_block_region): Likewise.
23427 * tracer.c (tail_duplicate): Likewise.
23429 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23431 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
23432 longer requires x_flag_profile_use.
23434 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23436 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
23437 instead of flag_reorder_blocks_and_partition.
23438 * dbxout.c (dbxout_function_end): Likewise.
23439 * dwarf2out.c (gen_subprogram_die): Likewise.
23440 * haifa-sched.c (sched_create_recovery_edges): Likewise.
23441 * hw-doloop.c (reorg_loops): Likewise.
23442 * varasm.c (assemble_start_function,
23443 assemble_end_function): Likewise.
23444 (decide_function_section): Do not check for
23445 flag_reorder_blocks_and_partition.
23447 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
23449 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
23451 (chkp_get_hard_register_fake_addr_expr): Ditto.
23452 (chkp_build_addr_expr): Add check for hard reg case.
23453 (chkp_parse_array_and_component_ref): Ditto.
23454 (chkp_find_bounds_1): Ditto.
23455 (chkp_process_stmt): Don't generate bounds store for
23458 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23460 * predict.c (maybe_hot_bb_p): Do not check profile status.
23461 (maybe_hot_edge_p): Likewise.
23462 (probably_never_executed): Check for zero counts even if profile
23464 (unlikely_executed_edge_p): New function.
23465 (unlikely_executed_stmt_p): New function.
23466 (unlikely_executed_bb_p): New function.
23467 (set_even_probabilities): Use unlikely predicates.
23468 (combine_predictions_for_bb): Likewise.
23469 (predict_paths_for_bb): Likewise.
23470 (predict_paths_leading_to_edge): Likewise.
23471 (determine_unlikely_bbs): New function.
23472 (estimate_bb_frequencies): Use it.
23473 (compute_function_frequency): Use zero counts even if profile is
23475 * profile-count.h: Fix typo.
23477 2017-08-08 Julia Koval <julia.koval@intel.com>
23479 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
23480 _mm512_mask_cvtsepi16_storeu_epi8,
23481 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
23482 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
23483 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
23484 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
23485 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
23486 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
23487 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
23488 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
23489 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
23490 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
23491 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
23492 __builtin_ia32_pmovuswb256mem_mask,
23493 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
23494 __builtin_ia32_pmovwb512mem_mask): New builtins.
23496 2017-08-08 Julia Koval <julia.koval@intel.com>
23498 PR target/73350,80862
23499 * config/i386/subst.md (round): Fix round pattern.
23500 * config/i386/i386.c (ix86_erase_embedded_rounding):
23501 Fix erasing rounding for the fixed pattern.
23503 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23505 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
23507 2017-06-08 Martin Liska <mliska@suse.cz>
23509 PR gcov-profile/80911
23510 * gcov.c (block_info::block_info): New constructor.
23512 2017-06-07 Carl Love <cel@us.ibm.com>
23514 * config/rs6000/rs6000-c: The return type of the following
23515 built-in functions was implemented as int not long long. Fix sign
23516 of return value for the unsigned version of vec_mulo and vec_mule.
23517 vector unsigned long long vec_bperm (vector unsigned long long,
23518 vector unsigned char)
23519 vector signed long long vec_mule (vector signed int,
23521 vector unsigned long long vec_mule (vector unsigned int,
23522 vector unsigned int)
23523 vector signed long long vec_mulo (vector signed int,
23525 vector unsigned long long vec_mulo (vector unsigned int,
23526 vector unsigned int)
23527 * doc/extend.texi: Fix the documentation for the built-in
23530 2017-06-07 Carl Love <cel@us.ibm.com>
23533 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
23536 2017-06-07 Carl Love <cel@us.ibm.com>
23538 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
23539 support, Generate doublehv for signed int/float for BE case only.
23541 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
23543 * doc/invoke.texi (mcx16): Rewrite.
23545 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23547 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
23548 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
23549 *mov<mode>_softfloat, and an anonymous splitter): Use
23550 nonimmediate_operand instead of rs6000_nonimmediate_operand.
23552 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23554 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
23556 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
23557 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
23558 (rs6000_debug_reg_global): Adjust.
23559 (rs6000_init_hard_regno_mode_ok): Adjust.
23560 (rs6000_dbx_register_number): Adjust.
23561 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
23562 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
23563 Remove SPE_ACC and SPEFSCR.
23564 (REG_ALLOC_ORDER): Ditto.
23565 (FRAME_POINTER_REGNUM): Change to 111.
23566 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
23567 (REG_CLASS_NAMES): Ditto.
23568 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
23569 (REGISTER_NAMES): Ditto.
23570 (ADDITIONAL_REG_NAMES): Ditto.
23571 (rs6000_reg_names): Ditto.
23572 * config/rs6000/rs6000.md: Renumber some register number
23575 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23577 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
23579 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
23580 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
23582 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
23583 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
23584 Delete the SPE high registers.
23585 (REG_ALLOC_ORDER): Ditto.
23586 (enum reg_class): Remove SPE_HIGH_REGS.
23587 (REG_CLASS_NAMES): Ditto.
23588 (REG_CLASS_CONTENTS): Delete the SPE high registers.
23589 (REGISTER_NAMES): Ditto.
23590 (rs6000_reg_names): Ditto.
23591 * doc/tm.texi.in: Remove SPE as example.
23592 * doc/tm.texi: Regenerate.
23594 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23596 * config/rs6000/8540.md (ppc8540_brinc): Delete.
23597 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
23598 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
23599 * config/rs6000/rs6000.md (type): Remove "brinc".
23601 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23603 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
23604 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
23605 * config/rs6000/linuxspe.h: Delete file.
23606 * config/rs6000/rs6000.md: Don't include spe.md.
23607 * config/rs6000/spe.h: Delete file.
23608 * config/rs6000/spe.md: Delete file.
23609 * config/rs6000/t-rs6000: Remove spe.md.
23611 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23613 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
23614 (reg_or_none500mem_operand): Delete.
23615 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
23616 instead of reg_or_none500mem_operand.
23618 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23620 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23621 handling of SPE flags.
23622 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
23624 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23626 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
23628 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
23629 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
23630 paired_divv2sf3): Similar.
23631 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
23632 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
23633 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
23635 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
23636 Rename the paired_* instruction patterns.
23637 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
23639 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
23640 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
23641 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
23642 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
23643 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
23644 PAIRED_VECTOR_MODE.
23645 (struct machine_function): Delete field spe_insn_chain_scanned_p.
23646 (spe_func_has_64bit_regs_p): Delete.
23647 (spe_expand_predicate_builtin): Delete.
23648 (spe_expand_evsel_builtin): Delete.
23649 (TARGET_DWARF_REGISTER_SPAN): Do not define.
23650 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
23651 (invalid_e500_subreg): Delete.
23652 (rs6000_legitimize_address): Always force_reg op2 as well, for
23653 paired single memory accesses.
23654 (rs6000_member_type_forces_blk): Delete.
23655 (rs6000_spe_function_arg): Delete.
23656 (rs6000_expand_unop_builtin): Delete SPE handling.
23657 (rs6000_expand_binop_builtin): Ditto.
23658 (spe_expand_stv_builtin): Delete.
23659 (bdesc_2arg_spe): Delete.
23660 (spe_expand_builtin): Delete.
23661 (spe_expand_predicate_builtin): Delete.
23662 (spe_expand_evsel_builtin): Delete.
23663 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
23664 (spe_init_builtins): Delete.
23665 (spe_func_has_64bit_regs_p): Delete.
23666 (savres_routine_name): Delete "info" parameter. Adjust callers.
23667 (rs6000_emit_stack_reset): Ditto.
23668 (rs6000_dwarf_register_span): Delete.
23669 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
23670 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
23671 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
23673 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
23675 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
23676 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
23677 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
23678 mulv2sf3, divv2sf3): Delete expanders.
23680 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23682 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
23684 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23686 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
23687 * config/rs6000/rs6000.c: Ditto.
23689 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23691 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
23692 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
23693 comparison_operator.
23695 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23697 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
23698 * config/rs6000/rs6000.opt: Ditto.
23699 * config/rs6000/t-rtems: Ditto.
23701 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23703 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
23704 TARGET_E500_SINGLE by 0, simplify.
23705 * config/rs6000/rs6000.c: Ditto.
23706 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
23707 (spe_build_register_parallel): Delete.
23708 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
23709 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
23710 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
23711 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
23712 (E500_CONVERT): Delete.
23713 * config/rs6000/spe.md: Remove many patterns and all define_constants.
23715 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23717 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
23718 * config/rs6000/dfp.md: Ditto.
23719 (negdd2, *negdd2_fpr): Merge.
23720 (absdd2, *absdd2_fpr): Merge.
23721 (negtd2, *negtd2_fpr): Merge.
23722 (abstd2, *abstd2_fpr): Merge.
23723 * config/rs6000/e500.h: Delete file.
23724 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
23725 TARGET_FPRS by 1 and simplify.
23726 * config/rs6000/rs6000-c.c: Ditto.
23727 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
23728 TARGET_DF_SPE by 0.
23729 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
23731 * config/rs6000/rs6000.md: Ditto.
23732 (floatdidf2, *floatdidf2_fpr): Merge.
23733 (move_from_CR_gt_bit): Delete.
23734 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
23735 (E500_CR_IOR_COMPARE): Delete.
23736 (All patterns that require !TARGET_FPRS): Delete.
23737 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
23739 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23741 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
23743 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23745 * graphds.c (add_edge): Intitialize edge's attached data.
23746 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
23747 pointer parameter. Call pointed function on each edge during
23748 graph traversing. Skip traversing the edge when the function
23750 (graphds_dfs, graphds_scc): Ditto.
23751 (for_each_edge): New parameter. Pass the new parameter to callback
23753 * graphds.h (skip_edge_callback): New function pointer type.
23754 (graphds_dfs, graphds_scc): New function pointer parameter.
23755 (graphds_edge_callback, for_each_edge): New parameter.
23757 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23759 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
23760 out code checking if runtime alias check is possible to below ...
23761 Call the new function.
23762 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
23763 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
23765 2017-06-07 Marek Polacek <polacek@redhat.com>
23768 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
23769 TYPE_OVERFLOW_WRAPS check.
23771 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23773 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
23774 if versioning is required.
23775 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
23776 peeling with the check for versioning.
23778 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23780 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
23781 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
23782 Set true to new parameter if new ssa variable is defined.
23783 (vect_gen_vector_loop_niters): Refactor. Set range information
23784 for the new vector loop bound variable.
23785 (vect_do_peeling): Ditto.
23787 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23789 * tree-affine.c (ssa.h): Include header file.
23790 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
23791 has wrapping overflow behavior.
23793 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23795 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
23797 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23799 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
23800 (tree_to_aff_combination): ... here.
23802 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23804 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
23805 reg_pressure model function.
23806 (ivopts_global_cost_for_size): Delete.
23807 (determine_set_costs, iv_ca_recount_cost): Call new model function
23808 ivopts_estimate_reg_pressure.
23810 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23812 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
23813 expensive than udiv. Remove floating point cases from mod.
23815 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23817 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
23818 Increase idiv cost.
23820 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23822 * config/aarch64/aarch64.md
23823 (copysignsf3): Fix mask generation.
23825 2017-06-07 Jakub Jelinek <jakub@redhat.com>
23827 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
23829 (class dump_manager): Add register_dumps method.
23830 * dumpfile.c: Include langhooks.h.
23831 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
23832 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
23833 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
23834 (dump_manager::dump_register): Start with 512 entries instead of 32.
23835 (dump_manager::register_dumps): New method.
23836 * toplev.c (general_init): Instead of invoking register_dumps
23837 langhook, invoke register_dumps method on the dump manager.
23838 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
23841 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
23843 * doc/md.texi: Clarify the restrictions on a define_insn condition.
23844 Say that # requires an associated define_split to exist, and that
23845 the define_split must be suitable for use after register allocation.
23847 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23849 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
23850 (compute_outgoing_frequencies): Also initialize zero counts.
23851 (find_many_sub_basic_blocks): Do not produce uninitialized profile
23852 around loops; preserve more of profile when nothing changes.
23854 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
23856 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
23858 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
23859 * config/arm/arm-cpu-cdata.h: Regenerate.
23860 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
23861 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
23862 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
23863 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
23864 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
23866 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
23868 * config/arm/t-rmprofile: Likewise.
23869 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
23871 2017-06-06 David S. Miller <davem@davemloft.net>
23874 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
23875 blockage if function uses alloca.
23877 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23879 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
23880 New "uid" fields to hold pretty-print IDs of group and ref.
23881 Memory references are now identified as <group_id>:<ref_id>
23882 instead of using [random] addresses.
23883 (dump_mem_details): Simplify, no functional change.
23884 (dump_mem_ref): Simplify and make output more concise.
23885 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
23886 (find_or_create_group): Initialize group uid.
23887 (record_ref): Initialize ref uid. Improve debug output.
23888 (prune_group_by_reuse, should_issue_prefetch_p,)
23889 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
23890 (mark_nontemporal_store, determine_loop_nest_reuse):
23891 Improve debug output.
23893 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23895 * dbgcnt.def (prefetch): New debug counter.
23896 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
23897 (schedule_prefetches): Stop issueing prefetches if debug counter
23900 2017-06-06 Tom de Vries <tom@codesourcery.com>
23902 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
23903 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
23905 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23907 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
23908 Use aarch64_reg_or_zero predicate for operand 4.
23909 (aarch64_compare_and_swap<mode> define_insn_and_split):
23910 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
23911 (aarch64_store_exclusive<mode>): Likewise for operand 2.
23913 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
23915 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
23916 (arm_compute_save_core_reg_mask): This.
23917 (thumb1_compute_save_reg_mask): Rename into ...
23918 (thumb1_compute_save_core_reg_mask): This.
23919 (arm_compute_save_reg0_reg12_mask): Adapt comment.
23920 (arm_compute_frame_layout): Likewise.
23922 2017-06-06 Richard Biener <rguenther@suse.de>
23924 PR tree-optimization/80974
23925 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
23926 keep or clear leaders SSA info.
23928 2017-06-06 Tom de Vries <tom@codesourcery.com>
23930 * config/nvptx/nvptx.c (split_mode_p): New function.
23931 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
23933 2017-06-06 Tom de Vries <tom@codesourcery.com>
23935 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
23937 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23940 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
23943 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23945 * shrink-wrap.c (handle_simple_exit): Update profile.
23946 (try_shrink_wrapping): Upate profile.
23948 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23950 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
23951 (tree_guess_outgoing_edge_probabilities): New.
23952 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
23953 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
23955 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23957 * ipa-split.c (split_function): Initialize return bb profile.
23959 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23961 * profile.c (compute_branch_probabilities): Also initialize
23962 EXIT_BLOCK profile.
23964 2017-06-06 Richard Biener <rguenther@suse.de>
23966 PR tree-optimization/80928
23967 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
23968 (vect_analyze_loop_operations): Properly guard analysis for
23970 (vect_transform_loop): Likewise.
23971 (vect_analyze_loop_2): Also reset SLP type on PHIs.
23972 (vect_model_induction_cost): Do not cost for pure SLP.
23973 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
23974 of induction in inner loop vectorization.
23975 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
23976 (vect_get_and_check_slp_defs): Handle vect_induction_def.
23977 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
23979 (vect_analyze_slp_cost_1): Cost induction.
23980 (vect_detect_hybrid_slp_stmts): Handle PHIs.
23981 (vect_get_slp_vect_defs): Likewise.
23982 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
23983 (vect_transform_stmt): Handle SLP reductions.
23984 * tree-vectorizer.h (vectorizable_induction): Adjust.
23986 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23988 * config/rs6000/rs6000.c (make_resolver_func): Update
23989 init_lowered_empty_function call.
23991 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
23993 * doc/invoke.texi: Document the -fprofile-abs-path option.
23994 * common.opt (fprofile-abs-path): New option.
23995 * gcov-io.h (gcov_write_filename): Declare.
23996 * gcov-io.c (gcov_write_filename): New function.
23997 * coverage.c (coverage_begin_function): Use gcov_write_filename.
23998 * profile.c (output_location): Likewise.
24000 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24002 * shring-wrap.c: Revert accidental commit.
24004 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
24006 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
24008 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24010 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
24012 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
24013 profile in callgraph edge.
24014 * profile-count.h (apply_probability): If THIS is 0, then result is 0
24015 (apply_scale): Likewise.
24016 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
24017 Also scale profile when inlining function with zero profile.
24018 (initialize_cfun): Update exit block profile even when it is zero.
24019 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
24020 when profile is read.
24022 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24024 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
24025 (CLONE_*): New constants to define the processors we can generate
24026 code for with the target_clone attribute.
24027 (rs6000_clone_map): New array to identify which clone processors
24028 the current program is running on.
24029 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
24030 target_clone attribute.
24031 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
24032 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
24033 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
24034 (cpu_expand_builtin): Add support for target_clone attribute.
24035 (rs6000_valid_attribute_p): Allow "default" attribute.
24036 (get_decl_name): New debug function to simplify printing the
24037 current function name in debugging statements.
24038 (rs6000_clone_priority): New functions to support the target_clone
24039 attribute, and be able to generate code to switch between ISA 2.05
24040 through ISA 3.0 (power6 through power9).
24041 (rs6000_compare_version_priority): Likewise.
24042 (rs6000_get_function_versions_dispatcher): Likewise.
24043 (make_resolver_func): Likewise.
24044 (add_condition_to_bb): Likewise.
24045 (dispatch_function_versions): Likewise.
24046 (rs6000_generate_version_dispatcher_body): Likewise.
24047 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
24048 (fusion_gpr_load_p): Fix a spacing issue.
24049 * doc/extend.texi (Common Function Attributes): Document that the
24050 PowerPC supports the target_clone attribute.
24052 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24054 * config/arm/arm.h: explain F symbol found in description of ARM
24055 register allocation in its legend.
24057 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24059 * config/mips/frame-header-opt.c: Include profile-count.h.
24060 * config/riscv/riscv.c: Include profile-count.h
24062 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24064 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
24066 (sm_set_flag_if_changed): Add bbs field.
24067 (execute_sm_if_changed_flag_set): Pass BBS.
24068 (execute_sm): Update.
24070 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24072 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
24075 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24077 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
24078 (peephole2): New peephole2 to emit the above.
24079 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
24081 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24083 * config/aarch64/aarch64.c (define_peephole2 above
24084 *sub_<shift>_<mode>): New peephole.
24086 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24088 * config/i386/i386.c (make_resolver_func): Update.
24089 * Makefile.in: Add profile-count.h and profile-count.o
24090 * auto-profile.c (afdo_indirect_call): Update to new API.
24091 (afdo_set_bb_count): Update.
24092 (afdo_propagate_edge): Update.
24093 (afdo_propagate_circuit): Update.
24094 (afdo_calculate_branch_prob): Update.
24095 (afdo_annotate_cfg): Update.
24096 * basic-block.h: Include profile-count.h
24097 (struct edge_def): Turn count to profile_count.
24098 (struct basic_block_def): Likewie.
24099 (REG_BR_PROB_BASE): Move to profile-count.h
24100 (RDIV): Move to profile-count.h
24101 * bb-reorder.c (max_entry_count): Turn to profile_count.
24102 (find_traces): Update.
24103 (rotate_loop):Update.
24104 (connect_traces):Update.
24105 (sanitize_hot_paths):Update.
24106 * bt-load.c (migrate_btr_defs): Update.
24107 * cfg.c (RDIV): Remove.
24108 (init_flow): Use alloc_block.
24109 (alloc_block): Uninitialize count.
24110 (unchecked_make_edge): Uninitialize count.
24111 (check_bb_profile): Update.
24112 (dump_edge_info): Update.
24113 (dump_bb_info): Update.
24114 (update_bb_profile_for_threading): Update.
24115 (scale_bbs_frequencies_int): Update.
24116 (scale_bbs_frequencies_gcov_type): Update.
24117 (scale_bbs_frequencies_profile_count): New.
24118 * cfg.h (update_bb_profile_for_threading): Update.
24119 (scale_bbs_frequencies_profile_count): Declare.
24120 * cfgbuild.c (compute_outgoing_frequencies): Update.
24121 (find_many_sub_basic_blocks): Update.
24122 * cfgcleanup.c (try_forward_edges): Update.
24123 (try_crossjump_to_edge): Update.
24124 * cfgexpand.c (expand_gimple_tailcall): Update.
24125 (construct_exit_block): Update.
24126 * cfghooks.c (verify_flow_info): Update.
24127 (dump_bb_for_graph): Update.
24128 (split_edge): Update.
24129 (make_forwarder_block): Update.
24130 (duplicate_block): Update.
24131 (account_profile_record): Update.
24132 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
24133 (get_estimated_loop_iterations): Update.
24134 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
24135 (single_likely_exit): Update.
24136 * cfgloopmanip.c (scale_loop_profile): Update.
24138 (set_zero_probability): Update.
24139 (lv_adjust_loop_entry_edge): Update.
24140 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24141 (purge_dead_edges): Update.
24142 (rtl_account_profile_record): Update.
24143 * cgraph.c (cgraph_node::create): Uninitialize count.
24144 (symbol_table::create_edge): Uninitialize count.
24145 (cgraph_update_edges_for_call_stmt_node): Update.
24146 (cgraph_edge::dump_edge_flags): Update.
24147 (cgraph_node::dump): Update.
24148 (cgraph_edge::maybe_hot_p): Update.
24149 * cgraph.h: Include profile-count.h
24150 (create_clone), create_edge, create_indirect_edge): Update.
24151 (cgraph_node): Turn count to profile_count.
24152 (cgraph_edge0: Likewise.
24153 (make_speculative, clone): Update.
24154 (create_edge): Update.
24155 (init_lowered_empty_function): Update.
24156 * cgraphclones.c (cgraph_edge::clone): Update.
24157 (duplicate_thunk_for_node): Update.
24158 (cgraph_node::create_clone): Update.
24159 * cgraphunit.c (cgraph_node::analyze): Update.
24160 (cgraph_node::expand_thunk): Update.
24161 * final.c (dump_basic_block_info): Update.
24162 * gimple-streamer-in.c (input_bb): Update.
24163 * gimple-streamer-out.c (output_bb): Update.
24164 * graphite.c (print_global_statistics): Update.
24165 (print_graphite_scop_statistics): Update.
24166 * hsa-brig.c: Include basic-block.h.
24167 * hsa-dump.c: Include basic-block.h.
24168 * hsa-gen.c (T sum_slice): Update.
24169 (convert_switch_statements):Update.
24170 * hsa-regalloc.c: Include basic-block.h.
24171 * ipa-chkp.c (chkp_produce_thunks): Update.
24172 * ipa-cp.c (struct caller_statistics): Update.
24173 (init_caller_stats): Update.
24174 (gather_caller_stats): Update.
24175 (ipcp_cloning_candidate_p): Update.
24176 (good_cloning_opportunity_p): Update.
24177 (get_info_about_necessary_edges): Update.
24178 (dump_profile_updates): Update.
24179 (update_profiling_info): Update.
24180 (update_specialized_profile): Update.
24181 (perhaps_add_new_callers): Update.
24182 (decide_about_value): Update.
24183 (ipa_cp_c_finalize): Update.
24184 * ipa-devirt.c (struct odr_type_warn_count): Update.
24185 (struct decl_warn_count): Update.
24186 (struct final_warning_record): Update.
24187 (possible_polymorphic_call_targets): Update.
24188 (ipa_devirt): Update.
24189 * ipa-fnsummary.c (redirect_to_unreachable): Update.
24190 * ipa-icf.c (sem_function::merge): Update.
24191 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
24192 * ipa-inline.c (compute_uninlined_call_time): Update.
24193 (compute_inlined_call_time): Update.
24194 (want_inline_small_function_p): Update.
24195 (want_inline_self_recursive_call_p): Update.
24196 (edge_badness): Update.
24197 (lookup_recursive_calls): Update.
24198 (recursive_inlining): Update.
24199 (inline_small_functions): Update.
24200 (dump_overall_stats): Update.
24201 (dump_inline_stats): Update.
24202 * ipa-profile.c (ipa_profile_generate_summary): Update.
24203 (ipa_propagate_frequency): Update.
24204 (ipa_profile): Update.
24205 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24206 * ipa-utils.c (ipa_merge_profiles): Update.
24207 * loop-doloop.c (doloop_modify): Update.
24208 * loop-unroll.c (report_unroll): Update.
24209 (unroll_loop_runtime_iterations): Update.
24210 * lto-cgraph.c (lto_output_edge): Update.
24211 (lto_output_node): Update.
24212 (input_node): Update.
24213 (input_edge): Update.
24214 (merge_profile_summaries): Update.
24215 * lto-streamer-in.c (input_cfg): Update.
24216 * lto-streamer-out.c (output_cfg): Update.
24217 * mcf.c (create_fixup_graph): Update.
24218 (adjust_cfg_counts): Update.
24219 (sum_edge_counts): Update.
24220 * modulo-sched.c (sms_schedule): Update.
24221 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
24222 * predict.c (maybe_hot_count_p): Update.
24223 (probably_never_executed): Update.
24224 (dump_prediction): Update.
24225 (combine_predictions_for_bb): Update.
24226 (propagate_freq): Update.
24227 (handle_missing_profiles): Update.
24228 (counts_to_freqs): Update.
24229 (rebuild_frequencies): Update.
24230 (force_edge_cold): Update.
24231 * predict.h: Include profile-count.h
24232 (maybe_hot_count_p, counts_to_freqs): UPdate.
24233 * print-rtl-function.c: Do not include cfg.h
24234 * print-rtl.c: Include basic-block.h
24235 * profile-count.c: New file.
24236 * profile-count.h: New file.
24237 * profile.c (is_edge_inconsistent): Update.
24238 (correct_negative_edge_counts): Update.
24239 (is_inconsistent): Update.
24240 (set_bb_counts): Update.
24241 (read_profile_edge_counts): Update.
24242 (compute_frequency_overlap): Update.
24243 (compute_branch_probabilities): Update; Initialize and deinitialize
24245 (branch_prob): Update.
24246 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
24247 (edge_gcov_count): New.
24248 (bb_gcov_count): New.
24249 * shrink-wrap.c (try_shrink_wrapping): Update.
24250 * tracer.c (better_p): Update.
24251 * trans-mem.c (expand_transaction): Update.
24252 (ipa_tm_insert_irr_call): Update.
24253 (ipa_tm_insert_gettmclone_call): Update.
24254 * tree-call-cdce.c: Update.
24255 * tree-cfg.c (gimple_duplicate_sese_region): Update.
24256 (gimple_duplicate_sese_tail): Update.
24257 (gimple_account_profile_record): Update.
24258 (execute_fixup_cfg): Update.
24259 * tree-inline.c (copy_bb): Update.
24260 (copy_edges_for_bb): Update.
24261 (initialize_cfun): Update.
24262 (freqs_to_counts): Update.
24263 (copy_cfg_body): Update.
24264 (expand_call_inline): Update.
24265 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
24266 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
24267 (try_unroll_loop_completely): Update.
24268 (try_peel_loop): Update.
24269 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24270 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
24271 * tree-ssa-loop-split.c (connect_loops): Update.
24272 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
24273 * tree-ssa-reassoc.c (branch_fixup): Update.
24274 * tree-ssa-tail-merge.c (replace_block_by): Update.
24275 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
24276 (compute_path_counts): Update.
24277 (update_profile): Update.
24278 (recompute_probabilities): Update.
24279 (update_joiner_offpath_counts): Update.
24280 (estimated_freqs_path): Update.
24281 (freqs_to_counts_path): Update.
24282 (clear_counts_path): Update.
24283 (ssa_fix_duplicate_block_edges): Update.
24284 (duplicate_thread_path): Update.
24285 * tree-switch-conversion.c (case_bit_test_cmp): Update.
24286 (struct switch_conv_info): Update.
24287 * tree-tailcall.c (decrease_profile): Update.
24288 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
24289 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24290 * value-prof.c (check_counter): Update.
24291 (gimple_divmod_fixed_value): Update.
24292 (gimple_mod_pow2): Update.
24293 (gimple_mod_subtract): Update.
24294 (gimple_ic_transform): Update.
24295 (gimple_stringop_fixed_value): Update.
24296 * value-prof.h (gimple_ic): Update.
24298 2017-06-02 Carl Love <cel@us.ibm.com>
24300 * config/rs6000/rs6000-c: Add support for built-in functions
24301 vector double vec_doublee (vector signed int);
24302 vector double vec_doublee (vector unsigned int);
24303 vector double vec_doublee (vector float);
24304 vector double vec_doubleh (vector signed int);
24305 vector double vec_doubleh (vector unsigned int);
24306 vector double vec_doubleh (vector float);
24307 vector double vec_doublel (vector signed int);
24308 vector double vec_doublel (vector unsigned int);
24309 vector double vec_doublel (vector float);
24310 vector double vec_doubleo (vector signed int);
24311 vector double vec_doubleo (vector unsigned int);
24312 vector double vec_doubleo (vector float);.
24313 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
24314 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
24316 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
24317 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
24318 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
24320 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
24321 vec_doublel, vec_doubleh.
24322 * doc/extend.texi: Update the built-in documentation file for the
24323 new built-in functions.
24325 2017-06-02 David Malcolm <dmalcolm@redhat.com>
24328 * ipa-inline-analysis.c (free_growth_caches): Set
24329 edge_removal_hook_holder to NULL after removing it.
24331 2017-06-02 Sudakshina Das <sudi.das@arm.com>
24333 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
24334 comparision with zero.
24336 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
24337 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24338 for early expansion of vec_min and vec_max builtins.
24339 (builtin_function_type): Add min/max unsigned variants to those
24340 identified as having unsigned arguments.
24342 2017-06-02 Olivier Hainque <hainque@adacore.com>
24344 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
24346 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24348 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
24349 Use VALL_F16 iterator rather than VALL.
24351 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24353 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
24354 Emit CBNZ inside loop when doing a strong exchange and comparing
24355 against zero. Generate the CC flags after the loop.
24357 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
24359 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
24360 (dl_section_ref): New.
24361 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
24362 On AIX, append an expression to subtract the size of the
24363 section length to dl_section_ref.
24365 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
24367 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24368 for early expansion of vector absolute builtins.
24370 2017-06-02 Richard Biener <rguenther@suse.de>
24372 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
24373 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
24375 2017-06-02 Richard Biener <rguenther@suse.de>
24377 PR tree-optimization/80948
24378 * tree-tailcall.c (find_tail_calls): Track stmts to move in
24379 stmt order as well.
24381 2017-06-02 Richard Biener <rguenther@suse.de>
24383 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
24385 * tree-vect-stmts.c (process_use): Do not mark backedge defs
24386 for inductions as relevant.
24388 2017-06-02 Richard Biener <rguenther@suse.de>
24390 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
24391 (vectorizable_induction): ... this. Remove dead code.
24393 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
24395 * builtins. (expand_builtin_alloca): Remove second parameter and
24396 infer its value from the first parameter instead.
24397 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
24399 2017-06-02 Jakub Jelinek <jakub@redhat.com>
24401 PR rtl-optimization/80903
24402 * loop-doloop.c (add_test): Unshare sequence.
24404 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
24406 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
24408 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
24410 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
24412 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
24413 xlogue_layout::get_instance, logue_layout::xlogue_layout,
24414 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
24415 (xlogue_layout::get_stub_rtx): Make static.
24416 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
24417 (xlogue_layout::compute_stub_managed_regs): Rename to...
24418 (xlogue_layout::count_stub_managed_regs): ...this.
24419 (xlogue_layout::is_stub_managed_reg): New function.
24420 (xlogue_layout::m_stub_names): Rename to...
24421 (xlogue_layout::s_stub_names): ...this, make static.
24422 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
24423 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
24424 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
24425 xlogue_layout::s_stub_names): Instantiate statics.
24426 (stub_managed_regs): Remove.
24427 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
24428 (disable_call_ms2sysv_xlogues): Rename to...
24429 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
24430 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
24432 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
24433 change after reload_completed.
24434 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
24436 (ix86_expand_prologue): Likewise.
24437 (ix86_expand_epilogue): Likewise.
24438 (ix86_expand_split_stack_prologue): Likewise.
24439 (ix86_compute_frame_layout): Remove frame parameter ...
24440 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
24441 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
24443 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
24444 (ix86_frame): Move from here ...
24445 * config/i386/i386.h (ix86_frame): ... to here.
24446 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
24447 complete ix86_frame data structure instead. Remove some_ld_name.
24449 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
24451 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
24452 symbols that hold a DECL_VALUE_EXPR.
24454 2017-06-01 Martin Jambor <mjambor@suse.cz>
24456 PR tree-optimization/80898
24457 * tree-sra.c (process_subtree_disqualification): Removed.
24458 (disqualify_candidate): Do not acll
24459 process_subtree_disqualification.
24460 (subtree_mark_written_and_enqueue): New function.
24461 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
24462 RHS has been disqualified and re-queue LHS if necessary. Apart
24463 from that, ignore disqualified RHS.
24465 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24467 * config/s390/s390.c (s390_emit_epilogue): Disable early return
24468 address fetch for z10 or later.
24470 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24472 * config/arc/arc.md (tst_movb): Add guard when splitting.
24474 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24476 * config/arc/arc.c (arc_can_eliminate): Test against
24477 arc_frame_pointer_needed.
24479 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24481 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
24482 to prevent store reordering.
24483 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
24484 (type): Add block type.
24485 (stack_tie): Define special instruction to be used in
24488 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24490 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
24491 constraint. It is not valid for the pattern.
24492 (noncommutative_binary_comparison): Likewise.
24494 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24496 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
24499 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24501 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
24502 be used by the reg-alloc.
24504 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24506 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
24507 reg-alloc when having mul64 or mul32x16 instructions.
24508 (mulsidi3): Likewise.
24509 (umulsidi3): Likewise.
24510 (mulsi32x16): New pattern.
24511 (mulsi64): Likewise.
24512 (mulsidi64): Likewise.
24513 (umulsidi64): Likewise.
24514 (MUL32x16_REG): Define.
24515 (mul64_600): Use MUL32x16_REG.
24516 (mac64_600): Likewise.
24517 (umul64_600): Likewise.
24518 (umac64_600): Likewise.
24520 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24522 * config/arc/arc.md (mulsi3_700): Make it commutative.
24524 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
24526 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
24528 (*sign_extendsidi2_insn): Likewise for movstosw.
24530 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
24532 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
24533 the type of the input discriminant value. Convert the
24534 discriminant value of signedness vary.
24536 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
24538 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
24539 Add to -Wall section.
24541 2017-06-01 Richard Biener <rguenther@suse.de>
24543 PR middle-end/66313
24544 * fold-const.c (fold_plusminus_mult_expr): If the factored
24545 factor may be zero use a wrapping type for the inner operation.
24546 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
24547 and handle moved defs.
24548 (process_assignment): Properly guard the unary op case. Return a
24549 tri-state indicating that moving the stmt before the call may allow
24550 to continue. Pass through to_move.
24551 (find_tail_calls): Handle moving unrelated defs before
24554 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
24557 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
24558 splitter result in the canonical way.
24560 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
24562 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
24563 also for 32bit target. Update insn attributes.
24564 (zero-extendsidi2 splitter): Allow all registers for operand 1.
24566 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
24568 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
24569 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
24570 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
24571 (_mm_maskz_min_ss): New intrinsics.
24573 2017-05-31 Martin Liska <mliska@suse.cz>
24575 * tree-vect-loop.c (vect_create_epilog_for_reduction):
24576 Change comment style to one we normally use.
24577 (vectorizable_reduction): Likewise.
24578 (vectorizable_induction): Likewise.
24579 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
24580 (vectorizable_call): Likewise.
24581 (vectorizable_simd_clone_call): Likewise.
24582 (vectorizable_conversion): Likewise.
24583 (vectorizable_assignment): Likewise.
24584 (vectorizable_shift): Likewise.
24585 (vectorizable_operation): Likewise.
24586 (vectorizable_store): Likewise.
24587 (vectorizable_load): Likewise.
24588 * tree-vectorizer.h: Likewise.
24590 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
24592 * passes.c (emergency_dump_function): New.
24593 * tree-pass.h (emergency_dump_function): Declare.
24594 * plugin.c (plugins_internal_error_function): Remove.
24595 * plugin.h (plugins_internal_error_function): Remove declaration.
24596 * toplev.c (internal_error_function): New static function. Use it...
24597 (general_init): ...here.
24599 2017-05-31 Graham Markall <graham.markall@embecosm.com>
24601 * config/arc/arc.c (arc_print_operand): Handle constant operands.
24602 (arc_rtx_costs): Add costs for new patterns.
24603 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
24604 * config/arc/predicates.md: Add _1_2_3_operand predicate.
24606 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
24608 * tree-ssa-strlen.c (get_next_strinfo): New function.
24609 (get_stridx_plus_constant): Use it.
24610 (zero_length_string): Likewise.
24611 (adjust_related_strinfos): Likewise.
24612 (adjust_last_stmt): Likewise.
24614 2017-05-31 Richard Biener <rguenther@suse.de>
24617 * config/i386/i386.c (ix86_expand_builtin): Remove assert
24618 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
24620 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
24622 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
24623 loop_vinfo argument and use of dependence distance vectors.
24624 Check instead whether the two references differ only in their
24625 initial value and assume that they have the same alignment if the
24626 difference is a multiple of the vector alignment.
24627 (vect_analyze_data_refs_alignment): Update call accordingly.
24629 2017-05-31 Martin Liska <mliska@suse.cz>
24632 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
24634 2017-05-31 Bin Cheng <bin.cheng@arm.com>
24636 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
24637 (create_intersect_range_checks): Move from ...
24638 * tree-data-ref.c (create_intersect_range_checks_index)
24639 (create_intersect_range_checks): ... to here.
24640 (create_runtime_alias_checks): New function factored from ...
24641 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
24642 here. Call above function.
24643 * tree-data-ref.h (create_runtime_alias_checks): New function.
24645 2017-05-31 Bin Cheng <bin.cheng@arm.com>
24647 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
24648 segment length for dr_b and compute it in wide_int.
24650 2017-05-31 Richard Biener <rguenther@suse.de>
24652 PR tree-optimization/80906
24653 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
24654 and pass through iv_map.
24655 (copy_bb_and_scalar_dependences): Adjust.
24656 (translate_pending_phi_nodes): Likewise.
24657 (copy_loop_close_phi_args): Handle code-generating IVs instead
24660 2017-05-30 David Malcolm <dmalcolm@redhat.com>
24662 * diagnostic-color.c (color_dict): Add "type-diff".
24663 (parse_gcc_colors): Update comment.
24664 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
24665 -fdiagnostics-show-template-tree and -fno-elide-type.
24666 (GCC_COLORS): Add type-diff to example.
24668 (-fdiagnostics-show-template-tree): New.
24669 (-fno-elide-type): New.
24670 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
24671 the pp_format_decoder callback. Call any m_format_postprocessor's
24673 (pretty_printer::pretty_printer): Initialize
24674 m_format_postprocessor.
24675 (pretty_printer::~pretty_printer): Delete any
24676 m_format_postprocessor.
24677 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
24678 (class format_postprocessor): New class.
24679 (struct pretty_printer::format_decoder): Document the new parameters.
24680 (struct pretty_printer::m_format_postprocessor): New field.
24681 * tree-diagnostic.c (default_tree_printer): Update for new
24682 bool and const char ** params.
24683 * tree-diagnostic.h (default_tree_printer): Likewise.
24685 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
24687 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
24688 (lwa_operand): Delete rs6000_gen_cell_microcode test.
24689 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
24690 rs6000_gen_cell_microcode code.
24691 (rs6000_final_prescan_insn): Delete.
24692 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
24693 "warn-cell-microcode" entries.
24694 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
24695 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
24696 throughout. Change cc_reg_not_micro_cr0_operand to
24697 cc_reg_not_cr0_operand throughout.
24698 (*extendhi<mode>2_noload): Delete.
24699 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
24700 (mwarn-cell-microcode): Delete.
24701 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
24702 -mgen-cell-microcode and -mwarn-cell-microcode.
24704 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
24707 * config/i386/constraints.md (Yd): New constraint.
24709 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
24710 and (?Yd, r) alternatives. Update insn attributes.
24711 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
24712 and (?*Yd, r) alternatives. Update insn attributes.
24713 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
24715 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
24717 * gimplify.c (gimplify_modify_expr): Don't create a
24718 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
24721 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
24723 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
24725 2017-05-30 Richard Biener <rguenther@suse.de>
24727 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
24728 and reduc_def fields.
24729 (STMT_VINFO_REDUC_TYPE): New define.
24730 (STMT_VINFO_REDUC_DEF): Likewise.
24731 (vect_force_simple_reduction): Adjust prototype.
24732 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
24733 (vect_is_simple_reduction): Remove check_reduction argument.
24734 (vect_force_simple_reduction): Adjust and set
24735 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
24736 (vectorizable_reduction): Do not re-do reduction analysis
24737 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
24738 * tree-parloops.c (gather_scalar_reductions): Adjust.
24740 2017-05-30 Richard Biener <rguenther@suse.de>
24742 PR middle-end/80901
24743 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
24746 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
24748 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
24749 Introduce unknown_misalignment parameter and remove vf.
24750 (vect_peeling_hash_get_lowest_cost):
24751 Pass unknown_misalignment parameter.
24752 (vect_enhance_data_refs_alignment):
24753 Fix unsupportable data ref treatment.
24755 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24757 * tree-vect-data-refs.c (vect_get_data_access_cost):
24758 Workaround for SLP handling.
24759 (vect_enhance_data_refs_alignment):
24760 Compute costs for doing no peeling at all, compare to the best
24761 peeling costs so far and avoid peeling if cheaper.
24763 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24765 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
24766 Return peeling info and set costs to zero for unlimited cost
24768 (vect_enhance_data_refs_alignment): Also inspect all datarefs
24769 with unknown misalignment. Compute and costs for unknown
24770 misalignment, compare them to the costs for known misalignment
24771 and choose the cheapest for peeling.
24773 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24775 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
24776 (vect_get_peeling_costs_all_drs): Create function.
24777 (vect_peeling_hash_get_lowest_cost):
24778 Use vect_get_peeling_costs_all_drs.
24779 (vect_peeling_supportable): Create function.
24780 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
24782 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24784 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
24785 DR_HAS_NEGATIVE_STEP.
24786 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
24787 (vect_enhance_data_refs_alignment): Use.
24788 (vect_duplicate_ssa_name_ptr_info): Use.
24789 * tree-vectorizer.h (dr_misalignment): Use.
24790 (known_alignment_for_access_p): Use.
24792 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
24795 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
24797 (has_section_name): New function.
24799 2017-05-30 Martin Liska <mliska@suse.cz>
24802 * auto-profile.c (get_function_decl_from_block): Fix
24805 2017-05-30 Richard Biener <rguenther@suse.de>
24807 PR middle-end/80876
24808 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
24810 2017-05-30 Martin Liska <mliska@suse.cz>
24812 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
24813 * dumpfile.h (struct dump_file_info): Remove ctors.
24815 2017-05-30 Martin Liska <mliska@suse.cz>
24817 * predict.def: Fix GNU coding style.
24819 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
24821 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24822 Mark 'to' argument with ATTRIBUTE_UNUSED.
24824 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
24826 * config/xtensa/xtensa.c (xtensa_emit_call): Use
24827 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
24828 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
24831 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
24833 * doc/install.texi (Options specification): Restore entry of
24834 --enable-sjlj-exceptions.
24836 2017-05-27 Michael Eager <eager@eagercon.com>
24839 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
24841 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
24843 * config/microblaze/microblaze.h
24844 (FIXED_REGISTERS): Update in macro.
24845 (CALL_USED_REGISTERS): Update in macro.
24847 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
24849 * doc/install.texi: Add links to macOS binary distributions.
24851 2017-05-27 Jakub Jelinek <jakub@redhat.com>
24855 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24857 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
24859 2017-05-26 Martin Liska <mliska@suse.cz>
24861 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
24863 2017-05-26 Martin Liska <mliska@suse.cz>
24865 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
24866 always leading ';; '.
24867 (dump_bb_info): Likewise.
24868 (brief_dump_cfg): Likewise.
24869 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
24870 * dumpfile.c: Remove usage of TDF_VERBOSE.
24871 * dumpfile.h (enum dump_kind): Likewise.
24872 (dump_gimple_bb_header): Do not use TDF_COMMENT.
24873 * print-tree.c (debug_verbose): Remove.
24874 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
24875 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
24876 * tree-diagnostic.c (default_tree_printer): Replace
24877 TDF_DIAGNOSTIC with TDF_SLIM.
24879 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24881 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
24882 in parameter loop, rather than loop_vinfo.
24883 (create_intersect_range_checks): Ditto.
24884 (vect_create_cond_for_alias_checks): Update call to above functions.
24886 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24888 PR tree-optimization/80815
24889 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
24890 for merging runtime alias checks. Handle negative DR_STEPs.
24892 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24894 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
24896 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
24897 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
24898 out code pruning runtime alias checks.
24899 * tree-data-ref.c (prune_runtime_alias_test_list): New function
24900 factored out from above.
24901 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
24903 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
24905 (prune_runtime_alias_test_list): New decalaration.
24907 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24909 * tree-vect-data-refs.c (compare_tree): Rename and move ...
24910 * tree-data-ref.c (data_ref_compare_tree): ... to here.
24911 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
24912 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
24913 (operator==, comp_dr_with_seg_len_pair): Ditto.
24914 (vect_prune_runtime_alias_test_list): Ditto.
24916 2017-05-26 Martin Liska <mliska@suse.cz>
24919 * params.def: Bound partial-inlining-entry-probability param.
24921 2017-05-26 Marek Polacek <polacek@redhat.com>
24924 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
24927 2017-05-26 Richard Biener <rguenther@suse.de>
24929 PR tree-optimization/80842
24930 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
24933 2017-05-26 Richard Biener <rguenther@suse.de>
24935 PR tree-optimization/80844
24936 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
24938 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
24940 * doc/md.texi (Machine Constraints): Update x86 family
24941 machine constraints section to match 'config/i386/constraints.md'.
24943 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
24945 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
24947 2017-05-25 Nathan Sidwell <nathan@acm.org>
24949 * doc/invoke.texi (--enable-languages): Update documentation.
24951 2017-05-25 Martin Liska <mliska@suse.cz>
24953 * dumpfile.c: Add TDF_FOLDING.
24954 * dumpfile.h (enum dump_kind): Likewise.
24955 * genmatch.c (dt_simplify::gen_1): Use it.
24957 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24959 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
24961 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24963 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
24964 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
24966 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24968 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
24969 * match.pd (X == C): Rewrite it here.
24970 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
24971 with_certain_nonzero_bits2): New predicates.
24972 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
24974 2017-05-24 Nathan Sidwell <nathan@acm.org>
24976 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
24979 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
24982 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
24984 * config/powerpcspe: New port. Files are copied from the rs6000
24985 port, with "rs6000" in filenames replaced by "powerpcspe".
24987 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
24989 PR rtl-optimization/80754
24990 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
24992 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
24994 * config/sparc/sparc.md (length): Return the correct value for -mflat
24995 sibcalls to match output_sibcall.
24997 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
25001 * config/rs6000/rs6000.c (struct machine_function): Add new field
25003 (rs6000_get_separate_components): Init that field, use it.
25004 (rs6000_components_for_bb): Use the field.
25006 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
25008 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
25010 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
25012 PR middle-end/80823
25013 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
25015 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25018 * config/s390/s390.c (s390_check_qrst_address): Check incoming
25019 address against address_operand predicate.
25020 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
25022 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
25024 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
25025 parameters passed indirectly.
25027 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
25029 * config/i386/i386.md (*movdi_internal): Remove SSE4
25030 alternative 18 (?r, *v). Update insn attributes.
25031 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
25032 Update insn attributes.
25033 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
25034 Update insn attributes.
25035 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
25036 alternative 1 (r, v). Remove isa attribute.
25037 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
25038 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
25039 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
25041 2017-05-23 Tom de Vries <tom@codesourcery.com>
25043 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
25046 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25048 * cgraphunit.c (symbol_table::process_new_functions): Update.
25049 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
25050 (inline_generate_summary): Rename to ...
25051 (ipa_fn_summary_generate): ... this one.
25052 (inline_read_summary): Rename to ...
25053 (ipa_fn_summary_read): ... this one.
25054 (inline_write_summary): Rename to ...
25055 (ipa_fn_summary_write): ... this one.
25056 (inline_free_summary): Rename to ...
25057 (ipa_free_fn_summary): ... this one.
25058 (pass_data_local_fn_summary, pass_local_fn_summary,
25059 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
25060 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
25061 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
25062 make_pass_ipa_fn_summary): New.
25063 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
25064 inline_write_summary, inline_free_summary): Remove.
25065 (ipa_free_fn_summary) : New.
25066 * ipa-inline.c (ipa_inline): Update.
25067 (pass_ipa_inline): Do not generate summaries.
25068 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
25070 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
25071 and add pass_ipa_fn_summary.
25072 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
25074 (make_pass_inline_parameters): Remove.
25076 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
25078 * omp-low.c (struct omp_context): Remove "default_kind" member.
25081 * omp-offload.c (execute_oacc_device_lower): Remove the
25082 parallelism dimensions function attributes for unparallelized
25083 OpenACC kernels constructs.
25085 2017-05-23 Martin Liska <mliska@suse.cz>
25087 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
25089 (cgraph_edge::make_speculative): Likewise.
25090 (cgraph_edge::resolve_speculation): Likewise.
25091 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
25092 (cgraph_node::dump): Likewise.
25093 * cgraph.h: Likewise.
25094 * cgraphunit.c (analyze_functions): Likewise.
25095 (symbol_table::compile): Likewise.
25096 * ipa-cp.c (print_all_lattices): Likewise.
25097 (determine_versionability): Likewise.
25098 (initialize_node_lattices): Likewise.
25099 (ipcp_verify_propagated_values): Likewise.
25100 (estimate_local_effects): Likewise.
25101 (update_profiling_info): Likewise.
25102 (create_specialized_node): Likewise.
25103 (perhaps_add_new_callers): Likewise.
25104 (decide_about_value): Likewise.
25105 (decide_whether_version_node): Likewise.
25106 (identify_dead_nodes): Likewise.
25107 (ipcp_store_bits_results): Likewise.
25108 * ipa-devirt.c (dump_targets): Likewise.
25109 (ipa_devirt): Likewise.
25110 * ipa-icf.c (sem_item::dump): Likewise.
25111 (sem_function::equals): Likewise.
25112 (sem_variable::equals): Likewise.
25113 (sem_item_optimizer::read_section): Likewise.
25114 (sem_item_optimizer::execute): Likewise.
25115 (congruence_class::dump): Likewise.
25116 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
25117 (dump_inline_summary): Likewise.
25118 (estimate_node_size_and_time): Likewise.
25119 (inline_analyze_function): Likewise.
25120 * ipa-inline-transform.c (inline_call): Likewise.
25121 * ipa-inline.c (report_inline_failed_reason): Likewise.
25122 (want_early_inline_function_p): Likewise.
25123 (edge_badness): Likewise.
25124 (update_edge_key): Likewise.
25125 (inline_small_functions): Likewise.
25126 * ipa-profile.c (ipa_profile): Likewise.
25127 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
25128 (ipa_make_edge_direct_to_target): Likewise.
25129 (remove_described_reference): Likewise.
25130 (ipa_impossible_devirt_target): Likewise.
25131 (propagate_controlled_uses): Likewise.
25132 (ipa_print_node_params): Likewise.
25133 (ipcp_transform_function): Likewise.
25134 * ipa-pure-const.c (pure_const_read_summary): Likewise.
25135 (propagate_pure_const): Likewise.
25136 * ipa-reference.c (generate_summary): Likewise.
25137 (read_write_all_from_decl): Likewise.
25138 (propagate): Likewise.
25139 (ipa_reference_read_optimization_summary): Likewise.
25140 * ipa-utils.c (ipa_merge_profiles): Likewise.
25141 * ipa.c (walk_polymorphic_call_targets): Likewise.
25142 (symbol_table::remove_unreachable_nodes): Likewise.
25143 (ipa_single_use): Likewise.
25144 * passes.c (execute_todo): Likewise.
25145 * predict.c (drop_profile): Likewise.
25146 * symtab.c (symtab_node::get_dump_name): New function.
25147 (symtab_node::dump_name): Likewise.
25148 (symtab_node::dump_asm_name): Likewise.
25149 (symtab_node::dump_references): Likewise.
25150 (symtab_node::dump_referring): Likewise.
25151 (symtab_node::dump_base): Likewise.
25152 (symtab_node::debug_symtab): Likewise.
25153 * tree-sra.c (convert_callers_for_node): Likewise.
25154 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
25155 * value-prof.c (init_node_map): Likewise.
25157 2017-05-23 Martin Liska <mliska@suse.cz>
25159 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
25160 and symtab_node::debug_symtab to symbol_table::debug.
25161 * cgraphunit.c (analyze_functions): Use the renamed function.
25162 (symbol_table::compile): Likewise.
25163 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
25164 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
25165 * passes.c (execute_todo): Likewise.
25166 * symtab.c (symbol_table::dump): New function.
25167 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
25169 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25171 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
25172 that nonconst implies exec.
25174 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25176 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
25177 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
25178 (inline_edge_summary_vec): Turn into ...
25179 (ipa_call_summaries): ... this one.
25180 (redirect_to_unreachable, edge_set_predicate,
25181 evaluate_properties_for_edge, inline_summary_alloc,
25182 reset_ipa_call_summary, reset_inline_summary,
25183 inline_summary_t::duplicate): Update.
25184 (inline_edge_duplication_hook): Turn to ...
25185 (ipa_call_summary_t::duplicate): ... this one.
25186 (inline_edge_removal_hook): Turn to ...
25187 (ipa_call_summary_t::remove): ... this one.
25188 (dump_inline_edge_summary): Turn to ...
25189 (dump_ipa_call_summary): ... this one.
25190 (estimate_function_body_sizes): Update.
25191 (inline_update_callee_summaries): Update.
25192 (remap_edge_change_prob): Update.
25193 (remap_edge_summaries): Update.
25194 (inline_merge_summary): Update.
25195 (do_estimate_edge_time): Update.
25196 (inline_generate_summary): Update.
25197 (inline_read_section): Update.
25198 (inline_read_summary): Update.
25199 (inline_free_summary): Update.
25200 * ipa-inline.c (can_inline_edge_p): Update.
25201 (compute_inlined_call_time): Update.
25202 (want_inline_small_function_p): Update.
25203 (edge_badness): Update.
25204 (early_inliner): Update.
25205 * ipa-inline.h (inline_edge_summary): Turn to ...
25206 (ipa_call_summary): ... this one.
25207 (ipa_call_summary_t): New class.
25208 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
25209 (ipa_call_summaries): New.
25210 (inline_edge_summary): Remove.
25211 (estimate_edge_growth): Update.
25212 * ipa-profile.c (ipa_propagate_frequency_1): Update.
25213 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25214 * ipa-split.c (execute_split_functions): Update.
25215 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
25217 2017-05-23 Tom de Vries <tom@codesourcery.com>
25219 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
25220 attributes): Document rdrand effective target.
25222 2017-05-23 Tom de Vries <tom@codesourcery.com>
25224 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
25225 attributes): Sort alphabetically.
25227 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
25229 * config/avr/genmultilib.awk: Use gsub instead of gensub.
25231 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
25234 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
25235 V2DF/V2DI splat into two separate patterns, one that handles
25236 registers, and the other that only handles memory. Drop support
25237 for splatting from a GPR on ISA 2.07 and then splitting the
25238 splat into direct move and splat.
25239 (vsx_splat_<mode>_reg): Likewise.
25240 (vsx_splat_<mode>_mem): Likewise.
25242 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
25244 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
25246 2017-05-22 Jakub Jelinek <jakub@redhat.com>
25248 PR middle-end/80809
25249 * omp-low.c (finish_taskreg_remap): New function.
25250 (finish_taskreg_scan): If unit size of ctx->record_type
25251 is non-constant, unshare the size expression and replace
25252 decls in it with possible outer var refs.
25254 PR middle-end/80809
25255 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
25256 GOVD_SHARED rather than GOVD_PRIVATE with it.
25257 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
25258 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
25260 PR middle-end/80853
25261 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
25262 as last argument to build_outer_var_ref for pointer bases of array
25263 section reductions.
25265 2017-05-19 Martin Sebor <msebor@redhat.com>
25267 * print-tree.c (print_node): Print DECL_READ_P flag.
25269 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25271 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
25272 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
25273 * cgraph.c: Likewise.
25274 * cgraphunit.c: Likewise.
25275 * gengtype.c: Likewise.
25276 * ipa-cp.c: Likewise.
25277 * ipa-devirt.c: Likewise.
25278 * ipa-icf.c: Likewise.
25279 * ipa-predicate.c: Likewise.
25280 * ipa-profile.c: Likewise.
25281 * ipa-prop.c: Likewise.
25282 * ipa-split.c: Likewise.
25284 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
25285 edge_predicate_pool, dump_inline_hints,
25286 inline_summary::account_size_time, redirect_to_unreachable,
25287 edge_set_predicate, set_hint_predicate,
25288 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
25289 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
25290 inline_summary_t::remove, remap_hint_predicate_after_duplication,
25291 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
25292 ipa_call_summary_t::remove, initialize_growth_caches,
25293 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
25294 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
25295 mark_modified, unmodified_parm_1, unmodified_parm,
25296 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
25297 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
25298 compute_bb_predicates, will_be_nonconstant_expr_predicate,
25299 will_be_nonconstant_predicate, record_modified_bb_info,
25300 get_minimal_bb, record_modified, param_change_prob,
25301 phi_result_unknown_predicate, predicate_for_phi_result,
25302 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
25303 estimate_function_body_sizes, compute_inline_parameters,
25304 compute_inline_parameters_for_curren, pass_data_inline_parameters,
25305 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
25306 inline_update_callee_summaries, remap_edge_change_prob,
25307 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
25308 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
25309 inline_analyze_function, inline_summary_t::insert,
25310 inline_generate_summary, read_ipa_call_summary, inline_read_section,
25311 inline_read_summary, write_ipa_call_summary, inline_write_summary,
25312 inline_free_summary): Move to ipa-fnsummary.h
25313 (predicate_t): Remove.
25314 * ipa-fnsummary.c: New file.
25315 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
25316 (enum inline_hints_vals, inline_hints, agg_position_info,
25317 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
25318 inline_summaries, ipa_call_summary, ipa_call_summary_t,
25319 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
25320 dump_inline_summary, dump_inline_hints, inline_generate_summary,
25321 inline_read_summary, inline_write_summary, inline_free_summary,
25322 inline_analyze_function, initialize_inline_failed,
25323 inline_merge_summary, inline_update_overall_summary,
25324 compute_inline_parameters): Move to ipa-fnsummary.h
25325 * ipa-fnsummary.h: New file.
25326 * ipa-inline-transform.h: Include ipa-inline.h.
25327 * ipa-inline.c: LIkewise.
25329 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25331 * ipa-inline.c (edge_badness): Use inlined_time instead of
25332 inline_summaries->get.
25334 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25336 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
25338 2017-05-22 Nathan Sidwell <nathan@acm.org>
25340 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
25341 (fdump-lang): Document 'raw' option.
25342 * dumpfile.h (TDI_tu): Delete.
25343 * dumpfile.c (dump_files): Remove translation-unit.
25344 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
25346 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
25348 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
25349 command option from $(AWK) call.
25350 * config/avr/genmultilib.awk: Simplify and rewrite so that it
25351 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
25352 [FORMAT]: Remove handling of variable.
25353 * config/avr/t-multilib: Regenerate.
25355 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25357 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
25359 (dump_inline_summary): Do not print self_time.
25360 (estimate_function_body_sizes): Do not set self_time.
25361 (compute_inline_parameters): Likewise.
25362 (inline_read_section, inline_write_summary): Do not stream self_time.
25363 * ipa-inline.h (inline_summary): Drop self_time.
25365 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25367 * ipa-inline-analysis.c (account_size_time): Rename to ...
25368 (inline_summary::account_size_time): ... this one.
25369 (reset_ipa_call_summary): Turn to ...
25370 (ipa_call_summary::reset): ... this one.
25371 (reset_inline_summary): Turn to ...
25372 (inline_summary::reset): ... this one.
25373 (inline_summary_t::remove): Update.
25374 (inline_summary_t::duplicate): Update.
25375 (ipa_call_summary_t::remove): Update.
25376 (dump_inline_summary): Update.
25377 (estimate_function_body_sizes): Update.
25378 (compute_inline_parameters): Update.
25379 (estimate_node_size_and_time): Update.
25380 (inline_merge_summary): Update.
25381 (inline_update_overall_summary): Update.
25382 (inline_read_section): Update.
25383 (inline_write_summary): Update.
25384 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
25385 add account_size_time and reset member functions.
25386 (ipa_call_summary): Add reset function.
25387 * ipa-predicate.h (predicate::operator &): Constify.
25389 2017-05-22 Richard Biener <rguenther@suse.de>
25391 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
25393 2017-05-19 Jason Merrill <jason@redhat.com>
25395 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
25397 2017-05-19 Marek Polacek <polacek@redhat.com>
25400 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
25401 TYPE_OVERFLOW_WRAPS checks.
25403 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
25405 * tree-core.h (enum omp_clause_default_kind): Add
25406 "OMP_CLAUSE_DEFAULT_PRESENT".
25407 * tree-pretty-print.c (dump_omp_clause): Handle it.
25408 * gimplify.c (enum gimplify_omp_var_data): Add
25409 "GOVD_MAP_FORCE_PRESENT".
25410 (gimplify_adjust_omp_clauses_1): Map it to
25411 "GOMP_MAP_FORCE_PRESENT".
25412 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
25414 * gimplify.c (oacc_default_clause): Clarify.
25416 2017-05-19 Nathan Sidwell <nathan@acm.org>
25418 LANG_HOOK_REGISTER_DUMPS
25419 * toplev.c (general_init): Call register dump lang hook.
25420 * doc/invoke.texi: Document -fdump-lang option family.
25421 * dumpfile.c (dump_files): Remove class dump here.
25422 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
25423 * dumpfile.h (tree_dump_index): Remove TDI_class.
25424 * langhooks-def.h (lhd_register_dumps): Declare.
25425 (LANG_HOOKS_REGISTER_DUMPS): Define.
25426 (LANG_HOOKS_INITIALIZER): Add it.
25427 * langhooks.c (lhd_register_dumps): Define.
25428 * langhooks.h (struct lang_hooks): Add register_dumps.
25430 2017-05-19 Nathan Sidwell <nathan@acm.org>
25432 * context.h (context::set_passes): New.
25433 * context.c (context::context): Do not create pass manager.
25434 * toplev.c (general_init): Create pass manager here.
25436 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
25438 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
25439 use this splitter if two add or or instructions would also work for
25440 the constant we want to generate.
25442 2017-05-19 Richard Biener <rguenther@suse.de>
25445 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
25446 predicate evaluation.
25448 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
25450 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
25452 * ipa-inline.c (want_inline_small_function_p): Do not cast to
25455 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
25457 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
25458 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
25459 (inline_edge_summary_vec): Turn into ...
25460 (ipa_call_summaries): ... this one.
25461 (redirect_to_unreachable, edge_set_predicate,
25462 evaluate_properties_for_edge, inline_summary_alloc,
25463 reset_ipa_call_summary, reset_inline_summary,
25464 inline_summary_t::duplicate): Update.
25465 (inline_edge_duplication_hook): Turn to ...
25466 (ipa_call_summary_t::duplicate): ... this one.
25467 (inline_edge_removal_hook): Turn to ...
25468 (ipa_call_summary_t::remove): ... this one.
25469 (dump_inline_edge_summary): Turn to ...
25470 (dump_ipa_call_summary): ... this one.
25471 (estimate_function_body_sizes): Update.
25472 (inline_update_callee_summaries): Update.
25473 (remap_edge_change_prob): Update.
25474 (remap_edge_summaries): Update.
25475 (inline_merge_summary): Update.
25476 (do_estimate_edge_time): Update.
25477 (inline_generate_summary): Update.
25478 (inline_read_section): Update.
25479 (inline_read_summary): Update.
25480 (inline_free_summary): Update.
25481 * ipa-inline.c (can_inline_edge_p): Update.
25482 (compute_inlined_call_time): Update.
25483 (want_inline_small_function_p): Update.
25484 (edge_badness): Update.
25485 (early_inliner): Update.
25486 * ipa-inline.h (inline_edge_summary): Turn to ...
25487 (ipa_call_summary): ... this one.
25488 (ipa_call_summary_t): New class.
25489 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
25490 (ipa_call_summaries): New.
25491 (inline_edge_summary): Remove.
25492 (estimate_edge_growth): Update.
25493 * ipa-profile.c (ipa_propagate_frequency_1): Update.
25494 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25495 * ipa-split.c (execute_split_functions): Update.
25496 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
25498 2017-05-19 Richard Biener <rguenther@suse.de>
25500 PR middle-end/80764
25501 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
25503 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
25505 * config/rs6000/rs6000.c (struct machine_function): Add field
25506 fpr_is_wrapped_separately.
25507 (rs6000_get_separate_components): Use 64 components. Handle the
25508 new FPR components.
25509 (rs6000_components_for_bb): Handle the FPR components.
25510 (rs6000_emit_prologue_components): Handle the FPR components.
25511 (rs6000_emit_epilogue_components): Handle the FPR components.
25512 (rs6000_set_handled_components): Handle the FPR components.
25513 (rs6000_emit_prologue): Don't output prologue code for those FPRs
25514 that are already separately shrink-wrapped.
25515 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
25516 that are already separately shrink-wrapped.
25518 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25521 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
25524 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
25525 (define_peephole2 for Altivec d-form load): Add peepholes to catch
25526 cases where the register allocator uses a move and an offsettable
25527 memory operation to/from a FPR register on ISA 2.06/2.07.
25528 (define_peephole2 for Altivec d-form store): Likewise.
25530 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
25533 * config/i386/mmx.md (*mov<mode>_internal): Enable
25534 alternatives 11, 12, 13 and 14 also for 32bit targets.
25535 Remove alternatives 15, 16, 17 and 18.
25536 * config/i386/sse.md (vec_concatv2di): Change
25537 alternative (!x, *y) to (x, ?!*Yn).
25539 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
25541 * dumpfile.h (enum dump_kind): Remove stray comma.
25543 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
25545 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
25546 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
25547 predicate::num_conditions
25548 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
25549 (CHANGED): turn into predicate::changed.
25550 (agg_position_info): Move to ipa-predicate.h
25551 (add_condition, predicate::add_clause, predicate::operator &=,
25552 predicate::or_with, predicate::evaluate, predicate::probability,
25553 dump_condition, dump_clause, predicate::dump,
25554 predicate::remap_after_duplication, predicate::remap_after_inlining,
25555 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
25556 (evaluate_conditions_for_known_args): Update.
25557 (set_cond_stmt_execution_predicate): Update.
25558 * ipa-inline.h: Include ipa-predicate.h
25559 (condition, inline_param_summary, conditions, agg_position_info,
25560 predicate): Move to ipa-predicate.h
25561 * ipa-predicate.c: New file.
25562 * ipa-predicate.h: New file.
25564 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
25566 * final.c (leaf_function_p): Check we are not in a sequence.
25568 2017-05-18 Martin Liska <mliska@suse.cz>
25570 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
25571 * dumpfile.c (dump_register): Use new enum dump_kind.
25572 (get_dump_file_name): Likewise.
25573 (dump_enable_all): Likewise.
25574 (dump_switch_p_1): Likewise.
25575 (enable_rtl_dump_file): Remove usage of TDF_RTL.
25576 * dumpfile.h (enum dump_kind): New enum type.
25577 (struct dump_file_info): Create constructor and
25578 format fields and comments.
25579 * passes.c (pass_manager::register_one_dump_file):
25581 * statistics.c (statistics_early_init): Likewise.
25582 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
25583 TDF_TREE with TDF_SLIM.
25584 (gather_memory_references_ref): Likewise.
25586 2017-05-18 Martin Liska <mliska@suse.cz>
25588 * vec.h (struct vnull): Use it.
25590 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
25592 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
25593 (true_predicate, false_predicate, true_predicate_p,
25594 false_predicate_p): Remove.
25595 (single_cond_predicate, not_inlined_predicate): Turn to member function
25597 (add_condition): Update.
25598 (add_clause): Turn to...
25599 (predicate::add_clause): ... this one; update; allow passing NULL
25601 (and_predicates): Turn to ...
25602 (predicate::operator &=): ... this one.
25603 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
25604 (or_predicates): Turn to ...
25605 (predicate::or_with): ... this one.
25606 (evaluate_predicate): Turn to ...
25607 (predicate::evaluate): ... this one.
25608 (predicate_probability): Turn to ...
25609 (predicate::probability): ... this one.
25610 (dump_condition): Update.
25611 (dump_predicate): Turn to ...
25612 (predicate::dump): ... this one.
25613 (account_size_time): Update.
25614 (edge_set_predicate): Update.
25615 (set_hint_predicate): UPdate.
25616 (evaluate_conditions_for_known_args): Update.
25617 (evaluate_properties_for_edge): Update.
25618 (remap_predicate_after_duplication): Turn to...
25619 (predicate::remap_after_duplication): ... this one.
25620 (remap_hint_predicate_after_duplication): Update.
25621 (inline_summary_t::duplicate): UPdate.
25622 (dump_inline_edge_summary): Update.
25623 (dump_inline_summary): Update.
25624 (set_cond_stmt_execution_predicate): Update.
25625 (set_switch_stmt_execution_predicate): Update.
25626 (compute_bb_predicates): Update.
25627 (will_be_nonconstant_expr_predicate): Update.
25628 (will_be_nonconstant_predicate): Update.
25629 (phi_result_unknown_predicate): Update.
25630 (predicate_for_phi_result): Update.
25631 (array_index_predicate): Update.
25632 (estimate_function_body_sizes): Update.
25633 (estimate_node_size_and_time): Update.
25634 (estimate_ipcp_clone_size_and_time): Update.
25635 (remap_predicate): Rename to ...
25636 (predicate::remap_after_inlining): ... this one.
25637 (remap_hint_predicate): Update.
25638 (inline_merge_summary): Update.
25639 (inline_update_overall_summary): Update.
25640 (estimate_size_after_inlining): Update.
25641 (read_predicate): Rename to ...
25642 (predicate::stream_in): ... this one.
25643 (read_inline_edge_summary): Update.
25644 (write_predicate): Rename to ...
25645 (predicate::stream_out): ... this one.
25646 (write_inline_edge_summary): Update.
25647 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
25648 (clause_t): Turn to uint32_t
25649 (predicate): Turn to class; implement constructor and operators
25651 (size_time_entry): Update.
25652 (inline_summary): Update.
25653 (inline_edge_summary): Update.
25655 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
25657 * fold-const.c (fold_binary_loc): Move transformation...
25658 * match.pd (C - X CMP X): ... here.
25660 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
25662 * config/sparc/sparc.c (sparc_option_override): Set function
25663 alignment for -mcpu=niagara7 to 64 to match the I$ line.
25664 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
25666 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
25668 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
25670 2017-05-18 Marek Polacek <polacek@redhat.com>
25673 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
25674 (pass_ubsan::execute): Call gimple_assign_single_p instead of
25675 gimple_assign_load_p.
25677 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
25679 PR middle-end/80692
25680 * real.c (do_compare): Give decimal_do_compare preference over
25681 comparing just the signs.
25683 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
25685 * doc/md.texi (Canonicalization of Instructions): Describe the
25686 canonical form of instructions that inherently set a condition
25689 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
25691 PR middle-end/80775
25692 * tree-cfg.c: Move deletion of unreachable case statements to after
25693 the merging of consecutive case labels.
25695 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
25697 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
25698 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
25699 restoring of callee-saved registers.
25701 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
25703 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
25704 * config/visium/visium.c (single_set_and_flags): Likewise.
25705 * config/visium/visium.md (Substitutions): Likewise.
25707 2017-05-17 Martin Liska <mliska@suse.cz>
25709 * cfg.c: Introduce dump_flags_t type and
25710 use it instead of int type.
25712 * cfghooks.c: Likewise.
25713 * cfghooks.h (struct cfg_hooks): Likewise.
25714 * cfgrtl.c: Likewise.
25715 * cfgrtl.h: Likewise.
25716 * cgraph.c (cgraph_node::get_body): Likewise.
25717 * coretypes.h: Likewise.
25718 * domwalk.c: Likewise.
25719 * domwalk.h: Likewise.
25720 * dumpfile.c (struct dump_option_value_info): Likewise.
25721 (dump_enable_all): Likewise.
25722 (dump_switch_p_1): Likewise.
25723 (opt_info_switch_p): Likewise.
25724 * dumpfile.h (enum tree_dump_index): Likewise.
25725 (struct dump_file_info): Likewise.
25726 * genemit.c: Likewise.
25727 * generic-match-head.c: Likewise.
25728 * gengtype.c (open_base_files): Likewise.
25729 * gimple-pretty-print.c: Likewise.
25730 * gimple-pretty-print.h: Likewise.
25731 * graph.c (print_graph_cfg): Likewise.
25732 * graphite-scop-detection.c (dot_all_sese): Likewise.
25733 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
25734 * loop-unroll.c (report_unroll): Likewise.
25735 * passes.c (pass_manager::register_one_dump_file): Likewise.
25736 * print-tree.c: Likewise.
25737 * statistics.c: Likewise.
25738 * tree-cfg.c: Likewise.
25739 * tree-cfg.h: Likewise.
25740 * tree-dfa.c: Likewise.
25741 * tree-dfa.h: Likewise.
25742 * tree-dump.c (dump_function): Likewise.
25743 * tree-dump.h (struct dump_info): Likewise.
25744 * tree-pretty-print.c: Likewise.
25745 * tree-pretty-print.h: Likewise.
25746 * tree-ssa-live.c: Likewise.
25747 * tree-ssa-live.h: Likewise.
25748 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
25749 * tree-vect-loop.c: Likewise.
25750 * tree-vect-slp.c: Likewise.
25752 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
25753 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25755 PR tree-optimization/80457
25756 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
25757 of all arguments to a statement as scalar_to_vec operations.
25758 (vectorizable_call): Adjust call to vect_model_simple_cost for
25760 (vectorizable_conversion): Likewise.
25761 (vectorizable_assignment): Likewise.
25762 (vectorizable_shift): Likewise.
25763 (vectorizable_operation): Likewise.
25764 (vectorizable_comparison): Likewise.
25765 (vect_is_simple_cond): Record the def types for operands.
25766 (vectorizable_condition): Likewise, call vect_model_simple_cost.
25767 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
25768 for statement argument count.
25770 2017-05-16 Carl Love <cel@us.ibm.com>
25772 * config/rs6000/rs6000-c: Add support for built-in functions
25773 vector unsigned long long vec_bperm (vector unsigned long long,
25774 vector unsigned char)
25775 vector signed long long vec_mule (vector signed int,
25777 vector unsigned long long vec_mule (vector unsigned int,
25778 vector unsigned int)
25779 vector signed long long vec_mulo (vector signed int,
25781 vector unsigned long long vec_mulo (vector unsigned int,
25782 vector unsigned int)
25783 vector signed char vec_sldw (vector signed char,
25784 vector signed char,
25786 vector unsigned char vec_sldw (vector unsigned char,
25787 vector unsigned char,
25789 vector signed short vec_sldw (vector signed short,
25790 vector signed short,
25792 vector unsigned short vec_sldw (vector unsigned short,
25793 vector unsigned short,
25795 vector signed int vec_sldw (vector signed int,
25798 vector unsigned int vec_sldw (vector unsigned int,
25799 vector unsigned int,
25801 vector signed long long vec_sldw (vector signed long long,
25802 vector signed long long,
25804 vector unsigned long long vec_sldw (vector unsigned long long,
25805 vector unsigned long long,
25807 * config/rs6000/rs6000-c: Add support for built-in functions
25808 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
25809 * config/rs6000/altivec.h: Add defintion for vec_sldw.
25810 * doc/extend.texi: Update the built-in documentation for the
25811 new built-in functions.
25813 2017-05-16 Marek Polacek <polacek@redhat.com>
25817 * tree.c (save_expr): Don't fold the expression.
25819 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
25821 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
25822 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
25823 and (?*y,m). Update insn attributes.
25825 2017-05-16 Martin Liska <mliska@suse.cz>
25827 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
25828 flags argument of print_gimple_stmt, print_gimple_expr,
25829 print_generic_stmt and print_generic_expr.
25830 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
25831 * coretypes.h: Likewise.
25832 * except.c (dump_eh_tree): Likewise.
25833 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
25834 * gimple-pretty-print.h: Likewise.
25835 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
25836 (backprop::push_to_worklist): Likewise.
25837 (backprop::pop_from_worklist): Likewise.
25838 (backprop::process_use): Likewise.
25839 (backprop::intersect_uses): Likewise.
25840 (note_replacement): Likewise.
25841 * gimple-ssa-store-merging.c
25842 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
25843 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
25844 (pass_store_merging::execute): Likewise.
25845 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
25846 (ssa_base_cand_dump_callback): Likewise.
25847 (dump_incr_vec): Likewise.
25848 (replace_refs): Likewise.
25849 (replace_mult_candidate): Likewise.
25850 (create_add_on_incoming_edge): Likewise.
25851 (create_phi_basis): Likewise.
25852 (insert_initializers): Likewise.
25853 (all_phi_incrs_profitable): Likewise.
25854 (introduce_cast_before_cand): Likewise.
25855 (replace_one_candidate): Likewise.
25856 * gimplify.c (gimplify_expr): Likewise.
25857 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
25858 (set_rename): Likewise.
25859 (rename_uses): Likewise.
25860 (copy_loop_phi_nodes): Likewise.
25861 (add_close_phis_to_merge_points): Likewise.
25862 (copy_loop_close_phi_args): Likewise.
25863 (copy_cond_phi_args): Likewise.
25864 (graphite_copy_stmts_from_block): Likewise.
25865 (translate_pending_phi_nodes): Likewise.
25866 * graphite-poly.c (print_pdr): Likewise.
25867 (dump_gbb_cases): Likewise.
25868 (dump_gbb_conditions): Likewise.
25869 (print_scop_params): Likewise.
25870 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
25871 (build_cross_bb_scalars_use): Likewise.
25872 (gather_bbs::before_dom_children): Likewise.
25873 * hsa-dump.c (dump_hsa_immed): Likewise.
25874 * ipa-cp.c (print_ipcp_constant_value): Likewise.
25875 (get_replacement_map): Likewise.
25876 * ipa-inline-analysis.c (dump_condition): Likewise.
25877 (estimate_function_body_sizes): Likewise.
25878 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
25879 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
25880 * ipa-prop.c (ipa_dump_param): Likewise.
25881 (ipa_print_node_jump_functions_for_edge): Likewise.
25882 (ipa_modify_call_arguments): Likewise.
25883 (ipa_modify_expr): Likewise.
25884 (ipa_dump_param_adjustments): Likewise.
25885 (ipa_dump_agg_replacement_values): Likewise.
25886 (ipcp_modif_dom_walker::before_dom_children): Likewise.
25887 * ipa-pure-const.c (check_stmt): Likewise.
25888 (pass_nothrow::execute): Likewise.
25889 * ipa-split.c (execute_split_functions): Likewise.
25890 * omp-offload.c (dump_oacc_loop_part): Likewise.
25891 (dump_oacc_loop): Likewise.
25892 * trans-mem.c (tm_log_emit): Likewise.
25893 (tm_memopt_accumulate_memops): Likewise.
25894 (dump_tm_memopt_set): Likewise.
25895 (dump_tm_memopt_transform): Likewise.
25896 * tree-cfg.c (gimple_verify_flow_info): Likewise.
25897 (print_loop): Likewise.
25898 * tree-chkp-opt.c (chkp_print_addr): Likewise.
25899 (chkp_gather_checks_info): Likewise.
25900 (chkp_get_check_result): Likewise.
25901 (chkp_remove_check_if_pass): Likewise.
25902 (chkp_use_outer_bounds_if_possible): Likewise.
25903 (chkp_reduce_bounds_lifetime): Likewise.
25904 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
25905 (chkp_mark_completed_bounds): Likewise.
25906 (chkp_register_incomplete_bounds): Likewise.
25907 (chkp_mark_invalid_bounds): Likewise.
25908 (chkp_maybe_copy_and_register_bounds): Likewise.
25909 (chkp_build_returned_bound): Likewise.
25910 (chkp_get_bound_for_parm): Likewise.
25911 (chkp_build_bndldx): Likewise.
25912 (chkp_get_bounds_by_definition): Likewise.
25913 (chkp_generate_extern_var_bounds): Likewise.
25914 (chkp_get_bounds_for_decl_addr): Likewise.
25915 * tree-chrec.c (chrec_apply): Likewise.
25916 * tree-data-ref.c (dump_data_reference): Likewise.
25917 (dump_subscript): Likewise.
25918 (dump_data_dependence_relation): Likewise.
25919 (analyze_overlapping_iterations): Likewise.
25920 * tree-inline.c (expand_call_inline): Likewise.
25921 (tree_function_versioning): Likewise.
25922 * tree-into-ssa.c (dump_defs_stack): Likewise.
25923 (dump_currdefs): Likewise.
25924 (dump_names_replaced_by): Likewise.
25925 (dump_update_ssa): Likewise.
25926 (update_ssa): Likewise.
25927 * tree-object-size.c (pass_object_sizes::execute): Likewise.
25928 * tree-parloops.c (build_new_reduction): Likewise.
25929 (try_create_reduction_list): Likewise.
25930 (ref_conflicts_with_region): Likewise.
25931 (oacc_entry_exit_ok_1): Likewise.
25932 (oacc_entry_exit_single_gang): Likewise.
25933 * tree-pretty-print.h: Likewise.
25934 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
25935 (get_scalar_evolution): Likewise.
25936 (add_to_evolution): Likewise.
25937 (get_loop_exit_condition): Likewise.
25938 (analyze_evolution_in_loop): Likewise.
25939 (analyze_initial_condition): Likewise.
25940 (analyze_scalar_evolution): Likewise.
25941 (instantiate_scev): Likewise.
25942 (number_of_latch_executions): Likewise.
25943 (gather_chrec_stats): Likewise.
25944 (final_value_replacement_loop): Likewise.
25945 (scev_const_prop): Likewise.
25946 * tree-sra.c (dump_access): Likewise.
25947 (disqualify_candidate): Likewise.
25948 (create_access): Likewise.
25949 (reject): Likewise.
25950 (maybe_add_sra_candidate): Likewise.
25951 (create_access_replacement): Likewise.
25952 (analyze_access_subtree): Likewise.
25953 (analyze_all_variable_accesses): Likewise.
25954 (sra_modify_assign): Likewise.
25955 (initialize_constant_pool_replacements): Likewise.
25956 (find_param_candidates): Likewise.
25957 (decide_one_param_reduction): Likewise.
25958 (replace_removed_params_ssa_names): Likewise.
25959 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
25960 * tree-ssa-copy.c (dump_copy_of): Likewise.
25961 (copy_prop_visit_cond_stmt): Likewise.
25962 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
25963 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
25964 (record_equivalences_from_stmt): Likewise.
25965 * tree-ssa-dse.c (compute_trims): Likewise.
25966 (delete_dead_call): Likewise.
25967 (delete_dead_assignment): Likewise.
25968 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
25969 (forward_propagate_into_cond): Likewise.
25970 (pass_forwprop::execute): Likewise.
25971 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
25972 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
25974 (move_computations_worker): Likewise.
25975 (execute_sm): Likewise.
25976 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
25977 (remove_exits_and_undefined_stmts): Likewise.
25978 (remove_redundant_iv_tests): Likewise.
25979 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
25980 (adjust_iv_update_pos): Likewise.
25981 * tree-ssa-math-opts.c (bswap_replace): Likewise.
25982 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
25983 (value_replacement): Likewise.
25984 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
25985 * tree-ssa-pre.c (print_pre_expr): Likewise.
25986 (get_representative_for): Likewise.
25987 (create_expression_by_pieces): Likewise.
25988 (insert_into_preds_of_block): Likewise.
25989 (eliminate_insert): Likewise.
25990 (eliminate_dom_walker::before_dom_children): Likewise.
25991 (eliminate): Likewise.
25992 (remove_dead_inserted_code): Likewise.
25993 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
25994 * tree-ssa-reassoc.c (get_rank): Likewise.
25995 (eliminate_duplicate_pair): Likewise.
25996 (eliminate_plus_minus_pair): Likewise.
25997 (eliminate_not_pairs): Likewise.
25998 (undistribute_ops_list): Likewise.
25999 (eliminate_redundant_comparison): Likewise.
26000 (update_range_test): Likewise.
26001 (optimize_range_tests_var_bound): Likewise.
26002 (optimize_vec_cond_expr): Likewise.
26003 (rewrite_expr_tree): Likewise.
26004 (rewrite_expr_tree_parallel): Likewise.
26005 (linearize_expr): Likewise.
26006 (break_up_subtract): Likewise.
26007 (linearize_expr_tree): Likewise.
26008 (attempt_builtin_powi): Likewise.
26009 (attempt_builtin_copysign): Likewise.
26010 (transform_stmt_to_copy): Likewise.
26011 (transform_stmt_to_multiply): Likewise.
26012 (dump_ops_vector): Likewise.
26013 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
26014 (print_scc): Likewise.
26015 (set_ssa_val_to): Likewise.
26016 (visit_reference_op_store): Likewise.
26017 (visit_use): Likewise.
26018 (sccvn_dom_walker::before_dom_children): Likewise.
26019 (run_scc_vn): Likewise.
26020 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
26022 (expr_hash_elt::print): Likewise.
26023 (const_and_copies::pop_to_marker): Likewise.
26024 (const_and_copies::record_const_or_copy_raw): Likewise.
26025 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
26026 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
26027 (dump_predicates): Likewise.
26028 (find_uninit_use): Likewise.
26029 (warn_uninitialized_phi): Likewise.
26030 (pass_late_warn_uninitialized::execute): Likewise.
26031 * tree-ssa.c (verify_vssa): Likewise.
26032 (verify_ssa): Likewise.
26033 (maybe_optimize_var): Likewise.
26034 * tree-vrp.c (dump_value_range): Likewise.
26035 (dump_all_value_ranges): Likewise.
26036 (dump_asserts_for): Likewise.
26037 (register_edge_assert_for_2): Likewise.
26038 (vrp_visit_cond_stmt): Likewise.
26039 (vrp_visit_switch_stmt): Likewise.
26040 (vrp_visit_stmt): Likewise.
26041 (vrp_visit_phi_node): Likewise.
26042 (simplify_cond_using_ranges_1): Likewise.
26043 (fold_predicate_in): Likewise.
26044 (evrp_dom_walker::before_dom_children): Likewise.
26045 (evrp_dom_walker::push_value_range): Likewise.
26046 (evrp_dom_walker::pop_value_range): Likewise.
26047 (execute_early_vrp): Likewise.
26049 2017-05-16 Richard Biener <rguenther@suse.de>
26051 * dwarf2out.c (loc_list_from_tree_1): Do not create
26052 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
26054 2017-05-16 Richard Biener <rguenther@suse.de>
26056 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
26058 (note_variable_value_in_expr): If we resolved the decl ref
26059 do not push to the stack.
26061 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
26063 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
26064 operations in fast-math mode.
26065 (vaddq_f16): Likewise.
26066 (vmul_f16): Likewise.
26067 (vmulq_f16): Likewise.
26068 (vsub_f16): Likewise.
26069 (vsubq_f16): Likewise.
26070 * config/arm/neon.md (add<mode>3): New.
26072 (fma:<VH:mode>3): New. Also remove outdated comment.
26075 2017-05-16 Martin Liska <mliska@suse.cz>
26079 * ipa-devirt.c (warn_types_mismatch): Fix typo.
26080 (odr_types_equivalent_p): Likewise.
26082 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
26084 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
26086 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
26089 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
26090 non-interunit SSE move alternatives with '?'.
26091 (zero-extendsidi peephole2): New peephole to skip intermediate
26092 general register in SSE zero-extend sequence.
26094 2017-05-15 Jeff Law <law@redhat.com>
26096 * reorg.c (relax_delay_slots): Create a new variable to hold
26097 the temporary target rather than clobbering TARGET_LABEL.
26099 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
26100 missing argument to extract_bit_field call.
26101 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
26103 2017-05-15 Martin Liska <mliska@suse.cz>
26106 * gcc.c (process_command): Do not allow empty argument of -o option.
26108 2017-05-15 Renlin Li <renlin.li@arm.com>
26110 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
26111 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
26112 * config/aarch64/constraints.md (Usf): Add long call check.
26113 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
26114 (call_value): Likewise.
26115 (sibcall): Likewise.
26116 (sibcall_value): Likewise.
26118 (call_value_insn): New.
26119 (sibcall_insn): Update rtx pattern.
26120 (sibcall_value_insn): Likewise.
26121 (call_internal): Remove.
26122 (call_value_internal): Likewise.
26123 (sibcall_internal): Likewise.
26124 (sibcall_value_internal): Likewise.
26125 (call_reg): Likewise.
26126 (call_symbol): Likewise.
26127 (call_value_reg): Likewise.
26128 (call_value_symbol): Likewise.
26130 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
26133 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
26135 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
26137 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
26138 compatible with CCGOCmode and with CCZmode.
26140 2017-05-14 Martin Sebor <msebor@redhat.com>
26142 PR middle-end/77671
26143 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
26144 (gimple_fold_builtin_snprintf): Same.
26145 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
26146 (gimple_fold_builtin_snprintf): Same.
26147 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
26148 of character types.
26149 (is_call_safe): New function.
26150 (try_substitute_return_value): Call it.
26151 (try_simplify_call): New function.
26152 (pass_sprintf_length::handle_gimple_call): Call it.
26154 2017-05-14 Martin Sebor <msebor@redhat.com>
26156 PR middle-end/80669
26157 * builtins.c (expand_builtin_stpncpy): Simplify.
26159 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
26161 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
26162 * config/i386/i386.h
26163 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
26164 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
26165 (struct machine_function): Add new members call_ms2sysv,
26166 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
26167 (struct machine_frame_state): New fields sp_realigned and
26168 sp_realigned_offset.
26169 * config/i386/i386.c
26170 (enum xlogue_stub): New enum.
26171 (enum xlogue_stub_sets): New enum.
26172 (class xlogue_layout): New class.
26173 (struct ix86_frame): New fields stack_realign_allocate_offset,
26174 stack_realign_offset and outlined_save_offset. Modify comments to
26175 detail stack layout when using out-of-line stubs.
26176 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
26177 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
26178 -mcall-ms2sysv-xlogues.
26179 (stub_managed_regs): New static variable.
26180 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
26181 registers managed by out-of-line stub.
26182 (disable_call_ms2sysv_xlogues): New function.
26183 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
26184 m->call_ms2sysv when appropriate and compute frame layout for
26186 (sp_valid_at, fp_valid_at): New inline functions.
26187 (choose_basereg): New function.
26188 (choose_baseaddr): Add align parameter, use choose_basereg and modify
26190 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
26191 Use align parameter of choose_baseaddr to generated aligned SSE movs
26193 (pro_epilogue_adjust_stack): Modify to track
26194 machine_frame_state::sp_realigned.
26195 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
26196 (ix86_nsaved_sseregs): Likewise.
26197 (ix86_emit_save_regs): Likewise.
26198 (ix86_emit_save_regs_using_mov): Likewise.
26199 (ix86_emit_save_sse_regs_using_mov): Likewise.
26200 (get_scratch_register_on_entry): Likewise.
26201 (gen_frame_set): New function.
26202 (gen_frame_load): Likewise.
26203 (gen_frame_store): Likewise.
26204 (emit_outlined_ms2sysv_save): Likewise.
26205 (emit_outlined_ms2sysv_restore): Likewise.
26206 (ix86_expand_prologue): Modify stack re-alignment code and call
26207 emit_outlined_ms2sysv_save when appropriate.
26208 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
26209 parameter rtx_insn *insn, which allows the function to be used to only
26210 generate the notes.
26211 (ix86_expand_epilogue): Modify validity checks of frame and stack
26212 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
26213 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
26214 * config/i386/predicates.md
26215 (save_multiple): New predicate.
26216 (restore_multiple): Likewise.
26217 * config/i386/sse.md
26218 (save_multiple<mode>): New pattern.
26219 (save_multiple_realign<mode>): Likewise.
26220 (restore_multiple<mode>): Likewise.
26221 (restore_multiple_and_return<mode>): Likewise.
26222 (restore_multiple_leave_return<mode>): Likewise.
26223 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
26225 2017-05-14 Julia Koval <julia.koval@intel.com>
26227 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
26228 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
26229 (__builtin_ia32_xsetbv): New builtins.
26230 * config/i386/i386.c (ix86_expand_special_args_builtin):
26232 (ix86_expand_builtin): Special expand for new intrinsics.
26233 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
26234 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
26235 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
26237 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26239 * cfganal.c (inverted_post_order_compute): Change argument type
26241 * cfganal.h (inverted_post_order_compute): Adjust prototype.
26242 * df-core.c (rest_of_handle_df_initialize): Adjust.
26243 (rest_of_handle_df_finish): Likewise.
26244 (df_analyze_1): Likewise.
26245 (df_analyze): Likewise.
26246 (loop_inverted_post_order_compute): Change argument to be a vec *.
26247 (df_analyze_loop): Adjust.
26248 (df_get_n_blocks): Likewise.
26249 (df_get_postorder): Likewise.
26250 * df.h (struct df_d): Change field to be a vec.
26251 * lcm.c (compute_laterin): Adjust.
26252 (compute_available): Likewise.
26253 * lra-lives.c (lra_create_live_ranges_1): Likewise.
26254 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
26255 * tree-ssa-pre.c (compute_antic): Likewise.
26257 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26259 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
26260 (depth_first_search::depth_first_search): Change structure init
26261 function to this constructor.
26262 (depth_first_search::add_bb): Rename function to this member.
26263 (depth_first_search::execute): Likewise.
26264 (flow_dfs_compute_reverse_finish): Adjust.
26266 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26268 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
26269 (longest_simple_path): Likewise.
26270 * shrink-wrap.c (spread_components): Likewise.
26271 (disqualify_problematic_components): Likewise.
26272 (emit_common_heads_for_components): Likewise.
26273 (emit_common_tails_for_components): Likewise.
26274 (insert_prologue_epilogue_for_components): Likewise.
26276 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26278 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
26281 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26283 * df-core.c (df_set_blocks): Start using auto_bitmap.
26284 (df_compact_blocks): Likewise.
26285 * df-problems.c (df_rd_confluence_n): Likewise.
26286 * df-scan.c (df_insn_rescan_all): Likewise.
26287 (df_process_deferred_rescans): Likewise.
26288 (df_update_entry_block_defs): Likewise.
26289 (df_update_exit_block_uses): Likewise.
26290 (df_entry_block_bitmap_verify): Likewise.
26291 (df_exit_block_bitmap_verify): Likewise.
26292 (df_scan_verify): Likewise.
26293 * lra-constraints.c (lra_constraints): Likewise.
26294 (undo_optional_reloads): Likewise.
26295 (lra_undo_inheritance): Likewise.
26296 * lra-remat.c (calculate_gen_cands): Likewise.
26297 (do_remat): Likewise.
26298 * lra-spills.c (assign_spill_hard_regs): Likewise.
26299 (spill_pseudos): Likewise.
26300 * tree-ssa-pre.c (bitmap_set_and): Likewise.
26301 (bitmap_set_subtract_values): Likewise.
26303 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26305 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
26306 management with auto_bitmap.
26307 (fix_inter_tick): Likewise.
26308 (fix_recovery_deps): Likewise.
26309 * ira.c (add_store_equivs): Likewise.
26310 (find_moveable_pseudos): Likewise.
26311 (split_live_ranges_for_shrink_wrap): Likewise.
26312 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
26313 (rtx_reuse_manager::seen_def_p): Likewise.
26314 (rtx_reuse_manager::set_seen_def): Likewise.
26315 * print-rtl.h (class rtx_reuse_manager): Likewise.
26317 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26319 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
26321 (migrate_btr_def): Likewise.
26322 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
26323 * df-core.c (loop_post_order_compute): Likewise.
26324 (loop_inverted_post_order_compute): Likewise.
26325 * hsa-common.h: Likewise.
26326 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
26327 * init-regs.c (initialize_uninitialized_regs): Likewise.
26328 * ipa-inline.c (resolve_noninline_speculation): Likewise.
26329 (inline_small_functions): Likewise.
26330 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
26331 * ira.c (combine_and_move_insns): Likewise.
26332 (build_insn_chain): Likewise.
26333 * loop-invariant.c (find_invariants): Likewise.
26334 * lower-subreg.c (propagate_pseudo_copies): Likewise.
26335 * predict.c (tree_predict_by_opcode): Likewise.
26336 (predict_paths_leading_to): Likewise.
26337 (predict_paths_leading_to_edge): Likewise.
26338 (estimate_loops_at_level): Likewise.
26339 (estimate_loops): Likewise.
26340 * shrink-wrap.c (try_shrink_wrapping): Likewise.
26341 (spread_components): Likewise.
26342 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
26343 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
26344 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
26345 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
26346 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
26347 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
26348 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
26349 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
26350 (mark_threaded_blocks): Likewise.
26351 (thread_through_all_blocks): Likewise.
26352 * tree-ssa.c (verify_ssa): Likewise.
26353 (execute_update_addresses_taken): Likewise.
26354 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
26356 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26358 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
26360 (post_order_compute): Likewise.
26361 (inverted_post_order_compute): Likewise.
26362 (pre_and_rev_post_order_compute_fn): Likewise.
26364 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26366 * genrecog.c (int_set::int_set): Explicitly construct our
26367 auto_vec base class.
26368 * vec.h (auto_vec::auto_vec): New constructor.
26370 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26372 * bitmap.h (class auto_bitmap): New constructor taking
26373 bitmap_obstack * argument.
26375 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26377 * bitmap.h (class auto_bitmap): Change type of m_bits to
26378 bitmap_head, and adjust ctor / dtor and member operators.
26380 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
26382 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
26383 when returned register mode doesn't match original mode.
26385 2017-05-12 Jeff Law <law@redhat.com>
26386 Jakub Jelinek <jakub@redhat.com>
26388 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
26389 we look for cc setter after the compare-elim changes.
26390 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
26391 within the vector to match what compare-elim now expects.
26392 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
26393 (xorsi3_flags, one_cmplsi2_flags): Likewise.
26395 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
26396 after the compare-elim changes.
26397 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
26398 the vector to match what compare-elim now expects.
26399 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
26400 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
26401 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
26402 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
26403 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
26405 * config/visium/visium.c (single_set_and_flags): Fix where
26406 we look for cc setter after the compare-elim changes.
26407 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
26408 with the vector to match what compare-elim now expects.
26409 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
26410 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
26411 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
26412 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
26413 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
26414 (neg<mode>2_insn_set_overflow): Likewise.
26416 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
26418 PR middle-end/79794
26419 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
26420 maybe_expand_insn call, set ops[0].target. If still set after call,
26421 set alt_rtl. Add extra arg to recursive calls.
26422 (extract_bit_field): Add alt_rtl argument. Pass to
26424 * expmed.h (extract_bit_field): Fix prototype.
26425 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
26426 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
26427 to extract_bit_field_calls.
26428 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
26429 Pass alt_rtl to extract_bit_field calls.
26430 * calls.c (store_unaligned_arguments_into_psuedos)
26431 load_register_parameters): Pass extra NULL to extract_bit_field calls.
26432 * optabs.c (maybe_legitimize_operand): Clear op->target when call
26434 * optabs.h (struct expand_operand): Add target bitfield.
26436 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
26438 * compare-elim.c (try_eliminate_compare): Canonicalize
26439 operation with embedded compare to
26440 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
26441 (set (reg) (operation)].
26443 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
26445 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
26448 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
26449 cost of adding a carry flag for ADC instruction.
26450 [case MINUS]: Ignore the cost of subtracting a carry flag
26451 for SBB instruction.
26453 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
26455 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
26457 * config/rs6000/bmiintrin.h: New file.
26458 * config/rs6000/bmi2intrin.h: New file.
26459 * config/rs6000/x86intrin.h: New file.
26461 2017-05-12 Jeff Law <law@redhat.com>
26463 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
26466 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
26468 PR middle-end/80707
26469 * tree-cfg.c: Remove cfg edges of unreachable case statements.
26471 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26473 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
26474 early expansion of vector divide builtins.
26475 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
26476 builtins identified as having unsigned arguments.
26478 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26480 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
26481 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
26482 expansion of vector logical operations (and, andc, or, xor,
26485 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26487 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
26488 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
26490 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26492 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
26493 early GIMPLE expansion of vector multiplies.
26495 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26497 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
26498 TARGET_HAVE_MOVT conditional.
26499 (movt splitter): Likewise.
26501 2017-05-12 Richard Biener <rguenther@suse.de>
26503 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
26504 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
26505 Fold all stmts not inplace.
26507 2017-05-12 Richard Biener <rguenther@suse.de>
26509 PR tree-optimization/80713
26510 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
26511 inserted_exprs bit for not removed stmts.
26513 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
26515 PR middle-end/69921
26516 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
26517 parallelized" attribute for parallelized OpenACC kernels.
26518 * omp-offload.c (execute_oacc_device_lower): Use it.
26520 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
26521 Set "oacc kernels" attribute.
26522 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
26523 parameter. Adjust all users.
26524 (oacc_fn_attrib_kernels_p): Remove function.
26525 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
26526 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
26527 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
26528 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
26529 assert "oacc kernels" attribute is set.
26531 2017-05-11 Carl Love <cel@us.ibm.com>
26533 * config/rs6000/rs6000-c: Add support for built-in functions
26534 vector unsigned char vec_popcnt (vector signed char)
26535 vector unsigned char vec_popcnt (vector unsigned char)
26536 vector unsigned short vec_popcnt (vector signed short)
26537 vector unsigned short vec_popcnt (vector unsigned short)
26538 vector unsigned int vec_popcnt (vector signed int)
26539 vector unsigned int vec_popcnt (vector unsigned int)
26540 vector unsigned long long vec_popcnt (vector signed long long)
26541 vector unsigned long long vec_popcnt (vector unsigned long long)
26542 vector signed long long vec_slo (vector signed long long,
26543 vector signed char)
26544 vector signed long long vec_slo (vector signed long long,
26545 vector unsigned char)
26546 vector unsigned long long vec_slo (vector unsigned long long,
26547 vector signed char)
26548 vector unsigned long long vec_slo (vector unsigned long long,
26549 vector unsigned char)
26550 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
26551 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
26552 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
26553 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
26554 * doc/extend.texi: Update the built-in documentation file for the
26555 new built-in functions.
26557 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
26559 * attribs.h (sorted_attr_string): Move machine independent
26560 functions for target clone support from the i386 port to common
26561 code. Rename ix86_function_versions to common_function_versions.
26562 Rename make_name to make_unique_name.
26563 (common_function_versions): Likewise.
26564 (make_unique_name): Likewise.
26565 (make_dispatcher_decl): Likewise.
26566 (is_function_default_version): Likewise.
26567 * attribs.c (attr_strcmp): Likewise.
26568 (sorted_attr_string): Likewise.
26569 (common_function_versions): Likewise.
26570 (make_unique_name): Likewise.
26571 (make_dispatcher_decl): Likewise.
26572 (is_function_default_version): Likewise.
26573 * config/i386/i386.c (attr_strcmp): Likewise.
26574 (sorted_attr_string): Likewise.
26575 (ix86_function_versions): Likewise.
26576 (make_name): Likewise.
26577 (make_dispatcher_decl): Likewise.
26578 (is_function_default_version): Likewise.
26579 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
26581 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26584 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
26585 Account for direct move costs for vec_construct of integer
26588 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
26591 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
26592 (UNSPEC_STX_ATOMIC): Ditto.
26593 (loaddi_via_sse): New insn.
26594 (storedi_via_sse): Ditto.
26595 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
26596 Update corresponding peephole2 patterns.
26597 (atomic_storedi_fpu): Ditto.
26599 2017-05-11 Julia Koval <julia.koval@intel.com>
26601 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
26602 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
26604 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
26605 (__builtin_ia32_rsqrt14ss_mask): New builtins.
26606 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
26608 2017-05-11 Nathan Sidwell <nathan@acm.org>
26610 * graphite-poly.c: Include dumpfile.h.
26612 * dumpfle.h (dump_function): Declare here ...
26613 * tree-dump.h (dump_function): ... not here.
26614 * dumpfile.c: #include tree-cfg.h.
26615 (dump_function): Move here from ...
26616 * tree-dump.c (dump_function): ... here.
26617 * gimplify.c: #include splay-tree.h, not tree-dump.h.
26618 * graphite-poly.c: Don't include tree-dump.h.
26619 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
26620 * print-tree.c: Likewise.
26621 * stor-layout.c: Likewise.
26622 * tree-nested.c: Likewise.
26624 * dumpfile.c (dump_start): Use TDF_FLAGS.
26625 (dump_enable_all): Fix TDF_KIND check thinko.
26627 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
26629 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26630 array entries to represent two legal parameterizations of the
26631 overloaded __builtin_cmpb function, as represented by the
26632 P6_OV_BUILTIN_CMPB constant.
26633 (altivec_resolve_overloaded_builtin): Add special case handling
26634 for the __builtin_cmpb function, as represented by the
26635 P6_OV_BUILTIN_CMPB constant.
26636 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
26637 (BU_P6_64BIT_2): New macro.
26638 (BU_P6_OVERLOAD_2): New macro
26639 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
26640 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
26641 (CMPB): Add overload support to represent both 32-bit and 64-bit
26642 compare-bytes function.
26643 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
26644 support for TARGET_CMPB.
26645 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
26646 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26647 documentation of the __builtin_cmpb overloaded built-in function.
26649 2017-05-11 Richard Biener <rguenther@suse.de>
26651 PR tree-optimization/80705
26652 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
26653 bases are not vectorizable.
26655 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26657 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
26658 when counting register pressure.
26660 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26662 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
26663 (iv_ca_more_deps): Renamed to ...
26664 (iv_ca_compare_deps): ... this.
26665 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
26667 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26669 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
26671 (determine_group_iv_costs): ... here.
26672 (find_inv_vars_cb): Record inv var if it's not recorded before.
26674 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26676 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
26677 (get_shiftadd_cost): Ditto.
26679 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26681 * tree-ssa-address.c: Include header file.
26682 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
26684 (add_to_parts): Refactor.
26685 (addr_to_parts): New parameter. Update use of move_hint_to_base.
26686 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
26689 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26691 PR tree-optimization/53090
26692 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
26694 (extract_cond_operands): Detect condition with IV on both sides
26695 and return COMP_IV_EXPR_2.
26696 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
26697 (rewrite_use_compare): Simplify by removing call to function
26698 extract_cond_operands.
26700 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26702 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
26703 (extract_cond_operands): Detect condition comparing against non-
26704 invariant bound and return appropriate enum value.
26705 (find_interesting_uses_cond): Update use of extract_cond_operands.
26706 Handle its return value accordingly.
26707 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
26709 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26711 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
26712 nonlinear iv_use computation in loop invariant sensitive way.
26714 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26716 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
26717 (find_iv_candidates): Call relate_compare_use_with_all_cands.
26719 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26721 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
26722 (dump_cand): Support iv_cand.inv_exprs.
26723 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
26725 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
26728 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26730 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
26732 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
26733 as local function. Include necessary header files.
26734 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
26736 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26738 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
26740 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26742 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
26743 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
26744 RSHIFT_EXPR and BIT_NOT_EXPR.
26746 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26748 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
26749 (adjust_setup_cost): New parameter supporting round up adjustment.
26750 (struct address_cost_data): Delete.
26751 (force_expr_to_var_cost): Don't bound cost with spill_cost.
26752 (split_address_cost, ptr_difference_cost): Delete.
26753 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
26754 (struct ainc_cost_data): New struct.
26755 (get_address_cost_ainc): New function.
26756 (get_address_cost, get_computation_cost): Reimplement.
26757 (determine_group_iv_cost_address): Record inv_expr for all uses of
26759 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
26760 (iv_ca_has_deps): Reimplemented to ...
26761 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
26763 (iv_ca_extend): Call iv_ca_more_deps.
26765 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26767 * tree-ssa-address.c (struct mem_address): Move to header file.
26768 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
26769 * tree-ssa-address.h (struct mem_address): Move from C file.
26770 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
26772 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26774 * tree-affine.h (aff_combination_type): New interface.
26775 (aff_combination_zero_p): Remove static.
26776 (aff_combination_const_p): New interface.
26777 (aff_combination_singleton_var_p): New interfaces.
26779 2017-05-11 Richard Biener <rguenther@suse.de>
26781 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
26782 Skip unreachable blocks and destinations.
26783 (eliminate): Move stmt removal and fixup ...
26784 (fini_eliminate): ... here. Skip inserted exprs.
26785 (pass_pre::execute): Move fini_pre after fini_eliminate.
26786 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
26787 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
26788 PRE to get rid of dead code that has invalid SSA form and
26789 split critical edges again.
26791 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26793 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26795 2017-05-11 Richard Biener <rguenther@suse.de>
26797 * passes.c (execute_function_todo): Verify loops if they are
26798 said to be up-to-date.
26799 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
26800 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
26802 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
26805 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
26806 handle calling assemble_external ourself.
26809 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
26810 modes with zero size. Enhance comment.
26812 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26814 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
26815 built-ins for vec_xl and vec_xst with short and char pointer
26818 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
26820 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
26821 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
26822 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
26823 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
26824 (_mm_maskz_min_round_ss): New intrinsics.
26825 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
26826 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
26827 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
26828 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
26829 (__builtin_ia32_minss_mask_round): New builtins.
26830 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26831 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26832 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
26834 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
26835 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
26837 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
26840 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
26842 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
26843 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
26844 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
26845 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
26846 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
26847 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
26848 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
26849 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26850 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26851 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
26852 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
26853 (__builtin_ia32_mulss_mask_round): New builtins.
26854 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26855 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26856 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
26858 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
26859 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26861 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26864 2017-05-10 Julia Koval <julia.koval@intel.com>
26866 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
26867 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
26868 (_mm256_setr_m128i): New intrinsics.
26870 2017-05-10 Julia Koval <julia.koval@intel.com>
26872 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
26873 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
26874 (_mm_maskz_rcp14_ss): New intrinsics.
26875 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
26876 (__builtin_ia32_rcp14ss_mask): New builtins.
26877 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
26879 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
26881 PR tree-optimization/51513
26882 * tree-cfg.c (gimple_seq_unreachable_p): New function.
26883 (assert_unreachable_fallthru_edge_p): Use it.
26884 (group_case_labels_stmt): Likewise.
26885 * tree-cfg.h: Prototype it.
26886 * stmt.c: Include cfghooks.h and tree-cfg.h.
26887 (emit_case_dispatch_table) <gap_label>: New local variable.
26888 Use it to fill dispatch table gaps.
26889 Test for default_label before updating probabilities.
26890 (expand_case) <default_label>: Remove unneeded initialization.
26891 Test for unreachable default case statement and remove its edge.
26892 Set default_label accordingly.
26893 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
26895 2017-05-10 Carl Love <cel@us.ibm.com>
26897 * config/rs6000/rs6000-c: Add support for built-in functions
26898 vector signed char vec_neg (vector signed char)
26899 vector signed short int vec_neg (vector short int)
26900 vector signed int vec_neg (vector signed int)
26901 vector signed long long vec_neg (vector signed long long)
26902 vector float vec_neg (vector float)
26903 vector double vec_neg (vector double)
26904 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
26906 * config/rs6000/altivec.h: Add define for vec_neg
26907 * doc/extend.texi: Update the built-in documentation for the
26908 new built-in functions.
26910 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26912 PR tree-optimization/77644
26913 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
26915 2017-05-10 Nathan Sidwell <nathan@acm.org>
26917 * dumpfile.h (TDI_lang_all): New.
26918 (TDF_KIND): New. Renumber others
26919 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
26921 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
26923 (get_dump_file_name): Adjust suffix generation.
26924 (dump_enable_all): Use TDF_KIND.
26925 * doc/invoke.texi (-fdump-lang-all): Document.
26927 * dumpfile.h: Tabify.
26929 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
26932 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
26933 Move member access before delete.
26935 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
26937 * tree-inline.c (expand_call_inline): Split block at stmt
26940 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
26943 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
26944 are now unused after splitting mov{sf,sd}_hardfloat.
26945 (f32_lr2): Likewise.
26946 (f32_lm): Likewise.
26947 (f32_lm2): Likewise.
26948 (f32_li): Likewise.
26949 (f32_li2): Likewise.
26950 (f32_lv): Likewise.
26951 (f32_sr): Likewise.
26952 (f32_sr2): Likewise.
26953 (f32_sm): Likewise.
26954 (f32_sm2): Likewise.
26955 (f32_si): Likewise.
26956 (f32_si2): Likewise.
26957 (f32_sv): Likewise.
26958 (f32_dm): Likewise.
26959 (f32_vsx): Likewise.
26960 (f32_av): Likewise.
26961 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
26962 For movsf, order stores so the VSX stores occur before the GPR
26963 store which encourages the register allocator to use a traditional
26964 FPR instead of a GPR. For movsd, order the stores so that the GPR
26965 store comes before the VSX stores to allow the power6 to work.
26966 This is due to the power6 not having a 32-bit integer store
26967 instruction from a FPR.
26968 (movsf_hardfloat): Likewise.
26969 (movsd_hardfloat): Likewise.
26971 2017-05-09 Martin Sebor <msebor@redhat.com>
26973 PR translation/80280
26974 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
26977 PR translation/80280
26978 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
26979 data member added in r247778.
26980 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
26982 2017-05-09 Nathan Sidwell <nathan@acm.org>
26984 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
26986 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
26989 2017-05-09 Marek Polacek <polacek@redhat.com>
26991 * doc/invoke.texi: Fix typo.
26993 2017-05-09 Richard Biener <rguenther@suse.de>
26995 * tree-vrp.c (vrp_val_is_max): Adjust comment.
26996 (vrp_val_is_min): Likewise.
26997 (set_value_range_to_value): Likewise.
26998 (set_value_range_to_nonnegative): Likewise.
26999 (gimple_assign_nonzero_p): Likewise.
27000 (gimple_stmt_nonzero_p): Likewise.
27001 (vrp_int_const_binop): Likewise. Remove unreachable case.
27002 (adjust_range_with_scev): Adjust comments.
27003 (compare_range_with_value): Likewise.
27004 (extract_range_from_phi_node): Likewise.
27005 (test_for_singularity): Likewise.
27007 2017-05-09 Richard Biener <rguenther@suse.de>
27009 * tree-vrp.c (get_single_symbol): Add assert that we don't
27010 get overflowed constants as invariant part.
27011 (compare_values_warnv): Add comment before the TREE_NO_WARNING
27012 checks. Use wi::cmp instead of recursing for integer constants.
27013 (compare_values): Just ignore whether we assumed undefined
27014 overflow instead of failing the compare.
27015 (extract_range_for_var_from_comparison_expr): Add comment before the
27016 TREE_NO_WARNING sets.
27017 (test_for_singularity): Likewise.
27018 (extract_range_from_comparison): Do not disable optimization
27019 when we assumed undefined overflow.
27020 (extract_range_basic): Remove init of unused var.
27022 2017-05-09 Richard Biener <rguenther@suse.de>
27024 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
27025 (extract_range_from_multiplicative_op_1): Adjust.
27026 (extract_range_from_binary_expr_1): Use int_const_binop.
27028 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
27031 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
27032 rs6000_store_data_bypass_p in seven define_bypass directives and
27033 in several comments.
27034 * config/rs6000/rs6000-protos.h: Add prototype for
27035 rs6000_store_data_bypass_p function.
27036 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
27037 function implements slightly different (rs6000-specific) semantics
27038 than store_data_bypass_p, returning false rather than aborting
27039 with assertion error when arguments do not satisfy the
27040 requirements of store data bypass.
27041 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
27042 rs6000_store_data_bypass_p.
27044 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
27046 * config/xtensa/xtensa-protos.h
27047 (xtensa_initial_elimination_offset): New declaration.
27048 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
27049 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
27050 macro definition, add case for FRAME_POINTER_REGNUM when
27051 FRAME_GROWS_DOWNWARD.
27052 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
27053 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
27054 xtensa_initial_elimination_offset.
27056 2017-05-08 Nathan Sidwell <nathan@acm.org>
27058 * doc/invoke.texi: Alphabetize -fdump options.
27060 2017-05-08 Martin Sebor <msebor@redhat.com>
27062 PR translation/80280
27063 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
27065 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
27067 * target.def (compute_frame_layout): New optional target hook.
27068 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
27069 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
27070 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
27072 * reload1.c (verify_initial_elim_offsets): Likewise.
27073 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
27074 (use_simple_return_p): Call arm_compute_frame_layout if needed.
27075 (arm_get_frame_offsets): Split up into this ...
27076 (arm_compute_frame_layout): ... and this function.
27078 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
27080 * config/aarch64/constraints.md (Usa): New constraint.
27081 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
27083 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
27085 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
27086 with_multilib_list after it has been checked.
27088 2017-05-08 Richard Biener <rguenther@suse.de>
27090 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
27091 (bitmap_set_subtract_values): Likewise.
27093 2017-05-08 Richard Biener <rguenther@suse.de>
27095 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
27096 (gimple_assign_nonzero): ... this and remove strict_overflow_p
27098 (gimple_stmt_nonzero_warnv_p): Rename to ...
27099 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
27101 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
27102 (extract_range_basic): Adjust, do not disable propagation on
27103 strict overflow sensitive simplification.
27104 (vrp_visit_cond_stmt): Likewise.
27106 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
27108 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
27109 body size unconditionally.
27111 2017-05-07 Jeff Law <law@redhat.com>
27114 2017-05-06 Jeff Law <law@redhat.com>
27115 PR tree-optimization/78496
27116 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
27119 PR tree-optimization/78496
27120 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
27121 (simplify_stmt_using_ranges): Call it.
27122 (vrp_dom_walker::before_dom_children): Extract equivalences
27123 from an ASSERT_EXPR with an equality comparison against a
27126 2017-05-06 Jeff Law <law@redhat.com>
27128 PR tree-optimization/78496
27129 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
27132 PR tree-optimization/78496
27133 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
27134 (simplify_stmt_using_ranges): Call it.
27135 (vrp_dom_walker::before_dom_children): Extract equivalences
27136 from an ASSERT_EXPR with an equality comparison against a
27139 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
27141 * lra-constraints.c (lra_copy_reg_equiv): New function.
27142 (split_reg): Use it to copy equivalence information from the
27143 original register to the spill register.
27145 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
27147 PR rtl-optimization/75964
27148 * simplify-rtx.c (simplify_const_relational_operation): Remove
27149 invalid handling of comparisons of integer ABS.
27151 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
27153 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
27154 initialize to zero.
27155 (init_regs): Remove declaration.
27156 (function_arg_advance_32): Initialize error_p as boolean variable.
27158 2017-05-05 Nathan Sidwell <nathan@acm.org>
27160 * store-motion.c (remove_reachable_equiv_notes): Reformat long
27161 lines. Use for (;;).
27163 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27165 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
27166 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
27167 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
27168 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
27169 VF=2 that require versioning.
27171 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27173 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
27176 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27178 * diagnostic.h (diagnostic_override_option_index): Convert from
27179 macro to inline function.
27181 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27183 * diagnostic.c (last_module_changed_p): New function.
27184 (set_last_module): New function.
27185 (diagnostic_report_current_module): Convert macro usage to
27186 the above functions.
27187 * diagnostic.h (diagnostic_context::last_module): Strengthen
27188 from const line_map * to const line_map_ordinary *.
27189 (diagnostic_last_module_changed): Delete macro.
27190 (diagnostic_set_last_module): Delete macro.
27192 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27194 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
27195 with diagnostic_report_diagnostic.
27196 (diagnostic_n_impl_richloc): Likewise.
27197 * diagnostic.h (report_diagnostic): Delete macro.
27198 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
27199 with diagnostic_report_diagnostic.
27200 * substring-locations.c (format_warning_va): Likewise.
27202 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27204 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
27205 save/restor of format_spec. Move option-printing code to...
27206 (print_option_information): ...this new function, and
27207 reimplement by simply printing to the pretty_printer,
27208 rather than appending to the format string.
27210 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27212 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
27213 handling logic into...
27214 (update_effective_level_from_pragmas): ...this new function.
27216 2017-05-04 Andrew Waterman <andrew@sifive.com>
27218 * config/riscv/riscv.opt (mstrict-align): New option.
27219 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
27220 (SLOW_UNALIGNED_ACCESS): Define.
27221 (riscv_slow_unaligned_access): Declare.
27222 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
27224 (riscv_slow_unaligned_access): New variable.
27225 (rocket_tune_info): Set slow_unaligned_access to true.
27226 (optimize_size_tune_info): Set slow_unaligned_access to false.
27227 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
27228 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
27229 (riscv_option_override): Set riscv_slow_unaligned_access.
27230 * doc/invoke.texi: Add -mstrict-align to RISC-V.
27232 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
27234 * config/riscv/riscv.md: Unify indentation.
27236 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27241 * config/rs6000/rs6000.md (u code attribute): Add FIX and
27243 (extendsi<mode>2): Add support for doing sign extension via
27244 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
27245 don't have ISA 3.0 instructions.
27246 (extendsi<mode>2 splitter): Likewise.
27247 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
27248 generate the normal insns since SImode can now go in vector
27249 registers. Disallow the special UNSPECs needed for previous
27250 machines to hide SImode being used. Add new insns
27251 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
27252 (fix_trunc<mode>si2_stfiwx): Likewise.
27253 (fix_trunc<mode>si2_internal): Likewise.
27254 (fixuns_trunc<mode>si2): Likewise.
27255 (fixuns_trunc<mode>si2_stfiwx): Likewise.
27256 (fctiw<u>z_<mode>_smallint): Likewise.
27257 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
27258 of floating point to 32-bit integer from doing a direct move to
27259 the GPR registers to do a store.
27260 (fctiwz_<mode>): Break long line.
27262 2017-05-05 Bin Cheng <bin.cheng@arm.com>
27264 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
27265 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
27266 (addr_list, addr_offset_valid_p): New.
27267 (split_address_groups): Check offset validity with above function.
27268 (gt-tree-ssa-loop-ivopts.h): Include header file.
27270 2017-05-05 Nathan Sidwell <nathan@acm.org>
27272 * config.gcc (arm*-*-*): Add missing 'fi'.
27274 2017-05-05 Steve Ellcey <sellcey@cavium.com>
27276 * doc/invoke.texi (-fopt-info): Explicitly say order of options
27277 included in -fopt-info does not matter.
27278 * doc/optinfo.texi (-fopt-info): Fix description of default
27279 behavour. Explicitly say order of options included in -fopt-info
27282 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
27284 * config.gcc: Allow combinations of aprofile and rmprofile values for
27285 --with-multilib-list.
27286 * config/arm/t-multilib: New file.
27287 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
27288 variables. Remove setting of ISA and floating-point ABI in
27289 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
27290 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
27291 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
27292 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
27294 * config/arm/t-rmprofile: Likewise except for the matches changes.
27295 * doc/install.texi (--with-multilib-list): Document the combination of
27296 aprofile and rmprofile values and warn about pitfalls in doing that.
27298 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
27300 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
27301 (movdi_aarch64): Likewise.
27303 2017-05-05 Jakub Jelinek <jakub@redhat.com>
27305 PR tree-optimization/80632
27306 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
27308 (build_arrays): Initialize it for virtual phis.
27309 (fix_phi_nodes): Use it for virtual phis.
27311 PR tree-optimization/80558
27312 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
27313 [x, y] op z into [x op, y op z] for op & or | if conditions
27316 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
27317 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27320 * config/arm/arm.md (use_literal_pool): Remove.
27321 (64-bit immediate split): No longer takes cost into consideration
27322 if arm_disable_literal_pool is enabled.
27323 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
27324 used when arm_disable_literal_pool is enabled.
27325 (arm_max_const_double_inline_cost): Remove use of
27326 arm_disable_literal_pool.
27327 (push_minipool_fix): Add assert.
27328 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
27329 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
27330 (no_literal_pool_sf_immediate): New.
27332 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27334 PR tree-optimization/80613
27335 * tree-ssa-dce.c (propagate_necessity): Remove cases for
27336 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
27338 2017-05-05 Richard Biener <rguenther@suse.de>
27340 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
27342 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
27344 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
27345 of this flag from insn conditions due to removal from r247495.
27347 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
27349 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
27351 (arm_early_store_addr_dep_ptr): Likewise.
27352 * config/arm/aarch-common-protos.h
27353 (arm_early_load_addr_dep_ptr): Add prototype.
27354 (arm_early_store_addr_dep_ptr): Likewise.
27355 * config/arm/cortex-a53.md: Add new bypasses.
27357 2017-05-05 Jakub Jelinek <jakub@redhat.com>
27359 * tree.c (next_type_uid): Change type to unsigned.
27360 (type_hash_canon): Decrement back next_type_uid if
27361 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
27362 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
27365 2017-05-04 Martin Sebor <msebor@redhat.com>
27367 * builtins.c: Fix a trivial typo in a comment.
27369 PR middle-end/79234
27370 * builtins.c (check_sizes): Adjust to handle reading past the end.
27371 Avoid printing excessive upper bound of ranges. Use %E to print
27372 tree nodes instead of converting them to %wu.
27373 (expand_builtin_memchr): New function.
27374 (compute_dest_size): Rename...
27375 (compute_objsize): ...to this.
27376 (expand_builtin_memcpy): Adjust.
27377 (expand_builtin_mempcpy): Adjust.
27378 (expand_builtin_strcat): Adjust.
27379 (expand_builtin_strcpy): Adjust.
27380 (check_strncat_sizes): Adjust.
27381 (expand_builtin_strncat): Adjust.
27382 (expand_builtin_strncpy): Adjust and simplify.
27383 (expand_builtin_memset): Adjust.
27384 (expand_builtin_bzero): Adjust.
27385 (expand_builtin_memcmp): Adjust.
27386 (expand_builtin): Handle memcmp.
27387 (maybe_emit_chk_warning): Check strncat just once.
27389 2017-05-04 Martin Sebor <msebor@redhat.com>
27391 PR preprocessor/79214
27392 PR middle-end/79222
27393 PR middle-end/79223
27394 * builtins.c (check_sizes): Add inlining context and issue
27395 warnings even when -Wno-system-headers is set.
27396 (check_strncat_sizes): Same.
27397 (expand_builtin_strncat): Same.
27398 (expand_builtin_memmove): New function.
27399 (expand_builtin_stpncpy): Same.
27400 (expand_builtin): Handle memmove and stpncpy.
27402 2017-05-04 Bin Cheng <bin.cheng@arm.com>
27404 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
27405 which is not used any more.
27407 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27409 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
27411 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27413 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
27414 (cortexa53_tunings): Likewise.
27415 (cortexa57_tunings): Likewise.
27416 (cortexa72_tunings): Likewise.
27417 (cortexa73_tunings): Likewise.
27419 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27421 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
27422 Set loop alignment to 8.
27424 2017-05-04 Martin Sebor <msebor@redhat.com>
27426 PR translation/80280
27427 * builtins.c (expand_builtin_object_size): Add missing quoting to
27428 %D and like directives.
27429 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
27430 (hsa_type_for_tree_type): Same.
27431 (verify_function_arguments): Same.
27432 * symtab.c (symbol_table::change_decl_assembler_name): Same.
27433 * varasm.c (get_section): Same.
27436 2017-05-04 Martin Sebor <msebor@redhat.com>
27438 PR translation/80280
27439 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
27441 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27443 * config/aarch64/aarch64.c (generic_addrcost_table):
27444 Change HI/TI mode setting.
27446 2017-05-04 Martin Jambor <mjambor@suse.cz>
27448 PR tree-optimization/80622
27449 * tree-sra.c (comes_initialized_p): New function.
27450 (build_accesses_from_assign): Only set write lazily when
27451 comes_initialized_p is false.
27452 (analyze_access_subtree): Use comes_initialized_p.
27453 (propagate_subaccesses_across_link): Assert !comes_initialized_p
27454 instead of testing for PARM_DECL.
27456 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27458 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
27459 constraint on operand 0 to allow more general addressing modes.
27460 Adjust output template.
27461 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
27463 * config/aarch64/aarch64-protos.h
27464 (aarch64_address_valid_for_prefetch_p): Declare prototype.
27465 * config/aarch64/constraints.md (Dp): New address constraint.
27466 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
27469 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
27471 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
27472 update use of estimate_ipcp_clone_size_and_time.
27473 (estimate_local_effects): Update use of
27474 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
27475 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
27476 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
27477 Return nonspecialized time.
27479 2017-05-04 Richard Biener <rguenther@suse.de>
27481 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
27482 for the last VUSE which def dominates the PHI. Directly call
27484 (get_continuation_for_phi_1): Remove.
27486 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
27488 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
27489 to explain the use of truncating division. Cap the number of
27490 iterations to the maximum given by nb_iterations_upper_bound,
27493 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
27495 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
27496 * configure: Regenerate.
27497 * config.in: Regenerate.
27498 * config/i386/driver-mingw32.c: new file.
27499 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
27500 * config.host: Link driver-mingw32.o on MinGW host.
27501 * doc/install.texi: Document new --enable-mingw-wildcard configure
27504 2017-05-04 Marek Polacek <polacek@redhat.com>
27506 PR tree-optimization/80612
27507 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
27509 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27510 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
27512 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
27513 (movt splitter): Likewise.
27514 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
27515 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
27516 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
27517 block for Thumb-1 with MOVT.
27518 (thumb2_legitimate_address_p): Move code block ...
27519 (can_avoid_literal_pool_for_label_p): ... into this new function.
27520 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
27522 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
27523 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
27524 "M-profile targets with the MOVT instruction".
27526 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27528 * config/arm/arm-builtins.c (arm_init_builtins): Rename
27529 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
27530 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
27532 2017-05-04 Martin Liska <mliska@suse.cz>
27534 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
27535 variable cond_code.
27537 2017-05-04 Richard Biener <rguenther@suse.de>
27539 * tree.c (array_at_struct_end_p): Handle arrays at struct
27540 end with flexarrays more conservatively. Refactor and treat
27541 arrays of arrays or aggregates more strict. Fix
27542 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
27543 * tree.c (array_at_struct_end_p): Adjust prototype.
27544 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
27545 * gimple-fold.c (get_range_strlen): Likewise.
27546 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
27548 2017-05-04 Richard Biener <rguenther@suse.de>
27550 PR tree-optimization/31130
27551 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
27553 (supports_overflow_infinity): Likewise.
27554 (is_negative_overflow_infinity): Likewise.
27555 (is_positive_overflow_infinity): Likewise.
27556 (is_overflow_infinity): Likewise.
27557 (stmt_overflow_infinity): Likewise.
27558 (overflow_infinity_range_p): Likewise.
27559 (usable_range_p): Remove as always returning true.
27560 (make_overflow_infinity): Remove.
27561 (negative_overflow_infinity): Likewise.
27562 (positive_overflow_infinity): Likewise.
27563 (avoid_overflow_infinity): Likewise.
27564 (set_value_range): Adjust accordingly.
27565 (set_value_range_to_nonnegative): Likewise, remove now unused
27566 overflow_infinity arg.
27567 (vrp_operand_equal_p): Adjust.
27568 (update_value_range): Likewise.
27569 (range_int_cst_singleton_p): Likewise.
27570 (operand_less_p): Likewise.
27571 (compare_values_warnv): Likewise.
27572 (extract_range_for_var_from_comparison_expr): Likewise.
27573 (vrp_int_const_binop): Likewise.
27574 (zero_nonzero_bits_from_vr): Likewise.
27575 (extract_range_from_multiplicative_op_1): Likewise.
27576 (extract_range_from_binary_expr_1): Likewise.
27577 (extract_range_from_unary_expr): Likewise.
27578 (extract_range_from_comparison): Likewise.
27579 (extract_range_basic): Likewise.
27580 (adjust_range_with_scev): Likewise.
27581 (compare_ranges): Likewise.
27582 (compare_range_with_value): Likewise.
27583 (dump_value_range): Likewise.
27584 (test_for_singularity): Likewise, remove strict_overflow_p parameter
27586 (simplify_cond_using_ranges): Adjust.
27588 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
27590 * brig-builtins.def: Added a builtin for class_f64.
27591 * builtin-types.def: Added a builtin type needed by class_f64.
27593 2017-05-03 Jason Merrill <jason@redhat.com>
27595 * timevar.def: Add TV_CONSTEXPR.
27597 2017-05-03 David Malcolm <dmalcolm@redhat.com>
27599 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
27601 2017-05-03 Martin Jambor <mjambor@suse.cz>
27603 * ipa-prop.c (ipa_update_after_lto_read): Removed.
27604 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
27605 * ipa-cp.c (ipcp_propagate_stage): Do not call
27606 ipa_update_after_lto_read.
27607 * ipa-inline.c (ipa_inline): Likewise.
27609 2017-05-03 Martin Jambor <mjambor@suse.cz>
27611 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
27612 tag. Added a default constructor and a destructor.
27613 (ipa_edge_args_sum_t): New class;
27614 (ipa_edge_args_sum): Declare.
27615 (ipa_edge_args_vector): Remove declaration.
27616 (IPA_EDGE_REF): Use ipa_edge_args_sum.
27617 (ipa_free_edge_args_substructures): Remove declaration.
27618 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
27619 (ipa_edge_args_info_available_for_edge_p): Likewise.
27620 * ipa-prop.c (ipa_edge_args_vector): Removed.
27621 (edge_removal_hook_holder): Likewise.
27622 (edge_duplication_hook_holder): Likewise.
27623 (ipa_edge_args_sum): New variable.
27624 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
27625 ipa_edge_args_vector.
27626 (ipa_free_edge_args_substructures): Likewise.
27627 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
27628 ipa_edge_args_vector.
27629 (ipa_edge_removal_hook): Turned into method
27630 ipa_edge_args_sum_t::remove.
27631 (ipa_edge_duplication_hook): Turned into method
27632 ipa_edge_args_sum_t::duplicate.
27633 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
27634 registering edge hooks.
27635 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
27636 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
27637 ipa_edge_args_sum instead of ipa_edge_args_vector.
27638 * ipa-profile.c (ipa_profile): Likewise.
27640 2017-05-03 Martin Jambor <mjambor@suse.cz>
27642 * symbol-summary.h (function_summary): New method exists.
27643 (function_summary::symtab_removal): Deallocate through release.
27644 (call_summary): New class.
27645 (gt_ggc_mx): New overload.
27646 (gt_pch_nx): Likewise.
27647 (gt_pch_nx): Likewise.
27649 2017-05-03 Jeff Law <law@redhat.com>
27651 PR tree-optimization/78496
27652 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
27653 from simplify_cond_using_ranges. Split off code to walk
27654 backwards through casts into ...
27655 (simplify_cond_using_ranges_2): New function.
27656 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
27657 (execute_vrp): After identifying jump threads, call
27658 simplify_cond_using_ranges_2.
27660 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
27663 * ipa-inline.h (inline_summary): Add ctor.
27664 (create_ggc): Do not use ggc_cleared_alloc.
27666 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
27667 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27669 * gcc.c (handle_braces): Support escaping in switch matching
27671 * doc/invoke.texi (Spec Files): Document it.
27672 Remove superfluous @code markup in items.
27674 2017-05-03 David Malcolm <dmalcolm@redhat.com>
27676 * diagnostic-show-locus.c (struct column_range): New struct.
27677 (get_affected_columns): New function.
27678 (get_printed_columns): New function.
27679 (struct correction): New struct.
27680 (correction::ensure_capacity): New function.
27681 (correction::ensure_terminated): New function.
27682 (struct line_corrections): New struct.
27683 (line_corrections::~line_corrections): New dtor.
27684 (line_corrections::add_hint): New function.
27685 (layout::print_trailing_fixits): Reimplement in terms of the new
27687 (selftest::test_overlapped_fixit_printing): New function.
27688 (selftest::diagnostic_show_locus_c_tests): Call it.
27690 2017-05-03 Nathan Sidwell <nathan@acm.org>
27692 Canonicalize canonical type hashing
27693 * tree.h (type_hash_canon_hash): Declare.
27694 * tree.c (type_hash_list, attribute_hash_list): Move into
27695 type_hash_canon_hash.
27696 (build_type_attribute_qual_variant): Break out hash code calc into
27697 type_hash_canon_hash.
27698 (type_hash_canon_hash): New. Generic type hash computation.
27699 (build_range_type_1, build_array_type_1, build_function_type,
27700 build_method_type_directly, build_offset_type, build_complex_type,
27701 make_vector_type): Call it.
27703 2017-05-03 Richard Biener <rguenther@suse.de>
27705 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
27706 When all DRs have unknown misaligned do not always peel
27707 when there is a store but apply the same costing model as if
27708 there were only loads.
27710 2017-05-03 Richard Biener <rguenther@suse.de>
27713 PR tree-optimization/80492
27714 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27715 compare_base_decls returning dont-know properly.
27717 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
27719 * config/arm/iterators.md (CCSI): New mode iterator.
27720 (arch): New mode attribute.
27721 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
27722 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
27723 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
27724 code iterator for success result mode.
27725 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
27726 the corresponding new insn generators.
27728 2017-05-03 Bin Cheng <bin.cheng@arm.com>
27731 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27732 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27734 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
27736 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
27737 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
27738 (DDR_A): Wrap DDR argument in brackets.
27739 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
27740 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
27741 (DDR_REVERSED_P): Likewise.
27743 2017-05-03 Jakub Jelinek <jakub@redhat.com>
27745 PR tree-optimization/79472
27746 * tree-switch-conversion.c (struct switch_conv_info): Add
27747 contiguous_range and default_case_nonstandard fields.
27748 (collect_switch_conv_info): Compute contiguous_range and
27749 default_case_nonstandard fields, don't clear final_bb if
27750 contiguous_range and only the default case doesn't have the required
27752 (check_all_empty_except_final): Set default_case_nonstandard instead
27753 of failing if contiguous_range and the default case doesn't have empty
27755 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
27756 and only the default case doesn't have the required constants. Skip
27758 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
27759 if default_case_nonstandard.
27760 (build_constructors): Build constant 1 just once. Assert that default
27761 values aren't inserted in between cases if contiguous_range. Skip
27763 (build_arrays): Skip virtual phis.
27764 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
27765 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
27766 Handle virtual phis.
27767 (gen_inbound_check): Handle default_case_nonstandard case.
27768 (process_switch): Adjust check_final_bb caller. Call
27769 gather_default_values with the first non-default case instead of
27770 default case if default_case_nonstandard.
27772 2017-05-02 Nathan Sidwell <nathan@acm.org>
27774 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
27775 check. Fix formatting.
27777 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
27779 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
27780 errors when comparing specialized and unspecialized times.
27782 2017-05-02 David Malcolm <dmalcolm@redhat.com>
27784 * diagnostic-show-locus.c
27785 (layout::should_print_annotation_line_p): Make private.
27786 (layout::print_annotation_line): Make private.
27787 (layout::annotation_line_showed_range_p): Make private.
27788 (layout::show_ruler): Make private.
27789 (layout::print_source_line): Make private. Pass in line and
27790 line_width, rather than calling location_get_source_line. Drop
27792 (layout::print_leading_fixits): New method.
27793 (layout::print_any_fixits): Rename to...
27794 (layout::print_trailing_fixits): ...this, and make private.
27795 Don't print newline fixits.
27796 (diagnostic_show_locus): Move logic for printing one row into...
27797 (layout::print_line): ...this new function. Move the
27798 location_get_source_line call and error-handling from
27799 print_source_line to here. Call print_leading_fixits, and rename
27800 print_any_fixits to print_trailing_fixits.
27801 (selftest::test_fixit_insert_containing_newline): Update now that
27802 newlines are partially supported.
27803 (selftest::test_fixit_insert_containing_newline_2): New test.
27804 (selftest::test_fixit_replace_containing_newline): Update comments.
27805 (selftest::diagnostic_show_locus_c_tests): Call the new test.
27806 * edit-context.c (class added_line): New class.
27807 (class edited_line): Describe newline handling in comment.
27808 (edited_line::actually_edited_p): New method.
27809 (edited_line::print_content): Delete redundant decl.
27810 (edited_line::m_predecessors): New field.
27811 (edited_file::print_content): Call edited_line::print_content.
27812 (edited_file::print_diff): Update to support newlines.
27813 (edited_file::print_diff_hunk): Likewise.
27814 (edited_file::print_run_of_changed_lines): New function.
27815 (edited_file::print_diff_line): Convert to...
27816 (print_diff_line): ...this.
27817 (edited_file::get_effective_line_count): New function.
27818 (edited_line::edited_line): Initialize new field m_predecessors.
27819 (edited_line::~edited_line): Clean up m_predecessors.
27820 (edited_line::apply_fixit): Handle newlines.
27821 (edited_line::get_effective_line_count): New function.
27822 (edited_line::print_content): New function.
27823 (edited_line::print_diff_lines): New function.
27824 (selftest::test_applying_fixits_insert_containing_newline): New
27826 (selftest::test_applying_fixits_replace_containing_newline): New
27828 (selftest::insert_line): New function.
27829 (selftest::test_applying_fixits_multiple_lines): Add example of
27831 (selftest::edit_context_c_tests): Call the new tests.
27833 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27835 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
27836 parameter cand. Update dump information.
27837 (get_computation_cost): Update uses.
27839 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27841 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
27842 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
27843 (get_computation_at, rewrite_use_address): Update use of
27844 get_computation_aff.
27846 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27848 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
27849 (get_computation): Delete.
27850 (get_computation_cost): Implement like get_computation_cost_at.
27851 Use get_computation_at.
27852 (get_computation_cost_at): Delete.
27853 (rewrite_use_nonlinear_expr): Use get_computation_at.
27854 (rewrite_use_compare, remove_unused_ivs): Ditto.
27856 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27858 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
27860 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27862 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
27863 (ivopts_global_cost_for_size): Rename parameter and update uses.
27864 (iv_ca_recount_cost): Update uses.
27865 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
27866 candidates seperately in n_invs and n_cands.
27867 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
27869 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27871 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
27872 (find_inv_vars_cb): New.
27873 (find_depends): Renamed to ...
27874 (find_inv_vars): ... this.
27875 (add_candidate_1, force_var_cost): Call find_inv_vars.
27876 (split_address_cost, determine_group_iv_cost_cond): Ditto.
27878 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27880 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
27881 inv_vars. Add inv_exprs.
27882 (struct iv_cand): Rename depends_on to inv_vars.
27883 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
27884 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
27885 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
27886 (dump_cand): Dump inv_vars.
27887 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
27888 (record_invariant, find_depends, add_candidate_1): Ditto.
27889 (set_group_iv_cost, force_var_cost): Ditto.
27890 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
27891 (get_computation_cost_at, get_computation_cost): Ditto.
27892 (determine_group_iv_cost_generic): Ditto.
27893 (determine_group_iv_cost_address): Ditto.
27894 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
27895 (determine_group_iv_costs): Ditto.
27896 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
27897 (iv_ca_set_remove_invariants): Renamed to ...
27898 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
27899 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
27900 (iv_ca_set_add_invariants): Renamed to ...
27901 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
27902 (iv_ca_set_cp): Use iv_ca_set_add_invs.
27903 (iv_ca_has_deps): Support inv_vars and inv_exprs.
27904 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
27905 (create_new_ivs): Remove useless dump.
27907 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27909 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
27911 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
27912 (iv_ca_set_no_cp, create_new_iv): Ditto.
27914 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27916 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27918 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
27920 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
27921 function tree_check2.
27923 2017-05-02 Martin Liska <mliska@suse.cz>
27925 * doc/gcov.texi: Add missing preposition.
27926 * gcov.c (function_info::function_info): Properly fill up
27927 all member variables.
27929 2017-05-02 Tamar Christina <tamar.christina@arm.com>
27931 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
27933 2017-05-02 Tamar Christina <tamar.christina@arm.com>
27935 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
27937 2017-05-02 Martin Liska <mliska@suse.cz>
27940 * lto-streamer-in.c (lto_read_tree_1): Remove
27941 LTO_STREAMER_DEBUG.
27942 * lto-streamer.c (struct tree_hash_entry): Likewise.
27943 (struct tree_entry_hasher): Likewise.
27944 (tree_entry_hasher::hash): Likewise.
27945 (tree_entry_hasher::equal): Likewise.
27946 (lto_streamer_init): Likewise.
27947 (lto_orig_address_map): Likewise.
27948 (lto_orig_address_get): Likewise.
27949 (lto_orig_address_remove): Likewise.
27950 * lto-streamer.h: Likewise.
27951 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
27952 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
27954 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
27956 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
27957 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
27958 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
27959 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
27960 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
27961 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
27962 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
27963 (mm_maskz_sub_ss): New intrinsics.
27964 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27965 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
27966 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
27967 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
27968 (__builtin_ia32_subss_mask_round): New builtins.
27969 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27970 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27971 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
27973 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
27974 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
27976 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
27979 2017-05-02 Martin Jambor <mjambor@suse.cz>
27981 PR tree-optimization/78687
27982 * tree-sra.c (access): New field parent.
27983 (process_subtree_disqualification): New function.
27984 (disqualify_candidate): Call it.
27985 (build_accesses_from_assign): Reset write flag if creating an
27987 (build_access_subtree): Fill in parent field and also prpagate
27988 down grp_write flag.
27989 (create_artificial_child_access): New parameter set_grp_write, set
27990 grp_write to its value.
27991 (propagate_subaccesses_across_link): Also propagate grp_write flag
27993 (propagate_all_subaccesses): Push the closest parent back to work
27994 queue if add_access_to_work_queue returned true.
27996 2017-05-02 Richard Biener <rguenther@suse.de>
27998 * common.opt (fstrict-overflow): Alias negative to fwrapv.
27999 * doc/invoke.texi (fstrict-overflow): Remove all traces of
28000 -fstrict-overflow documentation.
28001 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
28002 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
28003 flag_strict_overflow.
28004 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
28005 * lto-opts.c (lto_write_options): Do not stream it.
28006 * lto-wrapper.c (merge_and_complain): Do not handle it.
28007 * opts.c (default_options_table): Do not set -fstrict-overflow.
28008 (finish_options): Likewise do not clear it when sanitizing.
28009 * simplify-rtx.c (simplify_const_relational_operation): Do not
28010 test flag_strict_overflow.
28012 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
28014 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
28015 using enabled attribute.
28016 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
28017 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
28018 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
28019 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
28020 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
28021 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
28022 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
28023 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
28024 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
28025 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
28027 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
28029 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
28031 2017-05-02 Richard Biener <rguenther@suse.de>
28033 PR tree-optimization/80591
28035 2017-04-10 Richard Biener <rguenther@suse.de>
28037 * tree-ssa-structalias.c (find_func_aliases): Properly handle
28040 2017-05-02 Richard Biener <rguenther@suse.de>
28042 PR tree-optimization/80549
28043 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
28044 (cleanup_tree_cfg_noloop): Create forwarders to known loop
28045 headers if they do not have a preheader.
28047 2017-05-02 Martin Liska <mliska@suse.cz>
28050 * common.opt: Fix typo.
28051 * doc/invoke.texi: Likewise.
28053 2017-05-01 Jan Beulich <jbeulich@suse.com>
28055 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
28056 swapping, add (x,x,m,x,n) alternative.
28058 2017-05-01 Nathan Sidwell <nathan@acm.org>
28060 * calls.c (combine_pending_stack_adjustment_and_call): Remove
28061 unnecessary unadjusted_alignment check.
28063 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
28066 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
28068 * gimplify.c (gimplify_cilk_detach): New function.
28069 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
28070 * tree-core.h: Document EXPR_CILK_SPAWN.
28071 * tree.h (EXPR_CILK_SPAWN): Define.
28073 2017-05-01 David Malcolm <dmalcolm@redhat.com>
28075 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
28076 to use new fixit_hint representation, using the "replace" logic.
28077 (get_line_span_for_fixit_hint): Likewise.
28078 (layout::print_any_fixits): Likewise.
28079 (selftest::test_one_liner_many_fixits): Rename to...
28080 (selftest::test_one_liner_many_fixits_1): ...this, and update
28081 comment and expected output to reflect that the multiple fix-it
28082 hints are now consolidated into one insertion.
28083 (selftest::test_one_liner_many_fixits_2): New test.
28084 (selftest::test_diagnostic_show_locus_one_liner): Update for
28086 (selftest::test_fixit_consolidation): Update for fix-it API
28088 * diagnostic.c (print_parseable_fixits): Likewise.
28089 * edit-context.c (edited_line::m_line_events): Convert from
28090 auto_vec <line_event *> to auto_vec <line_event>.
28091 (class line_event): Convert from abstract base class to a concrete
28092 class, taking over the role of replace_event.
28093 (class insert_event): Delete.
28094 (class replace_event): Rename to class line_event. Convert to
28096 (edit_context::add_fixits): Reimplement.
28097 (edit_context::apply_insert): Delete.
28098 (edit_context::apply_replace): Rename to...
28099 (edit_context::apply_fixit): ...this. Convert to half-open range.
28100 (edited_file::apply_insert): Delete.
28101 (edited_file::apply_replace): Rename to...
28102 (edited_file::apply_fixit): ...this.
28103 (edited_line::~edited_line): Drop deletion of events.
28104 (edited_line::apply_insert): Delete.
28105 (edited_line::apply_replace): Rename to...
28106 (edited_line::apply_fixit): ...this. Convert to half-open range.
28107 Update for change to type of m_line_events.
28108 * edit-context.h (edit_context::apply_insert): Delete.
28109 (edit_context::apply_replace): Rename to...
28110 (edit_context::apply_fixit): ...this.
28112 2017-05-01 Martin Sebor <msebor@redhat.com>
28114 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
28117 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
28120 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
28121 __get_cpuid_max returns 0.
28122 (__get_cpuid_count): Ditto.
28124 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
28126 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
28127 replacement expression is another instance of one of its arguments.
28129 2017-05-01 Jakub Jelinek <jakub@redhat.com>
28132 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
28133 check for stack push/pop autoinc.
28134 * config/i386/i386.c (ix86_agi_dependent): Return false
28135 if the only reason why modified_in_p returned true is that
28136 addr is SP based and set_insn is a push or pop.
28138 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
28140 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
28143 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
28146 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
28147 (account_size_time): Use two predicates - exec_pred and
28149 (evaluate_conditions_for_known_args): Compute both clause and
28151 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
28152 (inline_summary_t::duplicate): Update.
28153 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
28155 (compute_inline_parameters): Likewise.
28156 (estimate_edge_size_and_time): Update caluclation of time.
28157 (estimate_node_size_and_time): Compute both time and nonspecialized
28159 (estimate_ipcp_clone_size_and_time): Update.
28160 (inline_merge_summary): Update.
28161 (do_estimate_edge_time): Update.
28162 (do_estimate_edge_size): Update.
28163 (do_estimate_edge_hints): Update.
28164 (inline_read_section, inline_write_summary): Stream both new predicates.
28165 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
28167 (compute_inlined_call_time): Cleanup.
28168 (big_speedup_p): Update.
28169 (edge_badness): Update.
28170 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
28171 (size_time_entry): Replace predicate by exec_predicate and
28172 nonconst_predicate.
28173 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
28174 (estimate_edge_time): Return also nonspec_time.
28175 (reset_edge_growth_cache): Update.
28177 2017-04-29 Jakub Jelinek <jakub@redhat.com>
28179 PR rtl-optimization/80491
28180 * ifcvt.c (noce_process_if_block): When looking for x setter
28181 with missing else_bb, don't check only the insn right before
28182 cond_earliest, but look for the last insn that x is modified in
28183 within the same bb.
28185 PR rtl-optimization/80491
28186 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
28188 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
28190 PR tree-optimization/80487
28191 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
28193 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28195 PR tree-optimization/79697
28196 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
28197 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
28198 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
28200 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
28201 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
28203 2017-04-28 Martin Sebor <msebor@redhat.com>
28205 PR tree-optimization/80523
28206 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
28207 (init_target_to_host_charmap, target_to_host, target_strtol10): New
28209 (maybe_warn, format_directive, parse_directive): Use new functions.
28210 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
28212 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
28214 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
28216 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
28218 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
28219 target_header_dir): Set correctly.
28220 * configure: Regenerated.
28221 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28222 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28223 instead of SYSTEM_HEADER_DIR.
28225 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
28227 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
28228 (estimate_local_effects): Likewise.
28229 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
28230 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
28231 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
28232 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
28233 do_estimate_edge_time, estimate_edge_time): Likewise.
28234 * ipa-inline-analysis.c (estimate_node_size_and_time,
28235 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
28236 (estimate_time_after_inlining): Remove.
28238 2017-04-28 Martin Liska <mliska@suse.cz>
28240 * doc/gcov.texi: Enhance documentation of gcov.
28242 2017-04-28 Martin Liska <mliska@suse.cz>
28244 * doc/gcov.texi: Sort options in alphabetic order.
28245 * doc/gcov-dump.texi: Likewise.
28246 * doc/gcov-tool.texi: Likewise.
28247 * gcov.c (print_usage): Likewise.
28248 * gcov-dump.c (print_usage): Likewise.
28249 * gcov-tool.c (print_merge_usage_message): Likewise.
28250 (print_rewrite_usage_message): Likewise.
28251 (print_overlap_usage_message): Likewise.
28253 2017-04-28 Martin Liska <mliska@suse.cz>
28255 PR gcov-profile/53915
28256 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
28258 2017-04-28 Martin Liska <mliska@suse.cz>
28260 PR gcov-profile/79891
28261 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
28262 is marked by compiler as living on a line.
28263 (get_cycles_count): Remove usage of the union.
28264 (output_intermediate_file): Likewise.
28265 (find_source): Fix GNU coding style.
28266 (accumulate_line_counts): Remove old non-all block mode.
28267 (output_lines): Remove usage of the union.
28268 * profile.c (output_location): Include all BBs, even if
28269 belonging to a same line (and file) as a previous BB.
28271 2017-04-28 Martin Liska <mliska@suse.cz>
28273 * gcov.c (process_args): Handle new argument 'w'.
28274 (read_graph_file): Assign ID to BBs.
28275 (output_branch_count): Display BB # if verbose flag is set.
28276 (output_lines): Likewise for arcs.
28277 (print_usage): Add '--verbose' option help.
28278 * doc/gcov.texi: Document --verbose (-w) option.
28280 2017-04-28 Martin Liska <mliska@suse.cz>
28282 * gcov.c (struct block_location_info): New struct.
28283 (process_file): Fill up the new structure.
28284 (read_graph_file): Replace usage of encoding by the newly added
28286 (add_line_counts): Likewise.
28287 (accumulate_line_counts): Remove usage of the union.
28288 (function_info::function_info): New function.
28289 (function_info::~function_info): Likewise.
28290 (process_file): Call delete instead of release_function.
28291 (release_function): Release the function.
28292 (release_structures): Call delete instead of release_function.
28293 (solve_flow_graph): Replace usage of num_blocks.
28294 (find_exception_blocks): Likewise.
28295 (output_lines): Fix GNU coding style.
28297 2017-04-28 Martin Liska <mliska@suse.cz>
28300 * coverage.c (coverage_remove_note_file): New function.
28301 * coverage.h: Declare the function.
28302 * toplev.c (finalize): Clean if an error has been seen.
28304 2017-04-28 Martin Liska <mliska@suse.cz>
28306 PR gcov-profile/80031
28307 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
28308 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
28309 * gcov.c (read_graph_file): Read just number of blocks.
28310 * profile.c (branch_prob): Do not stream 0 flags per a basic
28313 2017-04-28 Martin Liska <mliska@suse.cz>
28315 * gcov-dump.c (tag_*): Add new argument to declarations.
28316 (dump_gcov_file): Likewise.
28317 (tag_blocks): Add and use new argument depth.
28318 (tag_arcs): Likewise.
28319 (tag_lines): Likewise.
28320 (tag_counters): Likewise.
28321 (tag_summary): Likewise.
28322 (dump_working_sets): Use depth to do a proper indentation.
28324 2017-04-28 Jakub Jelinek <jakub@redhat.com>
28327 * cgraph.h (symtab_node::debug_symtab): No longer inline.
28328 * symtab.c (symtab_node::debug_symtab): Move definition here.
28330 2017-04-28 Richard Biener <rguenther@suse.de>
28332 * lto-streamer.h (LTO_major_version): Bump to 7.
28334 2017-04-28 Richard Biener <rguenther@suse.de>
28336 * tree-vrp.c (assert_info): New struct.
28337 (add_assert_info): New helper.
28338 (register_edge_assert_for_2): Refactor to add asserts to a vector
28340 (register_edge_assert_for_1): Likewise.
28341 (register_edge_assert_for): Likewise.
28342 (finish_register_edge_assert_for): New helper actually registering
28343 asserts where live on edge.
28344 (find_conditional_asserts): Adjust.
28345 (find_switch_asserts): Likewise.
28346 (evrp_dom_walker::try_find_new_range): Generalize.
28347 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
28349 2017-04-27 Marek Polacek <polacek@redhat.com>
28352 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
28353 arg10 and arg11 to itype.
28355 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
28357 * doc/extend.texi (Object Size Checking): Improve grammar.
28359 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
28362 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
28363 that the logic for permitting reciprocal estimates matches that
28366 2017-04-27 Jakub Jelinek <jakub@redhat.com>
28369 * tree.c (type_cache_hasher::equal): Only compare
28370 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
28371 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
28372 non-aggregate element types.
28373 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
28374 about the flag on ARRAY_TYPEs in the comment, formatting fix.
28376 2017-04-27 Richard Biener <rguenther@suse.de>
28378 PR middle-end/80533
28379 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
28380 stripping ARRAY_REFs from MEM_EXPR make sure we're not
28381 keeping a reference to a trailing array.
28383 2017-04-27 Richard Biener <rguenther@suse.de>
28385 PR middle-end/80539
28386 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
28387 being in loop-closed SSA form conservatively.
28388 (chrec_fold_multiply_poly_poly): Likewise.
28390 2017-04-27 Tamar Christina <tamar.christina@arm.com>
28392 PR middle-end/79665
28393 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
28394 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
28396 2017-04-27 Jakub Jelinek <jakub@redhat.com>
28399 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
28400 (aarch64_function_arg_alignment): Return unsigned int again, but still
28401 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
28402 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
28403 Don't emit -Wpsabi note.
28404 (aarch64_function_arg_boundary): Likewise.
28405 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
28408 2017-04-26 Nathan Sidwell <nathan@acm.org>
28410 * tree.h (crc32_unsigned_n): Declare.
28411 (crc32_unsigned, crc32_unsigned): Make inline.
28412 * tree.c (crc32_unsigned_bits): Replace with ...
28413 (crc32_unsigned_n): ... this.
28414 (crc32_unsigned, crc32_byte): Remove.
28415 (crc32_string): Remove unnecessary braces.
28417 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
28419 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
28420 * ipa-inline-analysis.c (MAX_TIME): Remove.
28421 (account_size_time): Use sreal for time.
28422 (dump_inline_summary): Update.
28423 (estimate_function_body_sizes): Update.
28424 (estimate_edge_size_and_time): Update.
28425 (estimate_calls_size_and_time): Update.
28426 (estimate_node_size_and_time): Update.
28427 (inline_merge_summary): Update.
28428 (inline_update_overall_summary): Update.
28429 (estimate_time_after_inlining): Update.
28430 (inline_read_section): Update.
28431 (inline_write_summary): Update.
28432 * ipa-inline.c (compute_uninlined_call_time): Update.
28433 (compute_inlined_call_time): Update.
28434 (recursive_inlining): Update.
28435 (inline_small_functions): Update.
28436 (dump_overall_stats): Update.
28437 * ipa-inline.h: Include sreal.h.
28438 (size_time_entry): Turn time to sreal.
28439 (inline_summary): Turn self_time nad time to sreal.
28441 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
28443 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
28445 (sreal::stream_out, sreal::stream_in): New.
28446 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
28448 2017-04-25 Jakub Jelinek <jakub@redhat.com>
28450 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
28453 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
28456 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
28458 (dimode_scalar_chain::compute_convert_gain): Ditto.
28459 (dimode_scalar_chain::make_vector_copies): Ditto.
28460 (dimode_scalar_chain::convert_reg): Ditto.
28461 (dimode_scalar_chain::convert_insn): Ditto.
28462 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
28463 (VI248_AVX512BW_1): New mode iterator.
28464 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
28465 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
28468 2017-04-25 Martin Sebor <msebor@redhat.com>
28470 PR tree-optimization/80497
28471 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
28472 constants are representable in HOST_WIDE_INT.
28473 (parse_directive): Ditto.
28475 2017-04-25 Martin Sebor <msebor@redhat.com>
28478 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
28479 (new_zero_array): Adjust signature.
28480 (dom_info::dom_init): Used unsigned rather that size_t.
28481 (dom_info::dom_info): Same.
28483 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28484 Jakub Jelinek <jakub@redhat.com>
28487 * config/arm/arm.c: Include gimple.h.
28488 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
28489 returns negative, increment ncrn only if it returned positive.
28490 (arm_needs_doubleword_align): Return int instead of bool,
28491 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
28492 members, but if there is any such non-FIELD_DECL
28493 > PARM_BOUNDARY aligned decl, return -1 instead of false.
28494 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
28495 returns negative, increment nregs only if it returned positive.
28496 (arm_setup_incoming_varargs): Likewise.
28497 (arm_function_arg_boundary): Emit -Wpsabi note if
28498 arm_needs_doubleword_align returns negative, return
28499 DOUBLEWORD_ALIGNMENT only if it returned positive.
28501 2017-04-25 Marek Polacek <polacek@redhat.com>
28504 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
28505 first argument to type.
28507 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
28510 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
28511 type checks to test for compatibility instead of equality.
28513 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28514 Jakub Jelinek <jakub@redhat.com>
28517 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
28519 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
28520 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
28521 the alignment computation, but return their maximum in warn_alignment.
28522 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
28523 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
28525 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
28526 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
28529 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28531 * config/arc/simdext.md (dmpyh): Fix typo.
28533 2017-04-25 Richard Biener <rguenther@suse.de>
28535 PR tree-optimization/80492
28536 * alias.c (compare_base_decls): Handle registers with asm
28537 specification conservatively.
28538 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
28539 compare_base_decls returning dont-know properly.
28541 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28543 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
28544 (legitimate_offset_address_p): New function.
28545 (arc_legitimate_address_p): Use above function.
28547 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28549 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
28551 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28553 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
28554 ACCH registers whenever they are available.
28556 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28558 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
28559 double regs fix when not used.
28561 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28563 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
28565 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
28566 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
28568 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28570 * config/arc/arc.c (arc_output_addsi): Check for h-register class
28571 when emitting short ADD instructions.
28573 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28575 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
28577 (cmpsi_cc_c_insn): Likewise.
28578 (cbranchsi4_scratch): Compute proper instruction length using
28579 compact_hreg_operand.
28580 * config/arc/predicates.md (compact_hreg_operand): New predicate.
28582 2017-04-25 Richard Biener <rguenther@suse.de>
28584 PR middle-end/80509
28585 * passes.c (pass_manager::pass_manager): Initialize
28586 m_name_to_pass_map.
28588 2017-04-25 Richard Biener <rguenther@suse.de>
28590 PR tree-optimization/79201
28591 * tree-ssa-sink.c (statement_sink_location): Handle calls.
28593 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28596 * config/s390/vector.md: Split MEM->GPR vector moves for
28597 non-s_operand addresses.
28599 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28602 * config/s390/predicates.md (reload_const_wide_int_operand): New
28604 * config/s390/s390.md ("movti"): Remove d/P alternative.
28605 ("movti_bigconst"): New pattern definition.
28607 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28610 * s390-protos.h (s390_expand_cs_hqi): Removed.
28611 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
28612 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
28613 modes as well as CCZ1mode and CCZmode.
28614 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
28615 signature of s390_emit_compare_and_swap.
28616 (s390_expand_cs_hqi): Likewise, make static.
28617 (s390_expand_cs_tdsi): Generate an explicit compare before trying
28618 compare-and-swap, in some cases.
28619 (s390_expand_cs): Wrapper function.
28620 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
28622 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
28623 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
28624 patterns for small and large integers. Forbid symref memory operands.
28625 Move expander to s390.c. Require cc register.
28626 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
28627 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
28628 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
28629 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
28630 symref memory operands. Remove CC mode and call s390_match_ccmode
28632 ("atomic_exchange<mode>"): Allow and implement all integer modes.
28634 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28636 * config/s390/s390.md (define_peephole2): New peephole to help
28637 combining the load-and-test pattern with volatile memory.
28639 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28641 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
28642 with CCZmode for TARGET_Z196.
28644 2017-04-25 Jakub Jelinek <jakub@redhat.com>
28646 PR rtl-optimization/80501
28647 * combine.c (make_compound_operation_int): Set subreg_code to SET
28648 even for AND with mask of the sign bit of mode.
28650 PR rtl-optimization/80500
28651 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
28652 sum's initial value.
28654 2017-04-25 Julian Brown <julian@codesourcery.com>
28655 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28657 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
28659 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
28661 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
28663 2017-04-25 Julian Brown <julian@codesourcery.com>
28664 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28666 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
28667 (thunderx2t99_sha): New Reservation.
28669 2017-04-25 Julian Brown <julian@codesourcery.com>
28670 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28672 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
28673 type for 1-element load.
28675 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
28677 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
28679 2017-04-24 Martin Jambor <mjambor@suse.cz>
28681 PR tree-optimization/80293
28682 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
28683 char arrays not totally scalarizable if it is false.
28684 (analyze_all_variable_accesses): Pass correct value in the new
28685 parameter. Add a statistics counter.
28687 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
28689 PR middle-end/79931
28690 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
28692 2017-04-24 Richard Biener <rguenther@suse.de>
28694 PR tree-optimization/80494
28695 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
28696 out for complex types.
28698 2017-04-24 Richard Biener <rguenther@suse.de>
28700 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
28701 * tree-ssa-sccvn.c (print_scc): Print SCC size.
28702 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
28703 (DFS): Adjust and never fail.
28704 (sccvn_dom_walker::fail): Remove.
28705 (sccvn_dom_walker::before_dom_children): Adjust.
28706 (run_scc_vn): Likewise and never fail.
28707 * tree-ssa-pre.c (pass_pre::execute): Adjust.
28708 (pass_fre::execute): Likewise.
28710 2017-04-24 Richard Biener <rguenther@suse.de>
28712 PR tree-optimization/79725
28713 * tree-ssa-sink.c (statement_sink_location): Return whether
28714 failure reason was zero uses. Move that check later.
28715 (sink_code_in_bb): Deal with zero uses by removing the stmt
28718 2017-04-24 Richard Biener <rguenther@suse.de>
28721 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
28722 pointer-based references.
28724 2017-04-24 Richard Biener <rguenther@suse.de>
28727 * pass_manager.h (pass_manager::operator new): Remove.
28728 (pass_manager::operator delete): Likewise.
28729 * passes.c (pass_manager::operator new): Remove.
28730 (pass_manager::operator delete): Likewise.
28731 (pass_manager::pass_manager): Zero individual pass members.
28733 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
28736 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
28737 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
28738 Check "XEXP (src, 1)" operand here.
28739 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
28740 Check "XEXP (src, 1)" operand here.
28741 (dimode_scalar_chain::make_vector_copies): Detect count register
28742 of a shift instruction. Zero extend count register from QImode
28743 to DImode to satisfy vector shift pattern count operand predicate.
28744 Substitute vector shift count operand with a DImode copy.
28745 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
28748 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
28750 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
28751 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
28752 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
28753 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
28754 (UNSPEC_NOREX_MEM): Remove definition.
28756 2017-04-21 Richard Biener <rguenther@suse.de>
28758 PR tree-optimization/79547
28759 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
28760 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
28761 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
28762 without any constraints.
28764 2017-04-21 Richard Biener <rguenther@suse.de>
28766 PR tree-optimization/78847
28767 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
28769 2017-04-21 Richard Biener <rguenther@suse.de>
28771 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
28772 (build_distinct_type_copy): Likewise.
28773 (build_variant_type_copy): Likewise.
28774 * tree.c (build_qualified_type): Pass down mem-stat info.
28775 (build_distinct_type_copy): Likewise.
28776 (build_variant_type_copy): Likewise.
28778 2017-04-21 Richard Biener <rguenther@suse.de>
28780 PR tree-optimization/80237
28781 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
28783 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
28784 for a simplified result.
28786 2016-04-21 Richard Biener <rguenther@suse.de>
28788 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
28789 sth as strict as a simple_iv but a chrec without symbols and an
28790 operand defined in the loop we are peeling (and not some subloop).
28791 (propagate_constants_for_unrolling): Propagate all constants.
28793 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
28796 * config/i386/i386.c (print_reg): Remove assert for disalowed
28797 regno values, call output_operand_lossage instead.
28799 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
28802 * config/i386/constraints.md (Yc): New register constraint.
28803 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
28804 Use Yc constraint for alternative 2 of operand 0. Remove
28805 preferred_for_speed attribute.
28807 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
28809 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
28810 lastprivate clauses in SIMT case.
28812 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
28814 * doc/invoke.texi (-Wextra-semi): Document new warning option.
28816 2017-04-20 Richard Biener <rguenther@suse.de>
28818 PR tree-optimization/57796
28819 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
28820 as N scalar stores.
28821 (vect_model_load_cost): Cost gathers as N scalar loads.
28823 2017-04-20 Richard Biener <rguenther@suse.de>
28825 * ggc-page.c (ggc_allocated_p): Rename to ...
28826 (safe_lookup_page_table_entry): ... this and return the lookup
28828 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
28830 2017-04-20 Richard Biener <rguenther@suse.de>
28832 PR tree-optimization/80453
28833 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
28834 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
28835 from the conditions.
28836 (vn_phi_eq): Pass them down.
28837 (vn_phi_lookup): Record them.
28838 (vn_phi_insert): Likewise.
28840 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
28842 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
28843 uninitialized variable warning to avoid buffer overrun.
28845 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
28848 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
28849 is suppressed for '{ 0 }' in C.
28851 2017-04-20 Jakub Jelinek <jakub@redhat.com>
28853 * BASE-VER: Set to 8.0.0.
28855 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
28857 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
28858 priority .init_array and .fini_array section with SECTION_NOTYPE
28861 2017-04-20 Jakub Jelinek <jakub@redhat.com>
28863 PR middle-end/80423
28864 * tree.h (build_array_type): Add typeless_storage default argument.
28865 * tree.c (type_cache_hasher::equal): Also compare
28866 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
28867 (build_array_type): Add typeless_storage argument, set
28868 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
28870 (build_nonshared_array_type): Adjust build_array_type_1 caller.
28871 (build_array_type): Likewise. Add typeless_storage argument.
28873 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
28874 Jakub Jelinek <jakub@redhat.com>
28876 PR tree-optimization/80426
28877 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
28878 operation on symbolic operands, also compute the overflow for the
28879 invariant part when the operation degenerates into a negation.
28881 2017-04-19 Jakub Jelinek <jakub@redhat.com>
28884 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
28885 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
28888 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
28890 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
28893 * config/avr/avr.c (tree.h): Include it.
28894 (cgraph.h): Include it.
28895 (avr_encode_section_info): Don't warn for uninitialized progmem
28896 variable if it's just an alias.
28898 2017-04-19 Richard Biener <rguenther@suse.de>
28901 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
28902 when needed by AutoPGO.
28904 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
28907 * doc/lto.texi: Remove an extra 'that'.
28909 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
28911 PR rtl-optimization/80429
28912 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
28913 are only used in debug insns.
28915 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
28916 Vladimir Makarov <vmakarov@redhat.com>
28918 * config/sparc/predicates.md (input_operand): Add comment. Return
28919 true for any memory operand when LRA is in progress.
28920 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
28922 2017-04-18 Jeff Law <law@redhat.com>
28925 * mips.md ({return,simple_return}_internal): Do not overwrite
28928 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28930 PR tree-optimization/80443
28931 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
28932 instead of adding 1, subtract -1 and similarly instead of subtracting
28935 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
28937 PR rtl-optimization/80357
28938 * haifa-sched.c (tmp_bitmap): New variable.
28939 (model_recompute): Handle duplicate use records.
28940 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
28941 (free_global_sched_pressure_data): Free it.
28943 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
28946 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
28947 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28948 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28949 instead of SYSTEM_HEADER_DIR.
28951 2017-04-18 Jeff Law <law@redhat.com>
28953 PR middle-end/80422
28954 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
28955 predecessors after walking up the insn chain.
28957 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28960 * dwarf2out.c (modified_type_die): Try harder not to emit internal
28961 sizetype type into debug info.
28963 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
28966 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
28967 unneeded test for TARGET_UPPER_REGS_SF.
28968 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
28970 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28973 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
28974 instead of gsi_after_labels.
28976 2017-04-18 Jeff Law <law@redhat.com>
28978 * regcprop.c (maybe_mode_change): Avoid creating copies of the
28982 2017-04-13 Jeff Law <law@redhat.com>
28983 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
28984 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
28986 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
28989 * config/avr/avr.c (intl.h): Include it.
28990 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
28992 2017-04-18 Martin Liska <mliska@suse.cz>
28994 PR gcov-profile/78783
28995 * gcov-tool.c (gcov_output_files): Validate that destination
28996 file is either removed by the tool or by a user.
28998 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
28999 Guy Benyei <guybe@mellanox.com>
29001 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
29002 block, and do not negate it, the stored id is already negative.
29004 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
29006 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
29008 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29011 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
29012 masks of options that should be turned off if the VSX vector
29013 options are turned off.
29014 (OTHER_P8_VECTOR_MASKS): Likewise.
29015 (OTHER_VSX_VECTOR_MASKS): Likewise.
29016 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
29017 rs6000_disable_incompatible_switches to validate no type switches
29019 (rs6000_incompatible_switch): New function to disallow turning on
29020 other vector options if -mno-vsx, -mno-power8-vector, or
29021 -mno-power9-vector are specified.
29023 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29025 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
29027 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29029 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
29030 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
29031 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
29032 (ARG_POINTER_CFA_OFFSET): Likewise.
29034 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29036 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
29037 conditions to take advantage of various optimizations.
29039 2017-04-13 Jeff Law <law@redhat.com>
29041 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
29042 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
29043 (zero_extendsidi2_dext): Likewise.
29045 2017-04-13 Jakub Jelinek <jakub@redhat.com>
29048 * fold-const.c (fold_ternary_loc): Revert
29049 use op0 instead of fold_convert_loc (loc, type, arg0) part of
29052 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
29054 PR rtl-optimization/80343
29055 * lra-remat.c (update_scratch_ops): Assign original hard reg to
29056 new scratch pseudo.
29058 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
29061 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
29062 to ubsan_encode_value.
29064 2017-04-13 Jeff Law <law@redhat.com>
29066 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
29067 appearing in DEBUG_INSNs.
29069 2017-04-13 Martin Liska <mliska@suse.cz>
29071 PR gcov-profile/80413
29072 * gcov-io.c (gcov_write_string): Copy to buffer just when
29073 allocated size is greater than zero.
29075 2017-04-13 Jakub Jelinek <jakub@redhat.com>
29078 * dwarf2out.c (decls_for_scope): Ignore declarations of
29079 current_function_decl in BLOCK_NONLOCALIZED_VARS.
29081 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
29084 * ipa-visibility.c (non_local_p): Fix typos.
29085 (localize_node): When localizing symbol in same comdat group,
29086 dissolve the group only when we know external symbols are going
29088 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
29090 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29092 PR tree-optimization/79390
29093 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
29094 order does not result in usable sequence, retry with reversed operand
29100 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
29101 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
29102 op0 instead of fold_convert_loc (loc, type, arg0).
29104 2017-04-12 Jeff Law <law@redhat.com>
29106 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
29107 has a delay slot in the generated code.
29109 * config/cris/cris.md (cris_preferred_reload_class): Return
29110 GENNONACR_REGS rather than GENERAL_REGS.
29112 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29115 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
29116 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
29117 signedness of the result type.
29119 2017-04-12 Richard Biener <rguenther@suse.de>
29120 Jeff Law <law@redhat.com>
29122 PR tree-optimization/80359
29123 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
29124 trim stores to TARGET_MEM_REFs.
29126 2017-04-12 Richard Biener <rguenther@suse.de>
29128 PR tree-optimization/79390
29129 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
29130 threading case even more.
29132 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
29135 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
29136 for quad_address_p for TImode, instead of just not indexed_address.
29138 2017-04-12 Richard Biener <rguenther@suse.de>
29139 Bernd Edlinger <bernd.edlinger@hotmail.de>
29141 PR middle-end/79671
29142 * alias.c (component_uses_parent_alias_set_from): Handle
29143 TYPE_TYPELESS_STORAGE.
29144 (get_alias_set): Likewise.
29145 * tree-core.h (tree_type_common): Add typeless_storage flag.
29146 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
29147 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
29148 for types containing members with TYPE_TYPELESS_STORAGE.
29149 (place_field): Likewise.
29150 (layout_type): Likewise for ARRAY_TYPE.
29151 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
29152 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
29153 TYPE_TYPELESS_STORAGE.
29154 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
29156 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29159 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
29160 first argument to type.
29162 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29166 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
29167 CONST0_RTX (mode) rather than const0_rtx where appropriate.
29168 (rs6000_expand_binop_builtin): Likewise.
29169 (rs6000_expand_ternop_builtin): Likewise; also add missing
29170 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
29171 vshasigma built-ins.
29172 * doc/extend.texi: Document that vec_xxpermdi's third argument
29173 must be a constant.
29175 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
29177 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
29178 Use shift_const cost parameter when calculating gain of STV shifts.
29180 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
29182 PR rtl-optimization/70478
29183 * lra-constraints.c (process_alt_operands): Check memory for
29184 disfavoring memory insn operand.
29186 2017-04-11 Jakub Jelinek <jakub@redhat.com>
29188 PR middle-end/80100
29189 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
29190 left shift in unsigned HOST_WIDE_INT type.
29192 PR rtl-optimization/80385
29193 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
29194 (not (neg X)) into (plus X -1) for complex or non-integral modes.
29197 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
29198 if they have any depend clauses.
29200 2017-04-11 Martin Liska <mliska@suse.cz>
29203 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
29204 * ipa-split.c (split_function): Create a local comdat symbol
29205 if caller is in a comdat group.
29207 2017-04-11 Martin Liska <mliska@suse.cz>
29210 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
29213 2017-04-11 Martin Sebor <msebor@redhat.com>
29215 PR middle-end/80364
29216 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
29217 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
29219 (directive::set_width, directive::set_precision, format_character):
29221 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
29224 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
29227 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
29228 conflict, set target->arch_name instead of target->cpu_name.
29230 2017-04-11 Richard Biener <rguenther@suse.de>
29232 PR tree-optimization/80374
29233 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
29234 build_zero_cst, remove fold_convertible_p check again.
29236 2017-04-11 Martin Liska <mliska@suse.cz>
29239 * ubsan.c (instrument_object_size): Do not instrument register
29242 2017-04-11 Jakub Jelinek <jakub@redhat.com>
29245 * config/i386/i386-builtin-types.def
29246 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
29247 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
29248 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
29249 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
29250 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
29251 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
29252 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
29253 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
29254 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
29255 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
29256 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
29257 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
29258 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
29259 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
29260 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
29261 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
29262 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
29263 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
29264 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
29265 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
29266 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
29267 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
29268 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
29269 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
29270 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
29271 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
29272 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
29273 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
29274 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
29275 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
29276 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
29277 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
29278 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
29279 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
29280 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
29281 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
29282 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
29283 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
29284 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
29285 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
29286 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
29287 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
29288 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
29289 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
29290 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
29291 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
29293 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
29294 flag to second_arg_count, handle 4 argument function type _COUNT
29295 aliases, handle second_arg_count on second argument rather than last.
29297 2017-04-10 Jeff Law <law@redhat.com>
29299 PR tree-optimization/80374
29300 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
29301 record anything if we can not convert integer_zero_node to the
29304 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29307 * config/rs6000/rs6000.c (rs6000_option_override_internal):
29308 Enhance special handling given to the TARGET_P9_MINMAX option in
29309 relation to certain other options.
29311 2017-04-10 Bin Cheng <bin.cheng@arm.com>
29313 PR tree-optimization/80153
29314 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
29315 remove POINTER_PLUS_EXPR's base part directly, rather than through
29318 2017-04-10 Richard Biener <rguenther@suse.de>
29319 Bin Cheng <bin.cheng@arm.com>
29321 PR tree-optimization/80153
29322 * tree-affine.c (aff_combination_to_tree): Get base pointer from
29323 the first element of pointer type aff_tree. Build result expr in
29325 (add_elt_to_tree): Convert to type unconditionally. Remove other
29326 fold_convert calls.
29327 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
29328 (rewrite_use_nonlinear_expr): Check invariant using iv information.
29330 2017-04-10 Richard Biener <rguenther@suse.de>
29332 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29335 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
29337 PR rtl-optimization/70478
29338 * lra-constraints.c (curr_small_class_check): New.
29339 (update_and_check_small_class_inputs): New.
29340 (process_alt_operands): Update curr_small_class_check. Disfavor
29341 alternative insn memory operands. Check available regs for small
29344 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
29347 * config/mips/mips.opt (-mvirt): Update description.
29348 * doc/invoke.texi (-mvirt): Likewise.
29350 2017-04-10 Richard Biener <rguenther@suse.de>
29352 PR middle-end/80362
29353 * fold-const.c (fold_binary_loc): Look at unstripped ops when
29354 looking for NEGATE_EXPR in -A / -B to A / B folding.
29356 2017-04-10 Martin Liska <mliska@suse.cz>
29358 PR gcov-profile/80224
29359 * gcov.c (print_usage): Fix usage string.
29360 (get_gcov_intermediate_filename): Remove.
29361 (output_gcov_file): Use both for normal and intermediate format.
29362 (generate_results): Do not initialize special file for
29363 intermediate format.
29365 2017-04-10 Richard Biener <rguenther@suse.de>
29367 PR tree-optimization/80304
29368 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
29371 2017-04-10 Nathan Sidwell <nathan@acm.org>
29374 * config/rs6000/rs6000.c (rs6000_vector_type): New.
29375 (rs6000_init_builtins): Use it.
29377 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29379 * config/arm/arm.md (<mrc>): Add mode to SET source.
29380 (<mrrc>): Likewise.
29382 2017-04-10 Richard Biener <rguenther@suse.de>
29384 PR middle-end/80344
29385 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
29387 2017-04-10 Jakub Jelinek <jakub@redhat.com>
29390 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
29391 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
29392 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
29393 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
29394 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
29395 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
29396 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
29397 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
29398 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
29399 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
29400 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
29401 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
29402 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
29403 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
29404 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
29405 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
29406 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
29407 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
29408 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
29409 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
29410 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
29411 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
29412 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
29414 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
29416 PR rtl-optimization/70478
29417 * lra-constraints.c: Reverse the last patch.
29419 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
29421 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
29422 Add comment for WCHAR_T.
29424 2017-04-08 Martin Liska <mliska@suse.cz>
29427 2017-04-07 Martin Liska <mliska@suse.cz>
29430 * ipa-split.c (split_function): Add function part to a same comdat
29433 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29436 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
29438 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
29440 * rs6000/rs6000.c (vec_load_pendulum): Rename...
29441 (vec_pairing): ...to this.
29442 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
29443 (rs6000_sched_init): Adjust for name change.
29444 (struct rs6000_sched_context): Likewise.
29445 (rs6000_init_sched_context): Likewise.
29446 (rs6000_set_sched_context): Likewise.
29448 2017-04-07 Jakub Jelinek <jakub@redhat.com>
29454 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
29456 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
29457 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
29458 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
29460 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
29462 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
29464 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
29466 PR rtl-optimization/70703
29467 * ira-color.c (update_conflict_hard_regno_costs): Use
29468 int64_t instead of HOST_WIDE_INT.
29470 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
29472 PR rtl-optimization/70478
29473 * lra-constraints.c (process_alt_operands): Disfavor alternative
29474 insn memory operands.
29476 2017-04-07 Jeff Law <law@redhat.com>
29478 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
29479 CALL and NOTE_INSN_CALL_ARG_LOCATION.
29481 2017-04-07 Martin Liska <mliska@suse.cz>
29484 * config/aarch64/aarch64.c (aarch64_process_target_attr):
29485 Show error message instead of an ICE.
29487 2017-04-07 Martin Liska <mliska@suse.cz>
29490 * ipa-split.c (split_function): Add function part to a same comdat
29493 2017-04-07 Richard Biener <rguenther@suse.de>
29495 PR middle-end/80341
29496 * tree.c (get_unwidened): Also handle ! for_type case for
29498 * convert.c (do_narrow): Split out from ...
29499 (convert_to_integer_1): ... here. Do not pass final truncation
29500 type to get_unwidened for TRUNC_DIV_EXPR.
29502 2017-04-07 Richard Biener <rguenther@suse.de>
29504 * tree-affine.c (wide_int_ext_for_comb): Take type rather
29506 (aff_combination_const): Adjust.
29507 (aff_combination_scale): Likewise.
29508 (aff_combination_add_elt): Likewise.
29509 (aff_combination_add_cst): Likewise.
29510 (aff_combination_convert): Likewise.
29511 (add_elt_to_tree): Likewise. Remove unused argument.
29512 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
29514 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
29516 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
29518 * config/arm/arm.c (arm_default_short_enums): Use
29519 ARM_DEFAULT_SHORT_ENUMS.
29520 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
29522 2017-04-06 Jakub Jelinek <jakub@redhat.com>
29525 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
29526 members with redundant out-of-class redeclaration.
29528 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
29531 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
29532 * config/i386/i386.md (*zero_extendsidi2):
29533 Add (?*x,*x) and (?*v,*v) alternatives.
29535 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
29538 * config/i386/i386.c (ix86_expand_builtin)
29539 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
29540 mode from insn data. Convert operands to insn operand mode.
29541 Copy operands that don't satisfy insn predicate to a register.
29543 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
29545 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
29548 2017-04-06 Richard Biener <rguenther@suse.de>
29550 PR tree-optimization/80334
29551 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
29552 preserve alignment of accesses.
29554 2017-04-06 Richard Biener <rguenther@suse.de>
29556 PR tree-optimization/80262
29557 * tree-sra.c (build_ref_for_offset): Preserve address-space
29559 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
29560 Drop useless address-space information on MEM_REF offsets.
29562 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
29564 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
29566 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
29568 PR rtl-optimization/70703
29569 * ira-color.c (update_conflict_hard_regno_costs): Use
29570 HOST_WIDE_INT instead of long.
29572 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
29575 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
29576 not defined for x86_64 target. Add -mmmx target option when __SSE2__
29578 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
29579 for x86_64 target. Handle -m3dnowa option.
29581 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
29583 PR rtl-optimization/70703
29584 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
29585 (update_conflict_hard_regno_costs): Use long instead of unsigned
29586 arithmetic for cost calculation.
29588 2017-04-05 Jakub Jelinek <jakub@redhat.com>
29589 Bernd Edlinger <bernd.edlinger@hotmail.de>
29592 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
29595 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
29598 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
29599 ptr_mode with Pmode throughout.
29600 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
29601 into probe_stack_range and use DImode.
29603 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
29606 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
29607 call_eh_return is true.
29609 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29611 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
29612 Initialize last_match_fntype_index.
29614 2017-04-05 Jakub Jelinek <jakub@redhat.com>
29617 * tree-nvr.c: Include internal-fn.h.
29618 (pass_return_slot::execute): Ignore internal calls without
29621 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29622 Richard Biener <rguenther@suse.de>
29625 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
29626 captures used multiple times, except for the last use.
29627 * generic-match-head.c: Include gimplify.h.
29629 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29631 PR tree-optimization/79390
29632 * target.h (struct noce_if_info): Declare.
29633 * targhooks.h (default_noce_conversion_profitable_p): Declare.
29634 * target.def (noce_conversion_profitable_p): New target hook.
29635 * ifcvt.h (struct noce_if_info): New type, moved from ...
29636 * ifcvt.c (struct noce_if_info): ... here.
29637 (noce_conversion_profitable_p): Renamed to ...
29638 (default_noce_conversion_profitable_p): ... this. No longer
29640 (noce_try_store_flag_constants, noce_try_addcc,
29641 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
29642 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
29643 instead of noce_conversion_profitable_p.
29644 * config/i386/i386.c: Include ifcvt.h.
29645 (ix86_option_override_internal): Don't override
29646 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
29647 (ix86_noce_conversion_profitable_p): New function.
29648 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
29649 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
29650 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
29651 * doc/tm.texi: Regenerated.
29653 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29655 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
29658 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29661 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
29662 instructions for small multiply cores.
29664 2017-04-04 Jeff Law <law@redhat.com>
29666 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
29668 (mips_expand_vec_perm_const): Initialize elements in orig_perm
29669 that are not set by the loop over the elements.
29671 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29674 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
29675 int mode, convert_modes it to mode as unsigned, otherwise use
29676 lowpart_subreg to mode rather than SImode.
29677 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
29678 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
29679 Use DImode instead of SImode for the shift count operand.
29680 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
29683 2017-04-04 Richard Biener <rguenther@suse.de>
29685 PR middle-end/80281
29686 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
29687 arithmetic done for the negate or the plus. Simplify.
29688 (A - (-B) -> A + B): Likewise.
29689 * fold-const.c (split_tree): Make sure to not negate pointers.
29691 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
29693 PR rtl-optimization/60818
29694 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
29695 a compare of comparisons with the thing compared if this results
29696 in a different machine mode.
29698 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
29700 * alias.c (base_alias_check): Fix typo in comment.
29701 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
29702 * cgraphunit.c (symbol_table::compile): Likewise.
29703 * collect2.c (maybe_run_lto_and_relink): Likewise.
29704 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
29705 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
29706 * config/avr/avr.c (avr_map_op_t): Likewise.
29707 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
29708 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
29709 * config/epiphany/epiphany.md (movcc): Likewise.
29710 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
29711 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
29713 * config/mips/mips.c (mips_save_restore_reg): Likewise.
29714 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
29715 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
29716 * config/sh/sh.c (sh_rtx_costs): Likewise.
29717 * fold-const.c (fold_truth_andor): Likewise.
29718 * genautomata.c (collapse_flag): Likewise.
29719 * gengtype.h (struct type::u::s): Likewise.
29720 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
29721 * input.c (FORMAT_AMOUNT): Likewise.
29722 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
29723 (known_aggs_to_agg_replacement_list): Likewise.
29724 * ipa-inline-analysis.c: Likewise.
29725 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
29726 * ipa-polymorphic-call.c
29727 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
29728 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
29729 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
29731 * modulo-sched.c (apply_reg_moves): Likewise.
29732 * omp-expand.c (build_omp_regions_1): Likewise.
29733 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
29734 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
29735 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
29736 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
29737 * value-prof.c: Likewise.
29738 * var-tracking.c (val_reset): Likewise.
29740 2017-04-03 Richard Biener <rguenther@suse.de>
29742 PR tree-optimization/80275
29743 * fold-const.c (split_address_to_core_and_offset): Handle
29746 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
29748 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
29749 descriptors is at least equal to that of functions.
29751 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
29753 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
29755 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
29758 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
29759 (mov<IMOD4:mode>): New expander.
29760 (*mov<IMOD4:mode>_internal): New insn and split pattern.
29762 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
29764 PR rtl-optimization/79405
29765 * fwprop.c (propagations_left): New variable.
29766 (forward_propagate_into): Decrement it.
29767 (fwprop_init): Initialize it.
29768 (fw_prop): If the variable has reached zero, stop propagating.
29769 (fwprop_addr): Ditto.
29771 2017-03-31 Jakub Jelinek <jakub@redhat.com>
29774 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
29775 a FUNCTION_DECL, pass it as decl instead of origin to
29778 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
29780 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
29783 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
29786 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
29787 TARGET_VSX_SMALL_INTEGER.
29789 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29791 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
29792 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
29794 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
29796 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
29797 extraction from odd-numbered MSA register.
29799 2017-03-31 Jakub Jelinek <jakub@redhat.com>
29801 PR middle-end/80173
29802 * expmed.c (store_bit_field_1): Don't attempt to create
29803 a word subreg out of hard registers wider than word if they
29804 have HARD_REGNO_NREGS of 1 for their mode.
29806 PR middle-end/80163
29807 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
29808 conversions to integer types wider than word and pointer.
29811 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
29812 (rtx_equal_for_cselib_p): Pass 0 to it.
29813 * cselib.c (cselib_hasher::equal): Likewise.
29814 (rtx_equal_for_cselib_1): Add depth argument. If depth
29815 is 128, don't look up VALUE locs and punt. Increment
29816 depth in recursive calls when walking VALUE locs.
29818 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
29820 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
29821 (make_gcov_file_name): Use the canonical path name for generating
29823 (read_line): Fix handling of files with ascii null bytes.
29825 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
29827 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
29828 to initialise a vector register instead
29829 of using a const_int.
29831 2017-03-30 Jakub Jelinek <jakub@redhat.com>
29833 PR translation/80189
29834 * gimplify.c (omp_default_clause): Use %qs instead of %s in
29835 diagnostic messages.
29837 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
29840 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
29841 (dfp_diex_<mode>): Update mode of operand 1.
29842 * doc/extend.texi (dxex, dxexq): Document change to return type.
29843 (diex, diexq): Document change to argument type.
29845 2017-03-30 Martin Jambor <mjambor@suse.cz>
29848 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
29849 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
29850 it reflects the signature changes performed at the callee side.
29851 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
29852 to cgraph_build_function_type_skip_args.
29853 (build_function_decl_skip_args): Adjust call to the above function.
29855 2017-03-30 Jakub Jelinek <jakub@redhat.com>
29858 * config/i386/sse.md
29859 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
29860 register as dest whenever it is a MEM not rtx_equal_p to the
29861 corresponding dup operand, and when forcing into reg move the
29862 reg into the memory afterwards.
29863 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
29864 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
29865 for the force_reg mode.
29866 (avx512vl_vextractf128<mode>): Use register as dest either
29867 always when a MEM, or when it is a MEM not rtx_equal_p to the
29868 corresponding dup operand, or even not when it is a CONST_VECTOR
29869 depending on the mode and lo vs. hi.
29870 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
29872 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
29873 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
29874 Likewise. Require that operands[2] is even.
29875 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
29876 Remove extraneous parens. Require that operands[2] is a multiple
29878 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
29879 operands[0] is a MEM if <mask_applied>, the predicates/constraints
29880 disallow memory then.
29882 2017-03-30 Richard Biener <rguenther@suse.de>
29884 PR tree-optimization/77498
29885 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
29886 to non-constants over backedges.
29888 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
29890 PR rtl-optimization/80233
29891 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
29892 as last_combined_insn. Do not test for BARRIER_P separately.
29894 2017-03-29 Andreas Schwab <schwab@suse.de>
29897 * calls.c (prepare_call_address): Convert funexp to Pmode before
29898 copying to temp reg.
29900 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29902 PR tree-optimization/80158
29903 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
29904 Handle possible future case of more than one alternate
29906 (replace_rhs_if_not_dup): Likewise.
29907 (replace_one_candidate): Likewise.
29909 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
29911 PR rtl-optimization/80193
29912 * ira.c (ira): Do not check allocation for LRA.
29914 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
29916 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
29917 (nvptx_output_simt_exit): Declare.
29918 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
29919 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
29920 (init_softstack_frame): Move initialization of crtl->is_leaf to...
29921 (nvptx_declare_function_name): ...here. Emit declaration of local
29922 memory space buffer for omp_simt_enter insn.
29923 (nvptx_output_unisimt_switch): New.
29924 (nvptx_output_softstack_switch): New.
29925 (nvptx_output_simt_enter): New.
29926 (nvptx_output_simt_exit): New.
29927 * config/nvptx/nvptx.h (struct machine_function): New fields
29928 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
29929 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
29930 (UNSPECV_SIMT_EXIT): Ditto.
29931 (omp_simt_enter_insn): New insn.
29932 (omp_simt_enter): New expansion.
29933 (omp_simt_exit): New insn.
29934 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
29936 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
29937 (expand_GOMP_SIMT_ENTER_ALLOC): New.
29938 (expand_GOMP_SIMT_EXIT): New.
29939 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
29940 (GOMP_SIMT_ENTER_ALLOC): Ditto.
29941 (GOMP_SIMT_EXIT): Ditto.
29942 * target-insns.def (omp_simt_enter): New insn.
29943 (omp_simt_exit): Ditto.
29944 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
29946 (lower_rec_simd_input_clauses): Implement SIMT privatization.
29947 (lower_rec_input_clauses): Likewise.
29948 (lower_lastprivate_clauses): Handle SIMT privatization.
29950 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
29951 (ompdevlow_adjust_simt_enter): New.
29952 (find_simtpriv_var_op): New.
29953 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
29954 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
29956 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
29957 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
29958 (copy_decl_for_dup_finish): Ditto.
29960 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
29962 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
29965 * config/i386/i386.c (ix86_option_override_internal): Always
29966 allow -mpreferred-stack-boundary=3 for 64-bit targets.
29968 2017-03-28 Bin Cheng <bin.cheng@arm.com>
29970 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
29972 2017-03-28 Bin Cheng <bin.cheng@arm.com>
29974 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
29975 mark new edge's irreducible flag accordign to it.
29976 (vect_do_peeling): Check loop preheader edge's irreducible flag
29977 and pass it to function slpeel_add_loop_guard.
29979 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
29981 PR tree-optimization/80218
29982 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
29983 Update block frequencies and counts.
29985 2017-03-28 Richard Biener <rguenther@suse.de>
29987 PR tree-optimization/78644
29988 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
29989 of a simplification result we may not use it at all.
29991 2017-03-28 Richard Biener <rguenther@suse.de>
29994 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
29995 without arguments, generate default definition of a SSA name.
29997 2017-03-28 Richard Biener <rguenther@suse.de>
29999 PR middle-end/80222
30000 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
30001 TYPE_REF_CAN_ALIAS_ALL references.
30002 * fold-const.c (fold_indirect_ref_1): Likewise.
30004 2017-03-28 Martin Liska <mliska@suse.cz>
30007 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
30008 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
30010 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
30011 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
30013 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
30014 (EXTRA_SPECS): Define.
30015 (SUBTARGET_EXTRA_SPECS): Likewise.
30016 (SUBTARGET_CPP_SPEC): Likewise.
30017 * config/arc/elf.h (EXTRA_SPECS): Renamed to
30018 SUBTARGET_EXTRA_SPECS.
30019 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
30021 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
30023 * config/arc/simdext.md (vst64_insn): Update pattern.
30024 (vld32wh_insn): Likewise.
30025 (vld32wl_insn): Likewise.
30026 (vld64_insn): Likewise.
30027 (vld32_insn): Likewise.
30029 2017-03-28 Marek Polacek <polacek@redhat.com>
30032 * fold-const.c (fold_comparison): Use protected_set_expr_location
30033 instead of SET_EXPR_LOCATION.
30035 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
30037 * tree.c (add_expr): Avoid name lookup warning.
30039 2017-03-27 Jeff Law <law@redhat.com>
30041 PR tree-optimization/80216
30042 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
30043 function name. Limit recursion depth.
30044 (record_temporary_equivalences): Corresponding changes.
30046 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
30048 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
30051 2017-03-27 Jakub Jelinek <jakub@redhat.com>
30054 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
30056 * cfgcleanup.c (reg_note_cfa_p): New array.
30057 (insns_have_identical_cfa_notes): New function.
30058 (old_insns_match_p): Don't cross-jump in between /f
30059 and non-/f instructions. If both i1 and i2 are frame related,
30060 verify all CFA notes, their order and content.
30062 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
30065 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
30066 HImode and SImode with zero extend to DImode to one insn.
30067 (bswap<mode>2_extenddi): Likewise.
30068 (bswapsi2_extenddi): Likewise.
30069 (bswaphi2_extendsi): Likewise.
30070 (bswaphi2): Combine bswap HImode and SImode into one insn.
30071 Separate memory insns from swapping register.
30072 (bswapsi2): Likewise.
30073 (bswap<mode>2): Likewise.
30074 (bswaphi2_internal): Delete, no longer used.
30075 (bswapsi2_internal): Likewise.
30076 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
30077 store, and gpr<-gpr swap insns.
30078 (bswap<mode>2_store): Likewise.
30079 (bswaphi2_reg): Register only splitter, combine with the splitter.
30080 (bswaphi2 splitter): Likewise.
30081 (bswapsi2_reg): Likewise.
30082 (bswapsi2 splitter): Likewise.
30083 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
30084 the insns into load, store, and register/register insns.
30085 (bswapdi2_ldbrx): Likewise.
30086 (bswapdi2_load): Likewise.
30087 (bswapdi2_store): Likewise.
30088 (bswapdi2_reg): Likewise.
30090 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
30092 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
30093 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
30095 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
30098 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
30100 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30101 special handling for target option conflicts between dform
30102 options (-mpower9-dform, -mpower9-dform-vector,
30103 -mpower9-dform-scalar) and -mno-direct-move.
30105 2017-03-27 Richard Biener <rguenther@suse.de>
30107 PR tree-optimization/80181
30108 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
30110 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30112 * config/arc/predicates.md (move_double_src_operand): Replace the
30113 call to move_double_src_operand with a call to address_operand.
30115 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30117 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
30118 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
30119 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
30121 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30123 * config/arc/predicates.md (long_immediate_loadstore_operand):
30124 Consider scaled addresses cases.
30126 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30128 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
30129 restored when in interrupt.
30130 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
30131 doesn't have delay slot.
30133 2017-03-27 Richard Biener <rguenther@suse.de>
30136 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
30137 inlined thunk clones.
30139 2017-03-27 Jakub Jelinek <jakub@redhat.com>
30142 * asan.c (instrument_derefs): Copy over last operand from
30143 original COMPONENT_REF to the new COMPONENT_REF with
30144 DECL_BIT_FIELD_REPRESENTATIVE.
30145 * ubsan.c (instrument_object_size): Likewise.
30147 2017-03-27 Richard Biener <rguenther@suse.de>
30149 PR tree-optimization/80170
30150 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
30151 sure DR/SCEV didnt fold in constants we do not see when looking
30152 at the reference base alignment.
30154 2017-03-27 Richard Biener <rguenther@suse.de>
30156 PR middle-end/80171
30157 * gimple-fold.c (fold_ctor_reference): Properly guard against
30158 NULL return value from canonicalize_constructor_val.
30160 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
30163 * config/i386/i386.c (ix86_expand_builtin)
30164 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
30165 flags reg setting and flags reg using instructions.
30166 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
30167 clobbering instructions to zero extend op2.
30169 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
30171 * doc/install.texi (Configuration) <--with-aix-soname>:
30172 Update link to AIX ld.
30174 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
30176 PR rtl-optimization/80160
30177 PR rtl-optimization/80159
30178 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
30179 reg_alternate_class into account.
30181 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
30184 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
30185 to consider in curr_insn_transform.
30187 2017-03-24 Jakub Jelinek <jakub@redhat.com>
30189 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
30190 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
30191 and emit_mode_inner.
30193 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30195 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
30196 argument to the overloaded builtin variants. Use the new flag to
30197 deprecate certain builtin variants.
30198 * config/s390/s390-builtin-types.def: Add new builtin types.
30199 * config/s390/s390-builtins.h: Support new flags field for
30200 overloaded builtins.
30201 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
30202 (s390_macro_to_expand): Enable vector float data type.
30203 (s390_cpu_cpp_builtins_internal): Indicate support of the new
30204 builtins by incrementing the __VEC__ version number.
30205 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
30207 (s390_resolve_overloaded_builtin): Emit error messages depending
30208 on the builtin flags.
30209 * config/s390/s390.c (s390_expand_builtin): Support additional
30210 flags argument. Change error message to match the messages
30211 emitted in s390-c.c.
30212 * config/s390/s390.md: New UNSPEC_* constants.
30213 (op_type): Add new instruction types.
30214 * config/s390/vecintrin.h: Add new builtins and test data class
30216 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
30217 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
30218 (VEC_INEXACT, VEC_NOINEXACT): New constants.
30219 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
30220 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
30221 ("vec_mergel<mode>"): V_HW -> VEC_HW.
30223 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
30224 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
30225 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
30226 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
30228 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
30229 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
30230 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
30231 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
30233 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
30234 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
30235 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
30236 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
30237 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
30238 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
30239 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
30241 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
30242 ("vec_scatter_element<V_HW_4:mode>_DI")
30243 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
30244 ("vec_fpint<mode>", "vflls")
30245 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
30246 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
30247 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
30248 ("*vec_cmphe<mode>_cc"): ... these.
30250 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
30251 mode constant instead of magic value.
30253 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30255 * config/s390/s390.c (s390_expand_vec_compare): Support other
30256 vector floating point modes than just V2DF.
30257 (s390_expand_vcond): Likewise.
30258 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
30259 (s390_cannot_change_mode_class): Prevent mode changes between TF
30260 and V1TF in vector registers.
30261 * config/s390/s390.md (DF, SF): New mode attributes.
30262 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
30263 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
30264 SFmode support for VRs.
30265 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
30267 (VFT, VF_HW): New mode iterators.
30268 (vw, sdx): New mode attributes.
30269 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
30270 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
30271 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
30272 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
30273 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
30274 also the new vector floating point modes. Renaming to ...
30276 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
30277 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
30278 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
30279 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
30280 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
30281 ("vec_unordered<mode>"): ... these.
30283 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
30284 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
30285 ("*vec_extendv2df"): New insn definitions.
30287 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30289 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
30290 ("mulditi3_2", "*muldi3_sign"): New patterns.
30291 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
30292 rename the pattern definition.
30294 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30296 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
30298 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
30300 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30302 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
30303 instruction if possible.
30304 * config/s390/vector.md (vec_halfnumelts): New mode
30306 ("*vec_vllezlf<mode>"): New pattern.
30308 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30310 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
30311 ("popcountv4si2", "popcountv2di2"): Rename to ...
30312 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
30313 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
30315 ("popcount<mode>2_vxe"): New pattern.
30317 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30319 * common/config/s390/s390-common.c (processor_flags_table): Add
30321 * config.gcc: Add arch12.
30322 * config/s390/driver-native.c (s390_host_detect_local_cpu):
30323 Default to arch12 for unknown CPU model numbers.
30324 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
30325 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
30326 PROCESSOR_max sanity check.
30327 * config/s390/s390-opts.h (enum processor_type): Add
30329 * config/s390/s390.c (processor_table): Add arch12.
30330 (s390_expand_builtin): Add check for B_VXE flag.
30331 (s390_issue_rate): Add PROCESSOR_ARCH12.
30332 (s390_get_sched_attrmask): Likewise.
30333 (s390_get_unit_mask): Likewise.
30334 (s390_sched_score): Enable z13 scheduling for arch12.
30335 (s390_sched_reorder): Likewise.
30336 (s390_sched_variable_issue): Likewise.
30337 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
30339 (s390_tune_attr): Use z13 scheduling also for arch12.
30340 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
30341 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
30342 (TARGET_VXE_P): New macros.
30343 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
30344 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
30345 * config/s390/s390.opt: Add arch12 as processor_type.
30347 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30349 * config/s390/s390.md
30350 ("fixuns_truncdddi2", "fixuns_trunctddi2")
30351 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
30352 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
30354 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
30355 Rename expanders to ...
30357 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
30358 ("fixuns_truncdddi2_emu"): ... these.
30360 ("fixuns_trunc<mode>si2_emu"): New expander.
30362 ("*fixuns_truncdfdi2_z13"): Rename to ...
30363 ("*fixuns_truncdfdi2_vx"): ... this.
30365 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30367 * config/s390/2964.md: Remove the single element vector compare
30368 instructions which are no longer used.
30369 * config/s390/s390.c (s390_select_ccmode): Remove handling of
30371 (s390_canonicalize_comparison): Remove handling of DFmode
30373 (s390_expand_vec_compare_scalar): Remove function.
30374 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
30375 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
30377 ("*cmp<mode>_ccs"): Add wfcdb instruction.
30379 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30381 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
30383 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
30384 will anyway by matched by mov<mode>_64dfp.
30386 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30388 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
30389 vlef/vstef. Add missing operand to vleif.
30391 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30393 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
30394 pair for all vector types with 64 bit elements.
30395 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
30396 * config/s390/vector.md (V_HW_64): ... here.
30397 (V_128_NOSINGLE): New mode iterator.
30398 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
30399 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
30400 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
30401 ("*vec_load_pairv2di"): Change to ...
30402 ("*vec_load_pair<mode>"): ... this one.
30404 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30406 * config/s390/constraints.md: Add comments.
30407 (jKK): Reject element sizes > 8 bytes.
30408 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
30410 * config/s390/s390.md: Add the s_operand checks formerly in
30411 s390_split_ok_p to various splitters where they are still
30413 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
30414 for 128 bit vectors. Plus two splitters.
30416 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30418 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
30421 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30424 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
30425 error if the boundary argument is not constant.
30427 2017-03-24 Jakub Jelinek <jakub@redhat.com>
30429 PR rtl-optimization/80112
30430 * loop-doloop.c (doloop_condition_get): Don't check condition
30431 if cmp isn't SET with IF_THEN_ELSE src.
30433 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30435 PR tree-optimization/80158
30436 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
30437 replacing a candidate statement, also replace it for the
30438 candidate's alternate interpretation.
30439 (replace_rhs_if_not_dup): Likewise.
30440 (replace_one_candidate): Likewise.
30442 2017-03-24 Richard Biener <rguenther@suse.de>
30444 PR tree-optimization/80167
30445 * graphite-isl-ast-to-gimple.c
30446 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
30448 (translate_isl_ast_to_gimple::get_rename): Likewise.
30450 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
30452 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
30453 handling of certain combinations of target options, including the
30454 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
30455 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
30457 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30460 * config/arm/arm.md (*load_multiple): Add reload_completed to
30461 matching condition.
30463 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30464 Richard Biener <rguenth@suse.de>
30466 PR tree-optimization/79908
30467 PR tree-optimization/80136
30468 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
30469 been cast away, gimplify_and_add suffices.
30471 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
30473 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
30475 2017-03-23 Richard Biener <rguenther@suse.de>
30477 PR tree-optimization/80032
30478 * gimplify.c (gimple_push_cleanup): Forced unconditional
30479 cleanups still have to go to the conditional_cleanups
30482 2017-03-22 Jakub Jelinek <jakub@redhat.com>
30484 PR tree-optimization/80072
30485 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
30487 (next_operand_entry_id): Change type to unsigned int.
30488 (sort_by_operand_rank): Make sure to return the right return value
30489 even if unsigned fields are bigger than INT_MAX.
30490 (struct oecount): Change cnt and id type to unsigned int.
30491 (oecount_hasher::equal): Formatting fix.
30492 (oecount_cmp): Make sure to return the right return value
30493 even if unsigned fields are bigger than INT_MAX.
30494 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
30497 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
30498 TREE_READONLY on result if writing it more than once.
30501 * doc/invoke.texi (-fsanitize=thread): Document that with
30502 -fnon-call-exceptions atomics are not able to throw
30506 * tsan.c: Include tree-eh.h.
30507 (instrument_builtin_call): Call maybe_clean_eh_stmt or
30508 maybe_clean_or_replace_eh_stmt where needed.
30509 (instrument_memory_accesses): Add cfg_changed argument.
30510 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
30511 if it returned true.
30512 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
30514 PR rtl-optimization/63191
30515 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
30516 wrapper function, moved the whole old content into ...
30517 (ix86_delegitimize_address_1): ... this. New inline function.
30518 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
30519 true as last argument instead of ix86_delegitimize_address.
30521 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
30523 * config/aarch64/aarch64.c (generic_branch_cost): Copy
30524 cortexa57_branch_cost.
30526 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
30528 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
30530 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30533 * doc/md.texi (Constraints): Document wA constraint.
30534 * config/rs6000/constraints.md (wA): New.
30535 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
30536 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
30537 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
30538 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
30540 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
30543 * gimplify.c (is_oacc_declared): New function.
30544 (oacc_default_clause): Use it to set default flags for acc declared
30545 variables inside parallel regions.
30546 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
30547 declared variables.
30548 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
30549 declare attribute to any decl as necessary.
30551 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
30554 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
30555 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
30556 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
30557 (arm_arch_lpae): This.
30558 * config/arm/arm.c (arm_arch7ve): Rename into ...
30559 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
30560 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
30563 2017-03-22 Martin Liska <mliska@suse.cz>
30566 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
30567 error message instead of an ICE.
30569 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30571 * doc/extend.texi (6.11 Additional Floating Types): Revise.
30573 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
30575 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30577 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30580 2017-03-21 Martin Sebor <msebor@redhat.com>
30582 * doc/extend.texi: Use "cannot" instead of "can't."
30583 * doc/hostconfig.texi: Same.
30584 * doc/install.texi: Same.
30585 * doc/invoke.texi: Same.
30586 * doc/loop.texi: Same.
30587 * doc/md.texi: Same.
30588 * doc/objc.texi: Same.
30589 * doc/rtl.texi: Same.
30590 * doc/tm.texi: Same.
30591 * doc/tm.texi.in: Same.
30592 * doc/trouble.texi: Same.
30594 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
30597 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
30598 (collect_checksum_attributes): Set it.
30599 (die_checksum_ordered): Use it.
30601 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30603 PR tree-optimization/79908
30604 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
30605 change: For a VA_ARG whose LHS has been cast away, use
30606 force_gimple_operand to construct the side effects.
30608 2017-03-21 David Malcolm <dmalcolm@redhat.com>
30610 PR translation/80001
30611 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
30612 more amenable to translation.
30613 (oacc_loop_auto_partitions): Likewise.
30615 2017-03-21 Marek Polacek <polacek@redhat.com>
30616 Martin Sebor <msebor@redhat.com>
30618 PR tree-optimization/80109
30619 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
30620 on INTEGRAL_TYPE_P.
30622 2017-03-21 Jakub Jelinek <jakub@redhat.com>
30623 Segher Boessenkool <segher@kernel.crashing.org>
30626 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
30627 check reg_used_between_p between insn and one of succ or succ2
30628 depending on if succ is artificial insn not inserted into insn
30631 2017-03-21 Martin Liska <mliska@suse.cz>
30633 PR gcov-profile/80081
30634 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
30635 * doc/gcc.texi: Include gcov-dump stuff.
30636 * doc/gcov-dump.texi: New file.
30638 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
30640 PR rtl-optimization/79150
30641 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
30642 conditional jump, if the jump is the last insn of the loop.
30644 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30645 Richard Biener <rguenth@suse.de>
30647 PR tree-optimization/79908
30648 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
30649 been cast away, use force_gimple_operand to construct the side
30652 2017-03-21 Martin Liska <mliska@suse.cz>
30654 PR libfortran/79956
30655 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
30658 2017-03-21 Brad Spengler <spender@grsecurity.net>
30661 * plugin.c (htab_hash_plugin): New function.
30662 (add_new_plugin): Use it and adjust.
30663 (parse_plugin_arg_opt): Adjust.
30664 (init_one_plugin): Likewise.
30666 2017-03-21 Richard Biener <rguenther@suse.de>
30668 PR tree-optimization/80032
30669 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
30670 if set force the cleanup to happen unconditionally.
30671 (gimplify_target_expr): Push inserted clobbers with force_uncond
30672 to avoid them being removed by control-dependent DCE.
30674 2017-03-21 Richard Biener <rguenther@suse.de>
30676 PR tree-optimization/80122
30677 * tree-inline.c (copy_bb): Do not expans va-arg packs or
30678 va_arg_pack_len when the inlined call stmt requires pack
30680 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
30682 2017-03-21 Jakub Jelinek <jakub@redhat.com>
30685 * tsan.c (instrument_builtin_call): If the memory model argument
30686 is not a constant, assume it is valid.
30689 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
30692 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
30694 PR rtl-optimization/79910
30695 * combine.c (can_combine_p): Do not allow combining an I0 or I1
30696 if its dest is used by an insn before I2 (other than the combined
30697 insns themselves, which are properly handled already).
30699 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
30702 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
30704 * combine.c (record_used_regs): New static function.
30705 (try_combine): Handle situations where there is an additional
30706 instruction between I2 and I3 which needs to have a LOG_LINK
30710 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
30712 * combine.c (try_combine): Delete redundant i1 test. Call
30713 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
30715 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30718 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
30719 alternatives 13/14.
30721 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30723 PR tree-optimization/80054
30724 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
30725 the optimization if a PHI or any of its arguments is not dominated
30726 by the candidate's basis. Use gphi* rather than gimple* as
30728 (replace_profitable_candidates): Clean up a gimple* variable that
30729 should be a gphi* variable.
30731 2017-03-20 Martin Sebor <msebor@redhat.com>
30734 * doc/extend.texi (attribute constructor): Document present limitation.
30736 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
30739 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
30740 __POWER9_VECTOR__ #ifdef control, change template definition to
30741 use Power9-specific built-in function.
30742 (vec_any_eq): Likewise.
30743 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
30744 to control outcomes from this test.
30745 (vector_ae_<mode>p): For VEC_F modes, likewise.
30747 2017-03-20 Ian Lance Taylor <iant@google.com>
30749 * config/i386/i386.c (ix86_function_regparm): Save an extra
30750 register for -fsplit-stack with DECL_STATIC_CHAIN.
30752 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
30755 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
30756 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
30758 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
30760 * config/riscv/riscv.c (riscv_print_operand): Use "fence
30762 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
30765 2017-03-20 Marek Polacek <polacek@redhat.com>
30768 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
30770 2017-03-20 Richard Biener <rguenther@suse.de>
30772 PR tree-optimization/80113
30773 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
30774 allocate extra SSA name for PHI def.
30775 (add_close_phis_to_outer_loops): Likewise.
30776 (add_close_phis_to_merge_points): Likewise.
30777 (copy_loop_close_phi_args): Likewise.
30778 (copy_cond_phi_nodes): Likewise.
30780 2017-03-20 Martin Liska <mliska@suse.cz>
30782 PR middle-end/79753
30783 * tree-chkp.c (chkp_build_returned_bound): Do not build
30784 returned bounds for a LHS that's not a BOUNDED_P type.
30786 2017-03-20 Martin Liska <mliska@suse.cz>
30790 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
30791 COMPLEX_CST and VECTOR_CST.
30793 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30796 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
30797 target operand. A new splitter adds the clobber statement in case
30798 the target operand is dead anyway.
30800 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
30802 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
30803 to age-old versions of binutils and glibc.
30805 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
30807 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
30809 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
30811 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
30813 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
30815 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
30816 requirement for binutils 2.13.
30818 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
30820 * combine.c (try_combine): Delete redundant i1 test. Call
30821 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
30823 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
30825 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
30826 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
30828 <riscv64-*-elf>: Re-arrange section
30829 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
30830 <riscv32-*-linux>: Likewise.
30831 <riscv64-*-elf>: Likewise
30832 <riscv64-*-linux>: Likewise.
30834 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
30837 * aarch64.opt(verbose-cost-dump): Fix typo.
30839 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
30842 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
30843 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
30845 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
30847 * reload.c (find_reloads): When reloading a nonoffsettable address,
30848 use RELOAD_OTHER for it and its address reloads.
30850 PR rtl-optimization/79910
30851 * combine.c (record_used_regs): New static function.
30852 (try_combine): Handle situations where there is an additional
30853 instruction between I2 and I3 which needs to have a LOG_LINK
30856 2017-03-17 Jeff Law <law@redhat.com>
30858 PR tree-optimization/71437
30859 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
30860 conditional in the hash table first.
30861 (vrp_dom_walker::before_dom_children): Extract condition from
30862 ASSERT_EXPR. Record condition, its inverion and any implied
30863 conditions as well.
30865 2017-03-17 Marek Polacek <polacek@redhat.com>
30866 Markus Trippelsdorf <markus@trippelsdorf.de>
30868 PR tree-optimization/80079
30869 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
30872 2017-03-17 Richard Biener <rguenther@suse.de>
30874 PR middle-end/80075
30875 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
30876 Properly verify the LHS before the RHS possibly claims to be
30878 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
30881 2017-03-17 Martin Jambor <mjambor@suse.cz>
30883 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
30884 (List of -O2 options): Likewise.
30885 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
30888 2017-03-17 Tom de Vries <tom@codesourcery.com>
30890 * gcov-dump.c (print_usage): Print bug_report_url.
30892 2017-03-17 Richard Biener <rguenther@suse.de>
30894 PR middle-end/80050
30895 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
30896 (parser::peek): Likewise.
30898 2017-03-17 Richard Biener <rguenther@suse.de>
30900 PR tree-optimization/80048
30901 * sese.c (free_sese_info): Properly release rename_map and
30902 copied_bb_map elements.
30904 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
30906 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
30907 Add linked-list forward and backlinks. Insert on
30908 construction, remove on destruction.
30909 (class pass_store_merging): Add m_stores_head field.
30910 (pass_store_merging::terminate_and_process_all_chains):
30911 Iterate over m_stores_head list.
30912 (pass_store_merging::terminate_all_aliasing_chains):
30914 (pass_store_merging::execute): Check for debug stmts first.
30915 Push new chains onto the m_stores_head stack.
30917 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
30920 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
30921 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
30922 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
30924 2017-03-16 Jeff Law <law@redhat.com>
30926 PR tree-optimization/71437
30927 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
30928 member function. Implementation moved into after_dom_children
30929 member function and into the threader's thread_outgoing_edges
30931 (dom_opt_dom_walker::after_dom_children): Simplify by moving
30932 some code into new thread_outgoing_edges.
30933 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
30934 definition. Simplify marker handling (do it here). Assume we always
30935 have the available expression and the const/copies tables.
30936 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
30938 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
30939 * tree-vrp.c (equiv_stack): No longer file scoped.
30940 (vrp_dom_walker): New class.
30941 (vrp_dom_walker::before_dom_children): New member function.
30942 (vrp_dom_walker::after_dom_children): Likewise.
30943 (identify_jump_threads): Setup domwalker. Use it rather than
30944 walking edges in a random order by hand. Simplify setup/finalization.
30945 (finalize_jump_threads): Remove.
30946 (vrp_finalize): Do not call identify_jump_threads here.
30947 (execute_vrp): Do it here instead and call thread_through_all_blocks
30950 PR tree-optimization/71437
30951 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
30953 (simplify_stmt_for_jump_threading): Add basic_block argument. All
30955 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
30956 (dom_opt_dom_walker::thread_across_edge): Remove
30957 handle_dominating_asserts argument. All callers changed.
30958 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
30959 changes. Remove calls to lhs_of_dominating_assert. Other
30960 uses of handle_dominating_asserts turn into unconditional code
30961 (simplify_control_stmt_condition_1): Likewise.
30962 (simplify_control_stmt_condition): Likewise.
30963 (thread_through_normal_block, thread_across_edge): Likewise.
30964 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
30965 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
30966 object if it is not an SSA_NAME.
30967 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
30968 before calling into the VRP specific simplifiers.
30969 (identify_jump_threads): Remove handle_dominating_asserts
30972 2017-03-16 Jakub Jelinek <jakub@redhat.com>
30975 * tree-diagnostic.c (default_tree_printer): No longer static.
30976 * tree-diagnostic.h (default_tree_printer): New prototype.
30978 2017-03-16 Tamar Christina <tamar.christina@arm.com>
30980 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
30981 Change ins into fmov.
30983 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30985 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
30986 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
30987 Use h_con constraint for operand 1.
30988 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
30989 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
30991 2017-03-15 Jeff Law <law@redhat.com>
30993 PR tree-optimization/71437
30994 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
30995 (record_temporary_equivalences): Use it.
30997 PR tree-optimization/71437
30998 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
30999 tree-ssa-scopedtables.
31000 (lookup_avail_expr, build_and_record_new_cond): Likewise.
31001 (record_conditions, record_cond, vuse_eq): Likewise.
31002 (record_edge_info): Adjust to API tweak of record_conditions.
31003 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
31004 (record_temporary_equivalences, optimize_stmt): Likewise.
31005 (eliminate_redundant_computations): Likewise.
31006 (record_equivalences_from_stmt): Likewise.
31007 * tree-ssa-scopedtables.c: Include options.h and params.h.
31008 (vuse_eq): New function, moved from tree-ssa-dom.c
31009 (build_and_record_new_cond): Likewise.
31010 (record_conditions): Likewise. Accept vector of conditions rather
31011 than edge_equivalence structure for first argument.
31012 for the first argument.
31013 (avail_exprs_stack::lookup_avail_expr): New member function, moved
31014 from tree-ssa-dom.c.
31015 (avail_exprs_stack::record_cond): Likewise.
31016 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
31017 from tree-ssa-dom.c.
31018 (avail_exprs_stack): Add new member functions lookup_avail_expr
31020 (record_conditions): Declare.
31022 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
31025 * lra-constraints.c (process_alt_operands): Increase reject for
31026 reloading an input/output operand.
31028 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
31031 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
31032 insns to convert from signed/unsigned char/short to IEEE 128-bit
31034 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
31036 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
31039 * config/i386/i386.c (ix86_vector_duplicate_value): Create
31040 subreg of inner mode for values already in registers.
31042 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
31044 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
31045 iteration reg is used after the loop.
31047 2017-03-14 Martin Sebor <msebor@redhat.com>
31049 PR tree-optimization/79800
31050 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
31051 precision in negative-positive range.
31052 (format_floating): Call non-const overload with adjusted precision.
31054 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
31057 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
31060 2017-03-14 Martin Sebor <msebor@redhat.com>
31062 PR middle-end/80020
31063 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
31064 * builtins.def (aligned_alloc): Use it.
31067 * Makefile.in (GTFILES): Add calls.c.
31068 * calls.c: Include "gt-calls.h".
31070 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
31072 PR rtl-optimization/79728
31073 * regs.h (struct target_regs): New field
31074 x_contains_allocatable_regs_of_mode.
31075 (contains_allocatable_regs_of_mode): New macro.
31076 * reginfo.c (init_reg_sets_1): Initialize it, and change
31077 contains_reg_of_mode so it includes global regs as well.
31078 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
31079 rather than contains_regs_of_mode.
31081 2017-03-14 Martin Liska <mliska@suse.cz>
31083 * doc/invoke.texi: Document options that can't be combined with
31084 -fcheck-pointer-bounds.
31086 2017-03-14 Martin Liska <mliska@suse.cz>
31088 PR middle-end/79831
31089 * doc/invoke.texi (-Wchkp): Document the option.
31091 2017-03-14 Martin Liska <mliska@suse.cz>
31093 * Makefile.in: Install gcov-dump.
31095 2017-03-14 Martin Liska <mliska@suse.cz>
31097 * multiple_target.c (expand_target_clones): Bail out for
31098 an invalid attribute.
31100 2017-03-14 Richard Biener <rguenther@suse.de>
31102 * alias.c (struct alias_set_entry): Pack properly.
31103 * cfgloop.h (struct loop): Likewise.
31104 * cse.c (struct set): Likewise.
31105 * ipa-utils.c (struct searchc_env): Likewise.
31106 * loop-invariant.c (struct invariant): Likewise.
31107 * lra-remat.c (struct cand): Likewise.
31108 * recog.c (struct change_t): Likewise.
31109 * rtl.h (struct address_info): Likewise.
31110 * symbol-summary.h (function_summary): Likewise.
31111 * tree-loop-distribution.c (struct partition): Likewise.
31112 * tree-object-size.c (struct object_size_info): Likewise.
31113 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
31114 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
31115 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
31116 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
31117 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
31118 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
31119 (struct _stmt_vec_info): Likewise.
31121 2017-03-14 Martin Liska <mliska@suse.cz>
31124 * multiple_target.c (create_dispatcher_calls): Check that
31125 a target can create a function dispatcher.
31127 2017-03-14 Martin Liska <mliska@suse.cz>
31130 * multiple_target.c (expand_target_clones): Drop local.local
31131 flag for default implementation.
31133 2017-03-14 Richard Biener <rguenther@suse.de>
31135 PR tree-optimization/80030
31136 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
31138 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
31140 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
31141 gcc_fallthrough() instead of __attribute__((fallthrough));
31143 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
31145 * doc/gcc.texi: Remove "up" link to (DIR).
31146 * doc/gccint.texi: Ditto.
31148 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
31150 * doc/install.texi (Specific) <avr>: Remove reference to
31153 2017-03-13 Jeff Law <law@redhat.com>
31155 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
31156 attribute rather than comments.
31158 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
31159 match_scratch operand is highest.
31161 2017-03-13 Martin Liska <mliska@suse.cz>
31163 PR middle-end/78339
31164 * ipa-pure-const.c (warn_function_noreturn): If the declarations
31165 is a CHKP clone, use original declaration.
31167 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31169 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
31170 (arc_conditional_register_usage): Use a different allocation order
31171 when optimizing for size.
31172 * common/config/arc/arc-common.c (arc_option_optimization_table):
31173 Section anchors default on when optimizing for size.
31175 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31177 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
31179 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31181 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
31182 * config/arc/arc.md (cpu_facility): Add cd variant.
31183 (*movqi_insn): Add code density variant.
31184 (*movhi_insn): Likewise.
31185 (*movqi_insn): Likewise.
31186 (*addsi3_mixed): Likewise.
31187 (subsi3_insn): Likewise.
31189 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31191 * config/arc/arc.md (movsi_cond_exec): Update constraint.
31193 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31195 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
31196 expressions with MINUS and UNARY ops.
31198 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31201 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
31203 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
31204 between vec_select and vector argument.
31205 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
31206 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
31207 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
31208 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
31209 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
31210 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
31212 2017-03-13 Richard Biener <rguenther@suse.de>
31215 * params.def (vect-max-peeling-for-alignment): Fix typo.
31217 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31219 * doc/install.texi (Specific) <mips-*-*>: Remove description of
31220 issue that only occurred with binutils below 2.18.
31222 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31224 * doc/install.texi (Specific) <cris-axis-elf>: No longer
31225 refer to binutils 2.11/2.12 minimum.
31227 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31229 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
31230 ftp.kernel.org and simplify binutils requirement.
31232 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
31234 * doc/invoke.texi (Warning Options): Fix spelling of link-time
31236 (Optimize Options): Ditto. Also remove redundancy.
31238 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31240 PR translation/79848
31241 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
31243 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
31244 to G_ to avoid double translation.
31246 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31248 PR translation/79923
31249 * auto-profile.c (get_combined_location): Convert leading
31250 character of diagnostics to lower case and remove trailing period.
31251 (read_profile): Likewise for various diagnostics.
31252 * config/arm/arm.c (arm_option_override): Remove trailing period
31253 from various diagnostics.
31254 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
31255 (msp430_expand_delay_cycles): Likewise.
31257 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31260 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
31261 full command-line argument, rather than just "str".
31262 (aarch64_validate_march): Likewise.
31263 (aarch64_validate_mtune): Likewise.
31265 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
31267 PR rtl-optimization/78911
31268 * lra-assigns.c (must_not_spill_p): New function.
31269 (spill_for): Use it.
31271 2017-03-10 Jakub Jelinek <jakub@redhat.com>
31273 PR tree-optimization/79981
31274 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
31275 ATOMIC_COMPARE_EXCHANGE ifn result.
31276 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
31277 IFN_ATOMIC_COMPARE_EXCHANGE.
31279 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31282 * opts.c (parse_sanitizer_options): Add missing question mark to
31283 "did you mean" message.
31285 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31287 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
31289 (VMULEUH_UNS): Likewise.
31290 (VMULOUB_UNS): Likewise.
31291 (VMULOUH_UNS): Likewise.
31292 * config/rs6000/rs6000.c (builtin_function_type): Remove
31293 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
31295 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31298 * read-rtl-function.c (function_reader::read_rtx_operand): Update
31299 x with result of extra_parsing_for_operand_code_0.
31300 (function_reader::extra_parsing_for_operand_code_0): Convert
31301 return type from void to rtx, returning x. When reading
31302 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
31303 larger size containing struct block_symbol.
31305 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
31307 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
31308 -mfloat128-hardware without -m64.
31310 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
31313 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
31314 entries to the case statement that marks unsigned arguments to
31315 overloaded functions.
31317 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31319 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
31320 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
31322 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
31325 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
31326 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
31328 2017-03-10 Martin Liska <mliska@suse.cz>
31332 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
31333 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
31336 2017-03-10 Olivier Hainque <hainque@adacore.com>
31338 * tree-switch-conversion (array_value_type): Start by resetting
31339 candidate type to it's main variant.
31341 2017-03-10 Jakub Jelinek <jakub@redhat.com>
31343 PR rtl-optimization/79909
31344 * combine.c (try_combine): Use simplify_replace_rtx on individual
31345 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
31346 of the whole CALL_INSN_FUNCTION_USAGE.
31348 PR tree-optimization/79972
31349 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
31350 get_range_info on SSA_NAMEs. Formatting fixes.
31352 2017-03-10 Richard Biener <rguenther@suse.de>
31353 Jakub Jelinek <jakub@redhat.com>
31355 PR tree-optimization/77975
31356 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
31357 edge to be constant.
31358 (get_val_for): For constant x return it. Formatting fix.
31359 (loop_niter_by_eval): Avoid pointless looping if the next iteration
31360 would use the same bases as the current one.
31362 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31364 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
31365 instead of vec_select for V1TImode.
31366 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
31368 (VSX_LE_128): Add V1TI to this mode iterator.
31369 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
31370 (*vsx_le_perm_store_<mode>): Likewise.
31371 (pre-reload splitter for VSX stores): Likewise.
31372 (post-reload splitter for VSX stores): Likewise.
31373 (*vsx_xxpermdi2_le_<mode>): Likewise.
31374 (*vsx_lxvd2x2_le_<mode>): Likewise.
31375 (*vsx_stxvd2x2_le_<mode>): Likewise.
31377 2017-03-09 Michael Eager <eager@eagercon.com>
31379 Correct failures with --enable-checking=yes,rtl.
31381 * config/microblaze/microblaze.c (microblaze_expand_shift):
31382 Replace GET_CODE test with CONST_INT_P and INTVAL test with
31383 test for const0_rtx.
31384 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
31385 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
31387 2017-03-09 Richard Biener <rguenther@suse.de>
31389 PR tree-optimization/79977
31390 * graphite-scop-detection.c (scop_detection::merge_sese):
31391 Handle the case of extra exits to blocks dominating the entry.
31393 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
31395 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
31398 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
31400 PR rtl-optimization/79949
31401 * lra-constraints.c (process_alt_operands): Check memory when
31402 trying to predict a cycle. Print about the overall increase.
31404 2017-03-09 Richard Biener <rguenther@suse.de>
31406 PR middle-end/79971
31407 * gimple-expr.c (useless_type_conversion_p): Preserve
31408 TYPE_SATURATING for fixed-point types.
31410 2017-03-09 Richard Biener <rguenther@suse.de>
31413 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
31414 alignment of BLKmode params.
31416 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31419 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
31420 (VALL_NO_V2Q): Likewise.
31422 * config/aarch64/aarch64-simd.md
31423 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
31425 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
31426 VALL_NO_V2Q mode iterator.
31427 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
31429 2017-03-09 Martin Liska <mliska@suse.cz>
31431 PR tree-optimization/79631
31432 * tree-chkp-opt.c (chkp_is_constant_addr): Call
31433 tree_int_cst_sign_bit just for INTEGER constants.
31435 2017-03-09 Martin Liska <mliska@suse.cz>
31439 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
31442 2017-03-09 Marek Polacek <polacek@redhat.com>
31445 * tree.c (inchash::add_expr): Handle TREE_VEC.
31447 2017-03-09 Martin Liska <mliska@suse.cz>
31450 (chkp_narrow_size_and_offset): New function.
31451 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
31452 (void chkp_parse_bit_field_ref): New function.
31453 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
31454 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
31456 2017-03-09 Martin Liska <mliska@suse.cz>
31459 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
31460 (chkp_find_bounds_1): Remove gcc_unreachable.
31462 2017-03-09 Jakub Jelinek <jakub@redhat.com>
31465 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
31466 BUILT_IN_SYNC*, determine the access type from the size suffix and
31467 always build a MEM_REF with that type. Handle forgotten
31468 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
31471 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
31472 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
31473 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
31474 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
31475 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
31476 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
31477 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
31478 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
31479 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
31480 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
31481 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
31482 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
31483 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
31484 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
31485 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
31486 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
31487 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
31488 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
31489 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
31490 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
31491 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
31492 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
31493 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
31494 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
31495 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
31496 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
31497 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
31498 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
31499 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
31500 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
31501 definitions outside of __OPTIMIZE__ guarded section.
31504 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
31505 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
31506 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
31507 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
31510 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31512 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
31513 ("vfenez<mode>"): Add missing constraints.
31515 2017-03-08 Martin Sebor <msebor@redhat.com>
31518 * config/nds32/nds32.c (nds32_option_override):
31519 Fix misspelled diagnostic.
31521 2017-03-08 Jakub Jelinek <jakub@redhat.com>
31524 * gimplify.c (gimplify_omp_for): Replace index var in outer
31525 taskloop statement with an artificial variable and add
31526 OMP_CLAUSE_PRIVATE clause for it.
31528 2017-03-08 Richard Biener <rguenther@suse.de>
31530 PR tree-optimization/79955
31531 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
31532 for accesses that are completely outside of the variable.
31534 2017-03-08 Andrew Haley <aph@redhat.com>
31536 PR tree-optimization/79943
31537 * tree-ssa-loop-split.c (compute_new_first_bound): When
31538 calculating the new upper bound, (END-BEG) should be added, not
31541 2017-03-08 Jakub Jelinek <jakub@redhat.com>
31543 * config/avr/avr.md (setmemhi): Make sure match_dup
31544 operand number comes before match_scratch.
31546 2017-03-08 Richard Biener <rguenther@suse.de>
31548 PR tree-optimization/79920
31549 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
31550 with ncopies == 1 to ...
31551 (vect_transform_slp_perm_load): ... here. Properly compute
31552 all element loads by iterating VF times over the group. Do
31553 not handle ncopies (computed in a broken way) in
31554 vect_create_mask_and_perm.
31556 2017-03-08 Jakub Jelinek <jakub@redhat.com>
31559 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
31560 is a uniform vector, use uniform_vector_p return value instead of
31561 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
31563 2017-03-07 Marek Polacek <polacek@redhat.com>
31565 PR middle-end/79809
31566 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
31567 (alloca_call_type): Likewise.
31569 2017-03-07 Martin Liska <mliska@suse.cz>
31571 * gcov.c (process_args): Put comment to correct location.
31573 2017-03-07 Martin Liska <mliska@suse.cz>
31575 PR middle-end/68270
31576 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
31577 Use array_at_struct_end_p instead of DECL_CHAIN (field).
31578 (chkp_narrow_bounds_for_field): Likewise.
31579 (chkp_parse_array_and_component_ref): Pass one more argument to
31582 2017-03-07 Richard Biener <rguenther@suse.de>
31584 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
31587 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
31589 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
31590 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
31592 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31595 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
31596 to end of description.
31597 (PARAM_MAX_STORES_TO_MERGE): Likewise.
31599 2017-03-07 Jakub Jelinek <jakub@redhat.com>
31601 PR rtl-optimization/79901
31602 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
31604 (*avx512f_<code><mode>3<mask_name>): ... this.
31605 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
31606 iterator instead of VI8_AVX2_AVX512BW.
31608 PR rtl-optimization/79901
31609 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
31610 min/max expander, expand it using expand_vec_cond_expr.
31613 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
31616 2017-03-06 Jakub Jelinek <jakub@redhat.com>
31619 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
31620 to void * for PCH reasons.
31621 * dwarf2out.c (output_loc_operands, output_die): Cast
31622 v.val_vec.array to unsigned char *.
31624 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
31627 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
31630 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
31632 PR rtl-optimization/79571
31633 * lra-constraints.c (process_alt_operands): Calculate static
31634 reject and subtract it from overall when only addresses will be
31637 2017-03-06 Julia Koval <julia.koval@intel.com>
31640 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
31641 incoming stack boundary to 128 for 64-bit targets.
31643 2017-03-06 Richard Biener <rguenther@suse.de>
31645 PR tree-optimization/79894
31646 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
31647 to NULL after folding it.
31649 2017-03-06 Richard Biener <rguenther@suse.de>
31651 PR tree-optimization/79824
31652 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
31653 check disabling peeling for gaps.
31655 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
31657 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
31658 attributes): Document gettimeofday.
31660 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
31662 * config/s390/s390.c (s390_option_override_internal): Set
31663 PARAM_MIN_VECT_LOOP_BOUND
31665 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
31667 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
31668 * config/s390/s390.md: Likewise.
31670 2017-03-06 Jakub Jelinek <jakub@redhat.com>
31673 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
31674 (<avx2_avx512>_perm<mode>): Rename to ...
31675 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
31677 (<avx512>_perm<mode>_mask): Rename to ...
31678 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
31680 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
31681 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
31682 instead of VI8F_256_512.
31683 (avx512f_perm<mode>): New define_expand.
31684 (avx512f_perm<mode>_mask): Likewise.
31685 (avx512f_perm<mode>_1<mask_name>): New define_insn.
31686 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
31688 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
31690 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
31691 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
31693 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
31695 2017-03-06 Martin Liska <mliska@suse.cz>
31698 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
31699 when having a SSA NAME w/o VAR_DECL assigned to it.
31701 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
31703 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
31704 msa_dpsub_<su>_d): Fix MODE for vec_select.
31706 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
31708 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
31710 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
31712 2017-03-06 Richard Biener <rguenther@suse.de>
31714 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
31715 * plugin.c (register_plugin_info): Likewise.
31716 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
31718 2017-03-05 Jakub Jelinek <jakub@redhat.com>
31720 * config/i386/sse.md (sse_storehps, sse_storelps,
31721 avx_<castmode><avxsizesuffix>_<castmode>,
31722 avx512f_<castmode><avxsizesuffix>_<castmode>,
31723 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
31724 in condition that at least one operand is not a MEM.
31726 2017-03-03 Jakub Jelinek <jakub@redhat.com>
31728 PR middle-end/79805
31729 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
31730 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
31732 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
31733 gimple_call_nothrow_p flag based on whether original builtin can throw.
31734 If it can, emit following stmts on the fallthrough edge.
31735 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
31736 don't create new bb if inserting just debug stmts on the edge, try to
31737 insert them on the fallthru bb or just reset debug stmts.
31739 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
31742 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
31743 restore recog_data (including the operand rtxes inside it) around
31744 the call to get_insn_template.
31746 2017-03-03 Martin Sebor <msebor@redhat.com>
31748 PR tree-optimization/79699
31749 * context.c (context::~context): Free MPFR caches to avoid
31750 a memory leak on program exit.
31752 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31754 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
31755 Use wide_int::ulow () instead of .elt (0).
31757 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
31759 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
31760 (*pushxf): Limit oF constraint to 32bit targets and add oC
31761 constraint for 64bit targets.
31762 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
31763 (*pushdf): Change rmF constraint to rmC.
31765 2017-03-03 Martin Liska <mliska@suse.cz>
31767 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
31768 Remove unused variable.
31770 2017-03-03 Jakub Jelinek <jakub@redhat.com>
31773 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
31774 is a memory operand, increase num_memory.
31775 (ix86_expand_args_builtin): Likewise.
31777 2017-03-03 Jan Hubicka <jh@suse.cz>
31780 * ipa-devirt.c (maybe_record_node): Properly handle
31781 __cxa_pure_virtual visibility.
31783 2017-03-03 Martin Liska <mliska@suse.cz>
31785 PR tree-optimization/79803
31786 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
31788 (pass_loop_prefetch::execute): Disabled optimization if an
31789 assumption about L1 cache size is not met.
31791 2017-03-03 Martin Liska <mliska@suse.cz>
31793 PR rtl-optimization/79574
31794 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
31795 (hash_scan_set): Likewise.
31796 (dump_hash_table): Likewise.
31797 (hoist_code): Likewise.
31799 2017-03-03 Richard Biener <rguenther@suse.de>
31801 * fixed-value.c (fixed_from_string): Restore use of elt (1)
31802 in place of uhigh ().
31803 (fixed_convert_from_real): Likewise.
31805 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
31808 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
31810 2017-03-03 Richard Biener <rguenther@suse.de>
31812 PR middle-end/79818
31813 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
31814 TYPE_OVERFLOW_UNDEFINED check.
31816 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31818 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
31820 (vector_ae_<mode>_p): Likewise.
31821 (vector_nez_<mode>_p): Likewise.
31822 (vector_ne_v2di_p): Likewise.
31823 (vector_ae_v2di_p): Likewise.
31824 (vector_ne_<mode>_p): Likewise.
31825 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
31827 (vsx_tsqrt<mode>2_fe): Likewise.
31829 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
31832 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
31834 2017-03-02 Jakub Jelinek <jakub@redhat.com>
31836 PR rtl-optimization/79780
31837 * cprop.c (one_cprop_pass): When second and further conditional trap
31838 in a single basic block is turned into an unconditional trap, turn it
31839 into a deleted note to avoid RTL verification failures.
31841 2017-03-02 Richard Biener <rguenther@suse.de>
31843 * fold-const.c (const_binop): Use ulow () instead of elt (0).
31845 2017-03-02 Richard Biener <rguenther@suse.de>
31847 PR tree-optimization/79345
31849 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
31850 param and abort the walk, returning -1 if it is hit.
31851 (walk_aliased_vdefs): Take a limit param and pass it on.
31852 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
31853 defaulting to 0 and return a signed int.
31854 * tree-ssa-uninit.c (struct check_defs_data): New struct.
31855 (check_defs): New helper.
31856 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
31857 about uninitialized memory.
31858 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
31859 bogus uninitialized warning.
31860 (fixed_convert_from_real): Likewise.
31862 2017-03-02 Bin Cheng <bin.cheng@arm.com>
31864 PR tree-optimization/66768
31865 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
31866 iv_use if base object can't be determined.
31868 2017-03-02 Jakub Jelinek <jakub@redhat.com>
31870 PR tree-optimization/79345
31871 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
31872 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
31873 (get_pattern_stats): Initialize it.
31874 * genemit.c (gen_expand): Verify match_scratch numbers come after
31875 match_operand/match_dup numbers.
31876 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
31877 match_scratch numbers.
31878 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
31880 * config/s390/s390.md (trunctdsd2): Likewise.
31882 2017-03-02 Richard Biener <rguenther@suse.de>
31884 * wide-int.h (wide_int_storage::operator=): Implement in terms
31887 2017-03-02 Richard Biener <rguenther@suse.de>
31889 PR tree-optimization/79777
31890 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
31891 the to insert expression to sth existing.
31893 2017-03-01 Martin Sebor <msebor@redhat.com>
31895 PR middle-end/79692
31896 * gimple-ssa-sprintf.c
31897 (directive::known_width_and_precision): New function.
31898 (format_integer): Use it.
31899 (get_mpfr_format_length): Consider the full range of precision
31900 when computing %g output with the # flag. Set the likely byte
31901 count to 3 rather than 1 when precision is indeterminate.
31902 (format_floating): Correct the lower bound of precision.
31904 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31906 * doc/invoke.texi: Document default code model for 64-bit Linux.
31908 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31911 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
31912 udiv rather than div since input pattern is unsigned.
31914 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
31916 * config/i386/i386.c (print_reg): Warn for values of
31917 unsupported size in integer register.
31919 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
31922 * config/rs6000/predicates.md (current_file_function_operand): Do
31923 not allow self calls to be local if the function is replaceable.
31925 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
31928 * config/rs6000/altivec.h (vec_ctz and others): Change the
31929 preprocessor macro that controls conditional compilation from
31930 _ARCH_PWR9 to __POWER9_VECTOR__.
31931 (vec_all_ne): Change parameterization of __altivec_scalar_pred
31932 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
31933 control (instead of _ARCH_PWR9 control) so that template
31934 definition uses power9-specific function.
31935 (vec_any_eq): Likewise.
31936 (vec_all_ne): Change macro definition to use a power9-specific
31937 expansion under #ifdef __POWER9_VECTOR__ control (instead of
31938 _ARCH_PWR9 control).
31939 (vec_any_eq) Likewise.
31940 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
31941 expansion for CMPNEF to remove support for xvcmpnesp instruction.
31942 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
31943 support for xvcmpnedp instruction.
31944 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
31945 macro expansion so that Power9 implementation of vec_all_ne does
31946 not use the AltiVec predicate framework.
31947 (VCMPNEH_P): Likewise.
31948 (VCMPNEW_P): Likewise.
31949 (VCMPNED_P): Likewise.
31950 (VCMPNEFP_P): Likewise.
31951 (VCMPNEDP_P): Likewise.
31952 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
31953 implementation of vec_any_eq to not use AltiVec predicate
31955 (VCMPAEH_P): Likewise.
31956 (VCMPAEW_P): Likewise.
31957 (VCMPAED_P): Likewise.
31958 (VCMPAEFP_P): Likewise.
31959 (VCMPAEDP_P): Likewise.
31960 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
31961 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
31962 not use the AltiVec predicate framework.
31963 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
31964 of vec_any_eq to not use AltiVec predicate framework.
31965 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
31966 support for predefined __POWER9_VECTOR__ macro to indicate that
31967 Power9 instruction selection is enabled.
31968 (altivec_overloaded_builtins): Remove extraneous
31969 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
31970 function argument types RS6000_BTI_bool_V16QI and
31971 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
31972 entry for overloaded function argument types RS6000_BTI_bool_V4SI
31973 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
31974 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
31975 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
31976 Power9 for implementations of vec_cmpne. Change the signature for
31977 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
31978 (representing vec_all_ne) to remove the previously described first
31979 argument of type RS6000_BTI_INTSI, as this was an artifact of
31980 reliance on the AltiVec predicate framework, which is no longer
31981 used in the implementation of these functions. Add
31982 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
31983 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
31984 since, unlike the AltiVec predicate framework implementation, we
31985 do not share function descriptors between vec_alle and vec_anyeq.
31986 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
31987 set of modes that receive special treatment even when
31988 TARGET_P9_VECTOR is true. The special treatment emits code that
31989 does not depend on Power9 instructions.
31990 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
31991 define_expand to not rely on AltiVec predicate framework.
31992 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
31994 (vector_ne_v2di_p): Change this define_expand to not rely on
31995 AltiVec predicate framework.
31996 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
31998 (vector_ne_<mode>_p): Change this define_expand to not rely on
31999 AltiVec predicate framework.
32000 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
32002 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
32003 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
32004 define_insn pattern.
32005 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
32006 define_insn pattern because the xvcmpne<VSs>. instruction is not
32008 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
32009 instruction is not supported.
32011 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32013 * config/nvptx/nvptx.c: Include intl.h.
32015 2017-03-01 Martin Jambor <mjambor@suse.cz>
32018 * ipa-prop.h (ipa_bits): Removed field known.
32019 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
32020 to pointers. Adjusted their comments to warn about their sharing.
32021 (ipcp_transformation_summary): Change bits to a vector of pointers.
32022 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
32023 (ipa_get_ipa_bits_for_value): Declare.
32024 * tree-vrp.h (value_range): Mark as GTY((for_user)).
32025 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
32026 (ipa_bits_hash_table): Likewise.
32027 (ipa_vr_ggc_hash_traits): Likewise.
32028 (ipa_vr_hash_table): Likewise.
32029 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
32030 being pointers and vr_known being removed.
32031 (ipa_set_jf_unknown): Likewise.
32032 (ipa_get_ipa_bits_for_value): New function.
32033 (ipa_set_jfunc_bits): Likewise.
32034 (ipa_get_value_range): New overloaded functions.
32035 (ipa_set_jfunc_vr): Likewise.
32036 (ipa_compute_jump_functions_for_edge): Use the above functions to
32037 construct bits and vr parts of jump functions.
32038 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
32039 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
32041 (ipcp_grow_transformations_if_necessary): Also allocate
32042 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
32044 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
32045 them. Fix too long lines.
32046 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
32047 vr_known being removed.
32048 (ipa_read_jump_function): Use new setter functions to construct bits
32049 and vr parts of jump functions or set them to NULL.
32050 (write_ipcp_transformation_info): Adjust for bits being pointers.
32051 (read_ipcp_transformation_info): Likewise.
32052 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
32054 Include gt-ipa-prop.h.
32055 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
32057 (ipcp_store_bits_results): Likewise.
32058 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
32059 Do not write to existing jump functions but use a temporary instead.
32061 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32064 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
32065 attempt to use its first operand as BIT_FIELD_REF base.
32067 2017-03-01 Richard Biener <rguenther@suse.de>
32069 PR middle-end/79721
32070 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
32071 interpolating formula in wrapping arithmetic.
32072 (chrec_apply): Convert chrec_evaluate return value to wanted type.
32074 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32076 PR tree-optimization/79734
32077 * tree-vect-generic.c (expand_vector_condition): Optimize
32078 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
32079 Handle VEC_COND_EXPR where comparison has different inner width from
32080 type's inner width.
32082 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
32084 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
32085 markup, and similar issues. Remove @opindex entries for things
32086 that aren't options. Add missing -mmpy-option entries.
32088 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32090 PR tree-optimization/79737
32091 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
32092 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
32093 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
32094 instead of byte_size. Formatting fix.
32095 (shift_bytes_in_array_right): Formatting fix.
32097 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
32100 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
32101 condition on optimize for the leaf function test.
32103 2017-02-28 Martin Liska <mliska@suse.cz>
32106 * read-rtl-function.c (function_reader::handle_unknown_directive):
32107 Bail out when one uses -flto.
32109 2017-02-28 Martin Liska <mliska@suse.cz>
32111 * common.opt: Replace space with tabular for options of <number>
32113 * config/i386/i386.opt: Show <number> value for
32114 -mlarge-data-threshold.
32115 * opts.c (print_filtered_help): Do not display number in hexadecimal
32118 2017-02-28 Martin Liska <mliska@suse.cz>
32120 * common.opt: Fix --help=option -Q for options which are of
32123 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
32125 * config/i386/i386.c (print_reg): Error out for values
32126 of 8-bit size in invalid integer register.
32128 2017-02-28 Martin Sebor <msebor@redhat.com>
32130 PR tree-optimization/79691
32131 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
32133 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32136 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
32137 gcc_unreachable with output_operand_lossage.
32139 2017-02-28 Richard Biener <rguenther@suse.de>
32141 PR tree-optimization/79740
32142 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
32144 (visit_nary_op): Insert the nary into the hashtable if we
32145 pattern-matched sth.
32146 * tree-ssa-pre.c (eliminate_insert): Robustify.
32148 2017-02-28 Richard Biener <rguenther@suse.de>
32150 PR middle-end/79731
32151 * fold-const.c (decode_field_reference): Reject out-of-bound
32154 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32156 * config/i386/i386.c: Include intl.h.
32157 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
32158 instead of just cond ? "..." : "...".
32159 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
32160 * coverage.c (read_counts_file): Likewise.
32161 * omp-offload.c: Include intl.h.
32162 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
32163 of just cond ? "..." : "...".
32164 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
32165 of just cond ? "..." : "...".
32167 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
32170 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
32172 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
32173 'tune for' CPU name.
32174 * config/arm/arm-cpu-data.h: Regenerated.
32176 2017-02-28 Richard Biener <rguenther@suse.de>
32178 PR tree-optimization/79732
32179 * tree-inline.c (expand_call_inline): Do not shadow var.
32181 2017-02-28 Richard Biener <rguenther@suse.de>
32183 PR tree-optimization/79723
32184 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
32185 address-space properly.
32187 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
32189 * doc/optinfo.texi (Optimization groups): Fix option used for
32191 * doc/invoke.texi (-fopt-info): Document "omp".
32192 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
32193 (OPTGROUP_ALL): Add OPTGROUP_OMP.
32194 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
32195 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
32196 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
32198 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
32200 * dumpfile.c (optgroup_options): Instead of "openmp", associate
32201 OPTGROUP_OMP with "omp".
32203 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
32206 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
32207 for arithmetic shift of unsigned V2DI.
32209 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
32211 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
32212 arc/linux.h headers.
32213 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
32214 (LINK_SPEC): Likewise.
32215 (ARC_TLS_EXTRA_START_SPEC): Likewise.
32216 (EXTRA_SPECS): Likewise.
32217 (STARTFILE_SPEC): Likewise.
32218 (ENDFILE_SPEC): Likewise.
32219 (LIB_SPEC): Likewise.
32220 (TARGET_SDATA_DEFAULT): Likewise.
32221 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
32222 (MULTILIB_DEFAULTS): Likewise.
32223 (DWARF2_UNWIND_INFO): Likewise.
32224 * config/arc/big.h: New file.
32225 * config/arc/elf.h: Likewise.
32226 * config/arc/linux.h: Likewise.
32227 * config/arc/t-uClibc: Remove.
32229 2017-02-27 Bin Cheng <bin.cheng@arm.com>
32231 PR tree-optimization/77536
32232 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
32233 (tree_transform_and_unroll_loop): Use above function to compute the
32234 estimated niter of unrolled loop and use it when scaling profile.
32235 Also use count info rather than frequency if it's non-zero.
32236 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
32237 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
32238 (vect_transform_loop): Call above function.
32240 2017-02-27 Richard Biener <rguenther@suse.de>
32242 PR tree-optimization/45397
32243 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
32244 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
32245 (visit_nary_op): Add pattern matching for CSEing sign-changed
32246 or truncated operations with wider ones.
32248 2017-02-27 Richard Biener <rguenther@suse.de>
32250 PR tree-optimization/79690
32251 * tree-vect-stmts.c (vectorizable_store): Use vector type
32252 built from the DR with address-space.
32254 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
32256 * doc/invoke.texi (Optimize Options): Refine the description
32257 of asan-use-after-return.
32259 2017-02-25 Alan Modra <amodra@gmail.com>
32261 PR rtl-optimization/79584
32262 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
32263 base, not ad->base_term, the reg within base. Remove assertion
32264 that ad->base == ad->base_term. Replace gen_int_mode using
32265 bogus mode with const0_rtx.
32267 2017-02-25 Jakub Jelinek <jakub@redhat.com>
32269 PR middle-end/79396
32270 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
32271 FMA_EXPR like tcc_binary or tcc_unary.
32273 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
32276 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
32278 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
32279 (output_loc_operands): Handle DW_OP_call_ref and
32280 DW_OP_GNU_variable_value.
32281 (struct variable_value_struct): New type.
32282 (struct variable_value_hasher): Likewise.
32283 (variable_value_hash): New variable.
32284 (string_types): Remove.
32285 (copy_loc_descr): New function.
32286 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
32287 (prepend_loc_descr_to_each): New function.
32288 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
32289 instead of add_loc_descr_to_each if the first argument is single
32290 location list and the second has multiple.
32291 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
32292 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
32293 when looking for variable value which doesn't have other location info.
32294 (loc_list_from_tree): Formatting fix.
32295 (gen_array_type_die): Simplify DW_AT_string_length handling.
32296 (adjust_string_types): Remove.
32297 (gen_subprogram_die): Don't call adjust_string_types nor test/set
32298 string_types. Call resolve_variable_values.
32299 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
32300 (resolve_addr_in_expr): Likewise. Add A argument.
32301 (copy_deref_exprloc): Remove deref argument. Adjust for the
32302 original expression being DW_OP_GNU_variable_value with optionally
32303 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
32304 optionally after it.
32305 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
32306 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
32307 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
32308 (variable_value_hasher::hash, variable_value_hasher::equal): New
32310 (resolve_variable_value_in_expr, resolve_variable_value,
32311 resolve_variable_values, note_variable_value_in_expr,
32312 note_variable_value): New functions.
32313 (dwarf2out_early_finish): Call note_variable_value on all toplevel
32316 2017-02-24 Jakub Jelinek <jakub@redhat.com>
32319 * opts.h (handle_generated_option): Add GENERATED_P argument.
32320 * opts-common.c (handle_option): Adjust function comment.
32321 (handle_generated_option): Add GENERATED_P argument, pass it to
32323 (control_warning_option): Pass false to handle_generated_option
32325 * opts.c (maybe_default_option): Pass true to handle_generated_option
32327 * optc-gen.awk: Likewise.
32329 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32331 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
32332 a REG, look at the REG it is a SUBREG of.
32333 (splitter for cmpeqsi_t): Ditto.
32335 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32337 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
32338 the special USEs with the pattern of the insn, not the insn itself.
32340 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
32343 * doc/invoke.texi: Document -mload-store-pairs.
32345 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32346 Sandra Loosemore <sandra@codesourcery.com>
32348 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
32349 argument isn't a CONST_INT.
32350 (nios2_alternate_compare_const): Assert op is a CONST_INT.
32351 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
32352 (nios2_validate_compare): Bypass alternate compare logic if *op2
32353 is not a CONST_INT.
32354 (ldstwm_operation_p): Return false if first_base is not a REG or
32355 if first_offset is not a CONST_INT.
32357 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32359 * config/cris/cris.md: Use correct operand in a define_peephole2.
32361 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32363 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
32365 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32367 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
32368 this_insn if it is an INSN or JUMP_INSN.
32369 (force_offsettable): Look at base, not at addr.
32370 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
32371 on things that aren't necessarily CONST_INTs.
32373 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
32375 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
32376 -mfpmath=sse is the default also for x86-32 targets with SSE2
32377 instruction set when @option{-ffast-math} is enabled
32379 2017-02-24 Jeff Law <law@redhat.com>
32381 PR rtl-optimizatoin/79286
32382 * ira.c (update_equiv_regs): Drop may_trap_p exception to
32385 2017-02-24 Richard Biener <rguenther@suse.de>
32387 PR tree-optimization/79389
32388 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
32391 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
32393 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
32394 function comment to reflect reality.
32395 (loop_exits_before_overflow): Fix typo in function description.
32397 2017-02-24 Richard Biener <rguenther@suse.de>
32399 PR tree-optimization/79389
32400 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
32401 properly that a threading opportunity exists. Detect conditional
32402 copy/constant propagation opportunities.
32404 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
32406 * config/visium/visium.md (type): Add trap.
32407 (b): New mode attribute.
32408 (*btst): Rename into...
32409 (*btst<mode>): ...this and adjust.
32410 (*cbranchsi4_btst_insn): Rename into...
32411 (*cbranch<mode>4_btst_insn): ...this and adjust.
32412 (trap): New define_insn.
32414 2017-02-23 Jakub Jelinek <jakub@redhat.com>
32416 PR tree-optimization/79389
32417 * ifcvt.c (struct noce_if_info): Add rev_cond field.
32418 (noce_reversed_cond_code): New function.
32419 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
32420 reversed_comparison_code. Formatting fix.
32421 (noce_try_store_flag): Test rev_cond != NULL in addition to
32422 reversed_comparison_code.
32423 (noce_try_store_flag_constants): Likewise.
32424 (noce_try_store_flag_mask): Likewise.
32425 (noce_try_addcc): Use rev_cond if non-NULL instead of
32426 reversed_comparison_code.
32427 (noce_try_cmove_arith): Likewise. Formatting fixes.
32428 (noce_try_minmax, noce_try_abs): Clear rev_cond.
32429 (noce_find_if_block): Initialize rev_cond.
32430 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
32431 instead of false as last argument never attempt to reverse it
32434 2017-02-23 Bin Cheng <bin.cheng@arm.com>
32436 PR tree-optimization/79663
32437 * tree-predcom.c (combine_chains): Process refs in reverse order
32438 only for ZERO length chains, and add explaining comment.
32440 2017-02-23 Jeff Law <law@redhat.com>
32442 PR tree-optimization/79578
32443 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
32444 in call to operand_equal_p.
32446 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
32449 * config/i386/cpuid.h: Fix another undefined behavior.
32451 2017-02-23 Richard Biener <rguenther@suse.de>
32453 PR tree-optimization/79683
32454 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
32455 vector types for data-refs.
32457 2017-02-23 Martin Liska <mliska@suse.cz>
32459 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
32461 2017-02-23 Jakub Jelinek <jakub@redhat.com>
32463 PR middle-end/79665
32464 * internal-fn.c (get_range_pos_neg): Moved to ...
32465 * tree.c (get_range_pos_neg): ... here. No longer static.
32466 * tree.h (get_range_pos_neg): New prototype.
32467 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
32468 are known to be in between 0 and signed maximum inclusive, try to
32469 expand both unsigned and signed divmod and use the cheaper one from
32472 2017-02-22 Jeff Law <law@redhat.com>
32474 PR tree-optimization/79578
32475 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
32476 to compare base operands.
32478 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
32481 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
32482 gpc_reg_operand instead of fpr_reg_operand.
32484 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
32486 * config/mips/mips.c (mips_return_in_memory): Force FP
32487 vector types to be returned in memory for o32 ABI.
32489 2017-02-22 Jakub Jelinek <jakub@redhat.com>
32491 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
32492 instead of DW_TAG_member for static data member declarations and don't
32493 set no_linkage_name for static inline data members.
32494 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
32497 2017-02-22 Martin Liska <mliska@suse.cz>
32499 * doc/invoke.texi: Replace inequality signs with square brackets
32502 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
32505 * lra-constraints.c (simplify_operand_subreg): Handle
32506 WORD_REGISTER_OPERATIONS targets.
32508 2017-02-22 Jakub Jelinek <jakub@redhat.com>
32511 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
32512 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
32513 elimination by swapping fld*.
32515 2017-02-22 Richard Biener <rguenther@suse.de>
32517 PR tree-optimization/79673
32518 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
32519 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
32520 irrelevant address-space qualifiers and avoiding a
32521 ADDR_SPACE_CONVERT_EXPR from fold_convert.
32523 2017-02-22 Richard Biener <rguenther@suse.de>
32525 PR tree-optimization/79666
32526 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
32527 to not symbolically negate if that may introduce undefined
32530 2017-02-22 Martin Liska <mliska@suse.cz>
32533 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
32534 * data-streamer-out.c (streamer_write_gcov_count_stream):
32536 * value-prof.c (stream_out_histogram_value): Make assert more
32537 precise based on type of counter.
32539 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
32542 * config/i386/i386.md (standard_x87sse_constant_load splitter):
32543 Use nonimmediate_operand instead of memory_operand for operand 1.
32544 (float-extend standard_x87sse_constant_load splitter): Ditto.
32546 2017-02-21 Jeff Law <law@redhat.com>
32548 PR tree-optimization/79621
32549 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
32550 blocks with edges to themselves.
32552 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32555 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
32556 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
32557 Use gimple_call_builtin_p.
32560 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
32561 on temporarily removed DEBUG_INSNs.
32563 PR tree-optimization/79649
32564 * tree-loop-distribution.c (classify_partition): Give up on
32565 non-generic address space loads/stores.
32567 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
32569 * doc/loop.texi (Loop manipulation): Remove nonexistent
32570 tree_ssa_loop_version from the documentation.
32571 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
32573 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32576 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
32577 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
32578 * config/rs6000/rs6000.c: Include except.h.
32579 (rs6000_expand_split_stack_prologue): Call
32580 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
32582 2017-02-21 Martin Jambor <mjambor@suse.cz>
32585 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
32586 have been analyzed.
32588 2017-02-21 Martin Jambor <mjambor@suse.cz>
32590 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
32591 for backward compatibility only.
32592 * doc/invoke.texi (Option Summary): Remove all references to
32593 -fipa-cp-alignment.
32595 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
32599 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32601 * lra-constraints.c (curr_insn_transform): Handle
32602 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
32604 2017-02-21 Martin Liska <mliska@suse.cz>
32606 * config/i386/i386.opt: Replace -masm-dialect with -masm.
32608 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
32610 PR translation/79638
32611 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
32613 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
32616 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
32617 (arm_function_ok_for_sibcall): Return false for an indirect call by
32618 descriptor if all the argument registers are used.
32619 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
32620 alignment of the function.
32622 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32624 PR tree-optimization/61441
32625 * simplify-rtx.c (simplify_const_unary_operation): For
32626 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
32627 the sNaN unmodified.
32629 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
32631 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32632 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32633 instead of SYSTEM_HEADER_DIR.
32635 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
32636 Martin Liška <mliska@suse.cz>
32638 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
32639 Fix typos and grammar, use active voice, and clarify.
32641 2017-02-20 Marek Polacek <polacek@redhat.com>
32643 PR middle-end/79537
32644 * gimplify.c (gimplify_expr): Handle unused *&&L;.
32647 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
32649 2017-02-20 Jakub Jelinek <jakub@redhat.com>
32652 * config/i386/i386.c (ix86_expand_builtin): Handle
32653 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
32654 ix86_builtins_isa[fcode].isa as a requirement of those
32655 flags and any other flag in the bitmask.
32656 (ix86_init_mmx_sse_builtins): Use 0 instead of
32657 ~OPTION_MASK_ISA_64BIT as mask.
32658 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
32659 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
32660 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
32661 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
32663 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32666 * lra-constraints.c (split_reg): Check requested split mode
32667 is supported by the register.
32669 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32671 * lra-constraints.c (simplify_operand_subreg): Remove early
32674 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32677 * lra-constraints.c (curr_insn_transform): Tighten condition
32678 for converting SUBREG reloads from OP_OUT to OP_INOUT.
32680 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32683 * lra-constraints.c (curr_insn_transform): Handle
32684 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
32686 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
32689 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
32691 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
32693 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
32696 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
32699 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32701 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
32704 2017-02-18 Jakub Jelinek <jakub@redhat.com>
32706 * final.c (last_columnnum, override_columnnum): New variables.
32707 (final_start_function): Set last_columnnum, pass it to begin_prologue
32708 hook and pass 0 to dwarf2out_begin_prologue.
32709 (final_scan_insn): Update override_columnnum. Pass last_columnnum
32710 to source_line debug hook.
32711 (notice_source_line): Compute last_columnnum and for debug_column_info
32712 return true on column changes.
32713 * debug.h (struct gcc_debug_hooks): Add column argument to
32714 source_line and begin_prologue hooks.
32715 (debug_nothing_int_charstar_int_bool): Remove prototype.
32716 (debug_nothing_int_int_charstar,
32717 debug_nothing_int_int_charstar_int_bool): New prototypes.
32718 (dwarf2out_begin_prologue): Add column argument.
32719 * debug.c (do_nothing_debug_hooks): Adjust source_line and
32720 begin_prologue hooks.
32721 (debug_nothing_int_charstar_int_bool): Remove.
32722 (debug_nothing_int_int_charstar,
32723 debug_nothing_int_int_charstar_int_bool): New functions.
32724 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
32725 through to dwarf2out_source_line.
32726 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
32727 (dwarf2out_source_line): Add column argument, emit it if requested.
32728 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
32730 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
32731 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
32732 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
32733 through to dwarf2out_begin_prologue.
32734 (vmsdbgout_source_line): Add column argument, pass it through to
32735 dwarf2out_source_line.
32736 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
32737 dbxout_source_line caller.
32738 (dbxout_source_line): Add column argument.
32740 * common.opt (gno-column-info, gcolumn-info): New options.
32741 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
32742 (check_die): Also test for multiple DW_AT_decl_column attributes.
32743 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
32744 DW_AT_decl_column if requested.
32745 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
32747 (gen_variable_die): Likewise.
32748 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
32749 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
32752 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
32753 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
32754 (ix86_handle_option): Handle OPT_m3dnowa.
32755 * doc/invoke.texi (-m3dnowa): Document.
32756 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
32757 -m3dnowa instead of -m3dnow -march=athlon.
32760 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
32761 instead of gcc_assert for K, r and R code checks. Formatting fixes.
32763 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32766 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
32767 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
32768 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
32769 generator for vsx_xxpermdi_<mode>_be.
32770 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
32771 force big-endian semantics.
32772 (vsx_xxpermdi_<mode>_be): New define_expand with same
32773 implementation as previous version of vsx_xxpermdi_<mode>.
32775 2017-02-17 Jakub Jelinek <jakub@redhat.com>
32777 PR tree-optimization/79327
32778 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
32779 variable, its initialization and use.
32781 2017-02-17 Julia Koval <julia.koval@intel.com>
32783 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
32784 (OPTION_MASK_ISA_PKU_UNSET): New.
32785 (ix86_handle_option): Handle -mrdpid.
32786 * config/i386/cpuid.h (bit_RDPID): New.
32787 * config/i386/driver-i386.c (host_detect_local_cpu):
32788 Detect RDPID feature.
32789 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
32790 * config/i386/i386-c.c (ix86_target_macros_internal):
32792 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
32793 (ix86_valid_target_attribute_inner_p): Add "rdpid".
32794 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
32795 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
32796 * config/i386/i386.md (define_insn "rdpid"): New.
32797 * config/i386/i386.opt Add -mrdpid.
32798 * config/i386/immintrin.h (_rdpid_u32): New.
32800 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
32802 PR rtl-optimization/79541
32803 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
32804 instead of transforming it into USE.
32806 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
32808 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
32809 If HONOR_SNANS (SFmode) force the input to a register.
32810 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
32811 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
32812 an frsp or similar insn.
32814 2017-02-17 Martin Liska <mliska@suse.cz>
32816 PR rtl-optimization/79577
32817 * params.def (selsched-max-sched-times): Increase minimum to 1.
32819 2017-02-17 Martin Liska <mliska@suse.cz>
32821 PR rtl-optimization/79574
32822 * gcse.c (want_to_gcse_p): Prevent integer overflow.
32824 2017-02-17 Martin Liska <mliska@suse.cz>
32826 PR tree-optimization/79529
32827 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
32828 ssa_defined_default_def_p to handle cases which are implicitly
32830 * tree-ssa.c (ssa_defined_default_def_p): New function.
32831 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
32832 which are implicitly defined.
32833 * tree-ssa.h (ssa_defined_default_def_p): Declare.
32835 2017-02-17 Richard Biener <rguenther@suse.de>
32837 PR middle-end/79576
32838 * params.def (max-ssa-name-query-depth): Limit to 10.
32840 2017-02-17 Richard Biener <rguenther@suse.de>
32842 PR tree-optimization/79552
32843 * tree-ssa-structalias.c (visit_loadstore): Properly verify
32846 2017-02-17 Richard Biener <rguenther@suse.de>
32849 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
32851 2017-02-17 Marek Polacek <polacek@redhat.com>
32853 PR middle-end/79536
32854 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
32855 (fold_negate_expr): New wrapper.
32857 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
32859 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
32860 Correct terminology and de-emphasize pre-standard behavior.
32862 2017-02-16 Alan Modra <amodra@gmail.com>
32864 PR rtl-optimization/79286
32865 * ira.c (def_dominates_uses): New function.
32866 (update_equiv_regs): Don't create an equivalence for insns that
32867 may trap where the register def does not dominate the use.
32869 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
32871 PR rtl-optimization/78127
32872 * lra.c (lra): Call lra_eliminate before finish the loop after
32875 2017-02-16 Richard Biener <rguenther@suse.de>
32877 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
32879 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
32880 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
32881 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
32882 (isl_val_int_from_wi): New function.
32883 (extract_affine_gmp): Rename to ...
32884 (extract_affine_wi): ... this, take a widest_int.
32885 (extract_affine_int): Just wrap extract_affine_wi.
32886 (add_param_constraints): Use isl_val_int_from_wi.
32887 (add_loop_constraints): Likewise, and extract_affine_wi.
32889 2017-02-15 Jeff Law <law@redhat.com>
32891 PR middle-end/79521
32892 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
32893 ira_init_register_move_cost_if_necessary.
32895 2017-02-15 Martin Sebor <msebor@redhat.com>
32897 PR middle-end/32003
32898 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
32899 removed in a prior commit.
32901 2017-02-15 Bin Cheng <bin.cheng@arm.com>
32903 PR tree-optimization/79347
32904 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
32905 counters during peeling.
32907 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
32909 * Makefile.in (site.exp): Remove "set ISLVER".
32911 2017-02-15 Jakub Jelinek <jakub@redhat.com>
32914 * real.c (real_from_integer): Call real_convert even for decimal.
32916 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
32919 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
32921 2017-02-14 Andrew Pinski <apinski@cavium.com>
32923 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
32924 cores and change the partno/implementer to be correct.
32925 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
32926 the 'B" as the implementer.
32927 * config/aarch64/aarch64-tune.md: Regenerate.
32929 2017-02-14 Carl Love <cel@us.ibm.com>
32931 * config/rs6000/rs6000.c: Add case statement entry to make the
32932 xvcvuxdsp built-in argument unsigned.
32933 * config/rs6000/vsx.md: Fix the source and return operand types so they
32934 match the instruction definitions from the ISA document. Fix typo
32935 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
32938 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
32941 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
32942 member early_clobber_alts.
32943 * lra-lives.c (reg_early_clobber_p): New.
32944 (process_bb_lives): Use it.
32945 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
32946 (debug_operand_data): Initialize early_clobber_alts.
32947 (setup_operand_alternative): Set up early_clobber_alts.
32948 (collect_non_operand_hard_regs): Ditto. Pass early clobber
32949 alternatives to new_insn_reg.
32950 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
32952 (lra_update_insn_regno_info): Pass the new arg.
32954 2017-02-14 Jakub Jelinek <jakub@redhat.com>
32956 PR middle-end/79505
32957 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
32958 (new_oacc_loop_raw): Don't clear already cleared fields.
32961 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
32962 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
32963 _mm512_prefetch_i64gather_ps): New inline functions and macros.
32965 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
32968 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
32970 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
32973 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
32974 the extra instruction to the right place to store 128-bit constant
32977 2017-02-14 Martin Sebor <msebor@redhat.com>
32979 PR middle-end/79448
32980 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
32981 warning for strings of unknown length.
32983 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
32985 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
32987 2017-02-14 Jeff Law <law@redhat.com>
32990 * ira-costs.c (scan_one_insn): Initialize register move costs
32991 for pseudos seen in USE/CLOBBER insns.
32993 PR tree-optimization/79095
32994 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
32995 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
32996 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
32997 if the operands are known to be not equal, then the resulting range
32999 (intersect_ranges): If the new range is ~[0,0] and the old range is
33000 wide, then prefer ~[0,0].
33001 * tree-vrp.c (overflow_comparison_p_1): New function.
33002 (overflow_comparison_p): New function.
33003 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
33004 if NAME is used in an overflow test.
33005 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
33006 overflow check that can be expressed as an equality test, then adjust
33007 ops to be that equality test.
33009 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33011 * config/s390/s390-builtin-types.def: Remove flags argument.
33012 * config/s390/s390.c (s390_init_builtins): Likewise.
33014 2017-02-14 Martin Liska <mliska@suse.cz>
33016 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
33017 vector. Fix trailing white spaces.
33019 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
33021 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
33024 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33026 PR rtl-optimization/68664
33027 * config/arm/arm.c (arm_sched_can_speculate_insn):
33028 New function. Declare prototype.
33029 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
33031 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33033 PR rtl-optimization/68664
33034 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
33036 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
33038 2017-02-14 Amit Pawar <amit.pawar@amd.com>
33040 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
33041 max skip bytes for function, loop and jump.
33043 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33045 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
33046 ABS_EXPR for gimple dump.
33048 2017-02-14 Jakub Jelinek <jakub@redhat.com>
33051 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
33053 PR tree-optimization/79408
33054 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
33055 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
33056 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
33057 also if rhs1 is INTEGER_CST.
33059 2017-02-14 Richard Biener <rguenther@suse.de>
33061 PR middle-end/79432
33062 * tree-into-ssa.c (insert_phi_nodes): When the function can
33063 have abnormal edges rewrite SSA names with broken use-def
33064 dominance out of SSA and register them for PHI insertion.
33066 2017-02-13 Martin Sebor <msebor@redhat.com>
33068 PR middle-end/79496
33069 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
33070 clearing info.nowrite flag when snprintf size argument is a range.
33072 2017-02-13 Jakub Jelinek <jakub@redhat.com>
33074 * cprop.c (cprop_jump): Add missing space in string literal.
33075 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
33076 (get_constraint_for_component_ref): Likewise.
33077 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
33078 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
33079 * lra-constraints.c (process_alt_operands): Likewise.
33080 * ipa-inline.c (inline_small_functions): Likewise.
33081 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
33082 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
33083 * trans-mem.c (diagnose_tm_1_op): Likewise.
33084 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
33085 (grid_parallel_clauses_gridifiable): Likewise.
33087 * config/nvptx/mkoffload.c (process): Add space in between
33090 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
33091 "MOD4_SSE_REGS" and "ALL_REGS".
33093 * spellcheck.c (test_data): Add , in between "foo" and "food".
33095 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33098 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
33099 boundary crossing check and subsequent code generation agree.
33101 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33103 * config/aarch64/aarch64.c (has_memory_op): Delete.
33104 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
33107 2017-02-13 Jakub Jelinek <jakub@redhat.com>
33109 PR rtl-optimization/79388
33110 PR rtl-optimization/79450
33111 * combine.c (distribute_notes): When removing TEM_INSN for which
33112 corresponding dest has last value recorded, invalidate that last
33115 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33117 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
33118 of explicit '@'. Add missing assembly comment marker on branch costs
33121 2017-02-13 Nathan Sidwell <nathan@acm.org>
33123 * gengtype-lex.l (<in_struct>): Add '/'.
33125 2017-02-13 Martin Liska <mliska@suse.cz>
33128 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
33130 2017-02-13 Richard Biener <rguenther@suse.de>
33132 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
33134 * configure: Re-generate.
33135 * config.in: Likewise.
33136 * graphite-dependences.c: Simplify as if
33137 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
33138 * graphite-isl-ast-to-gimple.c: Likewise.
33139 * graphite-optimize-isl.c: Likewise.
33140 * graphite-poly.c: Likewise.
33141 * graphite-sese-to-poly.c: Likewise.
33142 * graphite.h: Likewise.
33143 * toplev.c: Include isl/version.h and use isl_version () for
33144 printing the ISL version.
33145 * doc/install.texi: Update ISL requirement.
33147 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
33149 * doc/standards.texi (Standards): Update reference to
33152 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
33154 * doc/extend.texi (Named Address Spaces): sourceware.org now
33156 * doc/install.texi (Binaries): Ditto.
33159 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
33161 * doc/cpp.texi: Replace "stringify"/"stringification" with C
33162 standard terminology "stringize"/"stringizing" throughout.
33163 * doc/cppinternals.texi: Likewise.
33165 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
33167 * doc/extend.texi: Fix some spelling mistakes and typos.
33168 * doc/invoke.texi: Likewise.
33170 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
33173 * params.def (inline-min-speedup) Change from 10 to 8.
33175 2017-02-11 Jakub Jelinek <jakub@redhat.com>
33177 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
33180 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
33183 * ipa-inline-analysis.c (get_minimal_bb): New function.
33184 (record_modified): Use it.
33185 (remap_edge_change_prob): Handle also ancestor functions.
33187 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
33189 * doc/contrib.texi (Contributors): Remove broken link into
33190 the Mauve CVS repository.
33192 2017-02-11 Jakub Jelinek <jakub@redhat.com>
33194 PR middle-end/79454
33195 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
33196 result computation whenever lhs doesn't have vector mode, not
33197 just when it has BLKmode.
33199 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
33201 * doc/makefile.texi (profiledbootstrap): Refer to the
33202 installation instructions only in textual form.
33204 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33207 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
33209 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
33211 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
33212 (Specific): Update mingw-w64 reference.
33214 (Specific): Remove broken link to Renesas RX processor.
33216 2017-02-10 Richard Biener <rguenther@suse.de>
33218 * toplev.c (process_options): Do not mention obsolete graphite
33219 options when printing sorry message about missing graphite support.
33220 Mention -floop-nest-optimize.
33222 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
33224 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
33225 (vtst_p16): Likewise.
33226 (vtstq_p8): Likewise.
33227 (vtstq_p16): Likewise.
33229 (vtstq_p64): Likewise.
33230 * config/arm/arm_neon.h (vgetq_lane_p64): New.
33231 (vset_lane_p64): New.
33232 (vsetq_lane_p64): New.
33234 2017-02-10 Jakub Jelinek <jakub@redhat.com>
33236 PR tree-optimization/79411
33237 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
33238 stmt operands are SSA_NAMEs used in abnormal phis.
33239 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
33242 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
33245 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
33248 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
33250 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
33252 2017-02-09 Jakub Jelinek <jakub@redhat.com>
33254 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
33258 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
33259 not arbitrary TREE_CONSTANT.
33262 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
33263 "omp declare target link" attribute unless is_global_var.
33264 * omp-offload.c (find_link_var_op): Likewise.
33266 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
33267 Chung-Lin Tang <cltang@codesourcery.com>
33269 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
33271 (gimplify_adjust_omp_clauses): Don't delete TILE.
33272 (gimplify_omp_for): Deal with TILE.
33273 * internal-fn.c (expand_GOACC_TILE): New function.
33274 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
33276 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
33277 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
33279 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
33280 avoid DIV for outermost collapse var.
33281 (expand_oacc_for): Insert tile element loop as needed. Adjust.
33282 Remove out of date comments, fix whitespace.
33283 * omp-general.c (omp_extract_for_data): Deal with tiling.
33284 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
33285 adjust OLF_DIM_BASE value.
33286 (struct omp_for_data): Add tiling field.
33287 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
33288 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
33289 for auto loops. Remove default auto determining, moved to
33290 oacc_loop_fixed_partitions.
33291 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
33292 stmts, add e_mask field.
33293 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
33294 (oacc_thread_numbers): Use oacc_dim_call.
33295 (oacc_xform_tile): New.
33296 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
33297 (finish_oacc_loop): Adjust for ifns vector.
33298 (oacc_loop_discover_walk): Append loop abstraction sites to list,
33299 add case for GOACC_TILE fns.
33300 (oacc_loop_xform_loop): Delete.
33301 (oacc_loop_process): Iterate over call list directly, and add
33302 handling for GOACC_TILE fns.
33303 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
33305 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
33306 vector partitioning to outer loops. Assign 2 partitions to loops
33307 when available. Add TILE handling.
33308 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
33309 (execite_oacc_device_lower): Process GOACC_TILE fns,
33310 ignore unknown specs.
33311 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
33312 * tree.c (omp_clause_num_ops): Adjust TILE ops.
33313 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
33315 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
33317 * configure.ac (ACX_BUGURL): Update.
33318 * configure: Regenerate.
33320 2017-02-09 Richard Biener <rguenther@suse.de>
33322 PR tree-optimization/69823
33323 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
33324 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
33326 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
33328 * config/arc/arc-c.def: Add __NPS400__ definition.
33329 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
33330 (TARGET_NPS400): Define.
33332 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
33334 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
33336 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
33337 pointer, arch_info.
33338 (arc_cpu_types): Fill the arch_info field with a pointer into the
33339 arc_arch_types table.
33340 (arc_selected_cpu): Declare.
33341 * config/arc/arc.c (arc_selected_cpu): Make global.
33342 (arc_selected_arch): Delete.
33343 (arc_base_cpu): Delete.
33344 (arc_override_options): Remove references to deleted variables,
33345 update access to arch information.
33346 (ARC_OPT): Update access to arch information.
33347 (ARC_OPTX): Likewise.
33348 * config/arc/arc.h (arc_base_cpu): Remove declaration.
33349 (TARGET_ARC600): Update access to arch information.
33350 (TARGET_ARC601): Likewise.
33351 (TARGET_ARC700): Likewise.
33352 (TARGET_EM): Likewise.
33353 (TARGET_HS): Likewise.
33354 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
33357 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
33360 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
33361 condition/operands for integer GE/LE/GEU/LEU operations.
33363 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
33365 PR translation/79397
33366 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
33369 2017-02-08 Martin Jambor <mjambor@suse.cz>
33372 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
33373 whether allocation happened.
33374 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
33375 nothing was allocated.
33377 2017-02-08 Jakub Jelinek <jakub@redhat.com>
33379 PR tree-optimization/79408
33380 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
33381 constant, but SSA_NAME with a known integer range, use the minimum
33382 of that range instead of op1 to determine if modulo can be replaced
33383 with its first operand.
33385 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33387 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
33389 2017-02-08 Richard Biener <rguenther@suse.de>
33391 PR tree-optimization/71824
33392 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
33393 Check all loops contained in the merged region.
33395 2017-02-07 Andrew Pinski <apinski@cavium.com>
33397 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
33399 2017-02-07 Andrew Pinski <apinski@cavium.com>
33401 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
33402 (thunderxt88): Likewise.
33403 (thunderxt81): Disable LSE and change v8.1 to v8.
33404 (thunderxt83): Likewise.
33406 2017-02-07 Jakub Jelinek <jakub@redhat.com>
33407 Richard Biener <rguenther@suse.de>
33409 PR middle-end/79399
33410 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
33411 type from int to size_t.
33412 * ira-costs.c (struct_costs_size): Change type from int to size_t.
33414 2017-02-07 Jakub Jelinek <jakub@redhat.com>
33416 PR rtl-optimization/79386
33417 * cprop.c (bypass_conditional_jumps): Initialize
33418 bypass_last_basic_block already before splitting bbs after
33419 unconditional traps...
33420 (bypass_conditional_jumps): ... rather than here.
33423 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
33424 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
33425 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
33426 fix -masm=intel patterns.
33428 2017-02-07 Richard Biener <rguenther@suse.de>
33430 PR tree-optimization/79256
33431 PR middle-end/79278
33432 * builtins.c (get_object_alignment_2): Use min_align_of_type
33433 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
33434 and ADJUST_FIELD_ALIGN.
33436 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
33438 * doc/tm.texi: Regenerate.
33439 * stor-layout.c (layout_decl): Adjust.
33440 (update_alignment_for_field): Likewise.
33441 (place_field): Likewise.
33442 (min_align_of_type): Likewise.
33443 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
33444 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
33445 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
33446 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
33447 * config/frv/frv.c (frv_adjust_field_align): Likewise.
33448 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
33449 * config/i386/i386.c (x86_field_alignment): Likewise.
33450 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
33451 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
33452 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
33453 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
33454 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
33455 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
33459 2017-01-30 Richard Biener <rguenther@suse.de>
33461 PR tree-optimization/79256
33462 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
33463 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
33466 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
33468 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
33469 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
33470 builtins to SImode and emit a zero-extend, if necessary.
33472 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33474 * docs/invoke.texi (RISC-V Options): Alphabetize.
33476 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33478 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
33481 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33483 * config/riscv/riscv.c: New file.
33484 * common/config/riscv/riscv-common.c: Likewise.
33485 * config.gcc: Likewise.
33486 * config/riscv/constraints.md: Likewise.
33487 * config/riscv/elf.h: Likewise.
33488 * config/riscv/generic.md: Likewise.
33489 * config/riscv/linux.h: Likewise.
33490 * config/riscv/multilib-generator: Likewise.
33491 * config/riscv/peephole.md: Likewise.
33492 * config/riscv/pic.md: Likewise.
33493 * config/riscv/predicates.md: Likewise.
33494 * config/riscv/riscv-builtins.c: Likewise.
33495 * config/riscv/riscv-c.c: Likewise.
33496 * config/riscv/riscv-ftypes.def: Likewise.
33497 * config/riscv/riscv-modes.def: Likewise.
33498 * config/riscv/riscv-opts.h: Likewise.
33499 * config/riscv/riscv-protos.h: Likewise.
33500 * config/riscv/riscv.h: Likewise.
33501 * config/riscv/riscv.md: Likewise.
33502 * config/riscv/riscv.opt: Likewise.
33503 * config/riscv/sync.md: Likewise.
33504 * config/riscv/t-elf-multilib: Likewise.
33505 * config/riscv/t-linux: Likewise.
33506 * config/riscv/t-linux-multilib: Likewise.
33507 * config/riscv/t-riscv: Likewise.
33508 * configure.ac: Likewise.
33509 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
33510 Waterman as RISC-V maintainers.
33511 * doc/install.texi: Add RISC-V entries.
33512 * doc/invoke.texi: Add RISC-V options section.
33513 * doc/md.texi: Add RISC-V constraints section.
33514 * configure: Regenerated.
33516 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
33519 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
33520 false values to be constant vectors with all 0 or all 1 bits set.
33521 (vcondu<mode><mode>): Likewise.
33522 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
33524 (fpmask_comparison_operator): Update comment.
33525 (vecint_comparison_operator): New predicate.
33526 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
33527 vector conditionals when the true and false values are constant
33528 vectors with all 0 bits or all 1 bits set.
33530 2017-02-06 Martin Sebor <msebor@redhat.com>
33532 PR tree-optimization/79376
33533 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
33535 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
33537 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
33538 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
33539 to simplify split condition.
33541 2017-02-06 Jakub Jelinek <jakub@redhat.com>
33543 * omp-expand.c (oxpand_omp_atomic_fetch_op,
33544 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
33547 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
33549 PR rtl-optimization/68664
33550 * target.def (can_speculate_insn): New hook.
33551 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
33552 * doc/tm.texi: Regenerate.
33553 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
33554 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
33555 (rs6000_sched_can_speculate_insn): New function.
33557 2017-02-06 Jakub Jelinek <jakub@redhat.com>
33559 PR tree-optimization/79284
33560 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
33561 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
33562 vectorizable_mask_load_store, vectorizable_operation,
33563 vect_is_simple_cond, get_same_sized_vectype): Use it instead
33564 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
33565 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
33566 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
33567 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
33568 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
33569 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
33570 is_gimple_assign (stmt). Replace another such test with
33571 is_gimple_assign (stmt).
33573 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
33576 * config/avr/avr.c (rtl-iter.h): Include it.
33577 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
33578 (avr_legitimate_combined_insn): ...and implementation.
33580 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
33582 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
33583 * config/s390/s390.c (s390_const_operand_ok)
33584 (s390_canonicalize_comparison, s390_extract_part)
33585 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
33586 (s390_contiguous_bitmask_p, s390_rtx_costs)
33587 (legitimize_pic_address): Likewise.
33588 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
33589 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
33590 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
33591 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
33592 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
33594 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
33596 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
33597 REGNO($0) == REGNO($1).
33599 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33601 * config/s390/linux.h(SIZE_TYPE): Add comment.
33603 2017-02-06 Julian Brown <julian@codesourcery.com>
33604 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33605 Virendra Pathak <virendra.pathak@broadcom.com>
33607 * config/aarch64/aarch64-cores.def: Change the scheduler
33609 * config/aarch64/aarch64.md: Include thunderx2t99.md.
33610 * config/aarch64/thunderx2t99.md: New file.
33612 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33614 * doc/standards.texi (Go Language): Update link to language
33617 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
33619 * tree-eh.c (lower_resx): Sanitize profile.
33620 (cleanup_empty_eh_move_lp): Likewise.
33622 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
33625 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
33627 * cfgloopmanip.h (loop_version): Update prototype.
33628 * modulo-sched.c (sms_schedule): Update call of loop_version.
33629 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
33630 * tree-parloops.c (gen_parallel_loop): Likewise.
33631 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
33632 * tree-ssa-loop-split.c (split_loop): Likewise.
33633 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
33634 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
33636 2017-02-05 Martin Liska <mliska@suse.cz>
33639 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
33641 (print_operand_address): Initialize a struct to zero.
33643 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33645 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
33646 garbage collector only in textual form.
33648 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33650 * doc/extend.texi (x86 specific memory model extensions for
33651 transactional memory): Simplify a phrase.
33653 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
33656 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
33657 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
33658 (atomic_storedi_1): Likewise.
33660 2017-02-04 Jakub Jelinek <jakub@redhat.com>
33662 PR tree-optimization/79338
33663 * tree-parloops.c (gather_scalar_reductions): Don't call
33664 vect_analyze_loop_form for loop->inner before destroying loop's
33667 2017-02-03 Martin Sebor <msebor@redhat.com>
33669 PR tree-optimization/79327
33670 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
33671 when precision has resulted in leading zeros.
33672 (format_integer): Adjust the likely counter to assume an unknown
33673 argument that may be zero is non-zero.
33675 2017-02-03 Jason Merrill <jason@redhat.com>
33678 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
33679 avoid copying non-taken branch.
33681 2017-02-03 Jakub Jelinek <jakub@redhat.com>
33683 PR tree-optimization/79340
33684 * tree-vect-loop.c (vectorizable_reduction): Release
33685 vec_defs elements after safe_splicing them into other vectors.
33688 PR tree-optimization/79327
33689 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
33690 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
33692 (format_integer): Use wide_int_to_tree instead of build_int_cst
33693 + to_?hwi. If argmin is NULL, just set argmin and argmax to
33694 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
33695 of shortest and longest sequence.
33697 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
33699 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
33700 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
33702 2017-02-03 Walter Lee <walt@tilera.com>
33705 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
33706 after initial stackframe link reg save.
33707 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
33709 2017-02-03 Jakub Jelinek <jakub@redhat.com>
33712 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
33713 wu for stxssp alternative.
33715 2017-02-03 Martin Sebor <msebor@redhat.com>
33717 PR tree-optimization/79352
33718 * gimple-fold.c (get_range_strlen): Add argument.
33719 (get_range_strlen): Change return type to bool.
33720 (get_maxval_strlen): Pass in a dummy argument.
33721 * gimple-fold.h (get_range_strlen): Change return type to bool.
33722 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
33723 * tree.h (array_at_struct_end_p): Add argument.
33724 * tree.c (array_at_struct_end_p): Handle it.
33726 2017-02-03 Martin Liska <mliska@suse.cz>
33729 * multiple_target.c (create_dispatcher_calls): Redirect edge
33730 from a caller of a dispatcher.
33731 (expand_target_clones): Make the clones local.
33732 (ipa_target_clone): Do both target clones and resolvers.
33733 (ipa_dispatcher_calls): Remove the pass.
33734 (pass_dispatcher_calls::gate): Likewise.
33735 (make_pass_dispatcher_calls): Likewise.
33736 * passes.def (pass_target_clone): Put as very first IPA early
33739 2017-02-03 Martin Liska <mliska@suse.cz>
33741 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
33742 in case of a function with ifunc attribute.
33744 2017-02-03 Martin Liska <mliska@suse.cz>
33746 * cgraph.c (cgraph_node::dump): Dump function version info.
33747 * symtab.c (symtab_node::dump_base): Add missing new line.
33749 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
33751 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
33752 (ifcombine_ifandif): Use it.
33754 2017-02-03 Martin Liska <mliska@suse.cz>
33756 * doc/invoke.texi: Document default value for
33757 use-after-scope-direct-emission-threshold.
33759 2017-02-03 Martin Liska <mliska@suse.cz>
33761 PR tree-optimization/79339
33762 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
33763 (format_floating): Likewise.
33765 2017-02-03 Martin Liska <mliska@suse.cz>
33768 * ipa-prop.c (ipa_node_params_t::insert): Remove current
33770 (ipa_node_params_t::remove): Likewise.
33771 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
33772 initialization from removed ipa_node_params_t::insert.
33773 (ipa_node_params::~ipa_node_params): Move from removed
33774 ipa_node_params_t::release.
33775 * symbol-summary.h (symbol_summary::m_released): New member.
33776 Do not release a summary twice. Do not allow to call finalizer
33777 for types of a summary that live in GGC memory.
33779 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33781 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
33784 2017-02-02 Martin Sebor <msebor@redhat.com>
33786 PR middle-end/79275
33787 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
33788 (format_string): Tighten up the range of output for non-constant
33789 strings and correct the expected range for wide non-constant strings.
33791 2017-02-02 Martin Sebor <msebor@redhat.com>
33793 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
33795 PR middle-end/32003
33796 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
33798 (-fdump-tree-@var): Add to index and document how to come up
33799 with pass-specific option and dump file names.
33800 (-fdump-passes): Clarify where to look for output.
33802 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
33804 PR middle-end/77445
33805 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
33806 statistics of the analyzed path; allow threading for speed when
33807 any of BBs along the path are optimized for speed.
33809 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
33811 PR middle-end/78468
33812 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
33813 settings of the virtual registers.
33816 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
33818 * explow.c (get_dynamic_stack_size): Take known alignment of stack
33819 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
33822 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33824 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
33825 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
33827 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33829 * config/s390/s390.md: Add missing comments with the expanded
33831 * config/s390/vector.md: Likewise.
33832 * config/s390/vx-builtins.md: Likewise.
33834 2017-02-02 Jakub Jelinek <jakub@redhat.com>
33837 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
33838 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
33839 conditions on a single line.
33841 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33843 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
33844 __S390_VX__ to __VX__.
33846 2017-02-01 Andrew Pinski <apinski@cavium.com>
33848 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
33849 stmt_info to record_stmt_cost.
33850 (vect_get_known_peeling_cost): Pass stmt_info if known to
33852 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
33853 cpu_vector_cost field into
33854 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
33855 field into vec_int_stmt_cost and vec_fp_stmt_cost.
33856 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
33857 splitting of scalar_stmt_cost and vec_stmt_cost.
33858 (thunderx_vector_cost): Likewise.
33859 (cortexa57_vector_cost): LIkewise.
33860 (exynosm1_vector_cost): Likewise.
33861 (xgene1_vector_cost): Likewise.
33862 (thunderx2t99_vector_cost): Improve after the splitting of the two
33864 (aarch64_builtin_vectorization_cost): Update for the splitting of
33865 scalar_stmt_cost and vec_stmt_cost.
33867 2017-02-01 Torvald Riegel <triegel@redhat.com>
33868 Richard Henderson <rth@redhat.com>
33870 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
33871 conditional on existance of a fast atomic load.
33872 * optabs-query.c (can_atomic_load_p): New function.
33873 * optabs-query.h (can_atomic_load_p): Declare it.
33874 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
33875 no fast atomic load is available for the particular size of access.
33876 (expand_atomic_compare_and_swap): Likewise.
33877 (expand_atomic_load): Likewise.
33878 (expand_atomic_store): Likewise.
33879 (expand_atomic_fetch_op): Likewise.
33880 * testsuite/lib/target-supports.exp
33881 (check_effective_target_sync_int_128): Remove x86 because it provides
33882 no fast atomic load.
33883 (check_effective_target_sync_int_128_runtime): Likewise.
33885 2017-02-01 Richard Biener <rguenther@suse.de>
33887 * graphite.c: Include tree-vectorizer.h for find_loop_location.
33888 (graphite_transform_loops): Provide opt-info for optimized nests.
33889 * tree-parloop.c (parallelize_loops): Provide opt-info for
33890 parallelized loops.
33892 2017-02-01 Richard Biener <rguenther@suse.de>
33894 PR middle-end/79315
33895 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
33896 was not set before.
33898 2017-02-01 Richard Biener <rguenther@suse.de>
33900 PR tree-optimization/71824
33901 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
33902 Verify the loops are valid in the merged SESE region.
33903 (scop_detection::can_represent_loop_1): Check analyzing the
33904 evolution of the number of iterations in the region succeeds.
33906 2017-01-31 Ian Lance Taylor <iant@golang.org>
33908 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
33909 REG_ARGS_SIZE note to 32-bit push insns and call insn.
33911 2017-01-31 David Malcolm <dmalcolm@redhat.com>
33913 PR preprocessor/79210
33914 * input.c (get_substring_ranges_for_loc): Replace line_width
33915 assertion with error-handling.
33917 2017-01-31 Richard Biener <rguenther@suse.de>
33919 PR tree-optimization/77318
33920 * graphite-sese-to-poly.c (extract_affine): Fix assert.
33921 (create_pw_aff_from_tree): Take loop parameter.
33922 (add_condition_to_pbb): Pass loop of the condition to
33923 create_pw_aff_from_tree.
33925 2017-01-31 Jakub Jelinek <jakub@redhat.com>
33927 * config/s390/s390.c (s390_asan_shadow_offset): New function.
33928 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
33930 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
33934 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
33936 (convert_int_to_float128): Likewise.
33937 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
33938 (convert_int_to_float128): Likewise.
33939 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
33940 (UNSPEC_IEEE128_CONVERT): Likewise.
33941 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
33942 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
33943 Use local variables for IBM extended format.
33944 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
33945 (fix_trunc<mode>si2_fprs): Likewise.
33946 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
33947 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
33948 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
33949 to know that we can now have integers of all sizes in vector
33951 (fix<uns>_<mode>di2_hw): Likewise.
33952 (float<uns>_<mode>si2_hw): Likewise.
33953 (fix_<mode>si2_hw): Likewise.
33954 (fixuns_<mode>si2_hw): Likewise.
33955 (float<uns>_<mode>di2_hw): Likewise.
33956 (float_<mode>di2_hw): Likewise.
33957 (float_<mode>si2_hw): Likewise.
33958 (floatuns_<mode>di2_hw): Likewise.
33959 (floatuns_<mode>si2_hw): Likewise.
33960 (xscvqp<su>wz_<mode>): Delete, no longer used.
33961 (xscvqp<su>dz_<mode>): Likewise.
33962 (xscv<su>dqp_<mode>): Likewise.
33963 (ieee128_mfvsrd_64bit): Likewise.
33964 (ieee128_mfvsrd_32bit): Likewise.
33965 (ieee128_mfvsrwz): Likewise.
33966 (ieee128_mtvsrw): Likewise.
33967 (ieee128_mtvsrd_64bit): Likewise.
33968 (ieee128_mtvsrd_32bit): Likewise.
33970 2017-01-31 Martin Liska <mliska@suse.cz>
33973 * ipa-prop.c (ipa_free_all_node_params): Call release method
33974 instead of ~sumbol_summary to not to trigger double times
33977 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
33979 PR tree-optimization/71691
33980 * bitmap.h (class auto_bitmap): New.
33981 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
33982 is_maybe_undefined instead of ssa_undefined_value_p.
33984 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33986 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
33987 __S390_ARCH_LEVEL__ to __ARCH__.
33989 2017-01-31 Jakub Jelinek <jakub@redhat.com>
33991 PR tree-optimization/79267
33992 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
33993 if should_remove_lhs_p is true.
33995 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
33998 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
33999 (add_alignment_attribute): New.
34000 (base_type_die): Add alignment attribute.
34001 (subrange_type_die): Likewise.
34002 (modified_type_die): Likewise.
34003 (gen_array_type_die): Likewise.
34004 (gen_descr_array_type_die: Likewise.
34005 (gen_enumeration_type_die): Likewise.
34006 (gen_subprogram_die): Likewise.
34007 (gen_variable_die): Likewise.
34008 (gen_field_die): Likewise.
34009 (gen_ptr_to_mbr_type_die): Likewise.
34010 (gen_struct_or_union_type_die): Likewise.
34011 (gen_subroutine_type_die): Likewise.
34012 (gen_typedef_die): Likewise.
34013 (base_type_cmp): Compare alignment attribute.
34015 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34018 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
34019 (setb_unsigned) New pattern for setb with CCUNS.
34020 * config/rs6000/rs6000.c (expand_block_compare): Use a different
34021 subfc./subfe sequence to avoid overflow problems. Generate a
34022 shorter sequence with cmpld/setb for power9.
34023 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
34024 for generating subfc. instruction.
34025 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
34026 now uses this instruction.
34028 2017-01-30 Ian Lance Taylor <iant@google.com>
34031 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
34032 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
34034 2017-01-30 Martin Sebor <msebor@redhat.com>
34036 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
34037 Move constant to the right of a relational operator.
34038 (get_mpfr_format_length, format_character, format_string): Ditto.
34039 (should_warn_p, maybe_warn): Same.
34041 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
34043 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
34046 * asan.c (get_translation_unit_decl): Remove function.
34047 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
34049 2017-01-30 Martin Liska <mliska@suse.cz>
34051 PR gcov-profile/79259
34052 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
34053 -fprofile-generate.
34055 2017-01-30 Martin Liska <mliska@suse.cz>
34058 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
34059 Initialize variables with NULL value.
34061 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
34064 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
34066 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
34068 2017-01-30 Richard Biener <rguenther@suse.de>
34070 PR tree-optimization/79276
34071 * tree-vrp.c (process_assert_insertions): Properly adjust common
34072 when removing a duplicate.
34074 2017-01-30 Richard Biener <rguenther@suse.de>
34076 PR tree-optimization/79256
34077 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
34078 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
34080 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
34082 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
34085 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
34086 ("*r<noxa>sbg_<mode>_sll_bitmask")
34087 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
34088 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
34089 Use contiguous_bitmask_nowrap_operand.
34091 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34094 * config/rs6000/altivec.h (vec_xl): Revise #define.
34095 (vec_xst): Likewise.
34097 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
34099 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
34101 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
34103 PR rtl-optimization/79194
34104 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
34105 traps before call to bypass_conditional_jumps.
34107 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
34109 PR tree-optimization/71374
34110 * lra-constraints.c (check_conflict_input_operands): New.
34111 (match_reload): Use it.
34113 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
34116 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
34117 account to calculate conflict_set.
34119 2017-01-27 Bin Cheng <bin.cheng@arm.com>
34121 PR rtl-optimization/78559
34122 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
34123 other_insn in combine.
34125 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
34127 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
34128 uint16_type_node for BT_UINT16.
34130 2017-01-27 David Malcolm <dmalcolm@redhat.com>
34132 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
34133 "RTL Tests" to menu.
34134 (GIMPLE Tests): New node.
34135 (RTL Tests): New node.
34137 2017-01-27 Richard Biener <rguenther@suse.de>
34139 PR tree-optimization/79245
34140 * tree-loop-distribution.c (distribute_loop): Apply cost
34141 modeling also to detected patterns.
34143 2017-01-27 Richard Biener <rguenther@suse.de>
34145 PR tree-optimization/71433
34146 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
34147 (compare_assert_loc): New function.
34148 (process_assert_insertions): Sort and optimize assert locations
34149 to remove duplicates and push down identical assertions on
34150 edges to their destination block.
34152 2017-01-27 Richard Biener <rguenther@suse.de>
34154 PR tree-optimization/79244
34155 * tree-vrp.c (remove_range_assertions): Forcefully propagate
34156 out SSA names even if abnormal.
34158 2017-01-27 Jakub Jelinek <jakub@redhat.com>
34160 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
34161 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
34162 instead of MPFR_RNDN.
34164 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
34167 * arm.c (arm_option_override): Don't call build_target_option_node
34168 until after doing all option overrides.
34169 (arm_valid_target_attribute_tree): Likewise.
34171 2017-01-27 Martin Liska <mliska@suse.cz>
34173 * doc/invoke.texi (-fprofile-arcs): Document profiling support
34174 for {cd}tors and C++ {cd}tors.
34176 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
34178 * config/s390/s390.md ("*setmem_long_and")
34179 ("*setmem_long_and_31z"): Use zero_extend instead of and.
34181 2017-01-26 Martin Sebor <msebor@redhat.com>
34183 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
34186 2017-01-26 Martin Sebor <msebor@redhat.com>
34188 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
34189 HAVE_DFmode before using XFmode or DFmode.
34190 (parse_directive): Avoid using the z length modifier to avoid
34191 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
34193 PR middle-end/78703
34194 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
34195 to accept adjustment as an array.
34196 (get_int_range): New function.
34197 (struct directive): Make width and prec arrays.
34198 (directive::set_width, directive::set_precision): Call get_int_range.
34199 (format_integer, format_floating): Handle width and precision ranges.
34200 (format_string, parse_directive): Same.
34202 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34205 * dwarf2out.c (generate_skeleton_bottom_up): For children with
34206 comdat_type_p set, just clone them, but keep the children in the
34210 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
34211 which have direct callers with -fvar-tracking-assignments enabled
34213 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
34214 inside of type units.
34216 2017-01-26 Martin Sebor <msebor@redhat.com>
34218 PR middle-end/78703
34219 * gimple-ssa-sprintf.c (struct result_range): Add likely and
34221 (struct format_result): Replace number_chars, number_chars_min,
34222 and number_chars_max with a single member of struct result_range.
34224 (format_result::operator+=): Adjust.
34225 (struct fmtresult): Remove bounded. Handle likely and unlikely
34227 (fmtresult::adjust_for_width_or_precision): New function.
34228 (fmtresult:type_max_digits): New function.
34229 (bytes_remaining): Handle likely and unlikely counters.
34230 (min_bytes_remaining): Remove.
34231 (format_percent): Simplify.
34232 (format_integer, format_floating): Set likely and unlikely counters.
34233 (get_string_length, format_character, format_string): Same.
34234 (format_plain, should_warn_p): New function.
34235 (maybe_warn): Call should_warn_p. Update diagnostic messages
34236 and handle those for all directives, including plain strings.
34237 (format_directive): Handle likely and unlikely counters.
34238 Remove unnecessary quoting from diagnostics. Add an informational
34240 (add_bytes): Remove.
34241 (pass_sprintf_length::compute_format_length): Simplify.
34242 (try_substitute_return_value): Handle likely and unlikely counters.
34244 2017-01-26 Carl Love <cel@us.ibm.com>
34246 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
34247 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
34249 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
34252 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
34253 endianess for subregs into account.
34254 * lra-constraints.c (lra_constraints): Do risky transformations
34255 always on the first iteration.
34256 * lra-lives.c (check_pseudos_live_through_calls): Add arg
34257 last_call_used_reg_set.
34258 (process_bb_lives): Define and use last_call_used_reg_set.
34259 * lra.c (lra): Always continue after lra_constraints on the first
34262 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
34264 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
34266 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
34268 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34270 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
34271 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
34272 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
34273 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
34274 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
34275 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
34276 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
34277 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
34278 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
34280 2017-01-26 Marek Polacek <polacek@redhat.com>
34283 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
34284 for the third operand.
34286 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34288 PR middle-end/79236
34289 * omp-low.c (struct omp_context): Add simt_stmt field.
34290 (scan_omp_for): Return omp_context *.
34291 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
34292 context to the _simt_ SIMD stmt.
34293 (lower_omp_for): For combined SIMD with sibling _simt_
34294 SIMD, make sure to use the same decls in _looptemp_
34295 clauses as in the sibling.
34297 2017-01-26 David Sherwood <david.sherwood@arm.com>
34299 PR middle-end/79212
34300 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
34303 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34306 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
34307 emit fld b; fld a; if possible.
34309 * brig-builtins.def: Update copyright years.
34310 * config/arm/arm_acle_builtins.def: Update copyright years.
34312 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
34315 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
34316 constraint instead of o for the stxsd instruction.
34318 2017-01-25 Carl Love <cel@us.ibm.com>
34320 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
34321 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
34323 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
34325 * doc/invoke.texi (C++ Dialect Options): Fix typo.
34327 2017-01-25 Richard Biener <rguenther@suse.de>
34329 PR tree-optimization/69264
34330 * target.def (vector_alignment_reachable): Improve documentation.
34331 * doc/tm.texi: Regenerate.
34332 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
34334 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
34335 earlier changes with respect to TYPE_USER_ALIGN.
34336 (vector_alignment_reachable_p): Likewise. Improve dumping.
34338 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34341 * config/arm/arm.md (xordi3): Force constant operand into a register
34344 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34346 * doc/invoke.texi (-fstore-merging): Correct default optimization
34347 levels at which it is enabled.
34348 (-O): Move -fstore-merging from list to...
34351 2017-01-25 Richard Biener <rguenther@suse.de>
34354 * omp-expand.c: Include debug.h.
34355 (expand_omp_taskreg): Make sure to generate early debug before
34356 outlining anything from a function.
34357 (expand_omp_target): Likewise.
34358 (grid_expand_target_grid_body): Likewise.
34360 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
34363 * asan.c (get_translation_unit_decl): New function.
34364 (asan_add_global): Extract modules file name from globals
34365 TRANSLATION_UNIT_DECL name.
34367 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
34370 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
34371 for long calls with APCS frame and VFP.
34373 2017-01-24 David Malcolm <dmalcolm@redhat.com>
34375 * cfg.c (original_copy_tables_initialized_p): New function.
34376 * cfg.h (original_copy_tables_initialized_p): New decl.
34377 * cfgrtl.c (relink_block_chain): Guard the call to
34378 free_original_copy_tables with a call to
34379 original_copy_tables_initialized_p.
34380 * cgraph.h (symtab_node::native_rtl_p): New decl.
34381 * cgraphunit.c (symtab_node::native_rtl_p): New function.
34382 (symtab_node::needed_p): Don't assert for early assembly output
34383 for __RTL functions.
34384 (cgraph_node::finalize_function): Set "force_output" for __RTL
34386 (cgraph_node::analyze): Bail out early for __RTL functions.
34387 (analyze_functions): Update assertion to support __RTL functions.
34388 (cgraph_node::expand): Bail out early for __RTL functions.
34389 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
34391 * function.h (struct function): Update comment for field
34393 * gimple-expr.c: Include "tree-pass.h".
34394 (gimple_has_body_p): Return false for __RTL functions.
34395 * Makefile.in (OBJS): Add run-rtl-passes.o.
34396 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
34398 (gcc::pass_manager::get_clean_slate): New accessor.
34399 * passes.c: Include "insn-addr.h".
34400 (should_skip_pass_p): Add logging. Update logic for running
34401 "expand" to be compatible with both __GIMPLE and __RTL. Guard
34402 property-provider override so it is only done for gimple passes.
34404 (skip_pass): New function.
34405 (execute_one_pass): Call skip_pass when skipping passes.
34406 * read-md.c (md_reader::read_char): Support filtering
34407 the input to a subset of line numbers.
34408 (md_reader::md_reader): Initialize fields
34409 m_first_line and m_last_line.
34410 (md_reader::read_file_fragment): New function.
34411 * read-md.h (md_reader::read_file_fragment): New decl.
34412 (md_reader::m_first_line): New field.
34413 (md_reader::m_last_line): New field.
34414 * read-rtl-function.c (function_reader::create_function): Only
34415 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
34416 curr_properties. Set DECL_INITIAL to a dummy block.
34417 (read_rtl_function_body_from_file_range): New function.
34418 * read-rtl-function.h (read_rtl_function_body_from_file_range):
34420 * run-rtl-passes.c: New file.
34421 * run-rtl-passes.h: New file.
34423 2017-01-24 Jeff Law <law@redhat.com>
34425 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
34428 2017-01-24 Bin Cheng <bin.cheng@arm.com>
34430 PR tree-optimization/79159
34431 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
34432 (record_nonwrapping_iv): Improve boundary using above function if no
34433 value range information.
34435 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
34436 Martin Jambor <mjambor@suse.cz>
34438 * brig-builtins.def: New file.
34439 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
34440 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
34441 (DEF_HSAIL_SAT_BUILTIN): Likewise.
34442 (DEF_HSAIL_INTR_BUILTIN): Likewise.
34443 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
34444 * builtin-types.def (BT_INT8): New.
34445 (BT_INT16): Likewise.
34446 (BT_UINT8): Likewise.
34447 (BT_UINT16): Likewise.
34448 (BT_FN_ULONG): Likewise.
34449 (BT_FN_UINT_INT): Likewise.
34450 (BT_FN_UINT_ULONG): Likewise.
34451 (BT_FN_UINT_LONG): Likewise.
34452 (BT_FN_UINT_PTR): Likewise.
34453 (BT_FN_ULONG_PTR): Likewise.
34454 (BT_FN_INT8_FLOAT): Likewise.
34455 (BT_FN_INT16_FLOAT): Likewise.
34456 (BT_FN_UINT32_FLOAT): Likewise.
34457 (BT_FN_UINT16_FLOAT): Likewise.
34458 (BT_FN_UINT8_FLOAT): Likewise.
34459 (BT_FN_UINT64_FLOAT): Likewise.
34460 (BT_FN_UINT16_UINT32): Likewise.
34461 (BT_FN_UINT32_UINT16): Likewise.
34462 (BT_FN_UINT16_UINT16_UINT16): Likewise.
34463 (BT_FN_INT_PTR_INT): Likewise.
34464 (BT_FN_UINT_PTR_UINT): Likewise.
34465 (BT_FN_LONG_PTR_LONG): Likewise.
34466 (BT_FN_ULONG_PTR_ULONG): Likewise.
34467 (BT_FN_VOID_UINT64_UINT64): Likewise.
34468 (BT_FN_UINT8_UINT8_UINT8): Likewise.
34469 (BT_FN_INT8_INT8_INT8): Likewise.
34470 (BT_FN_INT16_INT16_INT16): Likewise.
34471 (BT_FN_INT_INT_INT): Likewise.
34472 (BT_FN_UINT_FLOAT_UINT): Likewise.
34473 (BT_FN_FLOAT_UINT_UINT): Likewise.
34474 (BT_FN_ULONG_UINT_UINT): Likewise.
34475 (BT_FN_ULONG_UINT_PTR): Likewise.
34476 (BT_FN_ULONG_ULONG_ULONG): Likewise.
34477 (BT_FN_UINT_UINT_UINT): Likewise.
34478 (BT_FN_VOID_UINT_PTR): Likewise.
34479 (BT_FN_UINT_UINT_PTR: Likewise.
34480 (BT_FN_UINT32_UINT64_PTR): Likewise.
34481 (BT_FN_INT_INT_UINT_UINT): Likewise.
34482 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
34483 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
34484 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
34485 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
34486 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
34487 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
34488 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
34489 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
34490 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
34491 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
34492 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
34493 * doc/frontends.texi: List BRIG FE.
34494 * doc/install.texi (Testing): Add BRIG tesring requirements.
34495 * doc/invoke.texi (Overall Options): Mention BRIG.
34496 * doc/standards.texi (Standards): Doucment BRIG HSA version.
34498 2017-01-24 Richard Biener <rguenther@suse.de>
34500 PR translation/79208
34501 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
34503 2017-01-24 Martin Jambor <mjambor@suse.cz>
34506 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
34507 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
34508 and known_contexts.
34510 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
34512 PR middle-end/79123
34513 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
34514 casts from signed to unsigned really don't have a range.
34516 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
34518 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
34519 GMP_RNDx for compatiblity.
34521 2017-01-24 Martin Liska <mliska@suse.cz>
34524 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
34525 that would prevent us to call alloca with -1 as argument.
34527 2017-01-24 Jakub Jelinek <jakub@redhat.com>
34529 * dwarf2out.c (output_compilation_unit_header, output_file_names):
34530 Avoid -Wformat-security warning.
34532 2017-01-23 Andrew Pinski <apinski@cavium.com>
34534 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
34537 2017-01-23 Martin Sebor <msebor@redhat.com>
34539 PR middle-end/78703
34540 * gimple-ssa-sprintf.c (warn_level): New global.
34541 (format_integer): Use it here and throughout the rest of the file.
34542 Use the same switch to compute sign as base.
34543 (maybe_warn): New function.
34544 (format_directive): Factor out warnings into maybe_warn.
34545 Add debugging output. Use warn_level.
34546 (add_bytes): Use warn_level.
34547 (pass_sprintf_length::compute_format_length): Add debugging output.
34548 (try_substitute_return_value): Same.
34549 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
34551 PR middle-end/78703
34552 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
34553 (struct fmtresult, format_integer, format_floating): Adjust.
34554 (fmtresult::fmtresult): Set max correctly in two argument ctor.
34555 (get_string_length, format_string,format_directive): Same.
34556 (pass_sprintf_length::compute_format_length): Same.
34557 (try_substitute_return_value): Simplify slightly.
34559 PR middle-end/78703
34560 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
34561 (fmtresult::operator+=): Outlined.
34562 (struct fmtresult): Add ctors.
34563 (struct conversion_spec): Rename...
34564 (struct directive): ...to this. Add and remove data members.
34565 (directive::set_width, directive::set_precision): New functions.
34566 (format_percent): Use fmtresult ctor.
34567 (get_width_and_precision): Remove.
34568 (format_integer): Make naming changes. Avoid computing width and
34570 (format_floating): Same. Adjust indentation.
34571 (format_character, format_none): New functions.
34572 (format_string): Moved character handling to format_character.
34573 (format_directive): Remove arguments, change return type.
34574 (parse_directive): New function.
34575 (pass_sprintf_length::compute_format_length): Move directive
34576 parsing to parse_directive.
34578 2017-01-23 Jakub Jelinek <jakub@redhat.com>
34580 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
34581 (assign_assembler_name_if_needed): ... this.
34582 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
34583 (assign_assembler_name_if_needed): ... this.
34584 (free_lang_data_in_cgraph): Adjust callers.
34585 * cgraphunit.c (cgraph_node::analyze): Likewise.
34586 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
34589 2017-01-23 Richard Biener <rguenther@suse.de>
34591 PR tree-optimization/79088
34592 PR tree-optimization/79188
34593 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
34594 resetting loop bounds after last path deletion. Reset loop
34595 bounds of the target loop, make code match the comments.
34596 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
34597 Make sure loops need no fixups.
34599 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34601 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
34602 exponent support with double type for first argument.
34603 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
34604 type returned by __builtin_vec_extract_sig,
34605 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
34606 functions from "vector int" to "vector unsigned int" or from
34607 "vector long long int" to "vector unsigned long long int".
34608 Changed type returned by __builtin_vec_extract_exp,
34609 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
34610 functions from "vector int" to "vector unsigned int" or from
34611 "vector long long int" to "vector unsigned long long int".
34612 Changed return type of __builtin_vec_test_data_class,
34613 __builtin_vec_test_data_class_sp, and
34614 __builtin_vec_test_data_class_dp from "vector int" to
34615 "vector bool int" or from "vector long long int" to "vector bool
34616 long long int" and changed second argument type from "unsigned
34617 int" to "int". Added new overloaded function forms "vector float
34618 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
34619 "vector float __builtin_vec_insert_exp_sp (vector float, vector
34620 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
34621 double, vector unsigned long long int)" and "vector double
34622 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
34623 long int)". Changed return type of
34624 __builtin_scalar_test_data_class and
34625 __builtin_scalar_test_data_class_sp and
34626 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
34627 int" and changed second argument from "unsigned int" to "int".
34628 Changed type returned by __builtin_scalar_test_neg,
34629 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
34630 from "int" to "bool int". Added new overloaded function form
34631 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
34632 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
34633 exponent double-precision with floating point first argument.
34634 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
34635 documentation of scalar_test_data_class, scalar_test_neg,
34636 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
34637 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
34638 vec_test_data_class built-in functions to reflect refinements in
34639 their type signatures.
34641 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
34643 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
34645 (aarch64_elf_asm_destructor): Likewise.
34647 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
34649 PR rtl-optimization/78634
34650 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
34651 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
34652 * ifcvt.c (noce_try_cmove): Add missing cost check.
34654 PR rtl-optimization/71724
34655 * combine.c (if_then_else_cond): Look for situations where it is
34656 beneficial to undo the work of one of the recursive calls.
34658 2017-01-23 Bin Cheng <bin.cheng@arm.com>
34660 PR tree-optimization/70754
34661 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
34662 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
34663 combined stmt before it if not NULL.
34664 (combine_chains): Process refs reversely and compute dominance point
34667 2017-01-23 Martin Liska <mliska@suse.cz>
34669 PR tree-optimization/79196
34670 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
34671 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
34673 (strlen_optimize_stmt): Call the renamed function.
34675 2017-01-23 Michael Matz <matz@suse.de>
34677 PR tree-optimization/78384
34678 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
34680 2017-01-23 Richard Biener <rguenther@suse.de>
34682 PR tree-optimization/79186
34683 * tree-vrp.c (register_new_assert_for): Make sure we've seen
34684 both incoming edges before moving an assert.
34686 2017-01-23 Martin Jambor <mjambor@suse.cz>
34688 * ipa-prop.c (load_from_param_1): Removed.
34689 (load_from_unmodified_param): Bits from load_from_param_1 put back
34691 (load_from_param): Removed.
34692 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
34693 with stmt. Reverted back to use of load_from_unmodified_param.
34695 2017-01-23 Martin Jambor <mjambor@suse.cz>
34698 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
34699 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
34700 field a pointer to garbage collected vector, mark lattices and
34701 ipcp_orig_node with GTY((skip)).
34702 (ipa_get_param_count): Adjust to descriptors being a pointer.
34703 (ipa_get_param): Likewise.
34704 (ipa_get_type): Likewise.
34705 (ipa_get_param_move_cost): Likewise.
34706 (ipa_set_param_used): Likewise.
34707 (ipa_get_controlled_uses): Likewise.
34708 (ipa_set_controlled_uses): Likewise.
34709 (ipa_is_param_used): Likewise.
34710 (ipa_node_params_t): Move into garbage collector. New methods insert
34712 (ipa_node_params_sum): Annotate wth GTY(()).
34713 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
34715 (ipa_load_from_parm_agg): Adjust declaration.
34716 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
34717 * ipa-profile.c (ipa_profile): Likewise.
34718 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
34719 (ipa_populate_param_decls): Make descriptors parameter garbage
34721 (ipa_dump_param): Adjust to descriptors being a pointer.
34722 (ipa_alloc_node_params): Likewise.
34723 (ipa_initialize_node_params): Likewise.
34724 (load_from_param_1): Make descriptors parameter garbage collected.
34725 (load_from_unmodified_param): Likewise.
34726 (load_from_param): Likewise.
34727 (ipa_load_from_parm_agg): Likewise.
34728 (ipa_node_params::~ipa_node_params): Removed.
34729 (ipa_free_all_node_params): Remove call to delete operator.
34730 (ipa_node_params_t::insert): New.
34731 (ipa_node_params_t::remove): Likewise.
34732 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
34733 copy known_csts and known_contexts vectors.
34734 (ipa_read_node_info): Adjust to descriptors being a pointer.
34735 (ipcp_modif_dom_walker): Make m_descriptors field garbage
34737 (ipcp_transform_function): Make descriptors variable garbage
34740 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
34742 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
34743 * config/i386/avx512dqintrin.h: Ditto.
34744 * config/i386/avx512fintrin.h: Ditto.
34745 * config/i386/i386.c: Handle new builtins.
34746 * config/i386/i386-builtin.def: Add new builtins.
34747 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
34748 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
34750 2017-01-23 Jakub Jelinek <jakub@redhat.com>
34751 Martin Liska <mliska@suse.cz>
34753 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
34754 * asan.c (asan_expand_poison_ifn): Support stores and use
34755 appropriate ASAN report function.
34756 * internal-fn.c (expand_ASAN_POISON_USE): New function.
34757 * internal-fn.def (ASAN_POISON_USE): Declare.
34758 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
34759 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
34760 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
34761 ASAN_POISON calls w/o LHS.
34762 * tree-ssa.c (execute_update_addresses_taken): Create clobber
34763 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
34764 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
34765 * gimplify.c (asan_poison_variables): Add attribute
34766 use_after_scope_memory to variables that really needs to live
34768 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
34769 having the attribute.
34771 2017-01-23 Martin Liska <mliska@suse.cz>
34773 * asan.c (create_asan_shadow_var): New function.
34774 (asan_expand_poison_ifn): Likewise.
34775 * asan.h (asan_expand_poison_ifn): New declaration.
34776 * internal-fn.c (expand_ASAN_POISON): Likewise.
34777 * internal-fn.def (ASAN_POISON): New builtin.
34778 * sanopt.c (pass_sanopt::execute): Expand
34779 asan_expand_poison_ifn.
34780 * tree-inline.c (copy_decl_for_dup_finish): Make function
34782 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
34783 * tree-ssa.c (is_asan_mark_p): New function.
34784 (execute_update_addresses_taken): Rewrite local variables
34785 (identified just by use-after-scope as addressable) into SSA.
34787 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
34789 * doc/install.texi (Specific): opensource.apple.com uses https
34790 now. Remove trailing slash.
34792 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
34794 * README.Portability: Remove note on an Irix compatibility issue.
34796 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
34798 * gcov.c (INCLUDE_ALGORITHM): Define.
34799 (INCLUDE_VECTOR): Define.
34800 No longer include <vector> and <algorithm> directly.
34802 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
34804 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
34806 * doc/invoke.texi (Code Gen Options): Ditto.
34808 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
34811 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
34813 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
34815 rtl-optimization/79125
34816 * cprop.c (local_cprop_pass): Handle cases where we make an
34817 unconditional trap.
34819 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
34823 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
34824 read from, for big endian.
34826 2017-01-20 Jiong Wang <jiong.wang@arm.com>
34828 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
34829 register pauth builtins for LP64 only.
34831 2017-01-20 Marek Polacek <polacek@redhat.com>
34834 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
34837 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
34839 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
34841 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
34842 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
34843 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
34845 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34848 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
34849 in big-endian mode when they are not a single duplicated value.
34851 2017-01-20 Richard Biener <rguenther@suse.de>
34853 * BASE-VER: Bump to 7.0.1.
34855 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
34857 * omp-low.c (omplow_simd_context): New struct. Use it...
34858 (lower_rec_simd_input_clauses): ...here and...
34859 (lower_rec_input_clauses): ...here to hold common data. Adjust all
34860 references to idx, lane, max_vf, is_simt.
34862 2017-01-20 Graham Markall <graham.markall@embecosm.com>
34864 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
34867 2017-01-20 Martin Jambor <mjambor@suse.cz>
34869 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
34870 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
34872 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
34873 (GTFILES): Rename hsa.c to hsa-common.c.
34874 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
34875 * hsa-dump.c: Likewise.
34876 * hsa-gen.c: Likewise.
34877 * hsa-regalloc.c: Likewise.
34878 * ipa-hsa.c: Likewise.
34879 * omp-expand.c: Likewise.
34880 * omp-low.c: Likewise.
34881 * toplev.c: Likewise.
34883 2017-01-20 Marek Polacek <polacek@redhat.com>
34886 * doc/invoke.texi: Document -Wduplicated-branches.
34887 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
34888 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
34889 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
34890 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
34891 return 0 only when not OEP_LEXICOGRAPHIC.
34892 (fold_build_cleanup_point_expr): Use the expression
34893 location when building CLEANUP_POINT_EXPR.
34894 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
34895 * tree.c (add_expr): Handle error_mark_node.
34897 2017-01-20 Martin Liska <mliska@suse.cz>
34900 * tree-profile.c (init_ic_make_global_vars): Do not call
34902 (gimple_init_gcov_profiler): Likewise.
34904 2017-01-20 Martin Liska <mliska@suse.cz>
34907 * cgraph.h (maybe_create_reference): Remove argument and
34909 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
34911 * ipa-cp.c (create_specialized_node): Likewise.
34912 * symtab.c (symtab_node::maybe_create_reference): Handle
34913 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
34915 2017-01-20 Martin Liska <mliska@suse.cz>
34917 * read-rtl-function.c (function_reader::create_function): Use
34918 build_decl instread of build_decl_stat.
34920 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
34922 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
34923 * config/i386/avx512dqintrin.h: Ditto.
34924 * config/i386/avx512fintrin.h: Ditto.
34925 * config/i386/i386-builtin-types.def: Add new types.
34926 * config/i386/i386.c: Handle new types.
34927 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
34928 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
34929 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
34930 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
34931 (__builtin_ia32_kshiftridi): New.
34932 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
34934 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
34938 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
34939 define to rs6000_init_stack_protect_guard.
34940 (rs6000_init_stack_protect_guard): New function.
34942 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
34943 Yunqiang Su <yunqiang.su@imgtec.com>
34945 * config.gcc (supported_defaults): Add madd4.
34946 (with_madd4): Add validation.
34947 (all_defaults): Add madd4.
34948 * config/mips/mips.opt (mmadd4): New option.
34949 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
34951 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
34953 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
34954 (ISA_HAS_FUSED_MADD4): Likewise.
34955 * doc/invoke.texi (-mmadd4): Document the new option.
34956 * doc/install.texi (--with-madd4): Document the new option.
34958 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34960 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
34961 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
34962 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
34963 (aarch64_init_pauth_hint_builtins): New.
34964 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
34965 (aarch64_expand_builtin): Expand new builtins.
34967 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34969 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
34970 * combine-stack-adj.c (no_unhandled_cfa): Handle
34971 REG_CFA_TOGGLE_RA_MANGLE.
34972 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
34973 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
34974 info for return address signing.
34975 (aarch64_expand_epilogue): Likewise.
34977 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34979 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
34980 * config/aarch64/aarch64-protos.h
34981 (aarch64_return_address_signing_enabled): New declaration.
34982 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
34984 (aarch64_expand_prologue): Sign return address before it's pushed onto
34986 (aarch64_expand_epilogue): Authenticate return address fetched from
34988 (aarch64_override_options): Sanity check for ILP32 and ISA level.
34989 (aarch64_attributes): New function attributes for "sign-return-address".
34990 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
34991 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
34992 ("*do_return"): Generate combined instructions according to key index.
34993 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
34994 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
34996 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
34997 * config/aarch64/aarch64.opt (msign-return-address=): New.
34998 * doc/extend.texi (AArch64 Function Attributes): Documents
34999 "sign-return-address=".
35000 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
35002 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
35004 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
35005 overall option summary.
35007 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35009 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
35010 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
35011 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
35012 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
35014 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
35016 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
35017 -mpower9-minmax by default for -mcpu=power9.
35018 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
35019 128-bit floating point.
35021 2017-01-20 Alan Modra <amodra@gmail.com>
35023 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
35024 optimizing for size.
35026 2017-01-20 Alan Modra <amodra@gmail.com>
35029 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
35030 for strcmp and strncmp from corresponding builtin decl.
35032 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
35034 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
35035 instead of i386/rtems-64.h.
35036 * config/i386/rtems-64.h: Remove.
35038 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
35042 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
35044 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
35046 2017-01-19 Tamar Christina <tamar.christina@arm.com>
35048 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
35049 Change int to HOST_WIDE_INT.
35050 * config/aarch64/aarch64-protos.h
35051 (aarch64_simd_gen_const_vector_dup): Likewise.
35052 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
35054 2017-01-19 David Malcolm <dmalcolm@redhat.com>
35056 * langhooks-def.h (lhd_type_for_size): New decl.
35057 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
35058 * langhooks.c (lhd_type_for_size): New function, taken from
35061 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
35063 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
35064 define_bypass for CR latency.
35065 (power9-cracked-alu): Update bypass latency and remove power9-branch.
35066 (power9-alu2): Add define_bypass for CR latency.
35068 (power9-mul): Update insn latency.
35069 (power9-mul-compare): Update insn latency, bypass latency and remove
35072 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35074 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
35076 * config/aarch64/aarch64.md
35077 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
35078 aarch64_nopcrelative_literal_loads.
35079 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
35081 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
35083 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
35084 TARGET_LOONGSON_3A.
35085 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
35087 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
35090 * config.gcc (supported_defaults): Add lxc1-sxc1.
35091 (with_lxc1_sxc1): Add validation.
35092 (all_defaults): Add lxc1-sxc1.
35093 * config/mips/mips.opt (mlxc1-sxc1): New option.
35094 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
35096 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
35097 __mips_no_lxc1_sxc1.
35098 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
35099 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
35100 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
35102 2017-01-19 Richard Biener <rguenther@suse.de>
35104 PR tree-optimization/72488
35105 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
35106 sure to restore SSA info.
35107 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
35109 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
35111 PR rtl-optimization/79121
35112 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
35113 of the inner type when shifting an extended value.
35115 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
35118 * symtab.c (symtab_node::equal_address_to): Fix comparing of
35119 interposable aliases.
35121 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
35124 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
35125 Use the evmergelohi instruction.
35126 (mov_si<mode>_e500_subreg4_2_le): Likewise.
35127 (mov_sitf_e500_subreg8_2_be): Likewise.
35128 (mov_sitf_e500_subreg12_2_le): Likewise.
35129 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
35130 (mov_si<mode>_e500_subreg4_2_be): Likewise.
35131 (mov_sitf_e500_subreg8_2_le): Likewise.
35132 (mov_sitf_e500_subreg12_2_be): Likewise.
35134 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35136 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
35137 attribute from vecsimple to vecperm.
35138 (altivec_vbpermq2): Likewise.
35140 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35143 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
35145 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35146 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
35147 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
35148 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
35149 case where N arg is SIZE_MAX.
35150 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
35151 (cmpstrsi): Add pattern.
35153 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
35155 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
35156 __builtin_vec_revb builtins.
35157 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
35158 built-in functions to support generation of the ISA 3.0 XXBR<x>
35159 vector byte reverse instructions.
35160 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
35161 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
35162 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
35163 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
35164 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
35165 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
35166 (P9V_BUILTIN_VEC_REVB): Likewise.
35167 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
35168 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
35169 (p9_xxbrq_v16qi): Likewise.
35170 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
35171 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
35172 (p9_xxbrh_v8hi): Likewise.
35173 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
35174 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
35175 vec_revb built-in functions.
35177 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
35179 PR rtl-optimization/78952
35180 * config/i386/i386.md (any_extract): New code iterator.
35181 (*insvqi_2): Use any_extract for source operand.
35182 (*insvqi_3): Use any_shiftrt for source operand.
35184 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
35186 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
35188 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
35190 2017-01-18 Matthias Klose <doko@ubuntu.com>
35192 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
35194 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35196 * config/rs6000/altivec.h (vec_bperm): Change #define.
35197 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
35198 (altivec_vbpermq2): New define_insn.
35199 (altivec_vbpermd): Likewise.
35200 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
35201 function interface.
35202 (VBPERMD): Likewise.
35203 (VBPERM): New polymorphic function interface.
35204 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
35205 Add entries for P9V_BUILTIN_VEC_VBPERM.
35206 * doc/extend.texi: Add interfaces for vec_bperm.
35208 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35210 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
35211 first letter of error messages.
35212 (s390_resolve_overloaded_builtin): Likewise.
35213 * config/s390/s390.c (s390_expand_builtin): Likewise.
35214 (s390_invalid_arg_for_unprototyped_fn): Likewise.
35215 (s390_valid_target_attribute_inner_p): Likewise.
35216 * config/s390/s390.md ("tabort"): Likewise.
35218 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
35220 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
35221 (ISA_AVOID_DIV_HILO): New macro.
35222 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
35223 (ISA_HAS_DDIV): Likewise.
35225 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35227 * doc/invoke.texi (fabi-version): Correct number of occurrences.
35229 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35231 * doc/invoke.texi (fabi-version): Spelling fix.
35233 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35236 * doc/invoke.texi (fabi-version): Mention mangling fix for
35239 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35242 * doc/invoke.texi (fabi-version): Document discriminator mangling.
35244 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
35247 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
35248 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
35250 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
35252 (stack_protect_test): Ditto.
35253 * config/rs6000/rs6000.opt (mstack-protector-guard=,
35254 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
35256 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
35257 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
35258 -mstack-protector-guard-offset=.
35259 (RS/6000 and PowerPC Options): Ditto.
35261 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
35263 * config/i386/i386.h (MASK_CLASS_P): New define.
35264 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
35265 there are no registers from different register sets also when
35266 mask registers are used. Update function comment.
35267 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
35268 to (*k/*r) and (*k/*km) alternatives.
35270 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
35272 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
35273 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
35274 (EH_RETURN_HANDLER_RTX): New define.
35275 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
35276 Force frame pointer in EH return functions.
35277 (aarch64_expand_epilogue): Add barrier for eh_return.
35278 (aarch64_final_eh_return_addr): Remove.
35279 (aarch64_eh_return_handler_rtx): New function.
35280 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
35282 (aarch64_eh_return_handler_rtx): New prototype.
35284 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35286 * config/rs6000/altivec.h (vec_rlmi): New #define.
35287 (vec_vrlnm): Likewise.
35288 (vec_rlnm): Likewise.
35289 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
35290 (UNSPEC_VRLNM): Likewise.
35291 (VIlong): New mode iterator.
35292 (altivec_vrl<VI_char>mi): New define_insn.
35293 (altivec_vrl<VI_char>nm): Likewise.
35294 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
35296 (VRLDNM): Likewise.
35297 (RLNM): New polymorphic function entry.
35298 (VRLWMI): New monomorphic function entry.
35299 (VRLDMI): Likewise.
35300 (RLMI): New polymorphic function entry.
35301 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
35302 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
35303 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
35306 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35309 * dwarf2out.c (field_byte_offset): Restore the
35310 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
35311 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
35312 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
35315 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
35318 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
35320 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35323 * dwarf2out.c (add_data_member_location_attribute): For constant
35324 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
35325 instead of DW_AT_data_member_location, DW_AT_bit_offset and
35326 DW_AT_byte_size attributes.
35328 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
35330 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
35331 after forcing to constant memory when the code model is medium.
35333 2017-01-17 Julia Koval <julia.koval@intel.com>
35336 * config/i386/avx512fintrin.h
35337 (_mm512_i32gather_ps): Change __addr type to void const*.
35338 (_mm512_mask_i32gather_ps): Ditto.
35339 (_mm512_i32gather_pd): Ditto.
35340 (_mm512_mask_i32gather_pd): Ditto.
35341 (_mm512_i64gather_ps): Ditto.
35342 (_mm512_mask_i64gather_ps): Ditto.
35343 (_mm512_i64gather_pd): Ditto.
35344 (_mm512_mask_i64gather_pd): Ditto.
35345 (_mm512_i32gather_epi32): Ditto.
35346 (_mm512_mask_i32gather_epi32): Ditto.
35347 (_mm512_i32gather_epi64): Ditto.
35348 (_mm512_mask_i32gather_epi64): Ditto.
35349 (_mm512_i64gather_epi32): Ditto.
35350 (_mm512_mask_i64gather_epi32): Ditto.
35351 (_mm512_i64gather_epi64): Ditto.
35352 (_mm512_mask_i64gather_epi64): Ditto.
35353 (_mm512_i32scatter_ps): Change __addr type to void*.
35354 (_mm512_mask_i32scatter_ps): Ditto.
35355 (_mm512_i32scatter_pd): Ditto.
35356 (_mm512_mask_i32scatter_pd): Ditto.
35357 (_mm512_i64scatter_ps): Ditto.
35358 (_mm512_mask_i64scatter_ps): Ditto.
35359 (_mm512_i64scatter_pd): Ditto.
35360 (_mm512_mask_i64scatter_pd): Ditto.
35361 (_mm512_i32scatter_epi32): Ditto.
35362 (_mm512_mask_i32scatter_epi32): Ditto.
35363 (_mm512_i32scatter_epi64): Ditto.
35364 (_mm512_mask_i32scatter_epi64): Ditto.
35365 (_mm512_i64scatter_epi32): Ditto.
35366 (_mm512_mask_i64scatter_epi32): Ditto.
35367 (_mm512_i64scatter_epi64): Ditto.
35368 (_mm512_mask_i64scatter_epi64): Ditto.
35369 * config/i386/avx512pfintrin.h
35370 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
35371 (_mm512_mask_prefetch_i32gather_ps): Ditto.
35372 (_mm512_mask_prefetch_i64gather_pd): Ditto.
35373 (_mm512_mask_prefetch_i64gather_ps): Ditto.
35374 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
35375 (_mm512_prefetch_i32scatter_ps): Ditto.
35376 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
35377 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
35378 (_mm512_prefetch_i64scatter_pd): Ditto.
35379 (_mm512_prefetch_i64scatter_ps): Ditto.
35380 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
35381 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
35382 * config/i386/avx512vlintrin.h
35383 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
35384 (_mm_mmask_i32gather_ps): Ditto.
35385 (_mm256_mmask_i32gather_pd): Ditto.
35386 (_mm_mmask_i32gather_pd): Ditto.
35387 (_mm256_mmask_i64gather_ps): Ditto.
35388 (_mm_mmask_i64gather_ps): Ditto.
35389 (_mm256_mmask_i64gather_pd): Ditto.
35390 (_mm_mmask_i64gather_pd): Ditto.
35391 (_mm256_mmask_i32gather_epi32): Ditto.
35392 (_mm_mmask_i32gather_epi32): Ditto.
35393 (_mm256_mmask_i32gather_epi64): Ditto.
35394 (_mm_mmask_i32gather_epi64): Ditto.
35395 (_mm256_mmask_i64gather_epi32): Ditto.
35396 (_mm_mmask_i64gather_epi32): Ditto.
35397 (_mm256_mmask_i64gather_epi64): Ditto.
35398 (_mm_mmask_i64gather_epi64): Ditto.
35399 (_mm256_i32scatter_ps): Change __addr type to void*.
35400 (_mm256_mask_i32scatter_ps): Ditto.
35401 (_mm_i32scatter_ps): Ditto.
35402 (_mm_mask_i32scatter_ps): Ditto.
35403 (_mm256_i32scatter_pd): Ditto.
35404 (_mm256_mask_i32scatter_pd): Ditto.
35405 (_mm_i32scatter_pd): Ditto.
35406 (_mm_mask_i32scatter_pd): Ditto.
35407 (_mm256_i64scatter_ps): Ditto.
35408 (_mm256_mask_i64scatter_ps): Ditto.
35409 (_mm_i64scatter_ps): Ditto.
35410 (_mm_mask_i64scatter_ps): Ditto.
35411 (_mm256_i64scatter_pd): Ditto.
35412 (_mm256_mask_i64scatter_pd): Ditto.
35413 (_mm_i64scatter_pd): Ditto.
35414 (_mm_mask_i64scatter_pd): Ditto.
35415 (_mm256_i32scatter_epi32): Ditto.
35416 (_mm256_mask_i32scatter_epi32): Ditto.
35417 (_mm_i32scatter_epi32): Ditto.
35418 (_mm_mask_i32scatter_epi32): Ditto.
35419 (_mm256_i32scatter_epi64): Ditto.
35420 (_mm256_mask_i32scatter_epi64): Ditto.
35421 (_mm_i32scatter_epi64): Ditto.
35422 (_mm_mask_i32scatter_epi64): Ditto.
35423 (_mm256_i64scatter_epi32): Ditto.
35424 (_mm256_mask_i64scatter_epi32): Ditto.
35425 (_mm_i64scatter_epi32): Ditto.
35426 (_mm_mask_i64scatter_epi32): Ditto.
35427 (_mm256_i64scatter_epi64): Ditto.
35428 (_mm256_mask_i64scatter_epi64): Ditto.
35429 (_mm_i64scatter_epi64): Ditto.
35430 (_mm_mask_i64scatter_epi64): Ditto.
35431 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
35432 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
35433 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
35434 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
35435 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
35436 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
35437 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
35438 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
35439 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
35440 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
35441 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
35442 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
35443 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
35444 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
35445 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
35446 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
35447 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
35448 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
35449 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
35450 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
35451 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
35452 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
35453 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
35454 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
35455 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
35456 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
35457 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
35458 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
35459 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
35460 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
35461 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
35462 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
35463 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
35464 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
35465 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
35466 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
35467 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
35468 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
35469 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
35470 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
35471 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
35472 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
35473 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
35474 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
35475 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
35476 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
35477 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
35478 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
35479 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
35480 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
35481 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
35482 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
35483 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
35484 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
35485 definitions accordingly.
35487 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
35488 Kuan-Lin Chen <kuanlinchentw@gmail.com>
35491 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
35494 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
35497 * ira-conflicts.c (ira_build_conflicts): Update total conflict
35498 hard regs for inner regno.
35500 2017-01-17 Martin Liska <mliska@suse.cz>
35503 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
35504 assumption and add comment.
35506 2017-01-17 Nathan Sidwell <nathan@acm.org>
35508 * ipa-visibility.c (localize_node): New function, broken out of ...
35509 (function_and_variable_visibility): ... here. Call it.
35511 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
35513 PR middle-end/77445
35514 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
35515 correctly set frequency of oudgoing edge.
35516 (duplicate_thread_path): Fix profile updating.
35518 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35521 * configure.ac: Add GCC_BASE_VER.
35522 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
35523 version from BASE-VER file.
35524 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
35525 (gcc.o): Depend on $(BASEVER).
35526 * common.opt (dumpfullversion): New option.
35527 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
35528 * doc/invoke.texi: Document -dumpfullversion.
35529 * doc/install.texi: Document --with-gcc-major-version-only.
35530 * configure: Regenerated.
35532 2017-01-17 Richard Biener <rguenther@suse.de>
35534 PR tree-optimization/71433
35535 * tree-vrp.c (register_new_assert_for): Merge same asserts
35536 on all incoming edges.
35537 (process_assert_insertions_for): Handle insertions at the
35540 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
35542 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
35543 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
35545 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
35548 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
35551 2017-01-17 Alan Modra <amodra@gmail.com>
35554 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
35555 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
35556 symbolic stack limit when pic.
35558 2017-01-16 Martin Sebor <msebor@redhat.com>
35560 PR tree-optimization/78608
35561 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
35563 2017-01-16 Jeff Law <law@redhat.com>
35566 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
35567 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
35568 for several include directories that may be relative to sysroot.
35569 * config/i386/x-mingw32 (gplus_includedir): Define.
35570 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
35571 (native_system_includedir): Likewise.
35572 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
35573 override if TARGET_SYSTEM_ROOT is defined.
35574 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
35576 PR tree-optimization/79090
35577 PR tree-optimization/33562
35578 PR tree-optimization/61912
35579 PR tree-optimization/77485
35580 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
35581 and computed trims into the dump file.
35583 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
35585 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
35587 2017-01-16 Jakub Jelinek <jakub@redhat.com>
35590 * gimplify.c (gimplify_init_constructor): If want_value and
35591 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
35595 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
35596 sequence. Formatting fixes.
35597 (doloop_optimize): Formatting fixes.
35600 * gcc.c (debug_level_greater_than_spec_func): New function.
35601 (static_spec_functions): Add debug-level-gt spec function.
35602 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
35604 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
35605 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
35606 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
35607 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
35608 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
35609 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
35611 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
35613 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
35614 QImode fixups to general and mask registers only.
35616 2017-01-16 Carl Love <cel@us.ibm.com>
35618 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
35619 for built-in functions
35620 vector signed char vec_nabs (vector signed char)
35621 vector signed short vec_nabs (vector signed short)
35622 vector signed int vec_nabs (vector signed int)
35623 vector signed long long vec_nabs (vector signed long long)
35624 vector float vec_nabs (vector float)
35625 vector double vec_nabs (vector double)
35626 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
35628 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
35629 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
35630 * doc/extend.texi: Update the documentation file for the new built-in
35633 2017-01-16 Martin Sebor <msebor@redhat.com>
35635 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
35638 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35640 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
35641 UNSPEC_VSX__XXSPLTD to require special splat handling.
35643 2017-01-16 David Malcolm <dmalcolm@redhat.com>
35646 * system.h: Poison strndup.
35648 2017-01-16 Alan Modra <amodra@gmail.com>
35651 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
35654 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
35656 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
35658 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
35660 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
35661 call recog here. Assert that INSN_CODE (insn) is non-negative.
35663 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
35666 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
35668 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
35669 in the currently scheduled RTL fragment.
35671 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
35673 PR rtl-optimization/78751
35674 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
35677 2017-01-14 Jeff Law <law@redhat.com>
35679 PR tree-optimization/79090
35680 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
35681 variable length stores.
35682 (compute_trims): Delete dead assignment to *trim_tail.
35683 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
35686 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
35688 PR rtl-optimization/78626
35689 PR rtl-optimization/78727
35690 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
35691 of a block, and split such blocks after everything else is finished.
35693 2017-01-14 Alan Modra <amodra@gmail.com>
35696 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
35697 target legitimate_combined_insn.
35698 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
35699 (rs6000_legitimate_combined_insn): New function.
35700 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
35702 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
35703 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
35704 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
35706 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
35708 * doc/frontends.texi (G++ and GCC): Remove references to Java.
35710 2017-01-13 Jeff Law <law@redhat.com>
35712 PR tree-optimization/33562
35713 PR tree-optimization/61912
35714 PR tree-optimization/77485
35715 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
35717 (delete_dead_assignment): Likewise.
35718 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
35719 statement to delete_dead_call and delete_dead_assignment.
35721 2017-01-13 David Malcolm <dmalcolm@redhat.com>
35724 * substring-locations.c (format_warning_va): Strengthen case 1 so
35725 that both endpoints of the substring must be within the format
35726 range for just the substring to be printed.
35728 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
35730 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
35731 * config/i386/i386.c (ix86_target_string): Add missing options
35732 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
35733 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
35734 flags_other and ix86_target_other to flags2_other. Display unknown
35736 (ix86_valid_target_attribute_inner_p): Add missing options and
35737 reorder options by implied ISAs, as in ix86_target_string.
35739 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
35741 * hash-table.h (hash_table::too_empty_p): New function.
35742 (hash_table::expand): Use it.
35743 (hash_table::traverse): Likewise.
35744 (hash_table::empty_slot): Use sizeof (value_type) instead of
35745 sizeof (PTR) to convert bytes to elements. Shrink the table
35746 if the current size is excessive for the current number of
35749 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
35751 * ira-costs.c (record_reg_classes): Break from the inner loop
35752 early once alt_fail is known to be true. Update outer loop
35753 handling accordingly.
35755 2017-01-13 Jeff Law <law@redhat.com>
35757 * tree-ssa-dse.c (decrement_count): New function.
35758 (increment_start_addr, maybe_trim_memstar_call): Likewise.
35759 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
35760 when we know the partially dead statement is a mem* function.
35762 PR tree-optimization/61912
35763 PR tree-optimization/77485
35764 * tree-ssa-dse.c: Include expr.h.
35765 (maybe_trim_constructor_store): New function.
35766 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
35768 PR tree-optimization/33562
35769 PR tree-optimization/61912
35770 PR tree-optimization/77485
35771 * doc/invoke.texi: Document new dse-max-object-size param.
35772 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
35773 * tree-ssa-dse.c: Include params.h.
35774 (dse_store_status): New enum.
35775 (initialize_ao_ref_for_dse): New, partially extracted from
35777 (valid_ao_ref_for_dse, normalize_ref): New.
35778 (setup_live_bytes_from_ref, compute_trims): Likewise.
35779 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
35780 (maybe_trim_partially_dead_store): Likewise.
35781 (maybe_trim_complex_store): Likewise.
35782 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
35783 Track what bytes live from the original store. Return tri-state
35784 for dead, partially dead or live.
35785 (dse_dom_walker): Add constructor, destructor and new private members.
35786 (delete_dead_call, delete_dead_assignment): New extracted from
35788 (dse_optimize_stmt): Make a member of dse_dom_walker.
35789 Use initialize_ao_ref_for_dse.
35791 PR tree-optimization/33562
35792 PR tree-optimization/61912
35793 PR tree-optimization/77485
35794 * sbitmap.h (bitmap_count_bits): Prototype.
35795 (bitmap_clear_range, bitmap_set_range): Likewise.
35796 * sbitmap.c (bitmap_clear_range): New function.
35797 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
35799 2017-01-13 Martin Liska <mliska@suse.cz>
35802 * function.c (set_cfun): Add new argument force.
35803 * function.h (set_cfun): Likewise.
35804 * ipa-inline-transform.c (inline_call): Use the function when
35805 strict alising from is dropped for function we inline to.
35807 2017-01-13 Richard Biener <rguenther@suse.de>
35809 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
35810 for dumping GIMPLE INTEGER_CSTs.
35812 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
35814 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
35815 to 201112L since C++17.
35817 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
35820 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
35821 if -fsanitize=kernel-address is present.
35823 2017-01-13 Richard Biener <rguenther@suse.de>
35825 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
35826 as _Literal ( type ) number in case usual suffixes do not
35827 preserve all information.
35829 2017-01-13 Richard Biener <rguenther@suse.de>
35831 PR tree-optimization/77283
35832 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
35833 and ssa-iterators.h.
35834 (is_feasible_trace): Implement a cost model based on joiner
35837 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
35840 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
35841 char or short to __float128/_Float128 directly.
35843 2017-01-12 Martin Sebor <msebor@redhat.com>
35845 to -Wformat-overflow.
35846 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
35847 (min_bytes_remaining): Same.
35848 (get_string_length): Same.
35849 (format_string): Same.
35850 (format_directive): Same.
35852 (pass_sprintf_length::handle_gimple_call): Same.
35854 2017-01-12 Jakub Jelinek <jakub@redhat.com>
35856 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
35857 info.nowrite calls with no lhs that can't throw. Return bool
35858 whether gsi_remove has been called or not.
35859 (pass_sprintf_length::handle_gimple_call): Return bool whether
35860 try_substitute_return_value called gsi_remove. Formatting fix.
35861 (pass_sprintf_length::execute): Don't use gsi_remove if
35862 handle_gimple_call returned true.
35865 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
35866 be removed due to side-effects, don't remove following barrier nor
35867 turn the successor edge into fallthru edge.
35869 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35872 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
35873 element-reversing loads and stores as not swappable.
35875 2017-01-12 Nathan Sidwell <nathan@acm.org>
35876 Nicolai Stange <nicstange@gmail.com>
35878 * combine.c (try_combine): Don't ignore result of overlap checking
35879 loop. Combine overlap & asm check into single loop.
35881 2017-01-12 Richard Biener <rguenther@suse.de>
35883 * tree-pretty-print.c (dump_generic_node): Provide -gimple
35884 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
35886 2017-01-12 Richard Biener <rguenther@suse.de>
35888 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
35889 and TS_TARGET_OPTION directly derive from TS_BASE.
35890 * tree-core.h (tree_optimization_option): Derive from tree_base.
35891 (tree_target_option): Likewise.
35893 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
35895 * config/i386/i386.c (memory_address_length): Increase len
35896 only when rip_relative_addr_p returns false.
35898 2017-01-11 Julia Koval <julia.koval@intel.com>
35900 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
35901 (OPTION_MASK_ISA_SGX_SET): New.
35902 (ix86_handle_option): Handle OPT_msgx.
35903 * config.gcc: Added sgxintrin.h.
35904 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
35905 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
35906 * config/i386/i386.c (ix86_target_string): Add -msgx.
35908 (ix86_option_override_internal): Handle new options.
35909 (ix86_valid_target_attribute_inner_p): Add sgx.
35910 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
35911 * config/i386/i386.opt: Add msgx.
35912 * config/i386/sgxintrin.h: New file.
35913 * config/i386/x86intrin.h: Add sgxintrin.h.
35915 2017-01-11 Jakub Jelinek <jakub@redhat.com>
35918 * fold-const.c (maybe_nonzero_address): Return 1 for function
35920 (tree_single_nonzero_warnv_p): Don't handle function local objects
35924 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
35927 2017-01-11 David Malcolm <dmalcolm@redhat.com>
35930 * opts.c: Include "spellcheck.h"
35931 (struct string_fragment): New struct.
35932 (struct edit_distance_traits<const string_fragment &>): New
35934 (get_closest_sanitizer_option): New function.
35935 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
35937 2017-01-11 Jakub Jelinek <jakub@redhat.com>
35939 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
35941 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
35942 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
35943 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
35944 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
35945 for initial die_offset if dwarf_split_debug_info.
35946 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
35947 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
35949 (output_skeleton_debug_sections): Formatting fix. Use
35950 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
35951 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
35953 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
35955 * config/arm/cortex-a53.md: Add bypasses for
35956 cortex_a53_r2f_cvt.
35957 (cortex_a53_r2f): Only use for transfers.
35958 (cortex_a53_f2r): Likewise.
35959 (cortex_a53_r2f_cvt): Add reservation for conversions.
35960 (cortex_a53_f2r_cvt): Likewise.
35962 2017-01-11 Tamar Christina <tamar.christina@arm.com>
35964 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
35965 to all inlined functions, change static to extern.
35967 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
35970 * config/arm/arm.c (legitimize_pic_address): Handle reference to
35972 (arm_assemble_integer): Likewise.
35974 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
35976 * config.gcc: Use new awk script to check CPU, FPU and architecture
35977 parameters for --with-... options.
35978 * config/arm/parsecpu.awk: New file
35979 * config/arm/arm-cpus.in: New file.
35980 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
35982 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
35984 * config/arm/t-arm: Update dependency rules.
35985 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
35986 of processing .def files.
35987 * config/arm/genopt.sh: Deleted.
35988 * config/arm/gentune.sh: Deleted.
35989 * config/arm/arm-cores.def: Deleted.
35990 * config/arm/arm-arches.def: Deleted.
35991 * config/arm/arm-fpus.def: Deleted.
35992 * config/arm/arm-tune.md: Regenerated.
35993 * config/arm/arm-tables.opt: Regenerated.
35994 * config/arm/arm-cpu.h: New generated file.
35995 * config/arm/arm-cpu-data.h: New generated file.
35996 * config/arm/arm-cpu-cdata.h: New generated file.
35998 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
36001 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
36003 (input_varpool_node): Unpack dynamically_initialized bit.
36005 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
36007 PR rtl-optimization/79032
36008 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
36009 the alignment of the adjusted memory reference against that of MODE,
36010 instead of the alignment of the original memory reference.
36012 2017-01-11 Martin Jambor <mjambor@suse.cz>
36014 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
36016 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
36017 decorated functions.
36019 2017-01-11 Richard Biener <rguenther@suse.de>
36021 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
36022 set range/nonnull info for PHI results. Do not set it on
36023 stmts marked for removal.
36025 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
36027 * expr.c (store_field): In the bitfield case, fetch the return value
36028 from the registers before applying a single big-endian adjustment.
36029 Always do a final load for a BLKmode value not larger than a word.
36031 2017-01-10 David Malcolm <dmalcolm@redhat.com>
36034 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
36035 that we correctly handle column numbers greater than
36036 LINE_MAP_MAX_COLUMN_NUMBER.
36038 2017-01-10 Martin Sebor <msebor@redhat.com>
36040 PR middle-end/78245
36041 * gimple-ssa-sprintf.c (get_destination_size): Call
36042 {init,fini}object_sizes.
36043 * tree-object-size.c (addr_object_size): Adjust.
36044 (pass_through_call): Adjust.
36045 (pass_object_sizes::execute): Adjust.
36046 * tree-object-size.h (fini_object_sizes): Declare.
36048 2017-01-10 Martin Sebor <msebor@redhat.com>
36050 PR tree-optimization/78775
36051 * builtins.c (get_size_range): Move...
36052 * calls.c: ...to here.
36053 (alloc_max_size): Accept zero argument.
36054 (operand_signed_p): Remove.
36055 (maybe_warn_alloc_args_overflow): Call get_size_range.
36056 * calls.h (get_size_range): Declare.
36058 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
36060 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
36061 from TI's devices.csv file as of September 2016.
36062 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
36064 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
36066 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
36067 * doc/invoke.texi: Likewise.
36068 * doc/md.texi: Likewise.
36069 * doc/objc.texi: Likewise.
36071 2017-01-10 Joshua Conner <joshconner@google.com>
36073 * config/arm/fuchsia-elf.h: New file.
36074 * config/fuchsia.h: New file.
36075 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
36076 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
36078 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
36080 2016-01-10 Richard Biener <rguenther@suse.de>
36082 PR tree-optimization/79034
36083 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
36084 Propagate out degenerate PHIs in the joiner.
36086 2017-01-10 Martin Liska <mliska@suse.cz>
36088 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
36089 (sort_congruence_classes_by_decl_uid): Likewise.
36090 (sort_congruence_class_groups_by_decl_uid): Likewise.
36091 (sem_item_optimizer::merge_classes): Sort class, groups in these
36092 classes and members in the groups by DECL_UID of declarations.
36093 This would make merge operations stable.
36095 2017-01-10 Martin Liska <mliska@suse.cz>
36097 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
36098 usage of m_classes_vec.
36099 (sem_item_optimizer::~sem_item_optimizer): Likewise.
36100 (sem_item_optimizer::get_group_by_hash): Likewise.
36101 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
36102 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
36103 (sem_item_optimizer::verify_classes): Likewise.
36104 (sem_item_optimizer::process_cong_reduction): Likewise.
36105 (sem_item_optimizer::dump_cong_classes): Likewise.
36106 (sem_item_optimizer::merge_classes): Likewise.
36107 * ipa-icf.h (congruence_class_hash): Rename from
36108 congruence_class_group_hash. Remove declaration of m_classes_vec.
36110 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
36112 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
36113 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
36114 * config.gcc: Add avx512vpopcntdqintrin.h.
36115 * config/i386/avx512vpopcntdqintrin.h: New.
36116 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
36117 * config/i386/i386-builtin-types.def: Add new types.
36118 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
36119 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
36120 __builtin_ia32_vpopcountq_v8di_mask): New.
36121 * config/i386/i386-c.c (ix86_target_macros_internal): Define
36122 __AVX512VPOPCNTDQ__.
36123 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
36124 (PTA_AVX512VPOPCNTDQ): Define.
36125 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
36126 TARGET_AVX512VPOPCNTDQ_P): Define.
36127 * config/i386/i386.opt: Add mavx512vpopcntdq.
36128 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
36129 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
36131 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36133 PR middle-end/77484
36134 * predict.def (PRED_CALL): Set to 67.
36136 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
36138 * expr.c (store_field): In the bitfield case, if the value comes from
36139 a function call and is of an aggregate type returned in registers, do
36140 not modify the field mode; extract the value in all cases if the mode
36141 is BLKmode and the size is not larger than a word.
36143 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
36146 * config/i386/cpuid.h: Fix undefined behavior.
36148 2017-01-04 Jeff Law <law@redhat.com>
36150 PR tree-optimization/79007
36151 PR tree-optimization/67955
36152 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
36153 conservative for pt.null when flag_non_call_exceptions is on.
36155 2017-01-09 Jakub Jelinek <jakub@redhat.com>
36157 PR translation/79019
36158 PR translation/79020
36159 * params.def (PARAM_INLINE_MIN_SPEEDUP,
36160 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
36161 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
36163 * config/avr/avr.opt (maccumulate-args): Likewise.
36164 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
36165 * common.opt (freport-bug): Likewise.
36166 * cif-code.def (CIF_FINAL_ERROR): Likewise.
36167 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
36168 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
36169 translatable string.
36170 * config/i386/i386.c (function_value_32): Likewise.
36171 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
36172 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
36174 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
36175 * common/config/msp430/msp430-common.c (msp430_handle_option):
36177 * symtab.c (symtab_node::verify_base): Likewise.
36178 * opts.c (set_debug_level): Likewise.
36179 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
36180 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
36181 missing whitespace to translatable strings.
36182 * config/avr/avr.md (bswapsi2): Fix typo in comment.
36183 * config/sh/superh.h: Likewise.
36184 * config/i386/xopintrin.h: Likewise.
36185 * config/i386/znver1.md: Likewise.
36186 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
36187 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
36188 * double-int.h (struct double_int): Likewise.
36189 * double-int.c (div_and_round_double): Likewise.
36190 * wide-int.cc: Likewise.
36191 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
36192 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
36193 * cfgcleanup.c (crossjumps_occured): Renamed to ...
36194 (crossjumps_occurred): ... this.
36195 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
36198 PR tree-optimization/78899
36199 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
36200 returning bool return struct loop *, NULL for failure and the new
36202 (versionable_outer_loop_p): Don't version outer loop if it has
36203 dont_vectorized bit set.
36204 (tree_if_conversion): When versioning outer loop, ensure
36205 tree_if_conversion is performed also on the inner loop of the
36206 non-vectorizable outer loop copy.
36207 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
36208 LOOP_VECTORIZED in inner loop of the scalar outer loop and
36209 prevent vectorization of it.
36210 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
36211 the outer loop vectorization of the non-scalar version is attempted
36212 before vectorization of the inner loop in scalar version. If
36213 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
36214 vectorization of its inner loop.
36215 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
36216 has 2 inner loops, rename also on edges from bb whose single pred
36217 is outer_loop->header. Fix typo in function comment.
36219 2017-01-09 Martin Sebor <msebor@redhat.com>
36222 * asan.c (asan_emit_stack_protection): Increase local buffer size
36223 to avoid snprintf truncation warning.
36225 2017-01-09 Andrew Pinski <apinski@cavium.com>
36227 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
36228 to reference thunderx2t99 for the tuning structure
36229 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
36231 (thunderx2t99_extra_costs): This.
36232 * config/aarch64/aarch64-tune.md: Regenerate.
36233 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
36234 (vulcan_addrcost_table): This.
36235 (vulcan_regmove_cost): Rename to ...
36236 (thunderx2t99_regmove_cost): This.
36237 (vulcan_vector_cost): Rename to ...
36238 (thunderx2t99_vector_cost): this.
36239 (vulcan_branch_cost): Rename to ...
36240 (thunderx2t99_branch_cost): This.
36241 (vulcan_tunings): Rename to ...
36242 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
36243 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
36245 2017-01-09 Martin Jambor <mjambor@suse.cz>
36249 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
36250 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
36251 (propagate_vr_accross_jump_function): Use the above function for all
36252 value range computations for pass-through jump functions and type
36253 converasion from explicit value range values.
36254 (ipcp_propagate_stage): Do not attempt to deduce types of formal
36255 parameters from TYPE_ARG_TYPES.
36256 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
36257 (ipa_write_node_info): Stream type of the actual argument.
36258 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
36260 2017-01-09 Martin Liska <mliska@suse.cz>
36263 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
36264 (lookup_compiler): Do not show error message with have_E.
36266 2017-01-09 Jakub Jelinek <jakub@redhat.com>
36268 PR tree-optimization/78938
36269 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
36270 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
36271 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
36272 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
36275 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36277 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
36280 2017-01-09 Richard Biener <rguenther@suse.de>
36282 PR tree-optimization/78997
36283 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
36284 name condition properly.
36286 2017-01-09 Richard Biener <rguenther@suse.de>
36289 * dwarf2out.c (is_cxx): New overload with context.
36290 (is_naming_typedef_decl): Use it.
36292 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36294 * invoke.texi (Option Summary): Correct spacing in option lists
36295 and add line breaks to fix over-long lines.
36297 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36299 PR middle-end/17660
36301 * extend.texi (Common Variable Attributes): Add xref to GCC
36302 Internals manual to explain mode attribute keywords.
36304 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36307 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
36308 and Preprocessor Options.
36309 (Options for Linking): Document -pthread here....
36310 (RS/6000 and PowerPC Options): ...not here.
36311 (Solaris 2 Options): ...or here.
36312 * doc/cppopts.texi: Document -pthread.
36314 2017-01-08 Martin Sebor <msebor@redhat.com>
36316 PR middle-end/77708
36317 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
36318 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
36319 New member functions.
36320 (format_directive): Used them.
36322 (pass_sprintf_length::handle_gimple_call): Same.
36323 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
36324 to avoid truncation for any argument.
36325 (extract_affine_mul): Same.
36326 * tree.c (get_file_function_name): Same.
36328 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36330 PR middle-end/77484
36331 * predict.def (PRED_INDIR_CALL): Set to 86.
36333 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36335 PR preprocessor/54124
36336 * doc/cppopts.texi: Reformat -d subtable to list the full name
36337 of the options. Add cross-reference to the docs for the general
36338 compiler -d options.
36339 * doc/invoke.texi (Developer Options): Add cross-reference to the
36340 preprocessor-specific -d option documentation.
36342 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36344 PR preprocessor/13498
36345 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
36346 redudant material, and reflect new command-line options.
36347 (System Headers): Likewise.
36349 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36351 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
36352 -isystem, and -idirafter. Copy-edit.
36353 * doc/cppopts.texi: Copy-edit. Remove contradiction about
36354 default for -ftrack-macro-expansion. Delete obsolete and
36355 badly-formatted implementation details about -fdebug-cpp output.
36356 * doc/cppwarnopts.texi: Copy-edit.
36358 2017-01-07 David Malcolm <dmalcolm@redhat.com>
36361 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
36362 that the transition from a max line width >= 1<<10 to narrower
36363 lines works correctly.
36365 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
36367 * doc/options.texi (PerFunction): New.
36368 * opt-functions.awk (switch_flags): Map both Optimization and
36369 PerFunction to CL_OPTIMIZATION.
36370 * opth-gen.awk: Test for PerFunction flag along with
36372 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
36373 it only when the latter is present. Skip those that don't in
36374 the hash function generator.
36375 * common.opt (fvar-tracking): Mark as PerFunction instead of
36377 (fvar-tracking-assignments): Likewise.
36378 (fvar-tracking-assignments-toggle): Likewise.
36379 (fvar-tracking-uninit): Likewise.
36381 2017-01-07 Jakub Jelinek <jakub@redhat.com>
36383 PR translation/79018
36384 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
36387 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
36390 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
36391 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
36392 TARGET_LONG_JUMP_TABLE_OFFSETS.
36393 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
36394 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
36395 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
36396 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
36397 * config/m68k/m68k.md (tablejump expander): Likewise.
36398 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
36399 TARGET_LONG_JUMP_TABLE_OFFSETS.
36400 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
36401 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
36403 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
36404 David Holsgrove <david.holsgrove@xilinx.com>
36406 * common/config/microblaze/microblaze-common.c
36407 (TARGET_EXCEPT_UNWIND_INFO): Remove.
36408 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
36410 * config/microblaze/microblaze.c (microblaze_must_save_register)
36411 (microblaze_expand_epilogue, microblaze_return_addr): Handle
36413 (microblaze_eh_return): New function.
36414 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
36415 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
36416 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
36417 * config/microblaze/microblaze.md (eh_return): New pattern.
36419 2017-01-06 Jakub Jelinek <jakub@redhat.com>
36421 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
36422 GCC_DIAGNOSTIC_STRINGIFY): Define.
36424 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
36426 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36428 * config/arm/arm.md (<mcrr>): New.
36430 * config/arm/arm.c (arm_arch5te): New.
36431 (arm_option_override): Set arm_arch5te.
36432 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
36434 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
36435 (arm_mcrr_qualifiers): ... this. New.
36436 (MRRC_QUALIFIERS): Define to...
36437 (arm_mrrc_qualifiers): ... this. New.
36438 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
36440 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
36441 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
36442 (MRRCI, mrrc, MRRC): New.
36443 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
36444 VUNSPEC_MRRC2): New.
36446 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36448 * config/arm/arm.md (<mcr>): New.
36450 * config/arm/arm.c (arm_coproc_builtin_available): Add
36451 support for mcr, mrc, mcr2 and mrc2.
36452 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
36453 (arm_mcr_qualifiers): ... this. New.
36454 (MRC_QUALIFIERS): Define to ...
36455 (arm_mrc_qualifiers): ... this. New.
36456 (MCR_QUALIFIERS): Define to ...
36457 (arm_mcr_qualifiers): ... this. New.
36458 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
36460 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
36461 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
36462 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
36463 VUNSPEC_MRC2): New.
36465 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36467 * config/arm/arm.md (*ldc): New.
36471 * config/arm/arm.c (arm_coproc_builtin_available): Add
36472 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
36473 (arm_coproc_ldc_stc_legitimate_address): New.
36474 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
36475 'qualifier_const_pointer'.
36476 (LDC_QUALIFIERS): Define to...
36477 (arm_ldc_qualifiers): ... this. New.
36478 (STC_QUALIFIERS): Define to...
36479 (arm_stc_qualifiers): ... this. New.
36480 * config/arm/arm-protos.h
36481 (arm_coproc_ldc_stc_legitimate_address): New.
36482 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
36483 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
36484 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
36485 stc2, stcl, stc2l): New.
36486 * config/arm/constraints.md (Uz): New.
36487 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
36488 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
36489 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
36490 VUNSPEC_STC2L): New.
36492 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36494 * config/arm/arm.md (<cdp>): New.
36495 * config/arm/arm.c (neon_const_bounds): Rename this ...
36496 (arm_const_bounds): ... this.
36497 (arm_coproc_builtin_available): New.
36498 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
36499 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
36500 (CDP_QUALIFIERS): Define to...
36501 (arm_cdp_qualifiers): ... this. New.
36503 (arm_expand_builtin_args): Add case for 6 arguments.
36504 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
36505 (arm_const_bounds): ... this.
36506 (arm_coproc_builtin_available): New.
36507 * config/arm/arm_acle.h (__arm_cdp): New.
36509 * config/arm/arm_acle_builtins.def (cdp): New.
36511 * config/arm/iterators.md (CDPI,CDP,cdp): New.
36512 * config/arm/neon.md: Rename all 'neon_const_bounds' to
36513 'arm_const_bounds'.
36514 * config/arm/types.md (coproc): New.
36515 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
36516 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
36517 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
36518 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
36520 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36522 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
36523 (UBINOP_QUALIFIERS): New.
36525 (acle_builtin_data): New. Change comment.
36526 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
36527 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
36528 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
36529 arm_acle_builtins.def.
36530 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
36531 (arm_init_acle_builtins): New.
36532 (CRC32_BUILTIN): Remove.
36533 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
36534 crc32cb, crc32ch and crc32cw.
36535 (arm_init_crc32_builtins): Remove.
36536 (arm_init_builtins): Use arm_init_acle_builtins rather
36537 than arm_init_crc32_builtins.
36538 (arm_expand_acle_builtin): New.
36539 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
36540 * config/arm/arm_acle_builtins.def: New.
36542 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36544 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
36545 (arm_builtin_datum): ... this.
36546 (arm_init_neon_builtin): Rename to ...
36547 (arm_init_builtin): ... this. Add a new parameters PREFIX
36548 and USE_SIG_IN_NAME.
36549 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
36550 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
36551 'arm_builtin_datum'.
36552 (arm_init_vfp_builtins): Likewise.
36553 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
36554 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
36555 (arm_expand_neon_args): Rename to ...
36556 (arm_expand_builtin_args): ... this. Rename builtin_arg
36557 enum values and differentiate between ARG_BUILTIN_MEMORY
36558 and ARG_BUILTIN_NEON_MEMORY.
36559 (arm_expand_neon_builtin_1): Rename to ...
36560 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
36561 values, arm_expand_builtin_args and add bool parameter NEON.
36562 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
36563 (arm_expand_vfp_builtin): Likewise.
36564 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
36566 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36568 PR middle-end/77484
36569 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
36570 * predict.c (tree_estimate_probability_bb): Reverse direction of
36571 polymorphic call predictor.
36573 2017-01-06 David Malcolm <dmalcolm@redhat.com>
36575 * passes.c (execute_one_pass): Split out pass-skipping logic into...
36576 (determine_pass_name_match): ...this new function and...
36577 (should_skip_pass_p): ...this new function.
36579 2017-01-06 Nathan Sidwell <nathan@acm.org>
36581 * ipa-visibility.c (function_and_variable_visibility): Reformat
36582 comments and long lines. Remove extrneous if.
36583 * symtab.c (symtab_node::make_decl_local): Fix code format.
36584 (symtab_node::set_section_for_node): Fix comment typo.
36586 2017-01-06 Martin Liska <mliska@suse.cz>
36589 * lra-constraints.c: Rename invariant to lra_invariant.
36590 * predict.c (set_even_probabilities): Initialize e to NULL.
36592 2017-01-05 Martin Sebor <msebor@redhat.com>
36594 PR tree-optimization/78910
36595 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
36596 (format_integer): Correct off-by-one error in the handling
36597 of precision with negative numbers in signed conversions..
36599 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
36601 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
36603 2017-01-05 Jakub Jelinek <jakub@redhat.com>
36605 PR tree-optimization/71016
36606 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
36607 factor_out_conditional_conversion. Formatting fix.
36608 (factor_out_conditional_conversion): Add cond_stmt argument.
36609 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
36610 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
36613 2017-01-05 David Malcolm <dmalcolm@redhat.com>
36615 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
36616 read-rtl-function.o, and selftest-rtl.o.
36617 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
36618 (selftest::aarch64_test_loading_full_dump): New function.
36619 (selftest::aarch64_run_selftests): New function.
36620 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
36621 selftest::aarch64_run_selftests.
36622 * config/i386/i386.c
36623 (selftest::ix86_test_loading_dump_fragment_1): New function.
36624 (selftest::ix86_test_loading_call_insn): New function.
36625 (selftest::ix86_test_loading_full_dump): New function.
36626 (selftest::ix86_test_loading_unspec): New function.
36627 (selftest::ix86_run_selftests): Call the new functions.
36628 * emit-rtl.c (maybe_set_max_label_num): New function.
36629 * emit-rtl.h (maybe_set_max_label_num): New decl.
36630 * function.c (instantiate_decls): Guard call to
36631 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
36632 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
36634 * gensupport.c (gen_reader::gen_reader): Pass "false"
36635 for new "compact" param of rtx_reader.
36636 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
36637 rather than an empty string for NULL strings.
36638 * read-md.c: Potentially include config.h rather than bconfig.h.
36639 Wrap include of errors.h with #ifdef GENERATOR_FILE.
36640 (have_error): New global, copied from errors.c.
36641 (md_reader::read_name): Rename to...
36642 (md_reader::read_name_1): ...this, adding "out_loc" param,
36643 and converting "missing name or number" to returning false, rather
36645 (md_reader::read_name): Reimplement in terms of read_name_1.
36646 (md_reader::read_name_or_nil): New function.
36647 (md_reader::read_string): Handle "(nil)" by returning NULL.
36648 (md_reader::md_reader): Add new param "compact".
36649 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
36650 (md_reader::read_file): New method.
36651 * read-md.h (md_reader::md_reader): Add new param "compact".
36652 (md_reader::read_file): New method.
36653 (md_reader::is_compact): New accessor.
36654 (md_reader::read_name): Convert return type from void to file_location.
36655 (md_reader::read_name_or_nil): New decl.
36656 (md_reader::read_name_1): New decl.
36657 (md_reader::m_compact): New field.
36658 (noop_reader::noop_reader): Pass "false" for new "compact" param
36660 (rtx_reader::rtx_reader): Add new "compact" param.
36661 (rtx_reader::read_rtx_operand): Make virtual and convert return
36662 type from void to rtx.
36663 (rtx_reader::read_until): New decl.
36664 (rtx_reader::handle_any_trailing_information): New virtual function.
36665 (rtx_reader::postprocess): New virtual function.
36666 (rtx_reader::finalize_string): New virtual function.
36667 (rtx_reader::m_in_call_function_usage): New field.
36668 (rtx_reader::m_reuse_rtx_by_id): New field.
36669 * read-rtl-function.c: New file.
36670 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
36671 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
36672 (selftest::verify_three_block_rtl_cfg): New decl.
36673 * read-rtl-function.h: New file.
36674 * read-rtl.c: Potentially include config.h rather than bconfig.h.
36675 For host, include function.h, memmodel.h, and emit-rtl.h.
36676 (one_time_initialization): New function.
36677 (struct compact_insn_name): New struct.
36678 (compact_insn_names): New array.
36679 (find_code): Handle insn codes in compact dumps.
36680 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
36681 (bind_subst_iter_and_attr): Likewise.
36682 (add_condition_to_string): Likewise.
36683 (add_condition_to_rtx): Likewise.
36684 (apply_attribute_uses): Likewise.
36685 (add_current_iterators): Likewise.
36686 (apply_iterators): Likewise.
36687 (initialize_iterators): Guard usage of apply_subst_iterator with
36688 #ifdef GENERATOR_FILE.
36689 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
36690 (md_reader::read_mapping): Likewise.
36691 (add_define_attr_for_define_subst): Likewise.
36692 (add_define_subst_attr): Likewise.
36693 (read_subst_mapping): Likewise.
36694 (check_code_iterator): Likewise.
36695 (rtx_reader::read_rtx): Likewise. Move one-time initialization
36697 (one_time_initialization): New function.
36698 (rtx_reader::read_until): New method.
36699 (read_flags): New function.
36700 (parse_reg_note_name): New function.
36701 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
36702 Handle reuse_rtx ids.
36703 Wrap iterator lookup within #ifdef GENERATOR_FILE.
36704 Add parsing support for RTL dumps, mirroring the special-cases in
36705 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
36706 values, and calling handle_any_trailing_information.
36707 (rtx_reader::read_rtx_operand): Convert return type from void
36708 to rtx, returning return_rtx. Handle case 'e'. Call
36709 finalize_string on XSTR and XTMPL fields.
36710 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
36711 "(nil)" values were omitted. Call the postprocess vfunc on the
36713 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
36714 class ctor. Initialize m_in_call_function_usage. Call
36715 one_time_initialization.
36716 * rtl-tests.c (selftest::test_uncond_jump): Call
36717 set_new_first_and_last_insn.
36718 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
36719 * selftest-rtl.c: New file.
36720 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
36721 (selftest::get_insn_by_uid): New decl.
36722 * selftest-run-tests.c (selftest::run_tests): Call
36723 read_rtl_function_c_tests.
36724 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
36725 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
36728 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
36730 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
36731 operands in a special way. Assert that pos+len <= mode precision.
36733 2017-01-05 Jakub Jelinek <jakub@redhat.com>
36735 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
36736 3 argument Alias with unlimited for the negative form.
36737 (fno-vect-cost-model): Removed.
36739 2017-01-05 Martin Liska <mliska@suse.cz>
36741 * hsa-gen.c (gen_hsa_divmod): New function.
36742 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
36744 2017-01-05 Martin Liska <mliska@suse.cz>
36747 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
36750 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36752 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
36753 small constant length operands.
36755 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36757 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
36758 between loop iterations.
36760 2017-01-05 Martin Liska <mliska@suse.cz>
36763 * gimplify.c (gimplify_decl_expr): Compare to
36764 asan_poisoned_variables instread of checking flags.
36765 (gimplify_target_expr): Likewise.
36766 (gimplify_expr): Likewise.
36767 (gimplify_function_tree): Conditionally initialize
36768 asan_poisoned_variables.
36770 2017-01-04 Jeff Law <law@redhat.com>
36772 PR tree-optimizatin/78812
36773 * rtl.h (contains_mem_rtx_p): Prototype.
36774 * ifcvt.c (containts_mem_rtx_p): Move from here to...
36775 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
36776 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
36777 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
36778 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
36780 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36782 * input.c (assert_char_at_range): Default-initialize actual_range.
36784 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36786 * df-scan.c (df_ref_create_structure): Make regno unsigned,
36787 to match the caller.
36789 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36791 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
36792 insns after final jump in test to emit dummy move.
36794 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36796 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
36797 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
36799 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36801 * multiple_target.c (create_dispatcher_calls): Init e_next.
36802 * tree-ssa-loop-split.c (split_loop): Init border.
36803 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
36806 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
36811 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
36812 (altivec_register_operand): Do not return true if the operand
36813 contains a SUBREG mixing SImode and SFmode.
36814 (vsx_register_operand): Likewise.
36815 (vsx_reg_sfsubreg_ok): New predicate.
36816 (vfloat_operand): Do not return true if the operand contains a
36817 SUBREG mixing SImode and SFmode.
36818 (vint_operand): Likewise.
36819 (vlogical_operand): Likewise.
36820 (gpc_reg_operand): Likewise.
36821 (int_reg_operand): Likewise.
36822 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
36823 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
36824 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
36826 (rs6000_emit_move_si_sf_subreg): New helper function.
36827 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
36828 fixup SUBREGs involving SImode and SFmode.
36829 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
36830 numbers for the new peephole2 optimization.
36831 (peephole2 for SFmode unions): New peephole2 to optimize cases in
36832 the GLIBC math library that do AND/IOR/XOR operations on single
36833 precision floating point.
36834 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
36835 target macros to say whether we need to avoid SUBREGs mixing
36837 (TARGET_ALLOW_SF_SUBREG): Likewise.
36838 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
36839 (UNSPEC_SI_FROM_SF): Likewise.
36840 (iorxor): Change spacing.
36841 (and_ior_xor): New iterator for AND, IOR, and XOR.
36842 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
36843 (movdi_from_sf_zero_ext): Likewise.
36844 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
36845 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
36846 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
36847 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
36848 (fms<mode>4): Likewise.
36849 (fnma<mode>4): Likewise.
36850 (fnms<mode>4): Likewise.
36851 (nfma<mode>4): Likewise.
36852 (nfms<mode>4): Likewise.
36854 2017-01-04 Marek Polacek <polacek@redhat.com>
36857 * doc/invoke.texi: Document -Wpointer-compare.
36859 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36861 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
36864 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
36865 descriptions for -gdwarf-5 and emit them as uleb128 instead of
36868 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
36871 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
36872 documentation of the powerpc_popcntb_ok attribute.
36873 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
36874 code to issue warning messages if a requested CPU configuration is
36875 not supported by the binary (assembler and loader) toolchain.
36876 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
36877 made to define a built-in function that has been disabled.
36878 (paired_init_builtins): Add assertion to prevent ICE if attempt is
36879 made to define a built-in function that has been disabled.
36880 (altivec_init_builtins): Add comment explaining why definition
36881 of the DST built-in functions is not preceded by an assertion
36882 check. Add assertions to prevent ICE if attempts are made to
36883 define an altivec predicate or an abs* built-in function that has
36885 (htm_init_builtins): Add comment explaining why definition of the
36886 htm built-in functions is not preceded by an assertion check.
36888 2017-01-04 Jeff Law <law@redhat.com>
36890 PR tree-optimizatin/67955
36891 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
36892 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
36893 the points-to solution does not include pt_null. Use DECL_PT_UID
36896 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
36898 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
36899 Use gen_int_mode instead of gen_lopwart for const_int operands.
36901 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36903 PR tree-optimization/71563
36904 * match.pd: Simplify X << Y into X if Y is known to be 0 or
36905 out of range value - has low bits known to be zero.
36907 2017-01-04 Alan Modra <amodra@gmail.com>
36909 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
36910 * configure: Regenerate.
36911 * config.in: Regenerate.
36913 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36916 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
36917 a substring of the message, but strcmp with the whole message. Ifdef
36918 ENABLE_NLS, translate the message first using dgettext.
36920 2017-01-03 Jeff Law <law@redhat.com>
36922 PR tree-optimizatin/78856
36923 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
36924 (mark_threaded_blocks): Remove code to truncate thread paths that
36925 cross multiple loop headers. Instead invalidate the cached loop
36926 iteration information and handle case of a thread path walking
36927 into an irreducible region.
36929 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
36932 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
36933 assertions. Add support for doing the signbit if the IEEE 128-bit
36934 floating point value is in a GPR.
36935 * config/rs6000/rs6000.md (Fsignbit): Delete.
36936 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
36937 Update the length attribute if the value is in a GPR.
36938 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
36939 the sign or zero extension instruction, since the value is always 0/1.
36940 (signbit<mode>2_dm2): Delete using <Fsignbit>.
36943 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
36944 extracting SImode to a GPR register so that we can generate a
36945 store, limit the vector to be in a traditional Altivec register
36946 for the vextuwrx instruction.
36948 2017-01-03 Ian Lance Taylor <iant@google.com>
36950 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
36952 2017-01-03 Martin Sebor <msebor@redhat.com>
36954 PR tree-optimization/78696
36955 * gimple-ssa-sprintf.c (format_floating): Correct handling of
36956 precision. Use MPFR for %f for greater fidelity. Correct handling
36958 (pass_sprintf_length::compute_format_length): Set width and precision
36959 specified by asrerisk to void_node for vararg functions.
36960 (try_substitute_return_value): Adjust dump output.
36962 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
36964 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
36966 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
36968 * doc/invoke.texi (SPARC options): Document -mlra as the default.
36969 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
36970 -mlra/-mno-lra was passed to the compiler.
36972 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
36974 PR rtl-optimization/65618
36975 * emit-rtl.c (try_split): Move initialization of "before" and
36976 "after" to just before the call to emit_insn_after_setloc.
36978 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
36980 * doc/md.texi (Standard Names): Remove reference to Java frontend.
36982 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
36984 * dwarf2out.c (gen_enumeration_type_die): When
36985 -gno-strict-dwarf, add a DW_AT_encoding attribute.
36987 2017-01-03 Jakub Jelinek <jakub@redhat.com>
36989 PR tree-optimization/78965
36990 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
36991 Change first argument from const call_info & to call_info &. For %n
36992 set info.nowrite to false.
36994 PR middle-end/78901
36995 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
36996 possibly throwing calls.
36998 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
36999 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
37000 and fns handling, rather than in a separate case SSA_NAME.
37002 2017-01-02 Jeff Law <law@redhat.com>
37004 * config/darwin-driver.c (darwin_driver_init): Const-correctness
37005 fixes for first_period and second_period variables.
37007 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
37010 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
37011 (*insvqi_1): New insn pattern.
37012 (*insvqi_1_mem_rex64): Ditto.
37013 (*insvqi_2): Ditto.
37014 (*insvqi_3): Rename from *insvqi.
37016 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
37018 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
37020 * doc/cfg.texi (Edges): Remove reference to Java.
37021 (Maintaining the CFG): Ditto.
37023 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37025 PR middle-end/77674
37026 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
37027 transparent aliases.
37029 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37031 PR middle-end/77484
37032 * predict.def (PRED_CALL): Update hitrate.
37033 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
37034 * predict.c (tree_estimate_probability_bb): Split CALL predictor
37035 into direct/indirect/polymorphic variants.
37037 2017-01-01 Jakub Jelinek <jakub@redhat.com>
37039 Update copyright years.
37041 * gcc.c (process_command): Update copyright notice dates.
37042 * gcov-dump.c (print_version): Ditto.
37043 * gcov.c (print_version): Ditto.
37044 * gcov-tool.c (print_version): Ditto.
37045 * gengtype.c (create_file): Ditto.
37046 * doc/cpp.texi: Bump @copying's copyright year.
37047 * doc/cppinternals.texi: Ditto.
37048 * doc/gcc.texi: Ditto.
37049 * doc/gccint.texi: Ditto.
37050 * doc/gcov.texi: Ditto.
37051 * doc/install.texi: Ditto.
37052 * doc/invoke.texi: Ditto.
37054 Copyright (C) 2017 Free Software Foundation, Inc.
37056 Copying and distribution of this file, with or without modification,
37057 are permitted in any medium without royalty provided the copyright
37058 notice and this notice are preserved.